ENV 2A7Y, Community Analysis in EcologyInstructions for Ecological SurveyAlastair GrantCentre for Ecology, Evolution and Conservation, University of East Anglia, Norwich, UK |
 |
Introduction
There are two pieces of assessed work for this course. The second consists of a quantitative survey of animal or plant abundances, followed by a multivariate statistical analysis of the resulting data. For your data set, you will identify inter-relationships between species and groupings of samples that are similar to each other.
General
Aims
This piece of work has the following aims:
1. To give you experience of carrying out a quantitative ecological survey
2. To give you experience of carrying out statistical analyses on ecological data.
Criteria
by which the results will be judged
Skills in organism identification are assessed by the taxonomic collection, and will not be assessed again here. We will be assessing:
· Effort in collecting data (although above a threshold of minimum acceptable effort, more samples will not merit more marks).
· Correct and thorough use of statistical analysis.
· A written account that draws attention to the important features of the data, drawing sensible ecological conclusions about the reasons for the distributions of species and communities.
What you need to do independently
Choose a habitat in which to work. Sampling will be easiest with terrestrial plants and organisms on rocky shores, where all individuals are readily visible. But it is entirely possible to sample freshwater invertebrates using kick or grab samples; marine infaunal animals by taking cores and sieving animals out and a range of other possibilities.
Select several examples of the same habitat or identify an area where there is a range of environmental conditions. For best results you need to sample a moderate range of environmental conditions. You want to have some variation in the abundances of the species that occur, but not so much variation that there are no species in common between some pairs of samples. For example, for rocky shores you should probably restrict yourself to the mid-shore or the low shore, so that some species occur in all of your samples. If you are sampling rivers, then try to include sites that vary in the speed of water flow, or sample a number of rivers that you expect will have moderate differences in water chemistry. You need some replicate samples at each site, but as a general rule, it is better to sample a small number of replicates at a reasonable number of sites. So it is better to sample ten replicate quadrats at each of 10 sites rather than collect 50 replicates at each of two sites.
Collect quantitative data on animal or plant abundance. At a series of randomly located points, take at least two replicate samples. These may be quadrat samples in which you estimate percentage cover visually or count individuals. They may be individual kick samples or core samples from marine sedimentary environments. Record either counts of individuals or percentage cover values for each species. Presence/absence data is not enough. For surveys where you can use quadrat sampling, we will expect rather more samples than for those where sampling is more labour intensive. For a vegetation survey this probably means a total of at least 50-100 quadrat samples spread across several sites or along an environmental gradient. By contrast, if you choose a group like freshwater invertebrates there will be much more work per sample and 2 replicates kick samples from each of 10 river sites would be fine.
Collect some environmental data. It will be easier to make sense of your data if you have some information on environmental variation. For the purposes of this exercise, it can include subjective assessments of key environmental variables on perhaps a five point scale, but you can include quantitative data if you wish. Examples might include altitude; water depth, sediment type and velocity for samples in rivers; depth of anoxic layer for marine sediments etc. Ideally estimate these separately for each quadrat/sample, but it is OK to have a single estimate of the environmental variables for each cluster of samples. Don't get too obsessed with measuring environmental variables - choose things that can be measured very easily or estimated subjectively.
Tabulate your data in a form suitable for entering into statistical packages. You will need data in a table with one column for each species and one row for each sample, i.e. something like this:
|
Site id |
Replicate |
Altitude (m) |
Sp 1 |
Sp 2 |
Sp 3 |
Sp 4 |
Etc. |
|
Field 1 |
1 |
10 |
7 |
5 |
12 |
7 |
|
|
Field 1 |
2 |
10 |
10 |
9 |
6 |
12 |
|
|
Etc |
|
|
|
|
|
|
|
If you are doing this on computer before the start of next semester, and you do not have access to a copy of SPSS, then you are probably best off using Excel or another spreadsheet programme from which data can be cut and pasted into another application. You may find it simpler to put data onto computer if you collect it in tabular format.
