IUCNN has been updated to version 3.0 on github and will shortly be updated on CRAN to adapt to the retirement of sp and raster. The update may not be compatible with analysis-pipelines build with version 2.x

IUCNN

Batch estimation of species’ IUCN Red List threat status using neural networks.

Installation

1. Install IUCNN directly from Github using devtools.

install.packages("devtools")
library(devtools)

install_github("IUCNN/IUCNN")

2. Since some of IUCNNs functions are run in Python, IUCNN needs to set up a Python environment. This is easily done from within R, using the install_miniconda() function of the package reticulate (this will need c. 3 GB disk space). If problems occur at this step, check the excellent documentation of reticulate.

install.packages("reticulate")
library(reticulate)
install_miniconda()

3. Install the tensorflow python library. If you are using MacOS or Linux it is recommended to install tensorflow using conda:

reticulate::conda_install("r-reticulate","tensorflow=2.4")

If you are using Windows, you can install tensorflow using pip:

reticulate::py_install("tensorflow~=2.4.0rc4", pip = TRUE)

4. Finally install the npBNN python library from Github:

reticulate::py_install("https://github.com/dsilvestro/npBNN/archive/refs/tags/v0.1.11.tar.gz", pip = TRUE)

Usage

There are multiple models and features available in IUCNN. A vignette with a detailed tutorial on how to use those is available as part of the package: vignette("Approximate_IUCN_Red_List_assessments_with_IUCNN"). Running IUCNN will write files to your working directory.

A simple example run for terrestrial orchids (This will take about 5 minutes and download ~500MB of data for feature preparation into the working directory):

library(tidyverse)
library(IUCNN)

#load example data 
data("training_occ") #geographic occurrences of species with IUCN assessment
data("training_labels")# the corresponding IUCN assessments
data("prediction_occ") #occurrences from Not Evaluated species to prdict

# 1. Feature and label preparation
features <- iucnn_prepare_features(training_occ) # Training features
labels_train <- iucnn_prepare_labels(x = training_labels,
                                     y = features) # Training labels
features_predict <- iucnn_prepare_features(prediction_occ) # Prediction features

# 2. Model training
m1 <- iucnn_train_model(x = features, lab = labels_train)

summary(m1)
plot(m1)

# 3. Prediction
iucnn_predict_status(x = features_predict,
                     model = m1)

Additional features quantifying phylogenetic relationships and geographic sampling bias are available via iucnn_phylogenetic_features and iucnn_bias_features.

With model testing

library(tidyverse)
library(IUCNN)

#load example data 
data("training_occ") #geographic occurrences of species with IUCN assessment
data("training_labels")# the corresponding IUCN assessments
data("prediction_occ") #occurrences from Not Evaluated species to predict

# Feature and label preparation
features <- iucnn_prepare_features(training_occ) # Training features
labels_train <- iucnn_prepare_labels(x = training_labels,
                                     y = features) # Training labels
features_predict <- iucnn_prepare_features(prediction_occ) # Prediction features


# Model testing
# For illustration models differing in dropout rate and number of layers

mod_test <- iucnn_modeltest(x = features,
                            lab = labels_train,
                            logfile = "model_testing_results-2.txt",
                            model_outpath = "iucnn_modeltest-2",
                            mode = "nn-class",
                            dropout_rate = c(0.0, 0.1, 0.3),
                            n_layers = c("30", "40_20", "50_30_10"),
                            cv_fold = 5,
                            init_logfile = TRUE)

# Select best model
m_best <- iucnn_best_model(x = mod_test,
                          criterion = "val_acc",
                          require_dropout = TRUE)

# Inspect model structure and performance
summary(m_best)
plot(m_best)

# Train the best model on all training data for prediction
m_prod <- iucnn_train_model(x = features,
                            lab = labels_train,
                            production_model = m_best)

# Predict RL categories for target species
pred <- iucnn_predict_status(x = features_predict,
                             model = m_prod)
plot(pred)

Using a convolutional neural network

features <- iucnn_cnn_features(training_occ) # Training features
labels_train <- iucnn_prepare_labels(x = training_labels,
                                     y = features) # Training labels
features_predict <- iucnn_cnn_features(prediction_occ) # Prediction features

Citation

library(IUCNN)
citation("IUCNN")

Zizka A, Andermann T, Silvestro D (2022). “IUCNN - Deep learning approaches to approximate species’ extinction risk.” Diversity and Distributions, 28(2):227-241 doi: 10.1111/ddi.13450.

Zizka A, Silvestro D, Vitt P, Knight T (2021). “Automated conservation assessment of the orchid family with deep learning.” Conservation Biology, 35(3):897-908, doi: doi.org/10.1111/cobi.13616