Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Evaluating Marxan as a terrestrial conservation planning tool Munro, Krista Grace 2006

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2006-0270.pdf [ 6.97MB ]
Metadata
JSON: 831-1.0092583.json
JSON-LD: 831-1.0092583-ld.json
RDF/XML (Pretty): 831-1.0092583-rdf.xml
RDF/JSON: 831-1.0092583-rdf.json
Turtle: 831-1.0092583-turtle.txt
N-Triples: 831-1.0092583-rdf-ntriples.txt
Original Record: 831-1.0092583-source.json
Full Text
831-1.0092583-fulltext.txt
Citation
831-1.0092583.ris

Full Text

E V A L U A T I N G M A R X A N A S A T E R R E S T R I A L C O N S E R V A T I O N P L A N N I N G T O O L by KRISTA GRACE MUNRO B.A., Queen's University, 1994 B.Ed., Dalhousie University, 1995 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES (Planning) THE UNIVERSITY OF BRITISH COLUMBIA April 2006 © Krista Grace Munro, 2006 A B S T R A C T A variety of reserve design software programs are available to assist in the selection and spatial configuration of new protected areas. One such application, Marxan, produces spatially cohesive reserve configurations which meet representation targets efficiently. Conservation agencies worldwide are adopting Marxan as a conservation planning tool, however it is not currently used by Parks Canada when conducting feasibility studies for potential national park reserves. This thesis evaluates whether Marxan could be a useful decision-support tool for Parks Canada to use when selecting and designing potential park areas. The assessment is based on four usability criteria and three park selection criteria, developed in consultation with Parks Canada. Concurrent with this thesis, Parks Canada is conducting a feasibility study for a national park reserve in the South Okanagan-Lower Similkameen region of British Columbia, Canada. This region serves as a case study for the thesis. Marxan was used to create 36 unique reserve configuration options for the case study area and to help evaluate the performance of each reserve. Overall, Marxan fully satisfied three criteria, partially satisfied three, and failed to meet one. This study demonstrates that Marxan provides a useful means to design and explore a range of representative and scientifically defensible reserves. However, to use it effectively requires technical and ecological expertise, a comprehensive GIS infrastructure, good data and time. This analysis concludes that Marxan would be a very appropriate tool to assist Parks Canada in selecting and designing potential national park reserves. Marxan would be best used in conjunction with other decision-support tools, expert knowledge and public consultation. T A B L E O F C O N T E N T S A B S T R A C T II T A B L E O F C O N T E N T S Ill LIST O F T A B L E S V LIST O F F I G U R E S VI G L O S S A R Y VI M A R X A N G L O S S A R Y VIII A C K N O W L E D G E M E N T S X 1. INTRODUCTION I I.I OVERVIEW I 1.2. A BIOLOGICALLY DIVERSE & SENSITIVE ECOSYSTEM 2 1.3. INSTITUTIONAL C O N T E X T 4 1.4 A BRIEF HISTORY OF CONSERVATION PLANNING 5 1.5 EFFICIENT RESERVE DESIGN 5 1.6 KEY ATTRIBUTES OF G O O D RESERVE DESIGN T O O L S 8 1.7 EVALUATING M A R X A N 9 2. METHODS 12 2.1 INTRODUCTION 12 2.2 CONSERVATION G O A L S A N D ECOLOGICAL OBJECTIVES 12 2.3 CONSERVATION FEATURES, TARGETS A N D PENALTIES 12 2.4 COARSE-FILTER REPRESENTATION 14 2.5 FINE-FILTER REPRESENTATION 16 2.6 PREPARING M A R X A N 18 2.7 RUNNING M A R X A N 22 2.8 EVALUATION CRITERIA 29 3. RESULTS 30 3.1 USABILITY CRITERIA 30 3.2 PARK SELECTION CRITERIA 39 4. D I S C U S S I O N 47 4.1 M A R X A N ' S O V E R A L L PERFORMANCE 4 7 4.2 USABILITY - PROBLEMS A N D SOLUTIONS 4 7 4.3 FLEXIBILITY — BENEFITS A N D D R A W B A C K S 4 8 4.4 SENSITIVITY ANALYSIS 4 8 4.5 REPRESENTATION ASSESSMENT 4 9 4.6 E C O L O G I C A L INTEGRITY 5 0 4.7 ANALYSIS OF RESULTS 5 2 4.8 N E X T STEPS 5 4 4.9 O P E R A T I O N A L LIMITATIONS 55 4 .10 A D V A N T A G E S OF M A R X A N 56 4.11 R E C O M M E N D A T I O N S 5 7 4 . 1 2 C O N C L U S I O N 5 9 R E F E R E N C E S 6 0 A P P E N D I X I 65 -iv-L I S T O F T A B L E S Table 2.1: Summary of Conservation Features used in Marxan Spatial Analysis 14 Table 2.2: Neighbor Compatibility Rankings 20 Table 3.1: User-defined Parameters 33 Table 3.2: Summary data for the 36 Marxan Solutions 34 Table 3.3: Marxan Input Files 37 Table 3.4: Marxan Output Files 38 Table 3.5: Coarse-filter Representation Performance for 36 Marxan Solutions 41 Table 3.6: Fine-filter Representation Performance for 36 Marxan Solutions 43 -v-L I S T O F F I G U R E S Figure I.I: The South Okanagan Lower Similkameen Study Area 3 Figure 2.1: Watershed Integrity Ratings 19 Figure 2.2: Neighbor Compatibility and Boundary Cost '. 21 Figure 2.3: Influence of Boundary Modifier (BM) on Reserve Configurations 22 Figure 2.4: 20% Targets — Scenarios 1, 2 and 3 25 Figure 2.5: 20% Targets — Scenarios 4, 5 and 6 26 Figure 2.6: 12% Targets — Scenarios 1, 2 and 3 27 Figure 2.7: 12% Targets — Scenarios 4, 5 and 6 28 Figure 3.1: Marxan's Import File Editor 30 -vi-G L O S S A R Y General glossary terms appear in CAPS when first used in the text. Compactness : A dimensionless measure of the ratio of the boundary length of the reserve system to the circumference of a circle with the same area as the reserve (McDonnell et al. 2002). Decision Suppor t Software: A computer-based application that uses data, models, and structured decision processes to aid in the process of decision-making (Sullivan et al. 1997). Edge Effects: Changes in habitat conditions (such as exposure to light or wind) created at or near the more-or-less well-defined boundary between ecosystems (BC MOF 2005). Efficiency: The ability to meet all conservation targets (e.g. ecosystems, habitats, species) while minimizing the overall 'cost' of the reserve. 'Cost' may be a measure of size, cost of acquisition or another relative economic or ecological measure. Geographic Information Systems (GIS): A computer-based system for the capture, storage, management, analysis and presentation of geographic (spatial) data. Graphical U s e r Interface: A computer interface that presents information in a user-friendly way using graphics, menus and icons. Persistence: The ability of a protected area to support the long-term survival of biodiversity features (Margules and Pressey 2000). Reserve Design A lgor i thm: Computational methods that identify reserve design solutions that meet stated conservation goals. Reserve Design Software: Computer programs that deliver decision support for reserve selection and design. Sensit ivi ty Analysis: The process of varying the input parameters in a given model to assess the level of change in the model outputs. U s e r Interface: The means by which people interact with a particular application. -vii-M A R X A N G L O S S A R Y Marxan glossary terms are written in ITALICS AND CAPS when first used in the text. Boundary Cost : The 'cost' of the boundary between two adjacent planning units. It may be a simple measure of length or it may incorporate other ecological or economic factors. Boundary Modif ier: Controls the relative importance placed on minimizing the combined boundary cost of the reserve relative to minimizing the combined planning unit cost. Increasing the boundary modifier encourages Marxan to select larger, more cohesive areas to meet its targets. Conserva t ion Features: The biodiversity elements Marxan is trying to protect in the reserve (e.g. broad ecological units, species habitats, aquatic features). Conserva t ion Targets : Quantitative values that define how much of a particular conservation feature Marxan should protect in the reserve. Feature Penalty Factor: A user-defined weight which controls how much importance is placed on fully representing a particular conservation feature. Marxan will work harder to protect a conservation feature with a high penalty value than one with a low penalty value. M i n i m u m C l u m p Size: Defines the minimum size of contiguous planning units needed to count as a viable patch for a particular conservation feature. This parameter was not used in this study. N u m b e r of C lumps: The number of unique viable clumps of a feature required. This parameter is used to control the replication of conservation features in Marxan. This parameter was not used in this study. Planning Uni t : The building blocks of Marxan are the parcels of land that are compared to one another - these parcels of land are called planning units. Marxan selects a combination of planning units to build a reserve. Planning Un i t Cos t : The individual 'cost' of each planning unit. The 'cost' does not have to reflect the familiar notion of the cost of acquisition; it can reflect any relative economic or ecological measure. Separat ion Distance: Defines the minimum distance that distinct clumps of a feature should be from one another. This parameter was not used in this study. -viii-T o t a l Score: The combined 'cost' of the reserve. It is calculated by the following formula: Total Score = £ planning unit cost + (boundary modifier * £ boundary cost) + £ feature penalty Where, £ planning unit cost is the combined 'cost' of the selected planning units, the boundary modifier is a multiplicative factor that weights the importance of minimizing the boundary cost relative to the 'cost' of the selected planning units, £ boundary cost is the 'cost' of the boundary surrounding the reserve, and the feature penalty is the penalty imposed for failing to meet conservation targets, which diminishes as conservation targets are fulfilled. -ix-A C K N O W L E D G E M E N T S Several people have been instrumental in helping me complete this thesis. A special thank you to Dr. Wil l iam Rees, who acted as thesis supervisor, and provided inspiration, encouragement and helpful advice throughout this project. Special thanks also to Jeff Ardron, Scientific Advisor on Marine Protected Areas for the German Federal Agency for Nature Conservation, who as a member of my thesis committee, provided sound technical advice and posed excellent questions that pushed me to understand Marxan better. Dr. John Woods , Scientist Emeritus for Parks Canada, was a pleasure to work with, and as an external advisor provided practical and constructive feedback and direction for this research. Thank you also to Tony Dorcey who acted as external examiner and like Dr. Rees was as a mentor throughout my time at the School of Community and Regional Planning. Thanks also to Doug Harvey and Claude Mondor from the Parks Establishment Branch of Parks Canada who despite their very busy schedules provided support and helpful comments from the inception of this project through to the end. Tim Boumeester, GIS Coordinator for the Regional District Okanagan-Similkameen and Leanna Warman, University of British Columbia, provided the data for this thesis and offered timely and helpful answers to my questions. I