Package 'ecan'

Description

Helper function for clustering methods

Helper function for calculate distance

Add group names to hclust labels.

Add colors to dendrogram

Usage

cluster(x, c_method, d_method)

distance(x, d_method)

cls_add_group(cls, df, indiv, group, pad = TRUE)

cls_color(cls, df, indiv, group)

Arguments

Value

cluster() returns result of clustering. $clustering_method: c_method $distance_method: d_method

distance() returns distance matrix.

    Inputing cls return a color vector, 
            inputing dend return a dend with color.

Examples

library(dplyr)
df <- 
  tibble::tibble(
  stand = paste0("ST_", c("A", "A", "A", "B", "B", "C", "C", "C", "C")),
species = paste0("sp_", c("a", "e", "d", "e", "b", "e", "d", "b", "a")),
  abundance = c(3, 3, 1, 9, 5, 4, 3, 3, 1))
cls <- 
  df2table(df) %>%
  cluster(c_method = "average", d_method = "bray")
library(ggdendro)
# show standard cluster
ggdendro::ggdendrogram(cls)

# show cluster with group
data(dune, package = "vegan")
data(dune.env, package = "vegan")
cls <- 
  cluster(dune, c_method = "average", d_method = "bray")
df <- tibble::rownames_to_column(dune.env, "stand")
cls <- cls_add_group(cls, df, indiv = "stand", group = "Use")
ggdendro::ggdendrogram(cls)

Description

Calculating diversity

Usage

d(x)

h(x, base = exp(1))

Arguments

Value

A numeric vector.

Description

Convert data.frame and table to each other.

Usage

df2table(df, st = "stand", sp = "species", ab = "abundance")

table2df(tbl, st = "stand", sp = "species", ab = "abundance")

dist2df(dist)

Arguments

Value

df2table() return table, 
table2df() return data.frame,
dist2df() return data.frame.

Examples

tibble::tibble(
   st = paste0("st_", rep(1:2, times = 2)), 
   sp = paste0("sp_", rep(1:2, each = 2)), 
   ab = runif(4)) %>%
  dplyr::bind_rows(., .) %>%
  print() %>%
  df2table("st", "sp", "ab")

Description

Draw layer construction plot

Add mid point and bin width of layer heights.

Compute mid point of layer heights.

Compute bin width of layer heights.

Usage

draw_layer_construction(
  df,
  stand = "stand",
  height = "height",
  cover = "cover",
  group = "",
  ...
)

add_mid_p_bin_w(df, height = "height")

mid_point(x)

bin_width(x)

Arguments

Value

draw_layer_construction() returns gg object, add_mid_p_bin_w() returns dataframe including mid_point and bin_width columns. mid_point() and bin_width() return a numeric vector.

Examples

library(dplyr)
n <- 10
height_max <- 20
ly_list    <- c("B", "S", "K")
st_list    <- LETTERS[1]
sp_list    <- letters[1:9]
st_group   <- NULL
sp_group   <- rep(letters[24:26], 3)
cover_list <- 2^(0:4)
df <- gen_example(n = n, use_layer = TRUE,
                  height_max = height_max, ly_list = ly_list,
                  st_list  = st_list,  sp_list  = sp_list,
                  st_group = st_group, sp_group = sp_group,
                  cover_list = cover_list)

# select stand and summarise by sp_group
df %>%
  dplyr::group_by(height, sp_group) %>%
  dplyr::summarise(cover = sum(cover), .groups = "drop") %>%
  draw_layer_construction(group = "sp_group", colour = "white")

Description

Stand, species, and cover are basic. Layer, height, st_group, are sp_group optional.

Usage

gen_example(
  n = 300,
  use_layer = TRUE,
  height_max = 20,
  ly_list = "",
  st_list = LETTERS[1:9],
  sp_list = letters[1:9],
  st_group = NULL,
  sp_group = NULL,
  cover_list = 2^(0:6)
)

Arguments

Value

A dataframe with columns: stand, layer, species, cover, st_group and sp_group.

