rsa-unstructured-data-comp

Occurrence data in GBIF have various limitations and biases, due in part to biases in survey effort. Therefore, analyses that use these data to estimate biodiversity status and trends can be misleading.

This analysis forms part of a study to establish the necessary conditions for ensuring the accuracy of estimates on biodiversity status and trends. It compares results obtained when analyses use unstructured cube data from GBIF, with those from analyses using data from structured monitoring programs, which are designed to minimise biases and errors and provide a largely accurate representation of biodiversity.

The analysis focuses on birds in the Western Cape of South Africa. The structured data are from the Southern African Bird Atlas Programme 2 (SABAP2). The analysis is performed for the Western Cape and for the area of the Western Cape where the Hessequa Systematic Atlasing Subproject is undertaken (near Stillbaai). SABAP2 uses a pentad grid - grid cells at a 5-minute² spatial resolution (~ 8.2 km² in southern Africa). The analysis is performed at pentad resolution, and at quarter-degree grid cell resolution (QDGC) - nine pentads fit into one quarter-degree grid cell (15 min²)

This analysis forms part of the work of task 4.5 of the B-Cubed project.

Workflow

The workflow comprises 4 steps: (1) data acquisition, (2) data preparation, (3) data analysis, and (4) reporting.

1. data acquisition → 3. data analysis

Data are downloaded (Step 1), prepared (Step 2), and analysed (Step 3) by running two R Markdowns, one for each area covered by the analysis.

What to run

• prepare_&_compare_data_Hessequa.Rmd
• prepare_&_compare_data_WesternCape.Rmd

Where

R/

1. data acquisition

What happens

• The most recent versions of the data are downloaded. Most of these data are downloaded from GBIF, but some (e.g., SABAP2 survey effort data) are downloaded from SABAP2. Instructions are provided in the R markdowns.
• Alternatively, the exact datasets used in the analyses can be downloaded by following the links embedded in the Rmd files.

Outputs

• Raw data are stored in: data/raw/

2. data preparation

What happens

• Pentad cubes are created from downloaded unstructured occurrence data downloaded from GBIF.
• Pentad and QDGC cubes are created for the SABAP2 data, removing adhoc records.
• QDGC unstructured data cube downloaded from GBIF is subsetted as required.
• SABAP2 dataset with survey effort data is created for pentads and QDGC.
• Creates output directories automatically if they do not yet exist.
• Writes prepared data to file.

Outputs

• Prepared data are stored in: data/processed/ in .csv and .rds formats, as required for analysis.

3. data analysis

What happens

• Basic data exploration of structured and unstructured data
• Calculation of metrics and indicators using structured and unstructured data, and comparison of results

4. reporting

Performs the analysis and produces the outputs required for reporting, as identified based on the analysis.

What to run

• generate_outputs_for_reporting.Rmd

Where

R/

What happens

• Reads processed data generated during 2. data preparation
• Creates output directories automatically if they do not yet exist.
• Produces outputs (e.g., figures) of analysis for reporting.

Outputs

• Outputs are stored in: output/

Installation instructions

This repository can be downloaded onto your computer as a zip file, which will contain all the required folders. The main folder contains the Rstudio project file for the analysis (rsa-unstructured-data-comp.Rproj). This file should be opened to run the analysis.

Several R packages need to be installed from R Cran or GitHub. Executing the script below will install the required packages

Install packages from R Cran

packages = c("tidyverse", "here"", "sf", "rgbif", "iNEXT", "ggpubr", "RColorBrewer", "devtools", "magrittr", "ggplot2")

package.check <- lapply(
  packages,
  FUN = function(x) {
    if (!require(x, character.only = TRUE)) {
      install.packages(x, dependencies = TRUE)
      library(x, character.only = TRUE)
    }
  }
)

Install packages from GitHub

devtools::install_github("b-cubed-eu/b3gbi")
install.packages("gcube", repos = "https://b-cubed-eu.r-universe.dev")