MARXAN Conservation Planning. Decision Support Software. T u t o r i a l



Similar documents
INTRODUCTION to ESRI ARCGIS For Visualization, CPSC 178

Modeling Fire Hazard By Monica Pratt, ArcUser Editor

An Introduction to Point Pattern Analysis using CrimeStat

ArcGIS Tutorial: Adding Attribute Data

Spatial Adjustment Tools: The Tutorial

MultiExperiment Viewer Quickstart Guide

Intellect Platform - The Workflow Engine Basic HelpDesk Troubleticket System - A102

Creating and Managing Online Surveys LEVEL 2

University of Arkansas Libraries ArcGIS Desktop Tutorial. Section 4: Preparing Data for Analysis

User s Guide to ArcView 3.3 for Land Use Planners in Puttalam District

Objectives. Raster Data Discrete Classes. Spatial Information in Natural Resources FANR Review the raster data model

Web Ambassador Training on the CMS

How to Download Census Data from American Factfinder and Display it in ArcMap

Government 1009: Advanced Geographical Information Systems Workshop. LAB EXERCISE 3b: Network

GIS Procedural Guide Geocoding / Address Matching in ArcGIS Created by Steve Zuppa, Map Library Assistant - Serge A. Sauer Map Library, 2008

WFP Liberia Country Office

CREATING EXCEL PIVOT TABLES AND PIVOT CHARTS FOR LIBRARY QUESTIONNAIRE RESULTS

Custom Reporting System User Guide

INTRODUCTION TO ARCGIS SOFTWARE

Tutorial 5: Summarizing Tabular Data Florida Case Study

2. Look for a link to ODESI data portal. In the Key Links section in the left side and double click ODESI Data Retrieval.

Query 4. Lesson Objectives 4. Review 5. Smart Query 5. Create a Smart Query 6. Create a Smart Query Definition from an Ad-hoc Query 9

Tutorial Creating a regular grid for point sampling

Tutorial 8 Raster Data Analysis

Working with Data from External Sources

MICROSOFT ACCESS STEP BY STEP GUIDE

Personal Portfolios on Blackboard

Utility Billing Software Manual

MicroStrategy Desktop

Quick Start Guide to. ArcGISSM. Online

Web Intelligence User Guide

Texas Wildfire Risk Assessment Portal (TxWRAP) User Manual. Texas A&M Forest Service

Simply Accounting Intelligence Tips and Tricks Booklet Vol. 1

Getting Census Data into ArcMap or ArcView. Obtaining Shapefiles from ESRI and Data from the Census Bureau

Scribe Online Integration Services (IS) Tutorial

GIS Tools for Land Managers

A Method Using ArcMap to Create a Hydrologically conditioned Digital Elevation Model

Data source, type, and file naming convention

Installing Tri-Global Software

Module A2 Item Activities, Gantt Chart and Utilization Sheet

DPL. Portfolio Manual. Syncopation Software, Inc.

Pharmacy Affairs Branch. Website Database Downloads PUBLIC ACCESS GUIDE

ArcGIS Online. Visualizing Data: Tutorial 3 of 4. Created by: Julianna Kelly

Ohio University Computer Services Center August, 2002 Crystal Reports Introduction Quick Reference Guide

MAS 500 Intelligence Tips and Tricks Booklet Vol. 1

Create a folder on your network drive called DEM. This is where data for the first part of this lesson will be stored.

SAP BusinessObjects Financial Consolidation Web User Guide

Microsoft Access 2010 handout

ModelBuilder - Creating Tools Tutorial

A HYBRID APPROACH FOR AUTOMATED AREA AGGREGATION

1 P a g e. User Guide support.keytime.co.uk

Chapter 6: Data Acquisition Methods, Procedures, and Issues

Tutorial - PEST. Visual MODFLOW Flex. Integrated Conceptual & Numerical Groundwater Modeling

Groundwater Chemistry

How To Hydrologically Condition A Digital Dam

SimplyMap Canada Tutorial

Result Entry by Spreadsheet User Guide

Business Objects. Report Writing - CMS Net and CCS Claims

Converting GIS Datasets into CAD Format

MS Excel Template Building and Mapping for Neat 5

InfiniteInsight 6.5 sp4

Creating Figure Ground Maps in ArcMap 10.x: Basic procedures to download, open, manipulate and print spatial data

Microsoft Access Rollup Procedure for Microsoft Office Click on Blank Database and name it something appropriate.

Liferay Portal User Guide. Joseph Shum Alexander Chow

Content Author's Reference and Cookbook

Personal Geodatabase 101

Note: With v3.2, the DocuSign Fetch application was renamed DocuSign Retrieve.