Examples

n <- 300
height_max <- 20
ly_list    <- c("B1", "B2", "S1", "S2", "K")
st_list    <- LETTERS[1:9]
sp_list    <- letters[1:9]
st_group   <- rep(LETTERS[24:26], 3)
sp_group   <- rep(letters[24:26], 3)
cover_list <- 2^(0:6)
gen_example(n = n, use_layer = TRUE,
            height_max = height_max, ly_list = ly_list, 
            st_list  = st_list,  sp_list  = sp_list,
            st_group = st_group, sp_group = sp_group,
            cover_list = cover_list)

Description

Calculating diversity indices such as species richness (s), Shannon's H' (h), Simpson' D (d), Simpson's inverse D (i).

Usage

ind_val(
  df,
  stand = NULL,
  species = NULL,
  abundance = NULL,
  group = NULL,
  row_data = FALSE
)

Arguments

Value

A data.frame.

Examples

library(dplyr)
library(tibble)
data(dune, package = "vegan")
data(dune.env, package = "vegan")
df <- 
  dune %>%
  table2df(st = "stand", sp = "species", ab = "cover") %>%
  dplyr::left_join(tibble::rownames_to_column(dune.env, "stand"))
ind_val(df, abundance = "cover", group = "Moisture")

Description

Check cols one-to-one, or one-to-multi in data.frame

Usage

is_one2multi(df, col_1, col_2)

is_one2one(df, col_1, col_2)

is_multi2multi(df, col_1, col_2)

cols_one2multi(df, col, include_self = TRUE, inculde_self)

select_one2multi(df, col, include_self = TRUE, inculde_self)

unique_length(df, col_1, col_2)

Arguments

Value

is_one2multi(), is_one2one(), is_multi2multi() return a logical. cols_one2multi() returns strings of colnames that has one2multi relation to input col. unique_length() returns a list.

Examples

df <- tibble::tibble(
  x     = rep(letters[1:6], each = 1),
  x_grp = rep(letters[1:3], each = 2),
  y     = rep(LETTERS[1:3], each = 2),
  y_grp = rep(LETTERS[1:3], each = 2),
  z      = rep(LETTERS[1:3], each = 2),
  z_grp  = rep(LETTERS[1:3], times = 2))

unique_length(df, "x", "x_grp")

is_one2one(df, "x", "x_grp")
is_one2one(df, "y", "y_grp")
is_one2one(df, "z", "z_grp")

Description

Helper function for ordination methods

Usage

ordination(tbl, o_method, d_method = NULL, ...)

ord_plot(ord, score = "st_scores", x = 1, y = 2)

ord_add_group(ord, score = "st_scores", df, indiv, group)

ord_extract_score(ord, score = "st_scores", row_name = NULL)

Arguments

Value

ordination() returns result of ordination. $st_scores: scores for stand. $sp_scores: scores for species. $eig_val: eigen value for stand. $results_raw: results of original ordination function. $ordination_method: o_method. $distance_method: d_method. ord_plot() returns ggplot2 object. ord_extract_score() extracts stand or species scores from ordination result. ord_add_group() adds group data.frame into ordination scores. The columns that are one-to-multi to indiv are added, and the column named by group is always among them.

Examples

library(ggplot2)
library(vegan)
data(dune)
data(dune.env)

df <- 
  table2df(dune) %>%
  dplyr::left_join(tibble::rownames_to_column(dune.env, "stand"))
sp_dammy <- 
 tibble::tibble("species" = colnames(dune), 
                "dammy_1" = stringr::str_sub(colnames(dune), 1, 1),
                "dammy_6" = stringr::str_sub(colnames(dune), 6, 6))
df <- dplyr::left_join(df, sp_dammy)

ord_dca <- ordination(dune, o_method = "dca")
ord_pca <- 
  df %>%
  df2table() %>%
  ordination(o_method = "pca")

ord_dca_st <- 
  ord_extract_score(ord_dca, score = "st_scores")
ord_pca_sp <- 
  ord_add_group(ord_pca, 
  score = "sp_scores", df, indiv = "species", group = "dammy_1")

Description

Pad a string to the longest width of the strings.

Usage

pad2longest(string, side = "right", pad = " ")

Arguments

Value

    Strings.

Examples

x <- c("a", "ab", "abc")
pad2longest(x, side = "right", pad = " ")

Description

Expands each species into binary pseudospecies by cut levels. A pseudospecies of a cut level is present when the abundance is larger than zero and not less than the cut level. Pseudospecies that occur in no stand are dropped.