would also like to thank Patrick Lefebvre from Inform GIS — my former employer — and everyone from the Living Oceans Society, especially Kate Wil l is and Jennifer Lash who supported me in balancing a new job and the completion of this degree. Most importantly, I wish to thank my parents and the rest of my wonderful family including my soon to be husband Michael. -x-I. I N T R O D U C T I O N I . I O v e r v i e w A considerable body of conservation literature has been devoted to the selection and appropriate spatial configuration of new protected areas (for example, Margules and Pressey 2000; Warman 2001; Cabeza and Moilanen 2001; Groves et al. 2002; Ardron et al. 2002; Gonzales et al. 2003). Excessively simple strategies have proven to be methodologically inadequate (Reed 1983; Prendergast et al. 1993; Lomolino 1994; Simberloff 1997; Prendergast et al. 1999; Possingham et al. 2000), and in the past two decades, increasingly sophisticated DECISION-SUPPORT SOFTWARE (see Glossary) has been developed. Several of these programs leverage technical advancements in computers and GEOGRAPHIC INFORMATION SYSTEMS (GIS). One such application, Marxan, uses data on ecosystems, habitats, and other relevant conservation features to find efficient solutions to the problem of selecting a set of conservation areas that meet a suite of biodiversity targets (Ball and Possingham 2000; Possingham et al. 2000). Marxan provides a flexible and defensible tool with which to explore alternative management options and to facilitate negotiation amongst multiple stakeholders (Gonzales et al. 2003; Lewis et al. 2003; Airame et al. 2003). Marxan is gaining considerable support in the conservation community because it features an advanced RESERVE DESIGN ALGORITHM and generates efficient, spatially cohesive, and scientifically defensible results (Ball and Possingham 2000; Possingham et al. 2000; Leslie et al. 2003). However, Marxan is not without its shortcomings, and a number of application limitations have been raised in recent conservation literature (for example, Prendergast et al. 1999; Possingham et al. 2000; Cabeza and Moilanen 2001; Allison et al. 2003; Christensen 2004; Warman et al. 2004). Of particular concern is the inability of Marxan to explicitly consider spatial population dynamics (Cabeza and Moilanen 2001; Possingham et al. 2000), landscape dynamics (e.g. disturbance, catastrophes) (Allison et al. 2003), or the dynamics of human economic systems (e.g. ownership, site availability) (Prendergast et al. 1999; Christensen 2004). Additional concerns have been raised over the sensitivity of reserve design tools to decisions about scale, biological data and targets (Warman et al. 2004). In light of disagreement over the utility of Marxan, and the role this application is starting to play in applied conservation planning, it is important to investigate systematically how well Marxan functions. As part of that task, the present study evaluates whether Marxan could be a useful decision-support -I-tool for Parks Canada to use when selecting potential park areas. Concurrent with this study, Parks Canada is conducting a feasibility study for a national park reserve in the South Okanagan-Lower Similkameen (SOLS) region of British Columbia. This region serves as a case study for this thesis. The specific objectives of this study are threefold: (I) to apply a science-based systematic conservation planning framework to the design of a protected area in the SOLS; (2) to identity a portfolio of possible reserve configurations' for a new national park reserve in the SOLS region using Marxan; and (3) to assess Marxan using park selection criteria, and general usability standards. This research will shed some light on whether Marxan could be an effective program for Parks Canada to use in future feasibility studies. This thesis is divided into four chapters. The remainder of this chapter describes the SOLS study area, explains the institutional context for this study, provides a brief overview of the history of conservation planning and efficient reserve design, describes the qualities of an ideal reserve design application and introduces Marxan. Chapter 2 outlines the methods used to produce 36 unique reserve design options for the study areas and describes the criteria used to evaluate Marxan. Chapter 3 presents the results of the Marxan spatial analysis and evaluates the representational performance, ecological performance and usability of Marxan. Chapter 4 discusses the performance and usability of Marxan, and forwards a series of recommendations for Parks Canada's consideration. 1.2. A Biologically Diverse & Sensitive Ecosystem The SOLS study area (Fig I.I) is approximately 2,475 km2, and extends from the Canada/US border north to just south of Penticton, and from the west side of Snowy Mountain Protected Area east to the height of land of the Okanagan valley. The study area was defined by Parks Canada as part of the South Okanagan-Lower Similkameen feasibility study. Figure l.hThe South Okanagan Lower Similkameen Study Area The area is widely recognized as a unique, biologically diverse, ecologically rich, and sensitive natural region (Environment Canada 2000; SOSCP 2000). It is home to approximately 38 species of plants and animals that are federally listed as threatened, endangered or of special concern (COSEWIC 2005), and one-third of British Columbia's provincially red-listed1 species. The South Okanagan and Lower Similkameen watersheds serve as a critical ecological corridor linking the grasslands of interior British Columbia and the desert areas of the western United States (SOSCP 2000). In the past twenty years, the study area has experienced significant human population growth and accompanying development. Reduction and fragmentation of natural habitats due to rapid urbanization, agricultural development, and other human activities are posing a serious threat to native species and ecological processes in the region (Cannings and Durance 1998). Currently, the area is recognized as one of Canada's most endangered ecosystems (Environment Canada 2000). High levels of population, fragmented habitats, 1 Includes any ecological community, and indigenous species and subspecies that are extirpated, endangered, or threatened in British Columbia (BC MOE 2006). - 3 competing land uses and private ownership make the region a particularly challenging and constrained area in which to establish a national park reserve. 1.3. Institutional Context Under a Memoranda of Understanding (MOU) signed in 2003 by the Government of Canada and the Province of British Columbia, Parks Canada is currently investigating the feasibility of establishing a national park reserve in the South Okanagan - Lower Similkameen valleys. The reserve would preserve a representative area of the Interior Dry Plateau natural region (see Fig I; overview map), which spans from the US/Canada border to just south of Smithers, and is one of 12 natural regions not represented in Canada's system of national parks. Under the provisions of the Canada National Parks Act (2000), Parks Canada aims to establish ecologically viable national parks that protect in perpetuity representative examples of the Canadian landscape. According to Parks Canada's Guiding Principles and Operational Policies (Parks Canada 1997), the boundaries of a new national park reserve should strive to meet six ecological objectives, two socio-economic objectives and one visitor use/enjoyment objective. The six ecological objectives are: (I) protect ecosystems and landscape features representative of the natural region; (2) accommodate the habitat requirements of viable populations of native wildlife species; (3) include an undisturbed core which is relatively unaffected by impacts originating from the surrounding landscape; (4) ensure that the park boundaries do not fragment sensitive, highly diverse or productive natural communities; (5) maintain drainage basin integrity; and (6) protect exceptional natural phenomena and vulnerable threatened or endangered wildlife and vegetation. The two socio-economic objectives are: (I) minimize disruption of the social and economic life of the surrounding region; and (2) exclude permanent communities. The visitor use/enjoyment objective involves offering opportunities for public enjoyment. Despite these provisions, the majority of established national parks in Canada "deviate from optimum ecological limits and present significant challenges to park managers" (Manseau et al. 2001). Manseau et al. (2001) provide examples — Nahanni National Park and Fundy National Park cut across watersheds, Prince Edward Island National Park and St. Lawrence Islands National Park are highly fragmented, and Prince Albert National Park, and Riding Mountain National Park do not include the entire range of animal populations. Although the Parks Establishment Branch has examined the use of reserve design algorithms for the selection of new national park reserves and national marine conservation areas, they do not currently use Marxan or other algorithms for this task. This research will shed some light on -4-whether Marxan might be useful for locating and designing representative and ecologically viable national park reserves. 1.4 A Brief History of Conservation Planning Approaches to conservation planning are evolving rapidly in response to new insights into ecological systems and biodiversity (Poiani et al. 2000). Early in the conservation movement, protected areas were established in an ad hoc manner, on less productive sites, and managed as islands with little or no consideration of the surrounding area or the dynamic nature of ecological systems (Pressey 1994; Poiani et al. 2000). Today, most existing reserve systems have an over-representation of protected areas in isolated high mountain ecosystems, with little other resource value, and an under-representation of productive, low elevation ecosystems (Possingham et al. 2000). In the past 15 years, approaches to reserve selection and design have broadened in response to a more expansive definition of biodiversity, a more in-depth understanding of ecological systems and changing values (Poiani et al. 2000). Today biodiversity is viewed at multiple levels of biological organization, with each level displaying unique composition, structure and function (Poiani et al. 2000). This has resulted in the development of new, more systematic and inclusive approaches to reserve selection and design which target biodiversity at multiple spatial scales and levels of biological organization (for example, Soule and Terborgh 1999; Poiani et al. 2000; Margules and Pressey 2000; Groves et al. 2002). Systematic approaches to reserve design use explicit conservation goals, biodiversity surrogates, quantitative conservation targets, and decision-support tools to identify comprehensive reserves that promote the long term PERSISTENCE of biodiversity elements (Margules and Pressey 2000; Poiani et al. 2000). Although these systematic approaches take a lot of time and money, they are justified on the grounds that they are rigorous, scientifically defensible, and transparent (e.g. Margules and Pressey 2000; Groves et al. 2002; Gonzales et al. 2003). A number of different reserve design tools have been developed to provide decision-support for systematic reserve design. 