Assets, Groups & Networks

ICP Data Entry Module Training document. HHC Data Entry Module Training Document

Creating a File Geodatabase

What is FRAGSTATS? Scale Considerations Computer Requirements

Editing Common Polygon Boundary in ArcGIS Desktop 9.x

Wyoming Geographic Information Science Center & Wyoming Game & Fish Department

Microsoft PowerPoint 2008

Supervised Classification workflow in ENVI 4.8 using WorldView-2 imagery

IBM SPSS Statistics 20 Part 4: Chi-Square and ANOVA

Making Visio Diagrams Come Alive with Data

UPS System Capacity Management Configuration Utility

Invoice Quotation and Purchase Orders Maker

VEDATRAK CRM 2.1. User's Guide

Downloading & Using Data from the STORET Warehouse: An Exercise

Generating Open For Business Reports with the BIRT RCP Designer

The Conveyancer and PCLaw Quick Start Guide

USING THE UPSTREAM-CONNECT WEBSITE

Using Microsoft Access Databases

User Guide. Analytics Desktop Document Number:

Skills Funding Agency

Tips and Tricks SAGE ACCPAC INTELLIGENCE

SAP Business Intelligence (BI) Reporting Training for MM. General Navigation. Rick Heckman PASSHE 1/31/2012

Raster to Vector Conversion for Overlay Analysis

University of Arkansas Libraries ArcGIS Desktop Tutorial. Section 5: Analyzing Spatial Data. Buffering Features:

Studying Topography, Orographic Rainfall, and Ecosystems (STORE)

Intellicus Enterprise Reporting and BI Platform

Spatial Analyst Tutorial

Publishing Geoprocessing Services Tutorial

Downloading SSURGO Soil Data from Internet

Using Adobe Dreamweaver CS4 (10.0)

Abstract. For notes detailing the changes in each release, see the MySQL for Excel Release Notes. For legal information, see the Legal Notices.

Contouring and Advanced Visualization

Creating Online Surveys with Qualtrics Survey Tool

Transcription:

MARXAN Conservation Planning Decision Support Software Ian Ball & Hugh Possingham T u t o r i a l For GIS Experts Annette E Huggins

MARXAN Conservation Planning Software Structure of the Tutorial The introduction to the tutorial provides information regarding how marxan works, the design of planning units and the preparation of biodiversity distribution data to be used in the conservation planning analysis. Exercise 1 provides an introduction to the marxan software program. Section A describes the steps necessary to install marxan and examine the tutorial data. Section B allows the user to become familiar with running marxan using pre-prepared data tables and section C illustrates viewing the results using ArcView GIS software. Exercise 2 guides the user through methodologies in ArcView GIS and Microsoft Excel that can be used to create the datatables for running marxan. Introduction MARXAN is a software program that provides decision support to teams of conservation planners and local experts identifying efficient portfolios of planning areas that combine to satisfy a number of ecological, social and economic goals. It is readily available via the Internet at no cost (Ian Ball and Hugh Possingham, http://www.ecology.uq.edu.au/?page=20882&pid=). It is a stand-alone program that requires no other software to run, although a GIS is required to prepare the data, make the input files and to view the results. It is designed to help automate the planning process so that a team of planners can offer many different conservation plan scenarios. It can be used to offer planning scenarios that are alternatives to pre-conceived patterns of reserve or conservation area networks. It can also be used to offer alternatives and solutions where the input of local stakeholders is highly valued and a compromise with prospects for achievable results is sought. MARXAN offers decision support for teams choosing between hundreds of biodiversity targets and thousands of candidate areas (planning units). Using a transparent process and driven by quantitative goals, the analysis is repeatable and objective. A pattern of priority sites that satisfy explicit quantitative biodiversity goals can be identified that are of low political or social pressure, or where resources necessary to implement conservation strategies or threat abatement are forecast to be lower. Planning units are parts of the land and seascape that are analyzed as the potential building blocks of an expanded system of reserves or areas of conservation priority. They allow a comparison between candidate areas. Planning units can be natural, administrative or arbitrary sub divisions of the landscape. They differ widely in size between studies and within regions, dependant mostly on scale of analysis and data resolution. Marxan Algorithm MARXAN implements a simulated annealing site optimization algorithm, developed by Dr Ian Ball 1 and Professor Hugh Possingham 2. In order to design an optimal reserve network, each planning unit is examined for the values it contains. The features (biodiversity targets and a measure of cost, threat or opportunity) within one planning unit may be valuable alone but may not be the best choice overall, depending on the distribution and replication of those features in the wider area. The algorithm attempts to minimize portfolio total cost whilst meeting conservation goals in a spatially compact network of sites. This set of objectives constitutes the objective cost function : Total Cost = Planning Unit Cost + Species Penalties + Boundary Length 1 Australian Arctic Division, Environment Australia, Tasmania 2 Director of The Ecology Centre, Departments of Zoology and Mathematics, The University of Queensland The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 1

