calc_metrics {ecospace}  R Documentation 
Wrapper to FD::dbFD
that calculates common ecological
disparity and functional diversity statistics. When used with specieswise
simulations of community assembly or ecological diversification (and default
increm = 'TRUE'
), calculates statistical dynamics incrementally as a
function of species richness. Avoids filesharing errors so that can be used
in 'embarrasingly parallel' implementations in a highperformance computing
environment.
calc_metrics( nreps = 1, samples = NA, Smax = NA, Model = "", Param = "", m = 3, corr = "lingoes", method = "Euclidean", increm = TRUE, ... )
nreps 
Sample number to calculate statistics for. Default is the first
sample 
samples 
Data frame (if 
Smax 
Maximum number of 
Model 
Optional character string or numeric value naming simulation model. A warning issues if left blank. 
Param 
Optional numeric value or character string naming strength parameter used in simulation. A warning issues if left blank. 
m 
The number of PCoA axes to keep as 'traits' for calculating FRic and
FDiv in 
corr 
Character string specifying the correction method to use in

method 
Distance measure to use when calculating functional distances
between species. Default is 
increm 
Default 
... 
Additional parameters for controlling 
The primary goal of this function is to describe the statistical
dynamics of common ecological disparity (functional diversity) metrics as a
function of species richness (sample size). Statistics are calculated
incrementally within samples, first for the first row (species), second for
the first and second rows, ..., ending with the entire sample (by default,
or terminating with Smax
total species). The function assumes that
supplied samples are ecologically or evolutionary cohesive assemblages in
which there is a logical order to the rows (such that the sixth row is the
sixth species added to the assemblage) and that such incremental
calculations are sensible. See NovackGottshall (2016a,b) for additional
context. Samples must have species as rows and traits as columns (of many
allowed character types), and have class(data.frame)
or a list of
such data frames, with each data frame a separate sample.
Statistics calculated include four widely used in ecological disparity
studies (adapted from studies of morphological disparity) and four used in
functional diversity studies. See Foote (1993), Ciampaglio et al. (2001),
and Wills (2001) for definitions and details on morphological disparity
measures and NovackGottshall (2007; 2016a,b) for implementation as measures
of ecological disparity. See Mason et al. (2005), Anderson et al. (2006),
Villeger et al. (2008), Laliberte and Legendre (2010), Mouchet et al.
(2010), Mouillot et al. (2013) for definitions and details on functional
diversity statistics. For computation details of functional diversity
metrics, see Laliberte and Shipley (2014) package FD, and especially
FD::dbFD
, which this function wraps around to calculate
the functional diversity statistics.
Statistic (S
) is species (taxonomic) richness, or sample size.
When increm = 'FALSE'
, the function calculates statistics for the
entire sample(s) instead of doing so incrementally. In this case, the
implementation is essentially the same as FD::dbFD
with
default arguments (e.g., m, corr
) that reduce common calculation
errors, plus inclusion of common morphological disparity statistics.
Statistics that measure diversity (unique number of life habits / trait combinations within ecospace / functionaltrait space):
Life habit richness, the number of functionally unique trait combinations.
Statistics that measure disparity (or dispersion of species within ecospace / functionaltrait space) (Note these statistics are sensitive to outliers and sample size):
Maximum pairwise distance between
species in functionaltrait space, measured using the distance method
specified above.
Functional richness, the minimal convexhull volume in multidimensional principal coordinates analysis (PCoA) traitspace ordination.
Functional divergence, the mean distance of species from the PCoA traitspace centroid.
Statistics that measure internal structure (i.e., clumping or inhomogeneities within the traitspace):
Mean pairwise
distance between species in functionaltrait space, measured using the
distance method
specified above.
Total variance, the sum of variances for each functional trait across species; when using factor or ordered factor character types, this statistic cannot be measured and is left blank, with a warning.
Functional dispersion, the total deviance of species from the circle with radius equal to mean distance from PCoA traitspace centroid.
Statistics that measure spacing among species within the traitspace:
Functional evenness, the evenness of minimumspanningtree lengths between species in PCoA traitspace.