1.5 Efficient Reserve Design A decision-support system is an interactive, computer-based system that facilitates the use of data, models, and structured decision processes to support informed and scientifically defensible decisions (Sullivan et al. 1997). Reserve design software is a specific type of decision-support system that supports the analysis and evaluation of decisions pertaining to the location and design of protected areas. These applications help planning teams carry out systematic and iterative decision-making -5-processes, and effectively analyze the results of different conservation options. This thesis focuses on a particular class of reserve design software, where the goal is to represent all conservation features as EFFICIENTLY as possible (Kirkpatrick 1983; Ball and Possingham 2000). Efficient reserve design tools attempt to achieve all conservation targets (e.g. reserve 100 km 2 of the bunchgrass biogeoclimatic zone) while minimizing the overall 'cost'2 of the reserve (Pressey & Nichols 1989; McDonnell et al. 2002). The rationale for 'efficiency' is that competition between conservation and other forms of land use is considerable; therefore conservation areas should be as efficient as possible while still fulfilling all conservation objectives (Pressey et al. 1996; McDonnell et al. 2002). An algorithm is a process or set of rules used for problem solving. In conservation planning, reserve design algorithms are used to identify a set of potential conservation areas that best meet explicit conservation goals. Two general types of reserve design tools have been devised to efficiently solve reserve design problems: exact algorithms and heuristic (non-exact) algorithms. Exact algorithms, such as Integer Linear Programs (ILP), identify a single optimal solution, whereas heuristics provide a number of good, near-optimal solutions. Because most reserve design problems consider a large number of sites and conservation targets, it is difficult, and often impossible, to find an optimal solution in a reasonable amount of time using an exact algorithm (Possingham et al. 2000; Cabeza 2003). Currently, heuristics are preferred over exact algorithms because they provide timely solutions to complex reserve design problems, and they offer a range of near-optimal solutions for planners and stakeholders to consider (Possingham et al. 2000; McDonnell et al. 2002; Cabeza 2003). Two types of heuristic algorithms have historically been used to solve reserve design problems: iterative (Kirkpatrick 1983; Pressey et al. 1996; Possingham et al. 2000; McDonnell et al. 2002) and simulated annealing (Possingham et al. 2000; McDonnell et al. 2002). In order to use either of these heuristic algorithms it is necessary to divide the study area into small parcels of land, called PLANNING UNITS (see Marxan Glossary), and record how much of each conservation feature is contained in each planning unit. 2 The term 'cost' in the context of reserve design software does not have to reflect the familiar notion of the cost of acquisition; it can be a measure of size, number of sites or another economic, social or ecological measure. The user of the reserve design software decides how cost is used. Please note that this unconventional use of the term 'cost' is used throughout this thesis. -6-Iterative algorithms execute the same set of rules repeatedly until a termination criteria is met. They start with an empty reserve, and then at each iteration, one planning unit is added to the reserve system. The algorithm stops when no planning unit will improve the reserve. For example, the 'richness heuristic' adds the planning unit that has the most unrepresented conservation features (the richest parcel of land) (Pressey et al. 1996; Ball and Possingham 2000); it does this repeatedly until every conservation feature has been fully represented. Other iterative heuristic algorithms reflect different measures of biological value, such as species rarity (Ball and Possingham 2000). Although iterative algorithms run quickly and are easy to understand, they were not used in this study because they generate only one solution, and it is not necessarily the best one (Pressey et al. 1996; Possingham et al. 2000). This lack of optimization is due to the linear nature of the algorithm — it always acts in a predictable manner, seeking the best parcel of land first (i.e. the richest/the rarest), the next best second, and so forth until the reserve is designed. This approach means that the algorithm often gets stuck before it has fulfilled all conservation targets (McDonnell et al. 2002). Oftentimes, making less than optimal choices earlier on frees up parcels of land to be chosen later. Simulated annealing was used in this study because it reconciles the problems outlined above. The simulated annealing algorithm begins with a set of planning units that is either randomly selected or defined by the user3. Then, at each iteration (typically there are millions of iterations), a planning unit is randomly chosen and either added to, or removed from, the reserve (depending on whether it was already in the system). The total score of the reserve — the combined 'cost' of the planning units, the boundary 'cost,' and the penalty for not meeting conservation targets — is then calculated. If the total score of the reserve decreases or stays the same, the change is accepted. If it increases, the change may or may not be rejected. The strength of the simulated annealing algorithm is that it programs in a random element, so that early on in the process sub-optimal selections are permitted (selections where the total score increases). This prevents getting stuck prematurely in a 'local minimum' (Ball and Possingham 2000). As the algorithm progresses through successive iterations, it behaves more rationally — choosing progressively more optimal reserve configurations until the reserve is complete 3 The user can control what proportion of planning units make up the initial reserve. The starting proportion can be set to any value between 0 (no planning units start in the reserve) and I (all planning units start in the reserve). A value of 0.5 would mean that each planning unit had a 50% chance of being included in the initial reserve. There is no theoretical reason to set the default to any particular level (Ball and Possingham 2000). The user can also override what planning units are included in the initial reserve by controlling the 'status' of individual planning units. For example, a status of I means that the planning unit is guaranteed to be in the initial reserve. -7-(Ball and Possingham 2000). The random element means that each time the program is run it will generate a different solution. Running repeated runs of Marxan increase the likelihood of finding the best solution. Overall, simulated annealing has a number of benefits: (I) it out-performs other heuristic algorithms in terms of generating the most efficient and comprehensive solution to complex reserve design problems; (2) it generates a number of different reserve design options that meet the stated conservation objectives, rather than one single solution; and (3) it has the computation capacity to handle large amounts of data (Pressey et al. 1997; Possingham et al. 2000; McDonnell et al. 2002; Leslie et al. 2003; Stewart et al. 2003). 1.6 Key Attributes of Good Reserve Design Tools Existing conservation planning theory and expert opinion suggest that good reserve design applications should be efficient, flexible, transparent, technically proficient and easy to use. Efficiency was discussed above. Flexible reserve design applications integrate and analyze multiple conservation objectives and information sources, and derive diverse reserve design solutions. Good reserve design tools facilitate the investigation of different policy options (e.g. to include existing provincial parks, to exclude native reservations), and the comparison of different outputs. A range of viable reserve design solutions (outputs) is helpful when balancing economic, social and political considerations (Ball and Possingham 2000; Gonzales et al. 2003; Lewis et al. 2003; Leslie et al. 2003). Transparency refers to how well people understand the decision-making procedures and output products. Highly transparent reserve design methods are easy to communicate and to understand. For this reason, they tend to increase the accountability and credibility of decision-making. Technically proficient reserve design applications have the computational capacity to solve complex reserve design problems involving large amounts of data in a timely manner. They also respond predictably to changes in input parameters. Reserve design software is easy to use when procedures are logical, tools are easy to access, data can be easily queried, commands are automated, output products are easy to visualize and understand and documentation is comprehensive. Easy to use reserve design programs also integrate seamlessly with commercial GIS software and other decision-support tools. The effectiveness of a reserve design application also depends on how well it meets the goals of the initiating organization. Under the provisions of the National Parks Act (2000), Parks Canada aims to establish park areas that meet regional representation, ecological integrity, and visitor use/understanding objectives. Due to the ecological focus of this study, visitor use/understanding is not considered. Regional representation is the cornerstone of Parks Canada's national park systems plan. It involves establishing a national park reserve that is representative of the natural region. A successful reserve will capture the natural variability of biological and physical features at multiple spatial scales and levels of biological organization (i.e. ecosystems, habitats, species) (Parks Canada 1997; Manseau et al. 2001). Representing the full range of physical and biological diversity in large contiguous areas preserves biotic diversity and supports changes in species distributions in response to climate change (Noss 2001). However, representation alone cannot ensure the long-term success of new national parks. Reserves, once established must also support the persistence of key natural features by being functional, healthy and resilient. Ecological integrity refers to a condition that is determined to be characteristic of a natural region and likely to persist (Parks Canada 2002). To achieve ecosystem integrity, reserves must support viable populations of native species and the ecological and evolutionary processes they depend on. To meet these objectives, reserve design tools must meaningfully address spatial reserve design variables such as size, shape, replication, connectivity and appropriate alignment of boundaries (with for example watersheds) (Parks Canada 1997; Margules and Pressey 2000; Cabeza and Moilanen 2001; Stewart et al. 2003; McDonnell et al. 2002). 