Marxan: Input and Output - Land Change Modeler

The two Marxan panels within Land Change Modeler are meant to interface with the Marxan software which provides decision support for reserve system design. Marxan can be used to identify areas that meet biodiversity targets, taking into account minimum costs. Marxan is freely available software from the University of Queensland. See Note 2 for important downloading details. The IDRISI implementation of the Marxan interface works ONLY with Marxan Version 1.8.10. Note that all data layers need to have the same spatial resolution, projection, and dimension. Species distribution layers must have Boolean values with 1 representing species presence and 0 representing species absence. The planning unit layer must be in either integer or byte format. Please refer to the Marxan manual (available at same site noted above) for additional information.

Marxan: Input and Output Operation

1. Open the Marxan: Input and Output panel from the Planning tab.

2. Enter the name of the planning unit layer.

-- The planning unit layer determines the regions in the study area which may be selected as a new reserve area. The shapes of planning units can either be uniform, i.e., each pixel represents a planning unit, or larger and irregular shaped, i.e., created by natural boundaries.

-- Each planning unit ID must be unique.

-- See section 1.7.1 on choosing planning units or Appendix C-2 on creating the planning unit file of the Marxan User Manual for more information.

3. Specify the number of species distribution layers and enter their names in the grid. Alternately, enter the name of a raster group file.

-- Raster group files are created with IDRISI Explorer.

-- Species distribution layers must be Boolean images describing the presence or absence status of a single species. The pixel value is 1 if a species is present and 0 if it is absent.

4. Once the species distribution layers are specified, the name column of the species grid will populate. The default setting for the species type column is 1.

-- All species that will have the same target percentage and penalty factor should have the same type number. The same parameters will be applied to species of the same type by specifying the values under the species grid and clicking the Autofill button for quick input.

5. Specify a target percentage. The input should be a number between 0 and 100

-- The target percentage refers to the percentage of the presence cells in the species distribution layer. For example, if 70.5 is entered as the percentage, 70.5% of the presence cells will be calculated and the number will be placed in the corresponding row for the Target (no. cells) field of the grid.

6. Specify the species penalty factor (SPF).

-- The SPF is a multiplier that determines the size of the penalty that will be added to the objective function if the target for the species is not met in the current reserve scenario. A higher SPF means that it is more likely the species target will be met. See section 3.2.2.4 on the conservation feature penalty factor of the Marxan User Manual for more information.

7. Click the Autofill Spec. Type button to update the SPF and Target fields of the specified type of species in the species grid.

-- The values in the target field represents the number of cells.

-- If you have changed the species type, be sure to adjust the species type number next to the Autofill button using the arrow buttons. A new target % and SPF will need to be filled in for each species type.

8. Select whether to use a planning unit tenure layer and enter its name.

-- The planning unit tenure layer depicts current land use. In the tenure layer, a value of 0 represents available land, 2 represents existing reserves which cannot be removed, and 3 represents unfeasible land which cannot be added as new reserves.

9. Select whether to use a land cost layer and enter its name.

--The land cost layer represents the relative cost of pixels to determine where the lowest cost reserve can be planned. If there is no land cost layer, land area will be used as a surrogate for cost.

10. Select whether to use a boundary length file.

– This is an ASCII file with .dat extension that will be created automatically if the option is selected. The boundary length file records the length of the boundary between adjacent planning units. This file is necessary if you wish to use the boundary length modifier to control reserve compactness in the next panel.

11. Enter a prefix for all output files. See Notes for more information about output filenames.

12. Click Continue to open the Marxan: Parameters panel.

Notes

1. Output file types and names

Input parameter file: Input.dat

Planning unit file: prefix_pu.dat

Conservation feature file: prefix_spec.dat

Planning unit versus conservation feature file: prefix_matrix.dat

Solutions for each run: prefix_r001.txt

Best solution from all runs: prefix_best.txt

Missing value information for each run: prefix_mv001.txt

Missing value information for the best run: prefix_mvbest.txt

Scenario details: prefix_sen.txt

Summed solution: prefix_ssoln.txt

Screen log file: prefix_log.dat

Best solution raster map: prefix_bestsolution.rst

Summed solution raster map: prefix_ summedsolution.rst

See section 3.2.1 on input parameter files, 3.2.2 on conservation feature files, 3.2.3 on planning unit files, 3.2.4 on planning unit versus conservation feature files and 5.3 on output files in the Marxan User Manual for more information.

2. Marxan must be downloaded and installed before utilizing the IDRISI Marxan interface. After downloading Marxan, locate the marxan.exe file and move it to the IDRISI Selva Mods folder (default location: C:\Program Files\IDRISI Taig\Mods). If the exe file is not found in the folder, an error message will appear stating that the marxan.exe is not found in the directory. The full path of the directory will be shown to the user in this message.

3. References

Game, E. T. and H. S. Grantham. (2008). Marxan User Manual: For Marxan version 1.8.10. University of Queensland, St. Lucia, Queensland, Australia, and Pacific Marine Analysis and Research Association, Vancouver, British Columbia, Canada.

