R-CMD-check CRAN_Status_Badge CRAN RStudio mirror downloads

Automatically runs numerous processes from R packages ‘gbm’ and ‘dismo’ and script ‘gbm.utils.R’ which contains Elith et al.’s functions: roc, calibration, and gbm.predict.grids, as well as running my packages gbm.bfcheck, gbm.basemap,, gbm.rsb, gbm.cons, gbm.valuemap, and gbm.loop.

Especially on Linux systems it is recommended to type, in terminal: sudo apt install libgeos-dev sudo apt install libproj-dev sudo apt install libgdal-dev then manually install rgeos and rgdal in R/RStudio.

Also see each script’s Details section in the manual pages, as these frequently contain tips or common bugfixes.

I strongly recommend that you download papers 1 to 5 (or just the doctoral thesis) on, with emphasis on P4 (the guide) and P1 (statistical background). Elith et al 2008 ( is also strongly recommended. Also it’s imperative you read the R help files for each function before you use them. In RStudio: Packages tab, scroll to, click its name, the click the function to see its man (manual) page. Read the whole thing. Function man pages can also be accessed from the console by typing


Just because you CAN try every conceivable combination of tc, lr, bf, all, at once doesn’t mean you should. Try a range of lr in shrinking orders of magnitude from 0.1 to 0.000001, find the best, THEN try tc c(2, n.expvars), find the best THEN bf c(0.5, 0.75, 0.9) and then in between if either outperform 0.5.

Automated Boosted Regression Tree modelling and mapping suite

Automates delta log normal boosted regression trees abundance prediction. Loops through all permutations of parameters provided (learning rate, tree complexity, bag fraction), chooses the best, then simplifies it. Generates line, dot and bar plots, and outputs these and the predictions and a report of all variables used, statistics for tests, variable interactions, predictors used and dropped, etc. If selected, generates predicted abundance maps, and Unrepresentativeness surfaces.


Calculates minimum Bag Fraction size for

Provides minimum bag fractions for, preventing failure due to bf & samples rows limit.


Creates Basemaps for mapping from your data range

Downloads unzips crops & saves NOAAs global coastline shapefiles to user-set box. Use for ‘shape’ in If downloading in RStudio uncheck “Use secure download method for HTTP” in Tools > Global Options > Packages.

Maps of predicted abundance from Boosted Regression Tree modelling

Generates maps from the outputs of gbm.step then gbm.predict.grids, handled automatically within but can be run alone, and generates representativeness surfaces from the output of gbm.rsb.


Representativeness Surface Builder

Loops through explanatory variables comparing their histogram in ‘samples’ to their histogram in ‘grids’ to see how well the explanatory variable range in samples represents the range being predicted to in grids. Assigns a representativeness score per variable per site in grids, and takes the average score per site if there’s more than 1 expvar. Saves this to a CSV; it’s plotted by if called in This shows you which areas have the most and least representative coverage by samples, therefore where you can have the most/least confidence in the predictions from gbm.predict.grids. Can be called directly, and choosing a subset of expvars allows one to see their individual / collective representativeness.


Conservation Area Mapping

Runs for multiple subsets of the same overall dataset and scales the combined results, leading to maps which highlight areas of high conservation importance for multiple species in the same study area e.g. using juvenile and adult female subsets to locate candidate nursery grounds and spawning areas respectively.


Decision Support Tool that generates (Marine) Protected Area options using species predicted abundance maps

Scales response variable data, maps a user-defined explanatory variable to be avoided, e.g. fishing effort, combines them into a map showing areas to preferentially close. Bpa, the precautionary biomass required to protect the spawning stock, is used to calculate MPA size. MPA is then grown to add subsequent species starting from the most conservationally at-risk species, resulting in one MPA map per species, and a multicolour MPA map of all. All maps list the percentage of the avoid-variables total that is overlapped by the MPA in the map legend.


Calculate Coefficient Of Variation surfaces for predictions

Processes a user-specified number of loops through the same parameter combinations and calculates the Coefficient Of Variation in the predicted abundance scores for each site aka cell. This can be mapped to spatially demonstrate the output variance range.


ggplot-based update to PDP for factorial/categorical/character variables, allows changing order of categorical variables, and changing angle of x-axis labels to avoid them being cut off.


Linear plot of two variables.


Loops through lmplots for all expvars (x) against the same resvar (y).

roc & calibration

Internal functions authored by Elith & Leathwick, used by

Local copy of dismo’s gbm.step, with added functions to generate model evaluation metrics such as root mean squared error and amount of deviance explained relative to null.


You can install the released version of from CRAN with:


And the development version from GitHub with:

# install.packages("devtools")


(See each function’s help file for specific examples, and the documents listed above)

ToDo List

See GitHub issues section Feel free to contribute to this!