sdmbench

Species Distribution Modeling (SDM) is a field of increasing importance in ecology¹. Several popular applications of SDMs are understanding climate change effects on species², natural reserve planning³ and invasive species monitoring⁴. The sdmbench package solves several issues related to the development and evaluation of those models by providing a consistent benchmarking workflow:

• consistent species occurrence data acquisition and preprocessing
• consistent environmental data acquisition and preprocessing (both current data and future projections)
• consistent spatial data partitioning
• integration of a wide variety of machine learning models
• graphical user interface for non/semi-technical users

The end result of a sdmbench SDM analysis is to determine the model–data processing combination that results in the highest predictive power for the species of interest. Such an analysis is useful to researchers who want to avoid issues of model selection and evaluation, and want to rapidly test prototypes of species distribution models.

Installation

# add build_vignettes = TRUE if you want the package vignette
devtools::install_github("boyanangelov/sdmbench")

The package has two optional dependencies for full functionality:

Deep learning (Keras/TensorFlow). Requires a working Python installation.

# consult the keras documentation for GPU support
keras::install_keras()

MaxEnt. Requires Java and maxent.jar placed in the dismo Java directory. See ?dismo::maxent for installation instructions.

Examples

Download and prepare benchmarking data:

library(sdmbench)
#> sdmbench: Tools for benchmarking Species Distribution Models
#> ============================================================
#> For more information visit https://github.com/boyanangelov/sdmbench
#> To start the GUI: run_sdmbench()

benchmarking_data <- get_benchmarking_data("Loxodonta africana", limit = 1200, climate_resolution = 10)
#> Downloading occurrence data from GBIF...
#> Cleaning coordinates...
#> Downloading climate data...
#> Done!

head(benchmarking_data$df_data)
#>   bio1 bio2 bio3 bio4 bio5 bio6 bio7 bio8 bio9 bio10 bio11 bio12 bio13
#> 1  181  132   59 3078  283   60  223  187  141   220   141   425    48
#> 2  180  142   58 3336  292   51  241  210  136   222   136   410    50
#>   bio14 bio15 bio16 bio17 bio18 bio19 label
#> 1    25    19   122    87   114    87     1
#> 2    21    24   122    73   121    73     1

Apply spatial partitioning and set up learners:

benchmarking_data$df_data <- partition_data(
    dataset_raster = benchmarking_data$raster_data,
    dataset        = benchmarking_data$df_data,
    env            = benchmarking_data$raster_data$climate_variables,
    method         = "block"
)

learners <- list(
    mlr3::lrn("classif.ranger",  predict_type = "prob"),
    mlr3::lrn("classif.log_reg", predict_type = "prob"),
    mlr3::lrn("classif.rpart",   predict_type = "prob"),
    mlr3::lrn("classif.svm",     predict_type = "prob")
)

benchmarking_data$df_data <- na.omit(benchmarking_data$df_data)

Benchmark models and retrieve results:

bmr <- benchmark_sdm(benchmarking_data$df_data,
                     learners     = learners,
                     dataset_type = "block",
                     sample       = FALSE)

best_results <- get_best_model_results(bmr)
best_results
#> # A tibble: 4 × 3
#>   learner_id       iteration classif.auc
#>   <chr>                <int>       <dbl>
#> 1 classif.ranger           1       0.992
#> 2 classif.log_reg          2       0.915
#> 3 classif.rpart            3       0.900
#> 4 classif.svm              4       0.963

Plot the best model's predicted habitat suitability:

plot_sdm_map(raster_data = benchmarking_data$raster_data,
             bmr         = bmr,
             learner_id  = best_results$learner_id[1],
             iteration   = best_results$iteration[1],
             map_type    = "static")

Using custom data

If you prefer to bring your own occurrence data rather than downloading from GBIF, you can upload a CSV via the Shiny GUI sidebar. The file must have the following columns:

species	decimalLongitude	decimalLatitude
Lynx lynx	24.3	51.2
…	…	…

Custom data is currently supported only for General Models and does not include mapping functionality.

A good starting point to explore the full package is to launch the GUI:

run_sdmbench()

Screenshots:

Vignette

A thorough introduction to the package is available as a vignette in the package, and online.

vignette("sdmbench")

Tests

Tests are in the tests/ directory and can be run with:

devtools::test()

or Ctrl/Cmd + Shift + T in RStudio. Note that tests download live data from GBIF and require an internet connection.

Contributors

Contributions are welcome and guidelines are stated in CONTRIBUTING.md.

License

MIT (LICENSE.md)

References

1. Elith, J. & Leathwick, J. R. Species Distribution Models: Ecological Explanation and Prediction Across Space and Time. Annu. Rev. Ecol. Evol. Syst. 40, 677–697 (2009).

2. Austin, M. P. & Van Niel, K. P. Improving species distribution models for climate change studies: Variable selection and scale. J. Biogeogr. 38, 1–8 (2011).

3. Guisan, A. et al. Predicting species distributions for conservation decisions. Ecol. Lett. 16, 1424–1435 (2013).

4. Descombes, P. et al. Monitoring and distribution modelling of invasive species along riverine habitats at very high resolution. Biol. Invasions 18, 3665–3679 (2016).