Fits models to species abundance data (D.A. Murray).
|Controls printed output (
||The model or distribution fitted to the data (
||Defines the groups if there is more than one sample|
||Log base to use to form the octaves for the logseries, Poisson log-Normal and negative binomial distributions (
||Plots the fitted values (
||Number of individuals per species|
||Number of species|
||Saves the model estimates|
||Saves the grouping of the estimates|
ECFIT provides a range of distributions and models that can be used to describe species abundance data. For the log series, Poisson log-Normal and negative binomial distributions the species abundance data are grouped into “octaves” using a logarithmic scale. These distributions are then fitted using the
DISTRIBUTION directive using the octave classes. The geometric series, Zipf and Zipf-Mandelbrot models are fitted to the observed abundance data using the non-linear regression facilities.
The numbers of individuals per species are specified using the
INDIVIDUALS parameter. The
SPECIES parameter specifies a variate containing the number of species for the associated number of individuals specified in the corresponding element of
SPECIES can be omitted if each of the values in
INDIVIDUALS corresponds to one species. The
GROUPS option can be used to fit models for different samples.
The distribution or model to be fitted to the data is specified by the
MODELTYPE option. The log base for forming the octaves for the log series, Poisson log-normal and negative binomial distributions can be supplied using the
LOGBASE option. The default is to use log base 2, i.e. representing doubling in species abundance. The parameter estimates from the fitted model can be saved using the
ESTIMATES parameter. The
EGROUPS factor saves a factor indicating the group strucure of the estimates.
||summary of the analysis,|
||the parameter estimates,|
||the fitted values.|
PLOT option can be used to produce a plot of the fitted model or distribution. For the geometric series, Zipf and Zipf-Mandelbrot models, the fitted model can also be displayed on a rank/abundance plot on the log-scale.
The log series, negative binomial and poisson log normal are fitted using the
For the geometric series the abundances are ranked from the most to least abundant and fitted using
FITNONLINEAR where the series is given by
ai = N / (1 – (1 – k)S) × k × (1 – k)i-1
where ai is the total number of individuals in the ith species, N is the total number of individuals, k is the proportion of remaining niche space, and 1 / (1 – (1 – k)S) is a constant that ensures ∑i ai = N.
The Zipf and Zipf-Mandelbrot models are fitted using
FITNONLINEAR. The Zipf model is given by
Ai = A1 × i-γ
where A1 is the fitted abundance of the most abundant species, and γ is a constant representing the average probability of the appearance of a species.
The Zipf-Mandelbrot is an extension of the Zipf model and is expressed as
Ai = A1 × (i + β)-γ
where A1 and gamma are as before, and beta is a constant.
If a parameter is restricted the models will be fitted using only those units included in the restriction.
Kempton, R.A. & Taylor, L.R. (1974). Log-series and log-normal parameters as diversity determinants for the Lepidoptera. Journal of Animal Ecology, 43, 381-399
Magurran, A.E. (2003). Measuring Biological Diversity. Blackwell, Oxford.
Wilson, J.B. (1991). Methods for fitting dominance/diversity curves. Journal of Vegetation Science, 2, 35-46
Commands for: Ecological data.
CAPTION 'ECFIT example'; STYLE=meta VARIATE [VALUES=1...18,20...23,25,28,29,33,34,38,39,40,42,48,51,52,53,58,61,\ 64,69,73,75,83,87,88,105,115,131,139,173,200,223,232,294,323,603,\ 1799] Individuals VARIATE [VALUES=37,22,12,12,11,11,6,4,3,5,2,4,2,3,2,2,4,2,4,4,1,1,1,2,\ 2,2,2,1,1,3,2,2,1,1,1,1,1,2,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]\ NumSpecies ECFIT [PRINT=summary,estimates; MODELTYPE=logseries]\ Individuals; SPECIES=NumSpecies ECFIT [PRINT=summary,estimates; MODELTYPE=negativebinomial]\ Individuals; SPECIES=NumSpecies