where Total Cost is the objective (to be minimized), Planning Unit Cost is a cost assigned to each planning unit, Species Penalties are costs imposed for failing to meet biodiversity representation goals, and Boundary Length is a cost determined by the total outer boundary length of the planning units in the portfolio. During the run of the algorithm, an initial portfolio of planning units is selected and the total cost calculated. Planning units are then added and removed and the total cost re-evaluated through multiple iterations in an attempt to improve the total cost and efficiency of the portfolio. Attempts are made to minimize the total portfolio cost by selecting the fewest planning units with the lowest total unit cost needed to meet all biodiversity goals, and by selecting planning units that are clustered together rather than dispersed (thus reducing outer boundary length). Early in the procedure, changes in the portfolio that do not improve efficiency can be made in order to allow the possibility of finding a more efficient overall portfolio. The requirement to accept only those changes that improve efficiency becomes stricter as the algorithm progresses through a set of iterations. Many runs of the algorithm are used to find the most efficient portfolio and to calculate a measure of irreplaceability. MARXAN irreplaceability is the number of times a particular unit is chosen and offers a measure of the flexibility of including of each planning unit in a portfolio. Alternative scenarios can be evaluated by varying the inputs to the total cost function. The impact of the boundary length cost factor, for example, can be increased or decreased depending on the assumed importance of a spatially cohesive portfolio of sites using a boundary length multiplier (BLM). The relative importance of each biodiversity target can be set using the species penalty factors (SPF). A high value will ensure that the goal is met for that target. The planning unit cost can be any of a number of measures and may be used to introduce other non-biological factors into the analysis such as threat or suitability for conservation strategies, or can be a surrogate for actual cost, such as area. A cost threshold penalty can be set and applied if the total cost goal is exceeded. A minimum size for patches of target can be set to ensure that particular targets meet their goal with patches of large enough size to be thought of as viable, if the size is larger than that contained in one planning unit. Special interest areas can be locked into the portfolio before the algorithm is run. There is often many more candidate planning units available than needed to meet representation goals. Some locations will be irreplaceable, but for the majority there is some flexibility in which areas are selected to represent each target. Planning units can be removed from the portfolio and units that will provide a similar contribution to the goals can be identified. This aspect can be very useful in a forum where negotiation over which areas are to be selected is occurring, as the consequences of choosing or not choosing a particular candidate area can be explored. Designing Planning Units (from Pressey & Logan, 1998 3 ) Planning units are areas for which data on occurrence, frequency and extent of the targets exists. When performed at the local scale, planning unit groups identified by the analysis allow tentative identification of management sized parts of the land and seascape for later adjustment to the actual boundaries of reserves or use-zones, although are sometimes intended to serve as the units of management themselves. The final portfolio can consist of an array of clustered planning units, with others serving to connect isolated areas of existing or intended conservation management. Planning unit boundaries can be dictated by arbitrary cells such as squares or hexagons, by units of ownership or tenure, forestry management compartments or natural subdivisions such as watersheds. The area of distribution, length or number of occurrences of targets within each planning unit is calculated using a GIS and used to create MARXAN input files. The choice of planning units has important implications for the process of portfolio selection as well as implementation of its results. The choice of planning unit size and configuration for both wide and local scale analysis must be made with many factors in mind. These include: 3 Pressey, R.L. & Logan, V.S (1998) Size of selection units for future reserves and its influence on actual vs. targeted representation of features: a case study in Western New south Wales. Biological Conservation, 85, 305-319. The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 2

The size of the planning unit relative to the scale of the underlying features (e.g. planning units that are much larger than underlying fragments of vegetation can mask the size, shape and extent of fragmentation; planning units that are very small relative to the vegetation types will mostly be homogeneous, i.e. small and large patches of habitat will be indistinguishable). The number of planning units that can be handled by the analysis computer in a time that is reasonable for the intended process (e.g. calculation of target area within planning units, clustering tests etc. A limit of 20,000 is recommended although 40,000 have been used, but incurs an extended running time). The size of planning units in relation to the reliability of mapping (e.g. larger planning units could be needed where the locations of the targets to be represented are imprecise or where the boundaries of planning units are known to be inaccurate). The ability of regular grids or hexagons to show per unit area values for criteria such as richness of unprotected targets. Equality of the sizes of planning units over large geographic areas when factors such as map projections are an issue. Convenience of conversion of planning units to management units on the ground when analyzing at the fine scale. Appropriateness of boundaries for conservation management when analyzing at the fine scale. Whether the boundaries of some units are likely to change, for example as tenure parcels are exchanged and amalgamated. Size and shape related consideration such as edge effects, viability of populations and management overheads in the resulting reserves. Considerations for public presentation such as the potential sensitivity of mapping parcels of private tenure rather than arbitrary grid cells that do not identify specific holdings. Data Preparation and Considerations Screening: All target distributions to be entered into the analysis should be considered viable occurrences that are robust enough to influence the portfolio selection. Screening biodiversity distribution maps should be considered to improve the likelihood of only including viable occurrences. Methods previously used include screening vegetation by patch size, removing all patches below a threshold size, and screening by a threat surface such as a freshwater flow accumulation model. Stratification: Targets can be stratified to allow a geographic spread of representation or to represent biologically distinct target sub-groups. The portfolio will be greatly influenced by the way the targets are defined. For example, if the portfolio is required to hold both upland and lowland portions of a specific target, it must be defined in a way that includes this information. Examples include stratification by elevation, or by other geographic units that define biologically meaningful differences. Analysis extent: The areas of interest should be defined when the extent of all target and socio-economic data have been gathered and assessed. The extent should be defined with a shapefile rectangle or other polygon so that all targets can be clipped to the analysis area. Planning Unit Cost Range: The range of cost values must allow the desired influence on the choice of planning units. This must be considered with relation to boundary length, as the objective function is a combination of boundary length, boundary length modifier and cost. The effect of the BLM will also be affected by changes in cost. If a range of costs is to be used, the BLM value should be tested with the costs in place. Flexible / Low Irreplaceability Units: Areas that are not being indicated as highly irreplaceable are not unimportant for conservation; they are more flexible-- as there are other planning units that contain similar biodiversity. A highly irreplaceable area may not be known locally for its biodiversity value, but may, for example, contain 100% of the occurrences of a particular target and is therefore irreplaceable for meeting the representation goal for that target. Marxan is also useful for highlighting areas that have previously been overlooked- for example The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 3