The default number of PCoA axes used in calculating of FRic and FDiv equals
m = 3
. Because their calculation requires more species than traits
(here the m = 3
PCoA axes), the four functional diversity statistics
are only calculated when a calculated sample contains a minimum of m
species (S) or unique life habits (H). qual.FRic
is appended to the
output to record the proportion ('quality') of PCoA space retained by this
loss of dimensionality. Although including more PCoA axes allows greater
statistical power (Villeger et al. 2011, Maire et al. 2015), the use of
m = 3
here is computationally manageable, ecologically meaningful, and
allows standardized measurement of statistical dynamics across the wide
range of sample sizes typically involved in simulations of
ecological/evolutionary assemblages, especially when functionally redundant
data occur. Other integers greater than 1 can also be specified. See the
help file for FD::dbFD
for additional information.
Lingoes correction corr = 'lingoes'
, as recommended by Legendre and
Anderson (1999), is called when the speciesbyspecies distance matrix
cannot be represented in a Euclidean space. See the help file for
FD::dbFD
for additional information.
Note that the ecological disparity statistics are calculated on the raw
(unstandardized) distance matrix. The functional diversity statistics are
calculated on standardized data using standardizations in
FD::dbFD
. If all traits are numeric, they by default are
standardized to mean 0 and unit variance. If not all traits are numeric,
Gower's (1971) standardization by the range is automatically used.
Returns a data frame (if nreps
is a single integer or
samples
is a single data frame) or a list of data frames. Each
returned data frame has Smax
rows corresponding to incremental
species richness (sample size) and 12 columns, corresponding to:
Model 
(optional) 
Param 
(optional) strength parameter 
S 
Species richness (sample size) 
H 
Number of functionally unique life habits 
D 
Mean pairwise distance 
M 
Maximum pairwise distance 
V 
Total variance 
FRic 
Functional richness 
FEve 
Functional evenness 
FDiv 
Functional divergence 
FDis 
Functional dispersion 
qual.FRic 
proportion ('quality') of total PCoA traitspace used when calculating FRic and FDiv 
A bug exists within FD::gowdis
where nearestneighbor
distances can not be calculated when certain characters (especially numeric
characters with values other than 0 and 1) share identical traits across
species. The nature of the bug is under investigation, but the current
implementation is reliable under most uses. If you run into problems because
of this bug, a workaround is to manually change the function to call
cluster::daisy
using metric = "gower"
instead.
If calculating statistics for more than several hundred samples, it is
recommended to use a parallelcomputing environment. The function has been
written to allow usage (using lapply
or some other listapply
function) in 'embarrassingly parallel' implementations in such HPC
environments. Most importantly, overwriting errors during calculation of
convex hull volume in FRic are avoided by creating CPUspecific temporarily
stored vertices files.
See NovackGottshall (2016b) for recommendations for using random forest classification trees to conduct multimodel inference.
Phil NovackGottshall pnovackgottshall@ben.edu
Anderson, M. J., K. E. Ellingsen, and B. H. McArdle. 2006. Multivariate dispersion as a measure of beta diversity. Ecology Letters 9(6):683693.
Ciampaglio, C. N., M. Kemp, and D. W. McShea. 2001. Detecting changes in morphospace occupation patterns in the fossil record: characterization and analysis of measures of disparity. Paleobiology 27(4):695715.
Foote, M. 1993. Discordance and concordance between morphological and taxonomic diversity. Paleobiology 19:185204.
Gower, J. C. 1971. A general coefficient of similarity and some of its properties. Biometrics 27:857871.
Laliberte, E., and P. Legendre. 2010. A distancebased framework for measuring functional diversity from multiple traits. Ecology 91(1):299305.
Legendre, P., and M. J. Anderson. 1999. Distancebased redundancy analysis: testing multispecies responses in multifactorial ecological experiments. Ecological Monographs 69(1):124.
Maire, E., G. Grenouillet, S. Brosse, and S. Villeger. 2015. How many dimensions are needed to accurately assess functional diversity? A pragmatic approach for assessing the quality of functional spaces. Global Ecology and Biogeography 24(6):728740.
