bioclim_example: Example Bioclimatic Data for Ecological Modeling

In corrselect: Correlation-Based and Model-Based Predictor Pruning

Example Bioclimatic Data for Ecological Modeling

Description

A simulated dataset with the 19 WorldClim bioclimatic variables (https://www.worldclim.org/data/bioclim.html) measured at 100 geographic locations, with species richness as the response variable. Variables are organized into correlated blocks representing temperature (BIO1-BIO11) and precipitation (BIO12-BIO19).

Usage

bioclim_example

Format

A data frame with 100 rows and 20 variables:

species_richness

Integer. Number of species observed (response variable)

BIO1

Numeric. Annual Mean Temperature

BIO2

Numeric. Mean Diurnal Range

BIO3

Numeric. Isothermality

BIO4

Numeric. Temperature Seasonality

BIO5

Numeric. Max Temperature of Warmest Month

BIO6

Numeric. Min Temperature of Coldest Month

BIO7

Numeric. Temperature Annual Range

BIO8

Numeric. Mean Temperature of Wettest Quarter

BIO9

Numeric. Mean Temperature of Driest Quarter

BIO10

Numeric. Mean Temperature of Warmest Quarter

BIO11

Numeric. Mean Temperature of Coldest Quarter

BIO12

Numeric. Annual Precipitation

BIO13

Numeric. Precipitation of Wettest Month

BIO14

Numeric. Precipitation of Driest Month

BIO15

Numeric. Precipitation Seasonality

BIO16

Numeric. Precipitation of Wettest Quarter

BIO17

Numeric. Precipitation of Driest Quarter

BIO18

Numeric. Precipitation of Warmest Quarter

BIO19

Numeric. Precipitation of Coldest Quarter

Details

This dataset demonstrates a common problem in ecological modeling: bioclimatic predictors are highly correlated within groups (temperature variables BIO1-BIO11 are highly correlated; precipitation variables BIO12-BIO19 are moderately correlated), leading to multicollinearity issues. The species richness response depends on a subset of predictors.

Use case: Demonstrating corrPrune() and modelPrune() for reducing correlated environmental predictors before fitting species distribution models.

Source

Simulated data based on the 19 WorldClim bioclimatic variables

See Also

corrPrune(), modelPrune()

Examples

data(bioclim_example)

# The 19 WorldClim bioclimatic variables (https://www.worldclim.org/data/bioclim.html)
# Many are highly correlated, making them ideal for pruning

# Remove highly correlated variables
pruned <- corrPrune(bioclim_example[, -1], threshold = 0.7)
ncol(pruned)  # Reduced from 19 to 12 variables

# Model-based pruning with VIF
model_data <- modelPrune(species_richness ~ .,
                         data = bioclim_example,
                         limit = 5)
attr(model_data, "selected_vars")

corrselect documentation built on July 19, 2026, 1:06 a.m.