multidimensional trait space, taking into account the species relative abundances.
These authors implemented all these functional indices in the R package FD,
together with Rao quadratic entropy, community-weighted means of trait values
(CWMs) and functional group richness (FGR).
The Rao quadratic entropy has been advocated as a simple and universal diversity
index (see Sect. 8.5.2) and it is strongly related to functional dispersion. CWMs are
not strictly diversity indices but represent altogether the functional composition of
the community
2 . FGR is the number of functional groups in a given community,
these groups being defined from a clustering of species based on their functional
traits.
We cannot develop these theoretical explanations here and we invite the interested reader to refer to the cited articles. Let us now apply this distance-based
framework to the fish data. For this, we need two data frames, one containing species
abundances at each site (spe) and the other the functional traits of each species
(tra).
summary(fishtraits)
rownames(fishtraits)
names(spe)
names(fishtraits)
tra <- fishtraits [ , 6:15]
tra
The trait data frame contains four quantitative variables and six binary variables
describing the diet. Values are taken from various sources, mainly fishbase.org
(Froese and Pauly 2017). Quantitative functional traits are: BodyLength (the
average total body length of the adult fish), BodyLengthMax (the maximum
total body length of the adult fish), ShapeFactor (the ratio of total body length
to maximum body height) and TrophicLevel (the relative positions of the
species in the trophic chains).
The dbFD() function of the FD package will be used to compute all distancebased functional diversity indices, as well as CWMs and FGR. For FGR, we must
specify the number of data-defined functional groups (or the fusion level) after
looking at the dendrogram (select at least 5 groups to improve the resulting pattern):
2 Relating standard CWM matrices based on quantitative traits to environmental variables
representing gradients through linear models (correlation, regression or RDA) has been strongly
criticized by Peres-Neto et al. (2017). In particular, when only the traits or only the environmental
variables are important in structuring the species distributions, tests of correlations based on this
approach have strongly inflated type I error. The authors recommend the fourth-corner approach
(Sect. 6.11) instead.
8.5 Functional Diversity, Functional Composition and Phylogenetic. . .
405
These authors implemented all these functional indices in the R package FD,
together with Rao quadratic entropy, community-weighted means of trait values
(CWMs) and functional group richness (FGR).
The Rao quadratic entropy has been advocated as a simple and universal diversity
index (see Sect. 8.5.2) and it is strongly related to functional dispersion. CWMs are
not strictly diversity indices but represent altogether the functional composition of
the community
2 . FGR is the number of functional groups in a given community,
these groups being defined from a clustering of species based on their functional
traits.
We cannot develop these theoretical explanations here and we invite the interested reader to refer to the cited articles. Let us now apply this distance-based
framework to the fish data. For this, we need two data frames, one containing species
abundances at each site (spe) and the other the functional traits of each species
(tra).
summary(fishtraits)
rownames(fishtraits)
names(spe)
names(fishtraits)
tra <- fishtraits [ , 6:15]
tra
The trait data frame contains four quantitative variables and six binary variables
describing the diet. Values are taken from various sources, mainly fishbase.org
(Froese and Pauly 2017). Quantitative functional traits are: BodyLength (the
average total body length of the adult fish), BodyLengthMax (the maximum
total body length of the adult fish), ShapeFactor (the ratio of total body length
to maximum body height) and TrophicLevel (the relative positions of the
species in the trophic chains).
The dbFD() function of the FD package will be used to compute all distancebased functional diversity indices, as well as CWMs and FGR. For FGR, we must
specify the number of data-defined functional groups (or the fusion level) after
looking at the dendrogram (select at least 5 groups to improve the resulting pattern):
2 Relating standard CWM matrices based on quantitative traits to environmental variables
representing gradients through linear models (correlation, regression or RDA) has been strongly
criticized by Peres-Neto et al. (2017). In particular, when only the traits or only the environmental
variables are important in structuring the species distributions, tests of correlations based on this
approach have strongly inflated type I error. The authors recommend the fourth-corner approach
(Sect. 6.11) instead.
8.5 Functional Diversity, Functional Composition and Phylogenetic. . .
405