The
report
The report should be a description of the patterns that occurred in the communities that you sampled. The text should be in the region of 1000 words, accompanied by figures and text. Not much is needed in terms of an introduction. You do need a brief methods section describing your sites and sampling methods. The bulk of the write up will be the discussing the results of your statistical analyses. What differences are there between samples in species composition? If you have sampled more than one site, how does the amount of variation within sites compare with the variation between sites? Based on your own data on environmental variations or information on the environmental (and management) preferences, what do you think might be causing the differences? This should however be related very closely to your actual data and the species that occur, rather than speculations based on general ecological theory.
A couple of useful references on survey methods are:
Sutherland, W.J., 1996. Ecological
census techniques: a handbook.
Goldsmith, F.B., C.M.
Harrison and A.J. Morton, 1986. Description and analysis of
vegetation. Pp437-524 In: Moore,
P.D. and S.B. Chapman (eds),
Methods in Plant Ecology. Blackwell,
ENV 2A7Y community analysis in ecology
Statistical methods
The data that you have produced in your ecological survey consist of the abundances of a number of species at several sampling sites. We can consider the data as a rectangular table in which the individual samples are rows and the abundances of each species are columns. In most computer programs, the simplest way to carry out statistical analyses on these data is to treat the abundance of each species as a separate variable and to have some additional variables containing site identifiers; environmental variables etc.
You will soon realise that these data are not straightforward to make sense of. Producing scatter plots of the abundances of all pairs of species and calculating correlation coefficients for each pair would give you an unmanageable amount of printout! There are two broad approaches to making sense of this complexity, known in the ecological literature as classification and ordination. Classification is straightforward to understand. We seek to break up the data into groups of samples that are all similar to each other, and can then give each of these groups an informative name. We do this informally when we give names to plant communities (oak wood; beech wood etc), but in this course we will do this in a more objective way. Ordination tries to identify trends in the data, locating each sample on a small number of gradients in composition in a way that retains as much information as possible about the differences between samples.
Ordination
The simplest ordination method is principal components analysis. This works well when there is relatively little variation in floral or faunal composition between samples. To phrase this in terms of distributions along a gradient, PCA will work fine if there are some species that occur a one end of the gradient and others that occur at the other end, but the gradient is sufficiently short that there are no species that are common in the middle but rare at each end. The first principal component should then correspond to the largest environmental gradient in the data. When you data contain samples from a wider range of environments that this, the data points are often distributed in a horseshoe shape that corresponds to the environmental gradient. The first two principal components are then required to describe a single environmental gradient. The same problem occurs with other linear methods, such as correspondence analysis (CA). An example of the sort of plot that you get is shown on the next page. A series of non-linear methods, such as multidimensional scaling (MDS), have been proposed to overcome this so called "arch effect". These will find one axis that follows the horseshoe, although in the case of Detrended Correspondence analysis (DCA) this is just done by flattening the arch without stretching the ends. There is some useful material about ordination methods on the following web page: http://www.okstate.edu/artsci/botany/ordinate/.
Plot showing the relationship between abundances of two species that would generate an "arch effect" in PCA. MDS would find an axis that followed the horseshoe shape.
Classification
The idea of communities has a long history in ecology, often linked with the concept of succession. In applied ecology today there is considerable interest in classifying communities, as seen in the development of the National Vegetation Classification (NVC). The underlying reality is however that community composition usually changes gradually along environmental gradients and that any classification represents an artificial "dissection" of this continuum. It is however much easier to understand data broken up into a small number of categories that you can give names to, so classifying data up into groups is a useful exercise. Cluster analysis is the generic term for statistical methods that do this. The majority of cluster analysis methods first of compute the similarity between all pairs of points (using a method such as Euclidean distance):
The simplest approach then joins together the two most similar data points, then the next most similar pair etc. Etc., until all are joined into a single cluster.
Computer
programmes
There are three programmes available to you for carrying out statistical analysis. These are SPSS, Canoco and Primer. They are described briefly below, and more detailed instructions will be given next term.