where goals for many targets can be met in a spatially cohesive manor that increases the likelihood of strategy effectiveness Unexpected Results: Targets, goals and cost surface are all crucial to creating the best portfolios. If areas that are known to be important are not being reflected in the portfolio it may be because the targets or the cost surface are not defined in a way that distinguishes those areas as different from others. For example if no highland targets are represented in the portfolios it may be because the targets have not been stratified by elevation so those highland parts cannot be distinguished as needing representation separate from the lowland areas. The goal representation for that target will be met but it could be all within lowland areas. Exercise 1: An introduction to MARXAN This exercise describes how to use marxan to design a portfolio using case study data. The first section (a) examines the target distribution and planning unit shapefiles in ArcView and the pre-prepared marxan input files. The second section describes setting up a marxan run using input files that have been prepared from this data. Section A: Programs and Data A1) Copy the file TNC_marxan_tutorial.zip onto your computer and unzip it onto the root hard drive (straight onto C or D hard drive). A folder called marxan will unzip, and it should contain the marxan program and three folders named inputs, inputs_2, outputs and TNC_tutorial_data. The inputs folder should contain six dat files, the inputs_2 and the outputs folder should be empty. The TNC_tutorial_data folder should contain six further folders, the pu folder contains the planning unit shapefile, the targets folder contains biodiversity target shapefiles, the scripts folder contains several scripts, the pa folder contains the protected area shapefile, the land folder contains the coastal outline shapefile and the tables folder should be empty. Marxan must always be installed with a very short path name eg c:/marxan. Do not place it within a long pathname such as within the my documents folder. A2) Although MARXAN is included with the tutorial data it can be downloaded for free from The Ecology Centre at the University of Queensland: http://www.ecology.uq.edu.au/?page=20882&pid=. If the program is to be used in future, you should register as a user. A3) ArcView should already be installed on your computer. See http://www.esri.com/software/arcview/index.html for information about ArcView, and http://www.conservationgis.org/aaesrigrants.html for information concerning ESRI s conservation program, through which ArcView may be made available to conservation organizations. A4) Open ArcView GIS and open a new View. A5) Save the project as TNC_exercise_1.apr in the TNC_tutorial_data folder. A6) Open the target shapefiles called tnc_marine_targets.shp and tnc_terr_targets.shp from the c:\marxan\tnc_tutorial_data\targets folder and the planning units shapefile tnc_mxn_pu.shp from c:\marxan\tnc_tutorial_data\pu. They should display using the default legend files. The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 4

A7) These target distribution and planning unit files have been used to make marxan input files that are found in c:/marxan/inputs. Marxan uses up to 5 files that describe the planning units, the distributions of the conservation targets, the conservation goals for each target and the spatial arrangement of the planning units. A8) Open these text files files found in c:/marxan/inputs in a text editor such as notepad to examine them. Marxan uses a textfile called input.dat to identify the location of the input files and other running parameters. The program InEdit is used to create the input.dat file. The following section of the exercise will demonstrate how to set up marxan using these input files (creating the input files will be covered in exercise 2). Section B Using Inedit to set up MARXAN B1) Open InEdit by double clicking on InEdit.exe in c:\marxan. B2) The first screen is to set the problem. The options consist of the number of runs to be performed and the level of clustering desired. Type 100 runs in the first Miscellaneous Repeat Runs box. 200 runs are often used. 100 runs are used here due to time constraints. Type 0 into the Boundary Modifier box as shown here. The type of input file to be used is the new freeform style. B3) Click onto the Run Options tab positioned second in line at the top of the inedit screen. You should see options for the type of algorithm to be run by MARXAN. Click the box for simulated annealing and for iterative improvement as shown in the graphic below. Click on the save button at the bottom of the inedit screen. B4) Click on the Annealing tab positioned third from the left at the top of the inedit screen. The default value for the annealing controls should be used. These are 1000000 iterations of the algorithm with 10000 temperature decreases, used with adaptive annealing as shown below. The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 5