Usage

pseudospecies(x, cut_levels = c(0, 2, 5, 10, 20))

Arguments

Value

A binary matrix of stands by pseudospecies with "species", "level" and "cut_levels" attributes.

Examples

data(dune, package = "vegan")
psp <- pseudospecies(dune)
dim(psp)
head(colnames(psp))

Description

BiSS data is formatted as JSON.

Usage

read_biss(txt, join = TRUE)

Arguments

Value

A data frame.

Examples

path <- system.file("extdata", "biss_example.json", package = "ecan")
read_biss(path)

Description

Calculating diversity indices such as species richness (s), Shannon's H' (h), Simpson' D (d), Simpson's inverse D (i).

Usage

shdi(df, stand = NULL, species = NULL, abundance = NULL)

Arguments

Value

A data.frame. Including species richness (s), Shannon's H' (h), Simpson' D (d), Simpson's inverse D (i).

Examples

data(dune, package = "vegan")
df <- table2df(dune)
shdi(df)

Description

Gives a weight to each pseudospecies, so that the rare ones weigh less in the ordination. The original TWINSPAN downweights them before the correspondence analysis, and twinspan() does the same by default. The weights are used only in the ordination: the preference of the pseudospecies is counted on the raw occurrences.

Usage

tw_downweight(
  y,
  method = c("hill", "decorana"),
  fraction = 5,
  frq_lim = 0.2,
  w_min = 0.01,
  rw = NULL
)

Arguments

Details

Two ways are available. "hill" is the WEIGHT subroutine of the original TWINSPAN: a pseudospecies occurring in a smaller proportion of the stands than frq_lim is weighted in proportion to that shortfall, and no weight falls below w_min. "decorana" is the downweighting of decorana() and of vegan::downweight(), where the frequencies are compared with the most frequent pseudospecies instead of a fixed proportion.

Value

A numeric vector of the weight of each pseudospecies.

Examples

data(dune, package = "vegan")
psp <- pseudospecies(dune)
summary(tw_downweight(psp))
summary(tw_downweight(psp, method = "decorana"))

Description

The values that Hill's FORTRAN program sets for one division. They are used when twinspan(polish = "hill"), and are collected here so that the correspondence with the original is easy to check.

Usage

tw_hill_const()

Value

A named list of the constants. rat_lim, frq_lim, feeble, icw_exp, ipr_exp, cwt_min, cr_long, cr_cut, polish_iter, mz_crit, mz_out and mz_ind.

Examples

unlist(tw_hill_const())

Description

Used as the heterogeneity of a group in modified TWINSPAN.

Usage

tw_inertia(y, w = NULL)

Arguments

Value

A numeric of the total inertia (0 when it cannot be computed).

Examples

data(dune, package = "vegan")
tw_inertia(pseudospecies(dune))

Description

The preference is (f2 - f1) / (f2 + f1), where f1 and f2 are the relative frequencies of the pseudospecies in the negative and the positive group. It ranges from -1 (only in the negative group) to 1.

Usage

tw_preference(y, positive)

Arguments

Value

A numeric vector of the preference of each pseudospecies.

Examples

data(dune, package = "vegan")
psp <- pseudospecies(dune)
pos <- tw_ra(psp)$sample > 0
summary(tw_preference(psp, pos))

Description

Reciprocal averaging (first correspondence analysis axis)

Usage

tw_ra(y, w = NULL, rw = NULL, max_iter = 999, tol = 1e-10)

Arguments

Value

A list of stand scores ($sample), pseudospecies scores ($species), the eigenvalue ($eig) and $converged.

Examples

data(dune, package = "vegan")
ra <- tw_ra(pseudospecies(dune))
ra$eig

Description

The original TWINSPAN does not classify the species on the pseudospecies table itself, but on how faithful each species is to the groups of stands. Every group of the hierarchy, the terminal ones and the ones that were divided further, becomes three pseudo-quadrats, at the cut levels 0.8, 2 and 6 of the ratio between the frequency of the species inside the group and its frequency outside. A species weighs as much as it occurs, and a group weighs as much as it holds, multiplied by sqrt(2) for every level it stands above the deepest one, and doubled for the two upper cut levels.

Usage

tw_species_data(
  object,
  psp = object$pseudospecies,
  sp_map = attr(psp, "species"),
  levmax = object$max_depth
)

Arguments

Value

A list of the binary matrix ($y) of species by pseudo-quadrats, the weights of the species ($rw) and of the pseudo-quadrats ($cw), and the ratios ($ratio).