1.7 Evaluating Marxan Marxan, which was developed by Dr. Hugh Possingham and Dr. Ian Ball in Australia, comes from a respected lineage of reserve design programs, including SPEXAN and SITES (Ball and Possingham 2000). Of the decision-support software developed to achieve efficient reserve design, Marxan has emerged as a particularly promising program. Marxan was a key tool in the rezoning of the Great Barrier Reef Marine Park (Lewis et al. 2003) and in the design of the California Channel Islands (Airame et al. 2003). One of Marxan's greatest assets is that it uses the simulated annealing algorithm. However, Marxan has a number of additional features. For example, Marxan incorporates spatial considerations into the reserve design process. Without a spatial constraint, reserve configurations are highly fragmented and -9-have a high edge to area ratio, making them unsuitable from an ecological, socioeconomic and management perspective (Possingham et al. 2000; McDonnell et al. 2002). Compact reserves increase species persistence by reducing EDGE EFFECTS and facilitating the dispersal and recolonization of empty habitats; they are also preferable from a management perspective (Cabeza and Moilanen 2001; McDonnell et al. 2002). Marxan possesses a number of qualities of an ideal reserve design application. For example, it is capable of producing 'efficient' reserve design options that meet explicit representation and economic targets (Gonzales et al. 2003; Airame et al. 2003; Lieberknecht et al. 2004). Marxan also offers a flexible environment in which to explore and analyze alternative reserve configurations (Possingham et al. 2000; Airame et al. 2003; Stewart et al. 2003; Lieberknecht et al. 2004). It is however, critical to acknowledge that Marxan is a model — it operates according to a set of assumptions and is limited in its ability to simulate the complexity of ecological systems (Possingham et al. 2000; Margules and Pressey 2000; Cabeza and Moilanen 2001). Just because it finds solutions that are efficient and compact, does not mean these solutions are viable from an ecological perspective, or practical from a management perspective. In the last decade, there has been increasing interest in evaluating the effectiveness of reserve design projects and applications (Leslie 2005). To date, the performance of reserve design tools has been assessed using a number of measures including computational speed (Pressey et al. 1997), efficiency4 (Stewart and Possingham 2002), compactness5 (McDonnell et al. 2002; Possingham et al. 2000) and ecosystem representation6 (Gonzales et al. 2003). However, more comprehensive and case specific assessments of Marxan's real-world usefulness are needed to evaluate success. This thesis evaluates whether Marxan could be a useful decision-support tool for Parks Canada to use when selecting potential park areas. The assessment is based on usability criteria and park selection 4 Stewart and Possingham (2002) define efficiency as the "ability of a reserve design process to represent regional biodiversity in the least number of available sites". This is a more specific definition from the one in the Glossary. 5 McDonnell et al. (2002) define compactness as "the ratio of the boundary length of the reserve system to the circumference of a circle with the same area as the reserve" (McDonnell et al. 2002). This definition was adopted in this study. 6 A relative measure of the amount of the conservation target that is met in the final reserve (area reserved-area target)/area target) (Gonzales et al. 2002). Values greater than one indicate overrepresentation of a feature, whereas values less than one indicate underrepresentation of a feature. -10-criteria, developed in consultation with Parks Canada. The specific objectives of this study are threefold: (I) to apply a science-based systematic conservation planning framework to the design of a protected area in the SOLS; (2) to identity a portfolio of possible reserve configurations for a new national park reserve in the SOLS region using Marxan; and (3) to assess Marxan using park selection criteria, and general usability standards. This research will shed some light on whether Marxan could be an effective program for Parks Canada to use in future feasibility studies. 2 . M E T H O D S 2.1 I n t r o d u c t i o n The methods chapter of this thesis addresses two topics: (I) the methods used to identity a portfolio of 36 reserve configurations in the SOLS region using Marxan; and (2) the criteria used to evaluate Marxan as a decision-support tool. The Marxan spatial analysis was used to generate a number of representative reserve design options for a national park reserve in the SOLS study area. I adopted a systematic approach to designing the 36 reserve configurations which involved four stages: (I) establishing conservation goals and ecological objectives; (2) compiling data; (3) setting conservation targets; and (4) generating reserve configuration options using Marxan. Marxan employs a unique vocabulary, and at times, words have unconventional meanings (e.g. costs). For a dictionary of Marxan terminology please refer to the Marxan Glossary. Words that are included in the Marxan Glossary are in CAPS AND ITALICS. 2 . 2 C o n s e r v a t i o n G o a l s a n d E c o l o g i c a l O b j e c t i v e s Park's Canada's ecological goals of representation and ecological integrity were used to inform the reserve design process. In pursuit of the representation goal, I established two basic ecological objectives: (I) represent a broad spectrum of ecological variation; and (2) protect exceptional natural phenomena, including rare and endangered species and priority habitats. In pursuit of the ecological integrity goal, I established three ecological objectives: (I) protect ecologically intact areas; (2) preserve sites with compatible adjacent land use; and (3) create compact reserve configurations with low perimeter/surface area ratios. 2 .3 C o n s e r v a t i o n F e a t u r e s , T a r g e t s a n d P e n a l t i e s I entered eighty-eight CONSERVATION FEATURES (see Marxan Glossary) into Marxan in an effort to protect the physical and biological diversity of the study area and the region's special elements. Decisions regarding what features to include were based on the National Parks System Plan (1997), draft planning documents for the SOLS feasibility study (Parks Canada 2005) and available datasets. In this thesis, 'coarse-filter' refers to conservation efforts aimed at ecosystems or landscapes and 'fine-filter' defines efforts directed at habitats. Coarse-filter conservation features were grouped into four -12-categories: biogeoclimatic representation, geological representation, aquatic representation and physical relief representation. Fine-filter representation features were grouped into two categories: priority habitats and threatened, endangered and focal species habitats. Leanna Warman of the University of British Columbia provided the threatened, endangered and focal species habitat suitability data. The Regional District Okanagan Similkameen (RDOS) provided the remainder of the data. CONSERVATION TARGETS tell Marxan how much of each feature to protect in the reserve. Like conservation features, conservation targets are user-defined. In consultation with Parks Canada, I assigned each conservation feature two conservation targets, representing 20% and 12% of the area of each conservation feature within the study area respectively. A 20% target means that 20% of the area of the feature within the study area should be in the reserve. For example, if 100 km2 of interior Douglas fir is in the study area, and a 20% target is used, Marxan should create a reserve that includes at least 20 km2 of interior Douglas fir. Twenty percent and 12% representation targets were used in an attempt to configure reserves that were 500 km2 and 300 km 2 respectively. Although these reserve sizes are quite small in comparison to many other national parks, they are currently proposed as large enough for establishing a minimum fire control regime7. They are also practical for an area that is already subject to extensive development and private land ownership. The one exception to the targets defined above was rare, endangered and focal species targets. These targets were based on Warman's (2001) calculations of the amount of area required to maintain a species current population in the South Okanagan based on the density of individuals in suitable habitat. The method used to generate rare, endangered and focal species targets is discussed in more detail in Section 2.5.2. The FEATURE PENALTY FACTOR is a user-defined weight, which controls how much emphasis is placed on fully representing a particular conservation feature. Marxan will work harder to protect a conservation features with a high penalty factor than one with a low penalty factor. Using information derived from previous Marxan studies and informal discussions with Parks Canada, I chose to use feature penalty factors ranging from I (low) to 5 (high). I assigned high feature penalty factors to conservation features that had the following characteristics: (I) they play a vital role in achieving Park's 7 These values were proposed by Parks Canada early in the feasibility study process. They may change as the study progresses. -13-Canada's objectives; and (2) they are derived from accurate and complete datasets. I gave high feature penalties values to biogeoclimatic zones and aquatic features. I assigned low feature penalty factors to conservation features that had the following characteristics: (I) they are not vital to achieving Parks Canada's objectives; and/or (2) they are derived from less accurate, out-dated, or incomplete datasets. I gave low penalty factors to geological features (see Section 2.4.2), rare, endangered, and focal species habitats (see Section 2.5) and priority habitats (see Section 2.5). I assigned moderate penalty factors to physical relief classes. Table 2.1 summaries the breakdown of conservation features by representation goal, conservation feature category, layer(s), number of features, and feature penalty factor. For a detailed list of all conservation features, targets, and feature penalty factors refer to Appendix I. Table 2.1: Summary of Conservation Features used in Marxan Spatial Analysis R e p r e s e n t a t i o n G o a l C o n s e r v a t i o n Fea ture C a t e g o r y Layer(s) N u m b e r of Features