B5) Click on the input tab of the inedit screen. The input folder must be set first before navigating to the input files. Click on the input directory browse button towards the bottom of the screen and navigate to the folder c: \marxan\inputs. The input.dat files can then be specified by clicking on the browse button next to each input file name and navigating to the corresponding file. The necessary files are as follows; species file name = spec_goals.dat, planning unit file name = pu.dat and the planning unit versus species = puvspr_abun.dat. Un-check the block definitions file locator under the optional input files to indicate that block definitions are not being used in this example run. Click on the box to check the boundary length file option and browse to the boundary.dat file. Click on the save button to save the information in the input.dat file. B6) Click on the output tab of the inedit screen. This screen allows the user to specify the output files required to analyze the results. General screen progress gives a good idea how the algorithm is running. Click on the output files shown in the graphic below. Saving each run has minimal use and produces large amount of text files. The most important files are the overall best, the summary and the summed solution. Write 1 in the species missing if proportion of target lower than box. This option is to allow marxan to consider a target to have met its goal if it is very close. This can be useful if the goals are high and the cost to meet the last small part of the goal would be very expensive. The text within the box save file name will prefix all the output text files. Copy the text tnc_tutorial_run1 into this box. This text serves to remind the user of the specific run that the output files refer to. Browse to the outputs folder c:\marxan\outputs. This is the folder where the output files will be written by marxan. Click the save button to save the information to the input.dat file. B7) Click on the cost threshold tab on the inedit screen. This screen should be left with the default values and the threshold enabled box unchecked as shown. The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 6

B8) Click on the Misc tab on the inedit screen. This tab allows the user to specify a starting proportion of the planning units to be in the random starting portfolio. Type 0 into this box, and uncheck the specify random seed box. The clumping rule or the best score speed up option will not be in use, so can be left unchecked and as shown in the graphic. Click on the save button to update and save the input.dat file. Exit the inedit program by clicking on the exit button. B9) Double click on marxan.exe from within windows explorer to start marxan. A screen will appear to show the progress of the algorithm (shown in white here, but will appear as white writing on a black background). The program should run for 10 minutes or less. Error messages showing warnings about blockdefname and highdata are normal. For the tutorial there should be 3904 planning units, 39 species (targets), 11823 boundaries and 8828 conservation features read by the algorithm. The information screen will then print any further information. This can help identify any problems with the input files. A message will appear here if one or more targets are already adequately represented in a portfolio (e.g. if locking in any planning units at the beginning of the run), or if there are problems with the formatting of the dat files causing them to be unreadable. If this occurs, check the formatting, including the placing of commas, check all field headings are lowercase (no capital letters can appear in the headings). When it has finished running, close the window. Marxan writes the results to text files that can be found in the c:\marxan\outputs folder. They can be viewed in notepad. ArcView can be used to view maps of the results. B10) Examine the outputs: open tnc_tutorial_run1_mvbest.txt to check that each target met its goal (called target by marxan) in the best run. A high SPF value (species penalty factors- penalties for not representing targets to the set goals) for all targets will raise the relative importance of the target representation in comparison to overall portfolio cost (a combination of boundary length and modifier, cost per planning unit and SPF), thereby forcing the representation goals to be met. Also check tnc_tutorial_run1_sum.txt to view the score of each run. Section C: Mapping MARXAN Results There are seven possible types of file output. Those most often used are:, missing value information (x_mvbest.txt), best solutions of all runs (x_best.txt), summary information (x_sum.txt), scenario details (x_sen.txt), summed solutions over all runs (x_ssoln.txt). Screen log file, solutions for each run and snapshot files can also be requested. The solutions are in text file format and can be imported into excel, or as seen, viewed in a text editor. Summed runs and best run can be imported into ArcView and joined to a shapefile of the planning units to enable viewing of the portfolio and irreplaceability maps. Maps of the summed runs or irreplaceability (number or times each planning unit was chosen in a set of runs of the algorithm) can be extremely useful for conservation planning, allowing a guide for scheduling. Planning units that The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 7

have a lower irreplaceability score can be very important for biodiversity, but are flexible with other similar units in the analysis. These units are necessary for the goals to be met, but there is a choice about which ones are included. C1) Return to ArcView to display the irreplaceability of the planning units and the best portfolio. Open View 1 that contains the planning units theme. Open the pu attribute table. C2) Add the tnc_tutorial_run1_ssoln.txt text file from c:\marxan\outputs to the project as a table. This file contains information on the number of times each planning unit was chosen during the algorithm set of runs and is an indicator of irreplaceability. Join the text file to the planning units attribute table using the field heading planning_unit in the text file, and the field heading unit_id in the planning units shapefile attribute table. Make a new field in the planning unit attribute table named run1_irr (to indicate irreplaceability scores of run 1) and copy the values of the number field into the new field. Save the edits to the table and stop editing the table. The results text table must be removed from the attribute table before joining the next table (table/remove all joins). C3) Display the planning units shapefile using graduated red to yellow color with up to 64 classes according to the new run1_irr field. The map should look similar but not identical to the one shown above. Those units with dark red coloring have a higher irreplaceability score. This indicates that they are necessary to represent the targets to the goals efficiently. Those planning units with a lower irreplaceability score are not less important, but more flexible. Those areas that are highly irreplaceable may contain targets that have restricted distributions and therefore must be included to meet the representation goals. This can aid scheduling of conservation strategies within the portfolio. C4) To display the best portfolio, the results text file tnc_tutorial_run1_best.txt should be joined to the planning units shapefile and the values copied across from the solution field in the same way. The new field should be called run1_bst. Save the edits and remove the joined table. Display the planning units using unique value and the field run1_bst. It should have a portfolio of sites in a similar arrangement to those shown below. The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 8