Possingham, H. P., I. R. Ball and S. Andelman. (2000). Mathematical methods for identifying representative reserve networks. In: S. Ferson and M. Burgman (eds) Quantitative methods for conservation biology. Springer-Verlag, New York, pp. 291-305.

Marxan: Parameters - Land Change Modeler

Marxan: Parameters Operation

The Marxan: Input and Output panel, located in the Planning tab of Land Change Modeler, must be filled out prior to the Marxan: Parameters panel.

1. If the boundary length file option was selected in the previous Marxan panel, enter a non-negative number for the boundary length modifier (BLM).

-- A value of 0 will remove the boundary length from consideration. Higher BLM values will produce a more compact reserve system.

2. Enter a value for repeat runs.

-- Each run is a solution to the reserve problem and is independent of every other one. Marxan recommends that you begin with a small number (e.g. 5) to make sure the program is running with solutions meeting the required targets, and then increasing to a minimum of 100. A higher number will increase the processing time. See section 3.2.1.1.1 on repeat runs in the Marxan User Manual for more information.

3. Enter a value for species missing proportion.

-- This is the proportion of the target a species must reach in order for it to be reported as met. The value should be between 0 and 1. See section 3.2.1.5.4 on species missing proportion of the Marxan User Manual for more information.

-- A value of 1 means that if 100% of the target area specified in the previous Marxan panel is not conserved, the species is reported as having not been conserved. If a lower number such as .95 is used, a species is reported as having been conserved if 95% or higher of its target area is included in the reserve system.

4. Select a run mode, the method by which Marxan will find a solution. The choices include simulated annealing, iterative improvement, other heuristic algorithms, or a combination of these. Marxan recommends using simulated annealing followed only by iterative improvement for most applications.

-- Simulated annealing is an optimization algorithm which derives its name from the annealing technique in metallurgy. During the algorithm, the temperature starts at a high value and decreases gradually. When the temperature is high, both good and bad changes can be accepted or rejected. In Marxan, the change is either adding a new planning unit into the current reserve system or removing an existing planning unit from the current reserve system. The probability of accepting change is a function of the current temperature and the difference in the objective function’s values if the change is accepted. The probability is then compared to a random number to decide whether to accept the change. As the temperature decreases, the chance of accepting a bad change decreases until, finally, only good changes are accepted. See Appendix B-2.1 on simulated annealing in the Marxan User Manual for more information.

-- Iterative improvement is a simple optimization algorithm where a random change will be accepted if it will improve the value of the objective function. In Marxan, the type of change is dependent on the type of iterative improvement. See Appendix B-2.2 on iterative improvement in the Marxan User Manual for more information.

-- Heuristic algorithms mimic the way someone would choose a reserve system by hand by beginning with an empty reserve system and then adding planning units to the system sequentially until some stopping criteria is reached. See Appendix B-2.3 on heuristic algorithms in the Marxan User Manual for more information.

5. If simulated annealing will be used, specify the number of iterations and number of temperature decreases. Also indicate whether to use adaptive annealing. If adaptive annealing is not used, specify the initial temperature and cooling factor. If a heuristic will be used, choose the type from the drop-down menu. If iterative improvement will be used, choose the type from the drop-down menu.

-- Adjustments are rarely needed in annealing control parameters; the defaults will generally fit the analysis See section 3.2.1.3.1 on number of iterations, temperature decreases, initial temperature and cooling factor and Appendix B-2.1 on simulated annealing in the Marxan User Manual for more information.

6. Indicate whether to enable a cost threshold and enter values for the threshold and penalty factors A and B.

-- These parameters are used to set a fixed budget on the reserve system. See section 3.2.1.6 on cost threshold and Appendix B-1.4 on cost threshold penalty in the Marxan User Manual for more information.

-- Threshold is a number set in relation to the cost of the reserve system where if the reserve rises above the set cost it will be penalized.

-- Penalty Factor A controls the size of the penalty where the higher the number, the larger the penalty for exceeding the threshold.

-- Penalty Factor B controls how gradually the penalty is applied, where the higher the number the later in the run the penalty will be applied.

7. Optionally enter a value for the starting proportion and a random seed.

-- See section 3.2.1.7.1 on starting proportion, and section 3.2.1.7.2 on random seed in the Marxan User Manual for more information.

-- Starting proportion must be a number between 0 and 1 representing the percentage of planning units that will be randomly included in the initial reserve system.

-- Random seed determines whether the same ‘random’ selection of planning units is included in the initial reserve system for each run.

8. If you would like to revise any of the inputs, click Revise input files to return to the Input and Output panel.

9. Click Run Marxan.

Notes

1. The Marxan.exe must be located in the Mods folder under IDRISI's program directory. If the exe file is not found in the folder, an error message will appear stating that the marxan.exe is not found in the directory. The full path of the directory will be shown to the user in this message.

2. Generally if there is a default value in the input boxes, the value may be kept and will not need adjustment.

3. References