Fea ture Pena l ty F a c t o r Coarse-Filter Representation Biogeoclimatic Representation Biogeoclimatic Zones 6 5 Geological Representation Rock Type 5 1 Physical Relief Representation Physical Relief Classes 45 3 Aquatic Representation Definite Lakes Indefinite Lakes Definite Rivers Indefinite Rivers Wetlands 5 5 Total Coarse-Filter 61 Fine-Filter Representation Protection of Threatened, Endangered and Focal Species Habitats 1 layer for each threatened, endangered & focal species 14 2 Priority Habitat Representation 1 layer for each priority habitat 13 2 Total Fine-Filter 27 Total: 88 2.4 Coarse-Filter Representation Representing a broad spectrum of conservation features is a commonly stated objective of reserves. I used biogeoclimatic zones, aquatic features, physical relief classes, and rock type to represent the range of terrestrial and aquatic ecosystems in the study area. In total, these coarse-filter conservation features made up 61 of the 88 (69%) features input into the Marxan spatial analysis. The central premise for using coarse-filter conservation features is that by representing examples of all ecosystems, at all -14-successional stages, the majority of species and their supporting natural systems will also be preserved (Noss 1996; Schwartz 1999; Poiani et al. 2000). This approach is referred to by Soule and Noss (1998) as "a kind of habitat umbrella effect." Capturing a range of ecosystem variability in intact ecosystems also supports shifts in habitat preference of species at different life stages and provides maximum opportunity for biogeoclimatic zones to change over time — minimizing biodiversity loss if biogeoclimatic zones were to shift due to climatic change (Noss 2001). The sections below describe each of the coarse-filter conservation feature categories used in this study. 2.4.1 Biogeoclimatic Representation The Marxan spatial analysis sought to represent a broad spectrum of biogeoclimatic variation within the candidate reserve configurations. In British Columbia, the biogeoclimatic ecosystem classification (BEC) uses a combination of climate, soil and vegetation to delineate ecological zones that represent large geographic areas sharing a relatively uniform climate (Pojar et al. 1987; Meidinger and Pojar 1991). The spatial extent of each BEC zone is based on the distribution of climax and late-seral plant communities on sites that best reflect the regional climate. Fourteen biogeoclimatic zones have been delineated in British Columbia and 6 of these occur within the study area: bunch grass, ponderosa pine, interior Douglas fir, montane spruce, Engelmann spruce-subalpine fir, and alpine tundra. 2.4.2 Geological Representation The Marxan spatial analysis attempted to capture the full range of rock types within the study area. Rock type influences the composition and properties of soils, the type of vegetation, and the distribution and abundance of species. I delineated five different kinds of rock from BC Ministry of Energy Mines 1:250,000 geology mapping: volcanic rock, ultramafic rock, sedimentary rock, metamorphic rock and intrusive rock. I gave rock types a very low conservation penalty factor because they do not serve a critical role in fulfilling Parks Canada's representation or ecological integrity mandate. 2.4.3 Aquatic Representation Aquatic features receive a great deal of conservation attention in the SOLS region where water resources are under a great deal of pressure. In this study, five classes of aquatic features were used as high priority conservation targets: definite lakes, indefinite/intermittent lakes, definite rivers, indefinite/intermittent rivers and wetlands. I extracted water features from Terrain Resource Information Mapping (TRIM) 1:20,000 mapping produced by the British Columbia Ministry of -15-Sustainable Resource Management (MSRM). I converted linear rivers and streams into area features by applying a 5-metre buffer to definite rivers and a 2.5 metre buffer to indefinite/intermittent rivers. 2.4.4 Physical Relief Representation The goal of representing the full range of abiotic terrestrial diversity requires capturing the full range of physical relief from warm, flat, valley bottoms to cold, steep mountain tops. I developed a simple multidimensional relief classification using GIS to represent unique classes of slope, aspect and elevation. These physical relief classes were used to represent the fine-scale topographic variability of the study area. I derived slope and aspect from TRIM 1:20,000 Digital Elevation Model (DEM) points using tools provided by ESRI's ArcGIS 8.3 and the Spatial Analyst extension. Elevation data was delineated into 5 different classes (274-600, 600-1200,1200-1800,1800-2400, and >2400 metres). Slope was delineated into 5 categories based on natural breaks in the data (Flat: 0-3, Gentle: 4-15, Moderate: 16-26, Steep: 27-35, and Very Steep >35). Aspect values were generalized into 2 categories: cool (North facing: 285-135°) and warm (South facing: 135-285°). Level slopes did not include aspect. I combined the elevation, slope, and aspect layers in order to represent unique slope and aspect combinations within each of the 5 elevation classes. 2.5 Fine-Filter Representation The ability to support viable populations of native species and to protect the occurrence of threatened or endangered wildlife and vegetation are important criteria in the establishment of new national park reserves (Parks Canada 1997). These objectives are particularly important in the South Okanagan, which is home to the greatest number of species at risk in Canada (SOSCP 2004). In order to sustain a diversity of indigenous, rare and endangered species, I adopted the priority habitats identified by the South Okanagan Similkameen Conservation Program (SOSCP) and the wildlife habitat suitability predictions generated by Warman et al. (1998) and Warman and Hodges (in progress) as fine-filter conservation features. In total, fine scale conservation features comprised 27 of the 88 features (31%) input into Marxan. I set the feature penalty factor for each of the fine scale features to 2 (low) for three reasons: (I) incomplete data coverage (data existed for 43% of the study area); (2) Parks Canada policy prioritizes broad ecosystem representation over the protection of individual species and species habitats (Parks Canada 1994); and (3) the age of the Terrestrial Ecosystem Mapping (TEM) data8 (TEM data is 8 years old). 2.5.1 Priority Habitat Representation Due to the high number of threatened and endangered species in the SOLS valleys, conservation needs cannot be met solely on a species by species basis (SOSCP 2004). The SOSCP (2004) recommends fine-scale landscape level conservation efforts, such as those directed at priority habitats, as an effective strategy for protecting a wide array of non-target species and habitats. I adopted this strategy and used the priority habitats defined by the SOSCP as input into the Marxan spatial analysis. Priority habitats were identified by the SOSCP using the following criteria: (I) the percentage historic coverage still remaining (as of 1995); (2) how much of the remaining habitat is under some conservation protection; and (3) its importance to species at risk (SOSCP 2004). Priority habitat data came from 1:20,000 TEM data developed by the Ministry of Environment, Lands and Parks (MELP) and updated antelope-brush mapping data prepared by Iverson et al. (2005). I identified thirteen high priority habitats in the SOLS study area. I extracted priority habitat data from TEM polygons by selecting all polygons that contained 50% or more of a particular priority habitat in the TEM polygon layer. I obtained priority antelope brush habitats (Antelope Brush - Needle and thread grass and Antelope Brush - Selaginella) data from the core Antelope Brush units mapped by Iverson et al. (2005). 2.5.2 Representation of threatened, endangered and focal species habitats I used the wildlife habitat relationship models generated by Warman et al. (1998) and Warman and Hodges (in progress) to protect the habitats of 13 nationally threatened or endangered species and one focal species — the California Bighorn Sheep. For each species, the model identifies the amount of habitat available within each TEM polygon for the species life history requirements (e.g. breeding, foraging, nesting, escape, cover) (Warman 2001). I derived conservation targets for each of these species from Warman's (2001) calculations of the amount of area required to maintain a species' current population in the South Okanagan based on the density of individuals in suitable habitat. Because Warman's targets were based on a larger study area, I calculated the proportion of suitable habitat in the SOLS study area, and adjusted the targets accordingly. For example, if 50% of the species suitable habitat was in the SOLS study area, and Warman's target was 6 km2, I used a target of 3 km2. 8 TEM data uses information on soil type, slope, aspect, vegetation types, and successional stage to stratify the landscape into irregular biophysical habitat units (Lea et al. 1991; Resources Inventory Committee 2000). -17-2.6 Preparing Marxan 2.6. / Planning Units In order to use Marxan, it is necessary to divide the study area into small parcels of land, called PLANNING UNITS, and record how much of each CONSERVATION FEATURE is contained in each planning unit Planning units are the parcels of land that are available for selection by Marxan. The user determines the size and shape of the planning units. Planning units can be based on an arbitrary grid, administrative boundaries or more ecologically meaningful units (such as watersheds). I divided the study area into 23,826 hexagon shaped planning units. Each hexagon was 0.1039 km 2 or 10.39 ha. The size of each hexagon edge was 200 meters. I used uniformly sized planning units over irregularly sized units to avoid the area-related bias9 that can occur when using irregularly sized units such as watersheds (Round River Conservation Studies 2003), and hexagons over square grid cells because they have a smaller perimeter to area ratio. I locked out all planning units that were completely or primarily composed of urban areas because these areas are not likely to be restored to a natural state. I then created a file, which recorded how much of each conservation feature was contained in each planning unit. This file contained over 2 million values. 