Examples

data(dune, package = "vegan")
tw <- twinspan(dune)
str(tw_species_data(tw))

Description

Arranges the community data with the stands and the species in the order of their division paths, as in the printed output of TWINSPAN. The dichotomy of each stand is shown by the digits below the table.

Usage

tw_two_way(object, cells = c("level", "abundance"))

## S3 method for class 'tw_two_way'
print(x, ...)

Arguments

Value

tw_two_way() returns a character matrix with class "tw_two_way", "stand_path" and "species_path" attributes.

print() returns the object invisibly.

Examples

data(dune, package = "vegan")
tw_two_way(twinspan(dune))

Description

A native R implementation of TWINSPAN (Hill 1979) and modified TWINSPAN (Roleček et al. 2009). The algorithm divides stands (samples) hierarchically by the first axis of a correspondence analysis (reciprocal averaging) of pseudospecies, refines the division with differential species, and summarises it with a small set of indicator pseudospecies.

Usage

twinspan(
  x,
  cut_levels = c(0, 2, 5, 10, 20),
  min_size = 5,
  max_depth = 6,
  max_indicators = 7,
  diff_threshold = 1/3,
  refine_iter = 5,
  modified = FALSE,
  n_clusters = NULL,
  use_indicator = FALSE,
  downweight = TRUE,
  polish = c("hill", "ecan"),
  species = TRUE
)

## S3 method for class 'twinspan'
as.hclust(x, ...)

## S3 method for class 'twinspan'
print(x, ...)

Arguments

Details

The package is written in plain R and needs no compiler. It is not a port of Hill's FORTRAN program, but polish = "hill" (the default) follows the steps of that program: the rare pseudospecies are downweighted as WEIGHT does, the axis is polished twice as POLISH does, the stands are divided at the middle of the range of the polished axis, and the stands of the critical zone around that point are placed by the indicator pseudospecies, whose number and threshold are the ones that misclassify fewest stands. The constants of the original are in tw_hill_const().

The two halves of a division are put in the order of the original as well: the half that resembles the group next to the one being divided comes first, so that neighbouring groups stay together.

On the dune, sipoo, varespec, mite, BCI and pyrifos data of vegan this reproduces the original program exactly: the same groups with the same numbers, the same divisions and the same eigenvalues. It does so on twenty randomly generated data sets as well. The species are classified in the way of the original as well, that is on how faithful each of them is to the groups of stands rather than on the pseudospecies table itself, and without indicators: see tw_species_data(). The species groups are those of the original too.

polish = "ecan" keeps the earlier way of this package, which was written from the published description alone: the division is refined with the pseudospecies whose preference reaches diff_threshold, and the stands are divided at the centroid of the axis. It is kept because it needs no zone or indicator to place a stand, but it does not follow the original as closely.

If the results of the original program are needed, the twinspan package of Oksanen (https://github.com/jarioksa/twinspan, MIT licensed) calls Hill's FORTRAN code itself.

Value

twinspan() returns a list with class "twinspan". $classification: a tibble of stand, group, path and depth. $species_classification: a tibble of species, group, path and depth (of pseudospecies when polish = "ecan"). $nodes: a list of the nodes of the division tree. $pseudospecies: the pseudospecies matrix. $call, and the parameters above.

as.hclust() returns an "hclust" object so that cls_color(), cls_add_group() and ggdendro::ggdendrogram() can be used.

print() returns the object invisibly.

References

Hill, M.O. (1979) TWINSPAN: a FORTRAN program for arranging multivariate data in an ordered two-way table by classification of the individuals and attributes. Cornell University, Ithaca.

Roleček, J., Tichý, L., Zelený, D. and Chytrý, M. (2009) Modified TWINSPAN classification in which the hierarchy respects cluster heterogeneity. Journal of Vegetation Science 20: 596-602.

Examples

data(dune, package = "vegan")
tw <- twinspan(dune)
tw
tw$classification

# modified TWINSPAN with a fixed number of groups
tw_mod <- twinspan(dune, modified = TRUE, n_clusters = 4)
table(tw_mod$classification$group)

# use with the clustering helpers of ecan
library(ggdendro)
ggdendro::ggdendrogram(stats::as.hclust(tw))