SPSS
This is one of the most widely used general statistics packages, and includes the facility to do a number of statistical analyses. I suggest that you do initial analyses using this, then move on to Canoco or Primer if you need to do any more sophisticated analyses. It is widely available on computers throughout the University, and you can acquire a copy for your own computer for a nominal fee from CPC.
Canoco
This is one of the standard packages used by plant ecologists. It was developed by Cajo ter Braak (1986, 1987), and carries out a range of methods based around canonical correspondence analysis and detrended correspondence analysis. The accompanying graphics programme CanoDraw, plots results in graphical form.
For the author's information on the package, have a look at the following web page: http://www.plant.dlo.nl/default.asp?section=products&page=/products/canoco/right.htm
For some instructions on running Canoco see: http://www.okstate.edu/artsci/botany/ordinate/running.htm
For a detailed bibliography see: http://www.fas.umontreal.ca/biol/casgrain/cca_bib/introduction.html
It is available on the ENV Lab D and lab E machines. The easiest way to get data into Canoco is to place it in an Excel Spreadsheet, select all the data then copy it to the clipboard. Running the program WCanoImp then does the conversion to Canoco format.
Multiple copies of the manuals have been
placed in the ENV teaching collection.
Primer
This is rapidly becoming the standard package for the analysis of community data in the marine environment. Carries out a range of univariate and multivariate analyses, such as multidimensional scaling, and permutation methods such as partial Mantel tests (which it calls BIOENV). You will find a good bibliography of papers that have used the programmes at: http://www.pml.ac.uk/primer/Primary_papers.htm and a list of its statistical capabilities at: http://www.pml.ac.uk/primer/primer5.htm . I have extracted quite a lot of the literature references below from these pages
The Windows version of Primer (version 5) is very easy to use. We have a site licence that allows us to place the software on 10 PCs, and you will find it on machines 1 to 10 in Lab D. There are multiple copies of the manual in the ENV teaching collection. The software will read Excel files, and is very easy to use
References
I have deliberately given quite a long list, focussed on the statistical methods. You will find it helpful if you locate some literature that uses statistical methods on data from a similar community to the one that you have sampled, to see what others have done when faced with a similar problem. * In front of the reference indicates that we do not have the journal in the library, but a photocopy should be in the restricted loan collection (some of these may not have arrived yet).
General
Digby, P. G. N. and R.A. Kempton, 1987.
Multivariate analysis of ecological communities ;
P.G.N. Digby,
Gauch, H. G., 1982. Multivariate analysis in community ecology.
James, F.C. and C.E. McCulloch, 1990. Multivariate analysis in ecology and systematics: Panacea or Pandora’s box? Annual Review of Ecology and Systematics 21, 129-166. Available via Jstor.
Matthews, R.A, Matthews, G.B., Ehinger, W.J., 1991. Classification and ordination of limnological data - a comparison of analytical tools. Ecological Modelling, 53, 167-187
Pielou, E. C., 1984. The
interpretation of ecological data : a primer on
classification and ordination.
Plus there are many books on multivariate statistics in the library.
Canoco
Jongman, R. H. G., ter Braak, C. J. F. and van Tongeren, O. F. R. (1987/95). Data
analysis in community and landscape ecology.
ter Braak, C. J. F. (1986). Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology. 67, 1167-1179.
* ter Braak, C. J. F. (1987). The analysis of vegetation- environment relationships by canonical correspondence analysis. Vegetatio. 69, 69- 77.
ter Braak, C. J. F. (1990). Interpreting canonical correlation analysis through biplots of structural correlations and weights. Psychometrika. 55, 519-531.
* ter Braak, C. J. F. (1992). Multidimensional scaling and regression. Statistica Applicata. 4, 577-586.
* ter Braak, C. J. F. (1994). Canonical community ordination. Part I: Basic theory and linear methods. Ecoscience, 1, 127-140.
* ter Braak, C. J. F. (1995). Canonical correspondence analysis and related multivariate methods in aquatic ecology. Aquatic Sciences. 55, 255-289.
ter Braak, C. J. F. and Prentice, I. C. (1988). A theory of gradient analysis. Advances in ecological research. 18, 271-317.