2.6.2 Planning Unit Cost I assigned each planning unit a 'cost'. The 'cost' of including a planning unit in a reserve does not have to reflect the familiar notion of the cost of acquisition; it can reflect any relative measure, including, foregone revenue or ecological integrity. Marxan allows the user to define the cost measure. Regardless of the measure used, Marxan will always attempt to minimize the 'cost' of the reserve, while fulfilling conservation targets. In an attempt to achieve an ecologically viable park, I based planning unit 'cost' on watershed integrity (similar to Round River Conservation Studies 2003). I calculated the 'cost' of each planning unit using the following equation: Planning Unit Cost = Planning Unit Area * Watershed Integrity Rating 9 Because larger units tend to have a greater number and quantity of conservation features than smaller units, it is more likely that they will be selected over smaller units. -18-Where, planning unit area is the area of the watershed and the watershed integrity rating is a relative measure of the proportion of roads (buffered 200 meters), urban areas, clearings and. gravel pits per watershed. I used watershed integrity ratings that ranged from I (pristine) - 5 (heavily modified). In this way, planning units that were part of highly disturbed watersheds were five times more costly for Marxan to select than planning units that were part of pristine watersheds. Using this approach, highly disturbed planning units were not likely to be selected unless they contained conservation features that could not be found in other areas. Figure 2.1 shows a map of watershed integrity rankings. Lighter areas represent more pristine areas, whereas darker areas are more heavily modified by human activity. Figure 2.1 :Watershed Integrity Ratings Watershed Integrity Rating 2.6.3 The Boundary Cost Marxan users wishing to exert control over the spatial cohesiveness of reserves are required to do two things: (I) create a file which records BOUNDARY COSTS; and (2) set an appropriate BOUNDARY MODIFIER before running Marxan (section 2.6.4).The boundary cost is the'cost' of the boundary - 19-be tween t w o adjacent planning units. T h e user defines wha t measure is used t o r ep resen t the bounda ry cost . It may be a s imple measure o f length o r it may i nco rpo ra t e o t h e r ecologica l o r e c o n o m i c factors. M a r x a n uses the boundary cos t file to calculate the 'cos t ' o f the bounda ry of the reserve . W h e n the 'cos t ' is equal t o the length, reserves that are m o r e spatially cohes ive have a l o w e r bounda ry cos t than those that are f ragmented. T h e bounda ry file r e c o r d s o n e r e c o r d fo r each unique c o m b i n a t i o n o f neighbors . Because each hexagon that is n o t o n the edge o f the study area has six unique neighbors , the boundary file is ve ry large (23,826 hexagons * 6 = 142,956 r eco rds ) . In o r d e r t o encourage M a r x a n t o select sites w i t h compa t ib le ne ighbor ing land use, I assigned each boundary a relat ive 'cos t ' value based o n ne ighbor compat ib i l i ty using the fo l lowing f o r m u l a : B o u n d a r y C o s t = Length of Edge * N e i g h b o r C o m p a t i b i l i t y Ranking W h e r e , length of edge is the length of the edge (in meters) , and the neighbor compatibility ranking is a relat ive ranking based o n the ne ighbor compat ib i l i ty . I used ne ighbor compat ib i l i ty ratings that ranged f r o m 1-5, w i t h one being the m o s t compa t ib le ne ighbor and 5 being the least compa t ib le neighbor . Tab le 2.2 displays the b r e a k d o w n o f ne ighbor compat ib i l i ty rankings. Tab le 2.2: N e i g h b o r C o m p a t i b i l i t y Rankings Rank Description Land Use 1 Compatible neighbor I Incompatible neighbor existing protected area 2 open range, pasture, swamp, river, lake, forest 3 cultivated lands, gravel pits, clearing, golf course 5 urban areas, roads, canals Based o n the above rankings, a planning unit that shares a 2 0 0 m e t r e b o r d e r w i t h a p r o t e c t e d area 'cos ts ' 2 0 0 (the length o f that b o r d e r ) , whereas o n e that shares a b o r d e r w i t h an urban area 'cos ts ' 1000 (Fig 2.2). Th i s approach pushes M a r x a n t o c h o o s e compa t ib le neighbors o v e r those that are incompat ib le . -20 -Figure 2.2: Neighbor Compatibility and Boundary Cost 2.6.4 The Boundary Modifier The user controls the spatial cohesiveness of the reserve by defining a BOUNDARY MODIFIER prior to running Marxan. The boundary modifier controls the relative importance placed on minimizing the overall boundary 'cost' of the reserve relative to minimizing the 'cost' of selected planning units. Setting the boundary modifier to 0 means that the boundary 'cost' exerts no control over the reserve. If a zero boundary modifier is used, reserves are highly fragmented. Increasing the boundary modifier encourages Marxan to select fewer, larger contiguous areas to meet its targets. Because planning problems vary, and the cost measure of the planning unit and boundary is arbitrary, there is no predefined value to use for the boundary modifier. I experimented with a number of different values and analyzed the resulting maps. The maps showed that boundary lengths of I, 3 and 6 provided a good range of clustering (Fig 2.3). Figure 2.3: Influence of Boundary Modifier (BM) on Reserve Configurations Between 3 and 6, the influence of the boundary modifier diminished — reserves did not become appreciably more clustered although they did tend to get slightly larger and significantly more 'costly'. I ran Marxan scenarios using boundary modifier of I, 3 and 6 to include a range of cohesiveness in the reserve configurations. 2.6.5 Additional Marxan Parameters Three additional input parameters can be used to exert control over Marxan's outputs: MINIMUM CLUMP SIZE, SEPARATION DISTANCE, and SEPARATION NUMBER. The minimum clump size defines the minimum size of contiguous planning units needed to count as a viable patch for a particular conservation feature. The number of clumps defines the number of unique viable clumps of a feature required. This parameter is used to control the replication of conservation features in Marxan. The separation distance defines the minimum distance that distinct clumps of a feature should be from one another. Although promising in theory, these parameters were not used in this study because they push the limits of the simulated annealing algorithm when used with complex conservation problems such as the one considered in this thesis. Although I attempted to use these parameters in this study, these attempts failed. For this reason, these parameters are not considered further in this thesis. 2.7 Running Marxan Marxan V. 1.8.2 (Ball and Possingham 2000) was used to produce alternative reserve designs for the study area. Marxan is designed to identify representative, spatially compact and efficient reserves. The software offers a variety of reserve design algorithms. I used simulated annealing due to the benefits outlined in section 1.5. As explained in section 1.5, the simulated annealing algorithm attempts to minimize the TOTAL SCORE of the reserve while meeting user-defined conservation targets. The total -22-sco re is calculated by the fo l l ow ing fo rmu la : T o t a l S c o r e = £ Planning U n i t C o s t + (Boundary Mod i f i e r * ]T Bounda ry C o s t ) + £ Feature Penal ty W h e r e , £ planning unit cost is the c o m b i n e d ' cos t ' of all se lec ted planning units, the boundary modifier is a mul t ip l icat ive fac to r that weights the impor tance of min imiz ing the bounda ry ' cos t ' relat ive t o the ' cos t ' o f the se lec ted planning units, £ boundary cost is the 'cos t ' of the bounda ry su r round ing the reserve , and the Feature Penalty is the penalty i m p o s e d f o r failing t o m e e t conse rva t i on targets, w h i c h d imin ishes as the target approaches . In this thesis, the to ta l s c o r e wi l l reach a m i n i m u m w h e n the necessary amoun ts of all conserva t ion features are cap tu red in a reserve conf igurat ion that is smal l , cohes ive and ecolog ica l ly intact. 2.7. / Reserve Design Scenarios and Solutions Six un ique conserva t ion scenar ios w e r e run fo r the S O L S area t o genera te a var ie ty of reserve op t ions . T h e s e scenar ios w e r e deve loped in consu l ta t ion w i th Parks C a n a d a t o e x p l o r e di f ferent conse rva t i on values and management op t ions . I c rea ted the di f ferent scenar ios by modi fy ing the status ( locked- in , l ocked-ou t , available) of individual planning units in the Planning Un i t s file p r i o r t o runn ing M a r x a n . Planning units that are ' l ocked - in ' a re inc luded in the final reserve and planning units that are ' l o c k e d -ou t ' are no t inc luded in the final reserve . A s prev ious ly men t i oned , urban areas w e r e l ocked ou t o f all scenar ios . Each scenar io is desc r ibed b e l o w : • Scenar io I: M a r x a n was a l l owed t o c o n s i d e r all p lanning units (excep t urban areas). Th is scenar io was inc luded in o r d e r t o see wha t areas M a r x a n w o u l d p r io r i t i ze if no cons t ra in ts w e r e imposed , and t o help identify parcels o f land that may be a p r io r i t y f o r conse rva t i on par tnersh ips . • Scenar io 2: F i rs t N a t i o n s reserves w e r e l ocked out , t he reby d isa l lowing the i r se lec t ion in the final reserve . A l l o t h e r planning units w e r e available. R e m o v i n g F i rs t N a t i o n s reserves was essent ial because these areas are no t pe rm i t t ed in a nat ional pa rk reserve . • Scenar io 3: S n o w y Moun ta i n P r o t e c t e d A r e a was l o c k e d in, thus fo rc ing its inc lus ion in the final reserve . F i rs t N a t i o n s reserves w e r e l ocked out . Exis t ing p r o t e c t e d areas p rov ide a logical s tar t ing po in t f o r a nat ional pa rk reserve . S n o w y P r o t e c t e d A r e a is par t icu lar ly des i rab le because it conta ins a large c o m p a c t c o r e w i lde rness a rea and the major i ty o f adjacent lands have c o m p l i m e n t a r y - 2 3 -management objectives. • Scenario 4: Sough Okanagan Grasslands Protected Area was locked into the final reserve and First Nations reserves were locked out. Although fragmented, South Okanagan Grasslands contains extensive areas of bunchgrass and ponderosa pone — these biogeoclimatic zones are rich in species that are threatened, endangered or of special concern. • Scenario 5: Both Snowy Mountain Protected Area and South Okanagan Grasslands Protected Area were locked into the reserve, while First Nations reserves were locked out. • Scenario 6: Private property and First Nations reserves were locked out of the final reserve. Parks Canada can only buy private land on a 'willing buyer, willing seller' basis. For this reason, it is of interest to see what reserve configurations are possible when private land is locked out. Each scenario was run with three unique boundary modifiers (1,3, and 6) and two sets of conservation targets (12% and 20%) for a total of 36 unique simulations. For each simulation, Marxan was run 100 times. Because there is a random element in the simulated annealing algorithm, each of the 100 runs will result in a slightly different reserve configuration. Running the process multiple times increases the chances of finding the best reserve configuration. Marxan outputs a file containing the best solution of all runs. The best solution of all runs is the run with the lowest TOTAL SCORE. I mapped the 36 'best' solutions from each simulation using ArcGIS (Fig 2.4-2.7). Figure 2.4 displays the best solutions for scenarios I, 2 and 3, with 20% targets and boundary modifiers of I, 3 and 6. Figure 2.5 displays the best solutions for scenarios 4, 5 and 6, with 20% targets and boundary modifiers of 1, 3 and 6. Figure 2.6 displays the best solutions for scenarios I, 2 and 3, with 12 % targets and boundary modifiers of I, 3 and 6. Figure 2.7 displays the best solutions for scenarios 4, 5 and 6, with 12% targets and boundary modifiers of I, 3 and 6. In Fig 2.4-2.7, the scenarios run from top to bottom and are labeled with SI (Scenario I), S2 (Scenario 2), etc. Going across the page from left to right, the first map shows what planning units are locked in and/or out of the scenario. The second, third and fourth map, show the best solutions as the boundary modifier increases from I, to 3, to 6 respectively. -24-l i l t I 1 I I ( Reserve Design Options 20% Targets I . IflH!HI!