The portfolio represents all the targets to a 30% goal. C5) Run the algorithm again, using the same parameters except with a boundary length modifier of 0.06 and view the results in ArcView. This will encourage the clustering of planning units in the portfolio as the algorithm tries to minimize total cost, which includes the outer boundary length. To assess the appropriate value of the boundary length modifier, several values should be tested, as the magnitude of its effect will change with the flexibility of different data. The boundary length modifier, the cost of the planning units and the spf values all effect the outcome and should be balanced. C6) Run the algorithm again with the protected areas locked in by using the planning unit file pu_pa_lockin.dat. View the first steps of the marxan run to see how many species are already represented by the portfolio of locked in pu s. View the results in ArcView and assess the effect of locking these areas in. Find the number of units necessary to meet all goals when all PA units are locked in and compare with the number necessary without them locked in. Are the existing PAs efficient? C7) Try running the algorithm for 200 runs. This may provide a more efficient best portfolio (check the output file tnc_tutorial_run1_sum.txt). If it does not, 100 runs may be sufficient, although more runs have the advantage of providing a greater spread of values of irreplaceability. Exercise 2: Creating the input files for MARXAN using tutorial data. This exercise describes the GIS and excel procedures that can be used to create the files that describe the biodiversity and cost data for use in marxan. Two types of scripts can be used to extract the data in a GIS. Those to be described here are for use in ArcView with shapefiles. The second type are amls for use on coverages within ArcInfo. A description of these can be found within the TNC document MARXAN file preparation using ArcInfo. ArcView scripts that run on shapefiles take a little longer in processing, but the software is more readily available to conservation organizations (ArcView may be granted through ESRI s conservation program). The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 9

The order that the tables are created is important, as the methods to create the tables later in the list may rely on processes performed earlier. It is possible for several targets to be summarized separately and the tables joined together in a text editor. D Planning Unit versus Conservation Feature Table This file contains the information on the distribution of targets across the planning units. It has a default name of puvspr.dat, but any name can be used. puvspr_abun.dat will be used in this exercise. Vector shapefiles D1) The headings of the table are as follows (in bold). species pu amount Conservation target id must be a number Planning unit id Area of conservation target. The values are separated by commas. The file structure looks similar to this example: species,pu,amount 26,263,535739.34 26,271,228479.37 etc D2) The CLUZ abundance script will be used to calculate the area of distribution of all targets within each planning unit and build the table. This script can also be used to calculate the length of linear targets and the number of occurrences if target distributions are mapped in this way. Several shapefiles can be processed at once, providing they are loaded into in the view. Grids must be processed using the summarize zones tool in ArcView (see below). D3) The script requires the biodiversity distribution shapefile attribute table to contain a column containing target ids. The id must be numeric, under 6 figures, and unique for each target. The heading must be id. The tutorial data have been prepared in this way. The script requires an empty abundance table, containing only the planning unit id s in a column. This can be created by exporting the planning unit attribute table as a dbf and removing all columns except the planning unit id in excel or after re-adding to the ArcView project. D4) Open a new empty script. Open the CLUZ_abundance script (c:/marxan/tnc_tutorial_data/scripts) as a text file. Compile and run the script with the view active. D5) The script will ask which file is the planning unit file and which target files should be summarized. Choose both the marine target and the terrestrial target files. The script will ask for the location of the empty abundance table (tgt_abun.dbf) that was created in D2 in c:/marxan/tnc_tutorial_data/tables. It will also ask whether you would like the area to be divided by a number choose yes and type 10000 to convert meters squared to Ha. The script intersects the target files by the planning units, thereby cutting them and assigning each polygon portion the id of the planning unit it falls within. The area of each target within each planning unit is then calculated and processed into the abundance table. This can take considerable time depending on the type and size of target and planning unit files but is reduced by using a target shapefile that has been dissolved on the target id such as the tutorial target shapefiles. D6) When the script has finished, you will be able to load the abundance table into ArcView (tables/add) and view the abundances. If you need to look at a table for a second time, add it again to the project window, rather than viewing it straight from the project window, as it may appear distorted. D7) The abundance table is converted to the format required by marxan using the CLUZ_puvspr script. Open a new empty script box, and open the CLUZ_puvspr script from C;/marxan/TNC_tutorial_data/scripts as a text file. Compile and run the script with the abundance dbf table active. Create the abundance file in marxan/inputs_2. N.B. Grids can be summarized using the summarize zones tool in ArcView. Each target must be coded with a unique id using the gridcode. Load the grid and the planning unit shapefile into a view and choose The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 10

