Ordering species abundance by gradients
library(remotes)
remotes::install_github("Edbbioeco/ordenaR")
library(ordenaR)or simply:
install.packages("ordenaR")
library(ordenaR)This package works to better understand species ordering by gradients, and how gradients affect communities composition structure (Hill, 1973). {ordenaR} package can analyse species abundance under direct and indirect gradients (Magnusson & Bacchario, 2021):
-
Direct gradients: numeric values gradients, such as temperature, humidity, altitude, and such else numeric variables;
-
Indirect gradients, categorical values gradients, such as sample units, vegetation structure, location types, and such else categorical variables.
That package aload to use four functions:
-
data_ordenar: input data for exemples;
-
order_bar: ordering species by barplots
-
order_circles: ordering species by circlesplots
-
order_species: ordering species and ranking them into a dataframe.
Toour exemples, we use ordenaR package data. First, lets load package and next our data.
data("data_ordenar", package = "ordenaR")
data_ordenar## sample_unit gradient_1 gradient_2 species_1 species_2 species_3 species_4
## 1 p-1 0.780000 6.1200 0 0 2 0
## 2 p-2 1.232083 6.5625 0 0 3 0
## 3 p-3 1.684167 7.0050 0 0 4 0
## 4 p-4 2.136250 7.4475 0 0 3 0
## 5 p-5 2.588333 7.8900 0 0 2 0
## 6 p-6 3.040417 0.2700 0 0 0 4
## 7 p-7 3.492500 0.5725 0 0 0 7
## 8 p-8 3.944583 0.8750 0 0 0 9
## 9 p-9 4.396667 1.1775 0 0 0 8
## 10 p-10 4.848750 1.4800 0 0 0 8
## 11 p-11 5.300833 2.5000 0 0 0 0
## 12 p-12 5.752917 2.6675 0 0 0 0
## 13 p-13 6.205000 2.8350 0 0 0 0
## 14 p-14 6.657083 3.0025 0 0 0 0
## 15 p-15 7.109167 3.1700 0 0 0 0
## 16 p-16 7.561250 3.1300 0 1 0 0
## 17 p-17 8.013333 3.5300 0 4 0 0
## 18 p-18 8.465417 3.9300 0 7 0 0
## 19 p-19 8.917500 4.3300 0 6 0 0
## 20 p-20 9.369583 4.7300 0 2 0 0
## 21 p-21 9.821667 8.0100 4 0 0 0
## 22 p-22 10.273750 8.4800 5 0 0 0
## 23 p-23 10.725833 8.9500 6 0 0 0
## 24 p-24 11.177917 9.4200 4 0 0 0
## 25 p-25 11.630000 9.8900 3 0 0 0
## species_5
## 1 0
## 2 0
## 3 0
## 4 0
## 5 0
## 6 0
## 7 0
## 8 0
## 9 0
## 10 0
## 11 5
## 12 7
## 13 9
## 14 10
## 15 12
## 16 0
## 17 0
## 18 0
## 19 0
## 20 0
## 21 0
## 22 0
## 23 0
## 24 0
## 25 0
Next, lets get a summary about our data variables.
library(dplyr)##
## Anexando pacote: 'dplyr'
## Os seguintes objetos são mascarados por 'package:stats':
##
## filter, lag
## Os seguintes objetos são mascarados por 'package:base':
##
## intersect, setdiff, setequal, union
data_ordenar |> dplyr::glimpse()## Rows: 25
## Columns: 8
## $ sample_unit <chr> "p-1", "p-2", "p-3", "p-4", "p-5", "p-6", "p-7", "p-8", "p-9"…
## $ gradient_1 <dbl> 0.780000, 1.232083, 1.684167, 2.136250, 2.588333, 3.040417, 3…
## $ gradient_2 <dbl> 6.1200, 6.5625, 7.0050, 7.4475, 7.8900, 0.2700, 0.5725, 0.875…
## $ species_1 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 4…
## $ species_2 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 4, 7, 6, 2, 0…
## $ species_3 <dbl> 2, 3, 4, 3, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ species_4 <dbl> 0, 0, 0, 0, 0, 4, 7, 9, 8, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ species_5 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 5, 7, 9, 10, 12, 0, 0, 0, 0, 0,…
To our analysis, we use dataframe/tibble class objetos, containg:
-
A column contain direct gradient (numeric values) or indirect gradient (categorical values). Pay attention whether your indirect gradient are contained on row names (i.e. sample units). If it is true, turn it into a columns, such using tibble::rownames_to_column() function, previously;
-
A set of Columns containing species abundance data. Pay attention such species abundance data are conttained by a single column.