Mason, N. W. H., D. Mouillot, W. G. Lee, and J. B. Wilson. 2005. Functional richness, functional evenness and functional divergence: the primary components of functional diversity. Oikos 111(1):112118.
Mouchet, M. A., S. Villeger, N. W. H. Mason, and D. Mouillot. 2010. Functional diversity measures: an overview of their redundancy and their ability to discriminate community assembly rules. Functional Ecology 24(4):867876.
Mouillot, D., N. A. J. Graham, S. Villeger, N. W. H. Mason, and D. R. Bellwood. 2013. A functional approach reveals community responses to disturbances. Trends in Ecology and Evolution 28(3):167177.
NovackGottshall, P.M. 2007. Using a theoretical ecospace to quantify the ecological diversity of Paleozoic and modern marine biotas. Paleobiology 33: 274295.
NovackGottshall, P.M. 2016a. General models of ecological diversification. I. Conceptual synthesis. Paleobiology 42: 185208.
NovackGottshall, P.M. 2016b. General models of ecological diversification. II. Simulations and empirical applications. Paleobiology 42: 209239.
Villeger, S., N. W. H. Mason, and D. Mouillot. 2008. New multidimensional functional diversity indices for a multifaceted framework in functional ecology. Ecology 89(8):22902301.
Villeger, S., P. M. NovackGottshall, and D. Mouillot. 2011. The multidimensionality of the niche reveals functional diversity changes in benthic marine biotas across geological time. Ecology Letters 14(6):561568.
Wills, M. A. 2001. Morphological disparity: a primer. Pp. 55143. In J. M. Adrain, G. D. Edgecombe, and B. S. Lieberman, eds. Fossils, phylogeny, and form: an analytical approach. Kluwer Academic/Plenum Publishers, New York.
Laliberte, E., and B. Shipley. 2014. FD: Measuring functional diversity from multiple traits, and other tools for functional ecology, Version 1.012.
FD::dbFD
for details on the core function wrapped
here for calculating functional diversity statistics. neutral
,
redundancy
, partitioning
,
expansion
for building samples using simulations.
rbind_listdf
for efficient way to combine lists of data frames
for subsequent analyses.
# Build ecospace framework and a random 50species sample using neutral rule: ecospace < create_ecospace(nchar = 15, char.state = rep(3, 15), char.type = rep("numeric", 15)) sample < neutral(Sseed = 5, Smax = 50, ecospace = ecospace) # Using Smax = 10 here for fast example metrics < calc_metrics(samples = sample, Smax = 10, Model = "Neutral", Param = "NA") metrics # Plot statistical dynamics as function of species richness op < par() par(mfrow = c(2,4), mar = c(4, 4, 1, .3)) attach(metrics) plot(S, H, type = "l", cex = .5) plot(S, D, type = "l", cex = .5) plot(S, M, type = "l", cex = .5) plot(S, V, type = "l", cex = .5) plot(S, FRic, type = "l", cex = .5) plot(S, FEve, type = "l", cex = .5) plot(S, FDiv, type = "l", cex = .5) plot(S, FDis, type = "l", cex = .5) par(op) # Argument 'increm' switches between incremental and entiresample calculation metrics2 < calc_metrics(samples = sample, Smax = 10, Model = "Neutral", Param = "NA", increm = FALSE) metrics2 identical(tail(metrics, 1), metrics2) # These are identical # ... can further control 'FD::dbFD', here turning off calculation of FRic and FDiv metrics3 < calc_metrics(samples = sample, Smax = 10, Model = "Neutral", Param = "NA", calc.FRic = FALSE, calc.FDiv = FALSE) metrics3 rbind(metrics[10, ], metrics3[10, ]) ## Not run: # Can take a few minutes to run to completion # Calculate for 5 samples nreps < 1:5 samples < lapply(X = nreps, FUN = neutral, Sseed = 5, Smax = 50, ecospace) metrics < lapply(X = nreps, FUN = calc_metrics, samples = samples, Model = "Neutral", Param = "NA") alarm() str(metrics) ## End(Not run)