Most of these articles are collected together in:
Ter Braak, C.J.F., 1996. Unimodal models to relate species to environment. Wageninen
Publications
Relating To PRIMER
Methodology
Carr MR (1995) PRIMER User Manual.
Plymouth Marine Laboratory,
Clarke KR (1993) Non-parametric multivariate analyses of changes in community structure. Aust J Ecol 18: 117-143
Clarke KR, Ainsworth M (1993) A method of linking multivariate community structure to environmental variables. Mar Ecol Prog Ser 92: 205-219
Clarke KR, Green RH (1988) Statistical design and analysis for a 'biological effects' study. Mar Ecol Prog Ser 46: 213-226
Clarke KR, Warwick RM (1994) Similarity-based testing for community pattern: the 2-way layout with no replication. Mar Biol 118: 167-176
Clarke KR,
Field JG, Clarke KR, Warwick RM (1982) A practical strategy for analysing multispecies distribution patterns. Mar Ecol Prog Ser 8: 37-52
Somerfield PJ, Clarke KR (1995) Taxonomic levels, in marine community studies, revisited. Mar Ecol Prog Ser 127:113-119
Warwick RM (1986) A new method for detecting pollution effects on marine macrobenthic communities. Mar Biol 92: 557-562
Warwick RM (1988) The level of taxonomic discrimination required to detect pollution effects on marine benthic communities. Mar Pollut Bull 19: 259-268
Warwick RM (1993) Environmental impact studies on marine communities: pragmatical considerations. Aust J Ecol 18: 63-80
Warwick RM,
Clarke KR (1991) A comparison of some methods for
analysing changes in benthic community structure. J mar biol
Ass
Warwick RM, Clarke KR (1993) Comparing the severity of disturbance: a meta-analysis of marine macrobenthic community data. Mar Ecol Prog Ser 92: 221-231
Warwick RM, Clarke KR (1993) Increased variability as a symptom of stress in marine communities. J exp mar Biol Ecol 172: 215-226
PRIMER - some applications
* Agard JBR, Gobin J, Warwick RM
(1993) Analysis of marine macrobenthic community
tropical environment (
* Anderson MJ, Underwood AJ (1994) Effects of substratum on the recruitment and development of ... estuarine fouling assemblage. J exp mar Biol Ecol 184: 217-36
* Baranovic A, Solic M, Vucetic T, Krstulovic N (1993) Temporal fluctuations of zooplankton and bacteria in the middle Adriatic Sea. Mar Ecol Prog Ser 92: 65-75
* Chapman MG, Underwood AJ, Skilleter GA (1995) Variability at different spatial scales between a subtidal assemblage exposed to the discharge of sewage and two control assemblages. J exp mar Biol Ecol 189: 103-122
* Connolly RM (1994) A comparison of fish assemblages from seagrass and unvegetated areas of a Southern Australian estuary. Aust J Mar Freshwat Res 45:1033-44
* Connolly RM (1995) Effects of removal of seagrass canopy on assemblages of small motile invertebrates. Mar Ecol Prog Ser 118: 129-137
Danovaro R, Fabiano
M, Vincx M (1995) Meiofauna response to the Agip-Abruzzo oil-spill in subtidal
sediments of the
*
Dawson-Shepherd A, Warwick RM, Clarke KR, Brown BE (1991) An
analysis of fish community responses to coral mining in the
* Evans SM, Gill ME, Hardy FG, Seku FOK (1993) Evidence of change in some rocky shore communities on the coast of Ghana. J exp mar Biol Ecol 172: 129-141
* Gray JS, Aschan M, Carr MR, Clarke KR, Green RH, Pearson TH, Rosenberg R, Warwick RM (1988) Analysis of community attributes of the benthic macrofauna of Frierfjord/Langesundfjord and in a mesocosm experiment. Mar Ecol Prog Ser 46: 151-165
