Analysing Ecological Data (Statistics for Biology and Health)
Alain F. Zuur, Elena N. Ieno, Graham M. Smith
Format: PDF / Kindle (mobi) / ePub
This book provides a practical introduction to analysing ecological data using real data sets collected as part of postgraduate ecological studies or research projects.
The first part of the book gives a largely non-mathematical introduction to data exploration, univariate methods (including GAM and mixed modelling techniques), multivariate analysis, time series analysis (e.g. common trends) and spatial statistics. The second part provides 17 case studies, mainly written together with biologists who attended courses given by the first authors. The case studies include topics ranging from terrestrial ecology to marine biology. The case studies can be used as a template for your own data analysis; just try to find a case study that matches your own ecological questions and data structure, and use this as starting point for you own analysis. Data from all case studies are available from www.highstat.com. Guidance on software is provided in Chapter 2.
mathematical notation. Let Yj be the value of the response variable (richness) at the zth site, and Xt the value of the explanatory variable (NAP) for the same site. 5.1 Bivariate linear regression 51 0 o_ CN A o m - 0 o _ GO O O O O o 0 O oo m - o O 0 O o O O O o O an> o O o o o o oo o - o o o o ° -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 NAP Figure 5.1. A: scatterplot of species richness versus NAP for the RIKZ data. B: scatterplot and regression line for RIKZ data.
is no evidence to reject the null hypothesis, then it can be concluded that none of the explanatory variables is related to the response variable. For the RIKZ model, the F-statistic is 11.18, which is highly significant (p < 0 .001). This means the null hypothesis that all slope parameters are equal to 0 can be rejected. And consequently this means that at least one of the explanatory variables is significantly related to species richness. However, the Fstatistic does not say which explanatory
beach. This is a nominal variable with three classes, and one of the prime interests of the project was to know the effects of exposure. As (i) exposure can only fall into one of three pre-defined categories and (ii) one of the prime underlying questions is whether there is an exposure effect, it is modelled as a fixed effect and not as random effect and we have extended model 7 to include this new effect: YtJ=a+/7NAIV + exposure^. + aj + bjNAPy + €tj Model 8 where ay ~ N(0, a2a), bj ~ N(0,
as.factor($tation)=cd 3.157 n=25 as .factor(Month)=cd ef 27'.56 n=39 Figure 9.3. Regression tree for parrotfish of the Bahamas fisheries dataset. The observations are repeatedly split into two groups. The first and most important split is based on the nominal explanatory variable 'Method', where 'a' stands for method 1 and 4 b' for method 2. If a statement is true, follow the left side of a branch, and if false follow the right side. Numbers at the bottom of a terminal leaf represent the mean
better than the correlation coefficient and Chi-Square functions. Jongman et al. (1995) carried out a simulation study in which they looked at the sensitivity (sample total, dominant species, species richness) of nine measures of association. The Jaccard index, coefficient of community and the Chord distance