analysis/summarize zones. Choose the planning unit id to define the zones. The resulting table must be opened in excel, the unnecessary columns deleted, and the unit id and mean or sum values saved as a csv and then the extension changed to dat in windows explorer, and the table joined to any other vector or grid target summaries in a text editor. E Species File spec_goals.dat This file holds information about each target including the goals and names. Only id, type and target are essential, all other variables are optional, although a high (e.g. 10000) spf is highly recommended. If a column is missing, the default values will be used. For some columns, a value of 1 indicates either that the default is to be used or that value is given in the block definition file. The name column can contain spaces or other word separators, but any separator will be replaced by a single space. If there are any duplicate definitions, all but the last one will be ignored. Table format id type target spf target2 sepdistance Id of target-must Looks for block Goal Species penalty Minimum clump Minimum correspond to definitions representation of factor for each size optional separation puvspr_abun file the target target distance optional cont.. sepnum name targetocc Target number of mutually separated P.U.s in valid clump optional Name in words can include spaces all words must start with a letter optional Number of occurrences of the target required. optional The file structure looks similar to this example: id, type, target, spf, target2, sepdistance, sepnum, name, targetocc 334, 334, 8776898876.56, 10000, 443, 1000, 2, limestone, 0 E1) The file can be created in excel from ArcView tables. Open the marine target shapefile attribute table. Export the table as a dbf file called mar_target_id.dbf in c:/marxan/tnc_tutorial_data/tables. Repeat for the terrestrial target table and save as terr_target_id.dbf. E2) Open both tables in excel and delete all columns except the id and name columns. Rename the column headings in lower case text. Copy the terrestrial target ids into the marine target table in the same column under the marine ids. Save the table as c:/marxan/tnc_tutorial_data/tables/spec_goals.xls. E3) Insert three empty columns between the id and name columns (insert/columns). Type the headings type, target and spf in the top cell of these columns. Fill the type column with the value 1 to indicate that types are not being used. This option can be used if targets are to be grouped by type. The groups would then receive the same variables in the block definition file. This can also be a convenient way to define proportional goals in the block definition file, as only the proportion need be defined rather than the actual area, length or number of occurrences that are necessary in this (spec_goals) file. If no grouping of targets is required, but the proportion option is to be taken advantage of, the types can be the same identifying number as the id, and this used in the block definition file. For the purposes of this tutorial, types are not being used; the goals will be calculated directly. E4) To calculate a 30% goal for all targets, open the abundance dbf table created in section A1.4 (c:/marxan/ /inputs/puvspr_abun.dat) in Excel. Sum each column into a cell underneath the column to calculate the total area of each target. Calculate 30% of each target in a row below the total. The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 11

E5) Transform this row of figures into a column (paste special) and added to spec_goals.xls that contains the target ids. It is recommended that checks be made throughout the process. If other, non-proportional goals were being used, these areas and/or lengths could be entered into the table directly. E6) Fill the spf column with 10000. This value should ensure that all targets meet their goals. Save the xls file. The table should have the same values as the diagram opposite. E7) Save as a csv file called spec_goals.csv in marxan/inputs_2, then change the extension in windows explorer to.dat so that the file is called spec_goals.dat. F Planning Unit File pu.dat The default name for this file is pu.dat and it is a necessary file. This file contains all the information related to planning units except for the distribution of targets. The column headers can include: id, cost, status, xloc and yloc. The id column is the only one that is not optional. The cost and status will assume a default value of 1 and 0 respectively if the columns are not present. The xloc and yloc columns are critical if there are spatial separation requirements for any of the targets. These values represent the location of the planning units, and are usually the centroid of the polygon. Table format: id cost status xloc yloc Critical Of each pu Whether pu is locked in or out of the system Critical for Separation Critical for Separation and so takes this form: id, cost, status 1, 2.3, 0 2, 2.3, 0 etc The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 12

F1) In ArcView, open the attribute table of the planning units shapefile. Export this table to a dbf file called pu.dbf in marxan/tables. Open in excel and remove all other columns except id. Type cost in the top cell of the next column and status in the third column, making sure all column headings are lowercase. F2) If a measurement of cost, opportunity, threat or other cost surface is available, these values can be summarized by planning unit using the abundance script and inserted into the cost column. A cost surface is not being used here; the cost of each unit in this exercise will be equivalent to the area of each unit. Enter the value 260, which is the area in hectares. F3) The status of each planning unit can take one of 4 values: Status 0 1 2 3 Meaning The PU is not guaranteed to be in the initial or seed reserve. However it still may be. It s chance of being included in the initial reserve is exactly the starting proportion from the parameter input file. The PU will be included in the seed reserve or the initial reserve. It may or may not be in the final reserve. The PU is fixed in the reserve. It starts in the initial reserve and cannot be removed. The PU is fixed outside of the reserve. It is not included in the initial reserve and cannot be added. The planning units in this exercise will have an identifier of 0 to indicate that they not locked into or out of the portfolio or the seed reserve. Fill the status column with a value of 0. F4) Save the table as an xls file in case of dbf errors and then as a csv file named pu.csv in marxan/inputs_2, and then change the extension of the file in windows explorer to.dat so that the file is called pu.dat. F5) To make a pu.dat that will lock in all protected area planning units, the status identifier of 2 for all planning units that are over 50% protected area. Usually a planning unit is considered within a PA only if over 50% of its area is within a PA polygon. To change the status of all those units with over 50% of their area within a protected area, the area of PA within each unit must first be calculated. This can be done in the same way as summarizing target data (section D). The protected area vector polygon shapefile is treated as a target, with one id code for all PAs. Add the protected area shapefile tnc_mxn_pa.shp to the view from TNC_tutorial_data/pa. The PAs have an id of 999. Create a dbf table containing the unit ids (name it pa_area.dbf) and summarize the area of PA polygons within the planning unit hexagons to it (section D2). Join this table it to the planning unit shapefile attribute table using the unit id field as the join field. Create a new column in the attribute table (named prop_pa ) and copy the area of PA column (value_999 ) into the new column so that it is permanently in the table rather than a virtual join. The join can then be removed. Query the table to select all records that have a PA amount over 50% of the area of the units and assign those planning units a status of 2. The other units should keep their value of 0. Export the table to a dbf called pu_pa_lockin.dbf in marxan/inputs_2, open the table in Excel and complete the table, as described above in the first steps of this F section. G Boundary File boundary.dat The boundary length file contains information on the boundary costs of adjacent planning units. Whereas this cost is typically the actual length of the boundary it can be modified to a cost or effective length value to take into account boundaries that are particularly desirable or undesirable. Boundary.dat can be created automatically using the CLUZ boundary extension with the planning unit file. This table can have tabs or commas between the columns. The extension creates a table with tabs between columns. The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 13

