eco.scape {pez}

eco.space scape simulation with a macro-ecological focus

Description

eco.scape is a modified version of the Helmus et al. method implemented in scape. It produces phylogenetically structured communities. It allows phylogenetic signals in niche optima, but unlike scape, does not include the ability to specify niche optima signal type (attraction/repulsion) or phylogenetic signal in range size. Instead, the focus is on having more control over the macroecological characteristics of the resulting landscapes. In particular, eco.scape produces landscapes with fixed mean range sizes, reasonable range size and abundance distributions, and control over whether species present on a tree must be present in the landscape.

Usage

eco.scape(
  tree,
  scape.size = 10,
  g.center = 1,
  wd.all = 0.2 * (scape.size + 1)^2,
  signal.center = TRUE,
  center.scale = 1,
  site.stoch.scale = 0,
  sd.center = 1,
  sd.range = 1,
  K = 100,
  extinction = FALSE,
  rho = NULL
)

Arguments

tree

phylo object; must have branch lengths and be ultrametric

scape.size

edge dimension of square landscape

g.center

strength of phylogenetic signal in species range centers. See corBlomberg, 1=brownian,<1=rates of evol accelerate, >1=rates decelerate.

wd.all

niche width, larger values simulate broader range sizes

signal.center

simulate with phylosignal in range centers

center.scale

adjust strength of phylogenetic attraction in range centers independent of g.center

site.stoch.scale

adjust strength of random variation in species richness across sites

sd.center

sd in rnormrnorm for the range centers, increase to get more variation in center values across species

sd.range

sd in rnorm for the range sizes, increase to get more variation in range sizes across gradients

K

carrying capacity of a site in terms of maximum individuals that can be present. Currently a constant value. Used to scale the presence-absence matrix to include abundances.

extinction

TRUE/FALSE can species on the tree go extinct on the landscape? If the number of species present on the landscape should equal the number of tips on the tree, choose FALSE. See Details.

rho

Grafen branch adjustment of phylogenetic tree see corGrafen

Details

Simulates a landscape with species (i.e., tree tips) distributions dependent on a supplied phylogenetic tree. The amount and type of structure is determened by the signal parameter g.center. Parameters are based on an Ornstein-Uhlenbeck model of evolution with stabilizing selection. Values of g=1 indicate no stabilizing selection and correspond to the Brownian motion model of evolution; 01 corresponds to disruptive selection where phylogenetic signal for the supplied tree is amplified. See corBlomberg. Communities are simulated along two gradients where the positions along those gradients, g.center, can exhibit phylogenetic signal.

The function returns a landscape where the average range size is equivalent to the wd.all parameter - in the scape function, this parameter is not necessarily returned in the resulting landscape. To do this, the probability of presence (th) that returns the wd.all parameter is solved for. If there is no solution that can produce the wd.all given, the error "Error in uniroot(f, lower = 0, upper = max(X.), tol = 10^-200): f() values at end points not of opposite sign" will occur. This seems to mostly arise for extreme or unlikely parameter values (small species pools, low carrying capacities). Try adjusting parameter values first.

The extinction parameter specifies whether all of the species on the tree should be present in the final landscape. Some species will have probabilities of presence less than those required for presence. If extinctions is TRUE, these species will not be present. If FALSE, these species will be present in 1 site, that in which they have the highest probability of presence.

Value

cc

comparative.comm object with presence/absence results of simulations. The site names are the row.columns of the cells in the original grid cells that made up the data, and these co-ordinates are also given in the env slot of the object along with the environmental gradient information.

Y

presence/absence matrix

Yab

abundance matrix

index

spatial coordinates for X and Y (stacked columns)

X.joint

full probabilities of species at sites, used to construct Y

X1

probabilities of species along gradient 1

X2

probabilities of species along gradient 2

gradient1, gradient2

environmental gradient values

nichewd

average niche width of the assemblage

K

carrying capacity of each cell

environ

matrix depicting environmental values over the 2D landscape

sppXs

full probabilities of each species as an array arranged in a scape.size X scape.size matr ix

V.phylo

initial phylogenetic covariance matrix from tree, output of vcv.phylo(tree, corr=T)

V.phylo.rho

phylogenetic covariance matrix from tree scaled by Grafen if rho is provided, other wise just an output of vcv.phylo(tree, corr=T)

V.center

scaled (by g.center) phylo covariance matrix used in the simulations

bspp1

species optima for gradient 1

bspp2

pecies optima for gradient 2

Author(s)

Matt Helmus, Caroline Tucker, cosmetic edits by Will Pearse

See Also

scape sim.phy sim.meta

Examples

# Simulations
tree <- rcoal(64)

scape1 <- eco.scape(tree, scape.size=25, g.center=1,
    signal.center=FALSE, K=100, extinction=TRUE)
scape2 <- eco.scape(tree, scape.size=16, g.center=0.2,
    signal.center=TRUE, K=100, extinction=FALSE)
scape3 <- eco.scape(tree, scape.size=16, g.center=20,
    signal.center=TRUE, K=100, extinction=TRUE)

# Plotting distributions and landscape patterns
original_landscape <- scape1
abundmax <- original_landscape$K
PA_mat <- as.matrix(original_landscape$Y)
abund_mat <- original_landscape$Yab
site.size <- nrow(PA_mat)
species <- ncol(PA_mat)
mx <- original_landscape$gradient
env <- original_landscape$environ$env.gradient
par(mfrow=c(2,2), oma=c(0,0,2,0))
heatcol <- (colorRampPalette(c("yellow","red")))

image(matrix(env,sqrt(site.size),sqrt(site.size),byrow=TRUE),
    col=heatcol(max(env)),xaxt="n",yaxt="n",main="Env gradient")

image(matrix(rowSums(PA_mat),sqrt(site.size),sqrt(site.size),byrow=TRUE),
    col=heatcol(16),xaxt="n",yaxt="n",main="Species Richness")

hist(colSums(PA_mat),ylab="Number of species",xlab="Number of sites",
    main="Species Area Relationship",col="lightgrey")

hist(colSums(abund_mat),ylab="Number of species",xlab="Number of individuals",
    main="Species Abundance Relationship",col="lightgrey")
mtext("Env random, phy.signal=0.2, 32 species", outer=TRUE, side=3, cex=1.25)