{ 4 PIP • OT W KHHIIiq OT w r. HHIHIIil i f i 1* A OT w llfliniH ;« Hi - [{Jtflljllj OT ro OT UllflflJII Figure 2.5:20% Targets — Scenarios 4,5 and 6 s u o p d o u S i s a a e A j a s e n _ L L £ pUE 1' | SOUEU3DS — S13SjEJ_%J | '9'Z 3JnSjJ -81 - Reserve Design Options 12% Targets V> V) C/5 0) 01 4* 9 pUE C > SOUEU9DS — siagjEi%7j :iz a-inSy 2.8 E V A L U A T I O N C R I T E R I A My performance assessment of Marxan is based on six evaluation criteria. The first four criteria address software usability. The next three criteria address the representational and ecological performance of the park configurations derived by Marxan. Usability criteria were derived from literature on decision-support software and in consultation with Parks Canada personnel. Representation and ecological performance measures were based on the ecological goals and objectives of Parks Canada and were derived from the National Parks System Plan (Parks Canada 1997), draft planning documents for the South Okanagan-Lower Similkameen feasibility study (Parks Canada 2005), published articles (e.g. Manseau et al. 2001) and in consultation with key members of the Feasibility Study Working Group. 2.8. / Usability Criteria Criterion I: Marxan should be easy to use. Criterion 2: Marxan should allow users to exert some control over modeling parameters. Criterion 3: Marxan should respond predictably to changes in input parameters. Criterion 4: Marxan should be transparent. 2.8.2 Park Selection Criteria Criterion 5: Marxan should derive solutions that meet coarse-filter representation targets. Criterion 6: Marxan should derive solutions that meet fine-filter representation targets. Criterion 7: Marxan should derive park boundaries with a size and configuration that support the persistence of native species and ecological processes. I assessed Marxan's capabilities and limitations using these criteria. Data used in this assessment were derived from the Marxan spatial analysis, published literature and from personal correspondence with key Parks Canada personnel. -29-3. RESULTS 3.1 Usability Criteria 3.1.1 Criterion I: Marxan should be easy to use. In this thesis, usability refers to the ease with which the user can employ Marxan to design reserves. T o evaluate this criterion I considered five questions: (I) is the USER INTERFACE well designed?; (2) can users easily create input data?; (3) can users easily run Marxan simulations?; (4) can users easily communicate output data?; and (5) is the documentation informative and helpful? I. Is the user interface well designed? The term user interface is used to describe the means by which people interact with an application. Today, most applications have a G R A P H I C A L USER I N T E R F A C E (GUI), which allows the user to interact with the application using windows, graphics, menus and icons (e.g. Microsoft Windows). Marxan has a very limited GUI. Marxan includes a stand-alone windows program called the Import File Editor (inedit.exe) (Fig 3.1), which controls the execution of Marxan (see question 3). Although this program is useful and easy to use, its functionality is limited. The Import File Editor helps users populate one of five input files required to execute Marxan (see section 3.1.4 for an explanation of Marxan input files). The user is left to their own devices to create the other 4 input files, run Marxan and communicate Marxan outputs. Figure 3.1: Marxan's Import File Editor 4fobl™;|Run0ptiond^  its • M iscelaneoua^ sg r Repeat Runs ^ p Boundaries^  Boundary Modifier^ JQ-•Irpi.fae.typej™-™ Q^Tradional Formatted Styles This Program*edits an input file for MARXANvl 2 OeatedbylanBa£1993 Mod1ie1ibyJnBa!l20ra^ ^^ ^^ p^i fpNece ar> Input Filers pr- ** "Namexrs convaLdat Plannng Unt File Name^~ pudata.dat f^ lanran jftoVer us Species j p u v c v t ^ J~\ Block', Definitions ^ -ei-J M^ Bounday Lfingth^  j bound, dat Input Directory j| C\IJ eisMan\Marxan DownloadV j | -30-Another major design flaw is that Marxan is not integrated with GIS software. Consequently, users cannot see the spatial configuration of reserves (or other spatial information) in map form using Marxan alone. Moreover, Marxan does not automate the analysis or exploration of spatial data. For example, it is not possible to select a group of planning units in an output map and see a summary report of the amount of each conservation feature found in the selected units. Overall, Marxan fails to meet this standard — the interface is not well designed. 2. Can users easily create input data? Creating Marxan input files is time-consuming and cumbersome. As previously mentioned, a GUI is provided for only one file — the Input Parameter file. T o create the other four input files, it is necessary to use tools provided by other applications such as ArcGIS, Microsoft Excel, and/or C L U Z (see section 4.2). For example, in order to create the Distribution file — a large file with over 2 million values — I overlaid each conservation feature layer with the planning unit file using ArcGIS and used Microsoft Excel pivot tables to summarize the distribution of each feature in each planning unit. Although the methods I employed are not difficult for users with GIS and database expertise, they would be very difficult for non-technical users. In short, Marxan as a stand-alone application fails to satisfy this usability standard. 3. Can users easily run Marxan simulations? Setting-up and running different simulations is straightforward once the input files have been created. The Input File Editor (Fig 3.1) allows users to set up different scenarios by modifying input parameters. For example, I varied three parameters: (I) the boundary modifier; (2) the name of the Planning Unit file; and (3) the name of the Conservation Feature file in order to run 36 unique simulations. Running Marxan is straightforward using the Marxan executable. It is, however, quite time-consuming. O n average, it took approximately 5.5 hours to run each simulation — a total of 198 processing hours for 36 simulations. The speed of execution depended on the hardware I was using. Because I had four computers running Marxan, and they ranged in processing capabilities, the execution time ranged from 3 hours to 8.5 hours. In all, although Marxan is computationally intensive, it is easy for users to run and therefore satisfies this requirement. 4. Can users easily communicate output data? Marxan generates useful output data that is easily imported into GIS applications. However, it does not -31-provide reporting or mapping tools to convey this information in a readily understandable format. I used C L U Z (see Section 4.2) to import Marxan solutions into ArcView, and ArcView to analyze the spatial characteristics of the data and communicate the results. Collating the results into a format that was intuitive was a time-consuming and tedious process. In short, Marxan fails to satisfy this usability requirement. 5. Is the documentation informative and helpful? A 51 -page user manual details the concepts used in Marxan and the format of the input and output files. Although the manual contains useful information, it is fairly difficult for new users to understand because it uses unfamiliar concepts and terminology. The manual would be significantly more helpful if it included a list of procedures or a step-by-step tutorial explaining how to use Marxan from start to finish. Wi thout this information, users are left to their own resources to figure out how to create the various input files, execute Marxan and display output files. Due to these deficiencies, Marxan does not fulfill this requirement. In all, Marxan as a stand-alone application fails 4 of 5 usability requirements, and therefore does not satisfy criterion I. However, it is critical to mention that Marxan can be interfaced with free, user-friendly GUI-based software such as C L U Z or P . A . N . D . A — resolving many of the usability limitations discussed above (see Section 4.2). 3.1.2 Criterion 2; Marxan should allow users to exert some control over modeling parameters. Being able to exert some control over the parameters (variables, input data) that drive a reserve design model is important in the real world when trying to balance the goals and objectives of multiple stakeholders, explore different management options and cater to unique conservation problems. Overall, the reserve designer exerts significant control over the execution of Marxan. Table 3.1 summarizes the parameters that are defined by the user and provides some examples of how these parameters were used in this study. -32-Table 3.1: User-defined Parameters P a r a m e t e r D e s c r i p t i o n Conservation Features The user defines what conservation features are considered by Marxan. Biogeoclimatic zones, aquatic features, physical relief classes, rock type, rare and endangered species habitats and priority habitats were used in this study. Conservation Feature Targets The user sets the target to be met for each of the conservation features in the reserve. Varying the targets allows the user to explore how different levels of representation influence the area and configuration of the reserve. In this study 1 used two sets of representation targets — 2 0 % and 12%. Conservation Feature Penalty The user defines what priority Marxan places on achieving the conservation feature target. In this study biogeoclimatic zones and aquatic feature were given a high conservation feature penalty, whereas geological features were given a low feature penalty. Planning Unit Size and Shape Prior to using Marxan, it is necessary to divide the study area into planning units. The user determines the size and shape of these units. Planning units can be based on a regular grid, ecological boundaries (e.g. watersheds) or administrative boundaries. In this study 10 ha hexagons were used. Planning Unit Status The user controls what areas are locked in or out of the reserve by altering the status (locked-in, locked-out, available) of individual planning units in the Planning Unit file. In this manner users can explore different conservation and/or management scenarios. In this study, six unique scenarios were explored. Planning Unit 'Cost' The user determines what measure is used to represent the 'cost' of planning units. The 'cost' does not have to reflect the familiar notion of the cost of acquisition; it can reflect any relative measure, including foregone revenue or ecological integrity. Boundary Modifier The user controls the spatial cohesiveness of the reserve using the boundary modifier. Increasing the boundary modifier encourages Marxan to select large contiguous areas rather than small fragmented areas. Because the boundary modifier is data sensitive, testing must be done to derive reasonable values for this parameter. Boundary 'Cost' The user defines what measure is used to represent the boundary 'cost'. The measure can be the length, but it can also be another relative measure, such as neighbor compatibility. The above table demonstrates that Marxan allows the user to exert significant control over the parameters used to drive Marxan. Marxan fully satisfies Criteria 2. 