boundary id1 id2 Critical- the boundary Critical- planning unit id Critical -neighboring planning unit id or the same as id1 for length an irremovable boundary Form: boundary id1 id2 33453.5 1 2 334536.2 1 3 G1) Copy the boundary extension from TNC_tutorial_data/scripts into the ArcView extension folder, which resides within the ArcView program folder. By default this is usually c:/esri/av_gis/arcview/ext32. In ArcView, load the extension from File/extensions and click beside the name boundary file maker. The extension appears as an icon with a red B. G2) Make the view with the planning units file active and run the extension from the B icon. Follow the instructions on the screen to make boundary.dat in the marxan/inputs_2 folder. The marxan manual gives further information about moveable and irremovable boundaries, and the way boundaries can be used to identify two units are linked although no actual boundary cost exists. H Optional block definition file block.dat This optional file can be used to group targets together or to use the facility that allows the goal to be set as a proportion without calculating the actual amount in area, length or number of occurrences. When using this file, the type of each target must be set in the spec_goals file. If no grouping is necessary, the type names can remain the same as the target id. Table format: type target target2 targetocc sepnum sepdistance prop spf Critical-the type for which the other attributes are defined The goal for the target of the given type. Minimum clump size. If a clump of a number of planning units with the given target is below this size then it does not count toward the goal. The number of occurrences of the target required. This can be used in conjunction with or instead of target. Target number of mutually separated planning units in valid clumps. Minimum distance at which planning units holding the target are considered to be separated. An alternative to target. This is the proportion of the total amount of the target which must be preserved. The penalty factor for that target. H1) Copy the spec_goals file to a new file called spec_goals_type, open in excel and populate the type column with the same values as the id column. Fill the target column with a value of 1 which will indicate that the goal (called target by marxan) can be found in the block definition file. Save the file as spec_goals_type.csv in inputs_2 and rename to.dat as before. H2) Copy the spec_goals file again and name it block.csv and open in Excel. Remove the unnecessary columns and add the columns target, target2, targetocc, sepnum, sepdistance, prop and spf. Fill the prop column with 0.3 indicating a 30% goal and fill the spf column with a value of 10000. The other columns should have a value of 1 to The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 14

indicate they are not in use. Save the file as a csv file named block.csv and change the extension to.dat in windows explorer as before. Exercise 3: Running Marxan using new files J1) The tables created can now be used to run marxan. Follow exercise 1 section B to set up a series of marxan runs using the tables just created with the same parameters and a different output name, and save to marxan/outputs. If marxan fails and closes, it is most often because there is a comma out of place, or a capital letter in a column heading etc. Check all dat files in a text editor. The initial text in the marxan program screen gives useful information about how many targets (called species by marxan) have already or cannot met their goals (called targets by marxan). Check that this is correct for the analysis if some species have already met their target or cannot meet their target it may be because the species occurs in the spec_goals but not the abun_puvspr file or vice versa. Check the number of targets and the number of planning units reported here to make sure they are correct. J2) When marxan has finished running, check tnc_tutorial_run1_mvbest.txt to make sure that all targets met their goals in the best portfolio (you will recall goals are called targets by marxan). Then view the results using ArcView using exercise 1 steps C as a guide, checking that they are similar to the previous results. J3) Re-run the analysis using the spec_goals_type file and the block definition file made in section H to check they work (they will run the same analysis as the goals and all other parameters are the same). J4) Re-run the analysis with the protected areas locked into the portfolio by using the planning units text file pu_pa_lockin.dat in inputs_2 using exercise 1 steps B as a guide. J5) Re-run the analysis with the protected areas locked in and the BLM that gave the best results in the previous exercise.. These steps can now be followed using your own local data. When running an analysis consider the points on page 3 and 4 and assess the appropriateness of target mapping and stratification, cost surface values and range of values in addition to BLM values. The Nature Conservancy MARXAN Conservation Planning Decision Support Software Tutorial 15