There are three tests, on effects of 1. the shape of the SAD, 2. treatment/group-level density, 3. degree of aggregation. The user can specifically to conduct one or more of these tests.
get_delta_stats(
mob_in,
env_var,
group_var = NULL,
ref_level = NULL,
tests = c("SAD", "N", "agg"),
spat_algo = NULL,
type = c("continuous", "discrete"),
stats = NULL,
inds = NULL,
log_scale = FALSE,
min_plots = NULL,
density_stat = c("mean", "max", "min"),
n_perm = 1000,
overall_p = FALSE
)a "mob_out" object with attributes
an object of class mob_in created by make_mob_in()
a character string specifying the environmental variable in
mob_in$env to be used for explaining the change in richness
an optional character string
in mob_in$env which defines how samples are pooled. If not provided
then each unique value of the argument env_var is used define the
groups.
a character string used to define the reference level of
env_var to which all other groups are compared with. Only makes sense
if env_var is a factor (i.e. type == 'discrete')
specifies which one or more of the three tests ('SAD', N', 'agg') are to be performed. Default is to include all three tests.
character string that can be either: 'kNN' or
'kNCN' for k-nearest neighbor and k-nearest centroid neighbor
sampling respectively. It defaults to k-nearest neighbor which is a more
computationally efficient algorithm that closely approximates the
potentially more correct k-NCN algo (see Details of ?rarefaction).
"discrete" or "continuous". If "discrete", pair-wise comparisons are conducted between all other groups and the reference group. If "continuous", a correlation analysis is conducted between the response variables and env_var.
a vector of character strings that specifies what statistics to
summarize effect sizes with. Options include: c('betas', 'r2',
'r2adj', 'f', 'p') for the beta-coefficients, r-squared, adjusted
r-squared, F-statistic, and p-value respectively. The default value of
NULL will result in only betas being calculated when type ==
'discrete' and all possible stats being computed when type ==
'continuous'. Note that for a discrete analysis all non-betas stats are
meaningless because the model has zero degrees of freedom in this context.
effort size at which the individual-based rarefaction curves are
to be evaluated, and to which the sample-based rarefaction curves are to be
interpolated. It can take three types of values, a single integer, a vector
of integers, and NULL. If inds = NULL (the default), the curves are
evaluated at every possible effort size, from 1 to the total number of
individuals within the group (slow). If inds is a single integer, it is
taken as the number of points at which the curves are evaluated; the
positions of the points are determined by the "log_scale" argument. If inds
is a vector of integers, it is taken as the exact points at which the
curves are evaluated.
if "inds" is given a single integer, "log_scale" determines the position of the points. If log_scale is TRUE, the points are equally spaced on logarithmic scale. If it is FALSE (default), the points are equally spaced on arithmetic scale.
minimal number of plots for test 'agg', where plots are randomized within groups as null test. If it is given a value, all groups with fewer plots than min_plot are removed for this test. If it is NULL (default), all groups are kept. Warnings are issued if 1. there is only one group left and "type" is discrete, or 2. there are less than three groups left and "type" is continuous, or 3. reference group ("ref_group") is removed and "type" is discrete. In these three scenarios, the function will terminate. A different warning is issued if any of the remaining groups have less than five plots (which have less than 120 permutations), but the test will be carried out.
reference density used in converting number of plots to numbers of individuals, a step in test "N". It can take one of the three values: "mean", "max", or "min". If it is "mean", the average plot-level abundance across plots (all plots when "type" is "continuous, all plots within the two groups for each pair-wise comparison when "type" is "discrete") are used. If it is "min" or "max", the minimum/maximum plot-level density is used.
number of iterations to run for null tests, defaults to 1000.
Boolean defaults to FALSE specifies if overall across scale p-values for the null tests. This should be interpreted with caution because the overall p-values depend on scales of measurement yet do not explicitly reflect significance at any particular scale.
Dan McGlinn and Xiao Xiao
rarefaction
data(inv_comm)
data(inv_plot_attr)
inv_mob_in = make_mob_in(inv_comm, inv_plot_attr, coord_names = c('x', 'y'))
inv_mob_out = get_delta_stats(inv_mob_in, 'group', ref_level='uninvaded',
type='discrete', log_scale=TRUE, n_perm=3)
plot(inv_mob_out)
Run the code above in your browser using DataLab