To best ordering species under a gradient, we relocate them in a plot, to better understand how gradients affets species abundance. This is based on Hill (1973), calculating reciprocal average, where every species get a rank, and we relocate the species based on this rank. For indirect gradients, we assumes a theorical gradient (i.e. 1, 2. 3, 4,…, nbservation) and calcule them.
-
$P_i$ = abundancie for species$P$ to row$i$ ; -
$E_i$ = value to gradient$E$ to row$i$ .
Now, let’s to analyse our data to a direct gradient. We informe our
gradient column (gradient argument), a column named by gradient_1,
and species abundance data (species argument), a column set ranging
from 4th to 8th column in our data frame.
ordenaR::order_species(data = data_ordenar,
gradient = "gradient_1",
species = 4:8)## Warning: substituindo importação prévia 'magrittr::set_names' por
## 'purrr::set_names' quando carregando 'ordenaR'
## Warning: substituindo importação prévia 'magrittr::extract' por 'tidyr::extract'
## quando carregando 'ordenaR'
## # A tibble: 5 × 3
## specie `Reciprocal average` Rank
## <chr> <dbl> <int>
## 1 species_1 10.7 1
## 2 species_2 8.56 2
## 3 species_5 6.38 3
## 4 species_4 4.06 4
## 5 species_3 1.68 5
ordenaR::order_species(data = data_ordenar,
gradient = "gradient_2",
species = 4:8)## # A tibble: 5 × 3
## specie `Reciprocal average` Rank
## <chr> <dbl> <int>
## 1 species_1 8.89 1
## 2 species_3 7.00 2
## 3 species_2 4.01 3
## 4 species_5 2.90 4
## 5 species_4 0.951 5
And now to a indirect gradient, only informing direct = FALSE.
ordenaR::order_species(data = data_ordenar,
gradient = "sample_unit",
species = 4:8,
direct = FALSE)## # A tibble: 5 × 3
## specie `Reciprocal average` Rank
## <chr> <dbl> <int>
## 1 species_1 22.9 1
## 2 species_2 18.2 2
## 3 species_5 13.4 3
## 4 species_4 8.25 4
## 5 species_3 3 5
The main reason to ordering species is to know how to best visualize species abundance along a gradient. It remains the fact there is not a intrinsec reason to alocate a x-species abundance firts to a y-species abundance. That way, any species can be in any position in a plot. That best species postion configuration is setted by species rank, previously calculated.
Now, we build a barplot to a direct gradient, similar to
ordenaR::order_species() function.
ordenaR::order_bar(data = data_ordenar,
gradient = "gradient_1",
species = 4:8)
Notice the species are not ordered by its numbers, and that position configuration maximize species abundance ordering along the gradient. Now, let’s to build a barplot also, but to a indirect grtadient.
ordenaR::order_bar(data = data_ordenar,
gradient = "sample_unit",
species = 4:8,
direct = FALSE)
Notice not only species are ordered, but sample units too. Moreover, we can set bar width.
# Original
ordenaR::order_bar(data = data_ordenar,
gradient = "gradient_2",
species = 4:8,
width = 0.1)
# Set to width = 0.1
ordenaR::order_bar(data = data_ordenar,
gradient = "gradient_2",
species = 4:8,
width = 0.5)
# Set to width = 0.5
ordenaR::order_bar(data = data_ordenar,
gradient = "gradient_2",
species = 4:8,
width = 1)
# Set to width = 1.5
ordenaR::order_bar(data = data_ordenar,
gradient = "gradient_2",
species = 4:8,
width = 1.5)
We also can build those plots, but using circles instead bars. Instead
bar hight, use use circle diameter to understando species abundance
values. Similar to ordenaR::order_bar() function, we use
ordenaR::order_circle() function.
ordenaR::order_circle(data = data_ordenar,
gradient = "gradient_1",
species = 4:8)
And now to a indirect gradient.
ordenaR::order_circle(data = data_ordenar,
gradient = "sample_unit",
species = 4:8,
direct = FALSE)
Similar to bar width, we can set circle diameter differences range.
# Original
ordenaR::order_circle(data = data_ordenar,
gradient = "gradient_2",
species = 4:8)
# Set to range = 5
ordenaR::order_circle(data = data_ordenar,
gradient = "gradient_2",
species = 4:8,
range = 5)
# Set to range = 15
ordenaR::order_circle(data = data_ordenar,
gradient = "gradient_2",
species = 4:8,
range = 15)
# Set to range = 20
ordenaR::order_circle(data = data_ordenar,
gradient = "gradient_2",
species = 4:8,
range = 20)