* Gray JS, Clarke KR, Warwick RM, Hobbs G (1990) Detection of initial effects of pollution Ekofisk and Eldfisk oilfields, N Sea. Mar Ecol Prog Ser 66: 285-299
* Harriott VJ, Smith SDA,
* Hulley PA (1992) Upper-slope distributions of oceanic lanternfishes (family Myctophidae). Mar Biol 114: 365-383
* James RJ, Smith MPL, Fairweather PG (1995) Sieve mesh size and taxonomic resolution needed to describe natural spatial variation Mar Ecol Prog Ser 118:187-98
* Lasiak TA, Field JG (1995) Community-level attributes of
exploited rocky infratidal macrofaunal
assemblages in
* Lindley JA, Williams R (1994) Relating plankton assemblages to environmental variables using ships-of-opportunity. Mar Ecol Prog Ser 107: 245-262
* Nicolaidou A, Zenetos A, Pancucci MA, Simboura N (1993) Comparing ecological effects of pollution using multivariate techniques. PSZN I Mar Ecol 14: 113-128
* Olafsson E, Johnstone RW, Ndaro SGM (1995) Effects of intensive seaweed farming on the meiobenthos in a tropical lagoon. J exp mar Biol Ecol 191: 101-117
* Olsgard F, Gray JS (1995) A comprehensive analysis of the effects of offshore oil and gas exploration Norwegian continental-shelf. Mar Ecol Prog Ser 122:277-306
* Olsgard F, Hasle JR (1993) Impact of waste from titanium mining on benthic fauna. J exp Mar Biol Ecol 172: 185-213
* Pihl L, Wennhage H, Nilsson S (1994) Fish assemblage structure in shallow non-tidal rocky-bottom and soft-bottom habitats. Env Biol Fishes 39: 271-288
* Rundle SD, Attrill MJ (1995) Comparison of meiobenthic crustacean community structure across freshwater acidification gradients Arch Hydrobiol 133: 441-456
* Siokoufrangou I, Papathanassiou E (1991) Differentiation of zooplankton populations in a polluted area. Mar Ecol Prog Ser 76: 41-51
Smith SDA (1994) Impact of domestic sewage effluent versus natural
background variability
* Smith SDA,
* Somerfield PJ, Gee JM, Warwick RM (1994) Soft sediment meiofaunal community metal gradient in the Fal estuary system. Mar Ecol Prog Ser 105: 79-88
* Thrush SF,
Hewitt JE, Cummings VJ,
Villano N, Warwick RM (1995) Meiobenthic communities associated with Ulva rigida in the Palude Della Rosa, Lagoon of Venice. Est cstl shelf Sci 41: 181-194
* Warwick RM, Clarke KR, Gee JM (1990) The effect of disturbance by soldier crabs. on meiobenthic community structure. J exp mar Biol Ecol 135: 19-33
* Warwick RM,
Clarke KR, Suharsono (1990) A statistical analysis of
coral community responses to the 1982-3 El Nino.
Warwick RM,
Goss-Custard JD, Kirby R, George CL,
* Warwick RM, Platt HM, Clarke KR, Agard J, Gobin J (1990) Analysis of macrobenthic and meiobenthic community .Bermuda. J exp mar Biol Ecol 138: 119-142
Wright IA, Chessman BC, Fairweather PG, Benson LJ (1995) Measuring the impact of sewage effluent on the macroinvertebrate community of an upland stream - the effect of different levels of taxonomic resolution and quantification. Aust J Ecol 20: 142-149
Four volumes of this work have been published, and the final volume is still awaited. All are in the library at QK 306 ROD.
Rodwell, J.S. (Ed), 1992. British Plant Communities Volume 1,
Woodlands and
Rodwell, J.S. (Ed), 1992. British Plant Communities
Volume 2, Mires and Heaths.
Rodwell, J.S. (Ed), 1992. British Plant Communities
Volume 3, Grasslands and Montane Communities.
Rodwell, J.S. (Ed), 1992. British Plant Communities
Volume 4, Aquatic communities, swamps and tall-herb fens.