The same statistics for simulated and observed data, so that they can be compared or used
for inference. Each statistic that lacks its inputs is NA, so the result always has the
same names and results can be bound by rows.
Arguments
- x
An
mc_data()object, anmc_simobject, a matrix with one row per site and one column per species, or an array of sites x species x steps. In an array, the last step gives the statistics that are not temporal. From anmc_sim, these come fromabundanceand the temporal ones fromhistory, andenv,coordsandtraitdefault to those of the simulation, the trait being the environmental optimum. From anmc_dataobject, they default to its parts: the trait is the columnoptimum, or the only numeric column oftraits, andenvthe only numeric column ofenv.- tree
Optional
phyloobject, matched to species by name.- trait
Optional numeric vector with one value per species, named by species. For a simulation, the optimum of each species by default; with a niche of several axes, the optimum on the first axis.
- env
Optional numeric vector with one value per site. For a simulation, the environment that set growth by default; with a niche of several axes, the first axis.
- coords
Optional matrix or data frame with one row per site and the coordinates in columns.
- n_null
Number of shuffles of species for
ses_mpdandses_mntd.
Details
alpha, gamma and beta are the mean number of species per site, the number of species
in all sites and \(\gamma / \alpha\). occupancy and occupancy_sd are the
mean and the
standard deviation of the share of sites each species occupies, over the species present.
abundance is the mean number of individuals per site, evenness the mean Pielou
evenness of sites with two or more species, evenness_regional the same for all sites
together and sad_sd the standard deviation of the logarithm of regional abundances, which
describes the species abundance distribution. With 0/1 data, these last four describe
frequencies of occurrence.
ses_mpd and ses_mntd need tree. They are the standardised effect sizes of the mean
pairwise distance and of the mean nearest taxon distance, from presences, averaged over sites
with two or more species. The null model shuffles species across the tips, as
null.model = "taxa.labels" of picante::ses.mpd(), so the pool is all the species shared
by x and tree, including those without individuals. Compare results only between data
with the same pool. This is the only part that uses random numbers.
decay needs coords. It is the correlation between Bray-Curtis dissimilarity and distance
among occupied sites. trait_env needs trait and env: the correlation between the mean
trait of each site, weighted by abundance, and the environment. trait_sd is the standard
deviation of the trait among all individuals. trait_k needs trait and tree: Blomberg's
K (Blomberg et al. 2003) of the trait over all the species shared by trait and tree,
including those without individuals, as picante::Kcalc() gives it. For a simulation the
trait is the optimum, which is not observed in the field; the K of a measured trait carries
less of the signal of the optima.
The temporal statistics need at least two steps, from history or an array. cv_population,
cv_site and cv_regional are mean coefficients of variation over time: of each species
in each site, of the individuals of each site, and of each species in all sites, leaving out
those that are always absent. turnover is the mean Bray-Curtis dissimilarity of a site
between consecutive steps. Guzman et al. (2022) found that temporal statistics added
explanatory power, and that dispersal depended on them.
References
Blomberg, S. P., Garland, T., & Ives, A. R. (2003). Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution, 57, 717–745. doi:10.1111/j.0014-3820.2003.tb00285.x
Guzman, L. M., Thompson, P. L., Viana, D. S., Vanschoenwinkel, B., Horváth, Z., Ptacnik, R., Jeliazkov, A., Gascón, S., Lemmens, P., Anton-Pardo, M., Langenheder, S., De Meester, L., & Chase, J. M. (2022). Accounting for temporal change in multiple biodiversity patterns improves the inference of metacommunity processes. Ecology, 103, e3683. doi:10.1002/ecy.3683
See also
Other simulation studies:
as.data.frame.mc_sim(),
equilibrium_check(),
run_scenarios()
Examples
set.seed(1)
tree <- treeshapesim::sim_tree_shape(15)
sim <- simulate_metacommunity(landscape_grid(side = 5), species_traits(tree),
competition_matrix(tree, "phylo"), n_steps = 60,
seed_steps = 1:5, record_steps = seq(40, 60, by = 5))
round(summary_stats(sim, tree, n_null = 19), 3)
#> alpha gamma beta occupancy
#> 5.720 13.000 2.273 0.440
#> occupancy_sd abundance evenness evenness_regional
#> 0.334 239.720 0.766 0.841
#> sad_sd ses_mpd ses_mntd decay
#> 1.185 0.185 -0.251 0.163
#> trait_env trait_sd trait_k cv_population
#> 0.951 19.708 0.720 0.771
#> cv_site cv_regional turnover
#> 0.082 0.275 0.107
