mobr (version 3.0.0)

get_null_comm: Generate a null community matrix

Description

Three null models are implemented that randomize different components of community structure while keeping other components constant.

Usage

get_null_comm(comm, null_model, groups = NULL)

Value

a site-by-species matrix

Arguments

comm

community matrix of abundances with plots as rows and species columns.

null_model

a string which specifies which null model to use options include: 'rand_SAD', 'rand_N', and 'rand_agg'. See Details for description of each null model.

groups

optional argument that is a vector of group ids which specify which group each site is associated with. If is NULL then all rows of the community matrix are assumed to be members of the same group

Details

This function implements three different nested null models. They are considered nested because at the core of each null model is the random sampling with replacement of the relative abundance distribution (RAD) to generate a random sample of a species abundance distribution (SAD). Here we describe each null model:

• 'rand_SAD' ... A random SAD is generated using a sample with replacement of individuals from the species pool proportional to their observed relative abundance. This null model will produce an SAD that is of a similar functional form to the observed SAD (Green and Plotkin 2007). The total abundance of the random SAD is the same as the observed SAD but overall species richness will be equal to or less than the observed SAD. This algorithm ignores the group argument. This sampling algorithm is also used in the two other null models 'rand_N' and 'rand_agg'.

• 'rand_N' ... The total number of individuals in a plot is shuffled across all plots (within and between groups). Then for each plot that many individuals are drawn randomly from the group specific relative abundance distribution with replacement for each plot (i.e., using the 'rand_SAD' algorithm described above. This removes group differences in the total number of individuals in a given plot, but maintains group level differences in their SADs.

• 'rand_agg' ... This null model nullifies the spatial structure of individuals (i.e., their aggregation), but it is constrained by the observed total number of individuals in each plot (in contrast to the 'rand_N' null model), and the group specific SAD (in contrast to the 'rand_SAD' null model). The other two null models also nullify spatial structure. The 'rand_agg' null model is identical to the 'rand_N' null model except that plot abundances are not shuffled.

Replaces depreciated function `permute_comm`

References

Green, J. L., and J. B. Plotkin. 2007. A statistical theory for sampling species abundances. Ecology Letters 10:1037-1045.

Examples

S = 3
N = 20
nplots = 4
comm = matrix(rpois(S * nplots, 1), ncol = S, nrow = nplots)
comm
groups = rep(1:2, each=2)
groups
set.seed(1)
get_null_comm(comm, 'rand_SAD')
# null model 'rand_SAD' ignores groups argument
set.seed(1)
get_null_comm(comm, 'rand_SAD', groups)
set.seed(1)
get_null_comm(comm, 'rand_N')
# null model 'rand_N' does not ignore the groups argument
set.seed(1)
get_null_comm(comm, 'rand_N', groups)
# note that the 'rand_agg' null model is constrained by observed plot abundances
noagg = get_null_comm(comm, 'rand_agg', groups)
noagg
rowSums(comm)
rowSums(noagg)