GARP with Best Subsets - new openModeller implementation

openModeller id: GARP_BS

Current version: 3.0.4    Developer(s): Ricardo Scachetti Pereira

Accepts Categorical Maps: no

Requires absence points: yes

Author(s): Anderson, R. P., D. Lew, D. and A. T. Peterson.


GARP is a genetic algorithm that creates ecological niche models for species. The models describe environmental conditions under which the species should be able to maintain populations. For input, GARP uses a set of point localities where the species is known to occur and a set of geographic layers representing the environmental parameters that might limit the species' capabilities to survive. This algorithm applies the Best Subsets procedure using the new openModeller implementation in each GARP run. Please refer to GARP single run algorithm description for more information about the differences between DesktopGarp and the new GARP implementation.


Anderson, R. P., D. Lew, and A. T. Peterson. 2003. Evaluating predictive models of species' distributions: criteria for selecting optimal models.Ecological Modelling, v. 162, p. 211 232.


Training Proportion

openModeller id: TrainingProportion

Percentage of occurrence data to be used to train models.

Data type: real  Domain: [0.0, 100.0]  Typical value: 50

Total Runs

openModeller id: TotalRuns

Maximum number of GARP runs to be performed.

Data type: integer  Domain: [0.0, 10000.0]  Typical value: 20

Hard Omission Threshold

openModeller id: HardOmissionThreshold

Maximum acceptable omission error. Set to 100% to use only soft omission.

Data type: real  Domain: [0.0, 100.0]  Typical value: 100

Models Under Omission Threshold

openModeller id: ModelsUnderOmissionThreshold

Minimum number of models below omission threshold.

Data type: integer  Domain: [0.0, 10000.0]  Typical value: 20

Commission Threshold

openModeller id: CommissionThreshold

Percentage of distribution of models to be taken regarding commission error.

Data type: real  Domain: [0.0, 100.0]  Typical value: 50

Commission Sample Size

openModeller id: CommissionSampleSize

Number of samples used to calculate commission error.

Data type: integer  Domain: [1.0, oo]  Typical value: 10000

Maximum Number of Threads

openModeller id: MaxThreads

Maximum number of threads of executions to run simultaneously.

Data type: integer  Domain: [1.0, 1024.0]  Typical value: 1

Max generations

openModeller id: MaxGenerations

Maximum number of iterations (generations) run by the Genetic Algorithm.

Data type: integer  Domain: [1.0, oo]  Typical value: 400

Convergence limit

openModeller id: ConvergenceLimit

Defines the convergence value that makes the algorithm stop (before reaching MaxGenerations).

Data type: real  Domain: [0.0, 1.0]  Typical value: 0.01

Population size

openModeller id: PopulationSize

Maximum number of rules to be kept in solution.

Data type: integer  Domain: [1.0, 500.0]  Typical value: 50


openModeller id: Resamples

Number of points sampled (with replacement) used to test rules.

Data type: integer  Domain: [1.0, 100000.0]  Typical value: 2500

Sample models

The following image shows a possible model in the environmental space (temperature x precipitation) generated with the Thalurania furcata boliviana localities dataset. It is possible to notice the overlapping of different rulesets.

fig. 1: sample model