3.1.3 Criterion 3: Marxan should respond predictably to changes in input parameters. I investigated the influence of input parameters on Marxan solutions by varying three parameters: (I) the representation target (20% and 12%); (2) the boundary modifier (three values); and (3) the reserve design scenario (six in total). Figures 2.4-2.7 illustrate Marxan's best solution from each unique combination of parameters. I compared the spatial characteristics of the 36 reserves by calculating -33-the total area of the reserve, the total perimeter of the reserve, the number of unique contiguous patches that make up the reserve and the compactness ratio (Fig 3.2) using ArcGIS software. Table 3.2 displays the summary information generated for each of the reserve options. Figure 3.2: Compactness Ratio The compactness ratio measures the ratio of the boundary length of the reserve to the circumference of a circle with the same area as the reserve (theoretical minimum) (McDonnell et al. 2002, Stewart et al. 2003). „ . Boundary Length Ratio = 1 & -2v 7i x Area Lower values indicate more compact reserves; higher values represent more fragmented reserves. Table 3.2: Summary data for the 36 Marxan Solutions Notes: The first three columns describe the Marxan simulation (see Section 2.7.1). The next four columns describe: the area of the reserve, the perimeter of the reserve, the compactness ratio (see Figure 3.2) and the number of contiguous patches. iSillilllJiliB B o u n d a r y A r e a P e r i m e t e r C o m p a c t n e s s N o . of T a r g e t Scenar io Mod i f ie r (km2) (km) Rat io patches 1 532 480 5.87 19 SI 3 576 336 3.95 10 6 575 352 4.14 11 1 541 437 5.33 18 S2 3 552 356 4.29 11 6 576 340 4.00 9 1 658 448 4.92 20 S3 3 681 346 3.76 11 20% 6 696 350 3.76 10 1 544 509 6.13 28 S4 3 574 356 4.19 13 6 569 343 4.03 11 1 682 427 4.59 17 S5 3 697 354 3.77 8 6 715 332 3.49 9 1 541 959 11.69 > 100 S6 3 566 848 6.74 75 (approx) 6 615 826 9.39 75 (approx) -34-Table 3.2 (Continued): Summary data for the 36 Marxan Solutions Target Scenario Boundary Modifier Area (km2) Perimeter (km) Compactness Ratio No. of patches 1 309 364 5.87 23 SI 3 342 273 4.13 11 6 361 276 4.12 10 1 327 335 5.23 17 S2 3 348 254 3.85 17 6 348 256 3.88 11 1 485 354 4.54 17 S3 3 504 269 3.36 10 6 505 270 3.88 8 12% 1 328 388 6.06 25 S4 3 371 313 4.60 19 6 378 286 4.14 14 1 517 384 4.74 21 S5 3 557 276 3.29 7 6 585 274 3.19 9 1 330 550 8.59 43 S6 3 348 489 7.41 25 6 359 454 6.78 35 Both the spatial and tabular results show that as the representation target increased the total area of the reserve also increased. O n average, the total area needed to represent 20% targets (mean 605 km 2) was 200 k m 2 (49%) more than that needed to represent 12% targets (mean 405 km 2). Marxan required 1.49 (605/405) times more area to represent the 20% targets than it required to represent the 12% targets. Hence, the relationship between area and targets is slightly less than linear (if it were truly linear, Marxan would require 1.67 times more area (20/12)). The mean perimeter needed to represent 20% targets (mean 467 km) was 130 km greater than that needed to represent 12% targets (337 km). Overall, Marxan responds in a predictable manner to changes in representation targets and therefore satisfies the standard for this parameter. Table 3.2 shows the results obtained from Marxan using three different BOUNDARY MODIFIERS (see Section 2.6.4). O n average, changing the boundary modifier from I to 3 resulted in a 20.7% decrease in the total perimeter of the reserve, a 27.4% drop in the compactness ratio, a 42.9% decrease in the number of patches and a 5.6% increase in the total area of the reserve. Increasing the boundary modifier from 3 to 6 resulted in a 2.5% decrease in the total perimeter of the reserve, a 2.7% increase in the compactness ratio, a 3.6% decrease in the number of patches and a 2.7% increase in the total area of the reserve. The unanticipated increase in the mean compactness ratio that occurred when the boundary modifier was changed from 3 to 6 can be attributed to the erratic results of S6 at 20% -35-targets (Fig 2.5). In Scenario 6 several fragmented areas of the study area were locked out of the analysis, forcing Marxan to produce fragmented reserves regardless of the B L M . This apparently led to unpredictable behavior as the boundary modifier was increased. Note that changing the boundary modifier from I to 3 had a significantly larger influence on the spatial arrangement of the reserve than changing it from 3 to 6. Between 3 and 6 Marxan was much less sensitive to the boundary modifier than would be expected if the relationship were linear — the number of patches dropped by only 3.6%. Thus, Marxan responded only somewhat predictably to changes in the boundary modifier. Furthermore, the results became quite unpredictable when the solutions were constrained by a fragmented landscape. Marxan does not satisfy this standard. The major influence of the planning unit scenario on Marxan solutions at equivalent boundary modifiers and representation targets is clearly illustrated in Figures 2.4-2.7 and Table 3.2. Understandably, the size and location of areas locked in and out of the reserve exert significant control over Marxan solutions. Some general trends can be deduced from the spatial and tabular results: (I) running Marxan with fewer constraints (i.e. fewer areas locked in or out) produces smaller reserves (Scenarios I and 2); (2) locking in large compact areas (e.g. Snowy Mountain Protected Area) results in larger, more compact reserves (Scenarios 3 and 5); and (3) locking out fragmented patches of land (e.g. private land, South Okanagan Grasslands Protected Area) results in more fragmented reserves (Scenarios 4 and 6). In all, Marxan responded quite predictably to changes in the planning unit status, and therefore satisfies this requirement. In all, Marxan responded predictably to changes in 2 of 3 input parameters. For this reason, Marxan partially satisfies Criteria 3. 3.1.4 Criterion 4: Marxan should be transparent Transparency refers to how well people understand the decision-making procedures and output products. Transparency is critical to obtaining understanding, trust and buy-in when consulting with planning teams, experts and the public. T o evaluate this criterion, I examined the three components of Marxan — input files, methods, and output files. I then evaluated Marxan against two criteria: (I) are the components of Marxan open and available for examination; and (2) are the details of the decision-making easy to understand. As previously noted, four mandatory input files and one optional file drive Marxan. Together, these files -36-disclose all of the variables that drive Marxan. From a conceptual perspective, the role of four of these files is fairly easy to comprehend. The Planning Unit file (Table 3.3a) specifies the status and cost of each planning unit, the Conservation Feature file (Table 3.3b) identifies conservation features and their representation requirements, the Distribution file (Table 3.3c) contains information on how much of each conservation feature is contained in each planning unit, and the Parameter file controls the execution of Marxan. Less intuitive is the optional Boundary file (Table 3.3d), which works in conjunction with the boundary modifier to control the level of fragmentation in the reserve. Table 3.3: Marxan Input Files Table 3.3a: Planning Unit File This table displays six of 23,836 records from one of the Planning Unit files used in this study. The columns denote the planning unit ID, the 'cost' of the planning unit, and the planning unit status (I-Available, 2- Locked in, 3- Locked out). In this thesis planning unit 'cost' is a relative measure of ecological integrity. Planning units that are more ecologically intact have a lower value than those that are more disturbed. id cost status 1 " 0.1039 I 2 0.1039 I 3 0.2078 I 4 0.2078 I 5 0.2078 I 6 0.2078 I Table 3.3c: Distribution File This table displays six of 23,836 records and five of 88 features from the Distribution file used in this study. The first column identifies the planning unit ID; the remaining columns specify the amount of each conservation feature in each planning unit (matches the ID field in the Conservation Features file). §M • P i 2 3 4 5 1 0 0 0 0.1039 0 2 0 0 0 0.1039 0 3 0 0 0 0.1039 0 4 0 0 0 0.1039 0 5 0 0 0 0.1039 0 6 0 0 0 0.1039 0 Table 3.3b: Conservation Feature File This table displays six of 88 records from one of the Conservation Features files used in this study. The columns indicate the conservation feature ID, the name of the conservation feature, the amount of the conservation feature that will be represented in the final reserve (in km2), and the relative importance of reaching the target (feature penalty factor). id name target 2 Bunch Grass 53.86 5 3 Engelmann Spruce S.A Fir 39.40 5 4 Interior Douglas Fir 100.34 5 5 Montane Spruce 50.14 5 6 Ponderosa Pine 45.56 5 7 Intrusive Rock 133.85 1 Table 3.3d Boundary File This table displays six of approximately 150,000 records in the Boundary file used in this study. The columns denote the left and right planning unit ID and the boundary 'cost' which in this study is a measure of neighbor compatibility. Planning units that share a boundary with a more compatible neighbor have a lower value than those that share a boundary with a less compatible neighbor. i fl id I I d 2 jf^b'o u n d a ry 1 ' I | 600 I 2 200 I 12 200 1 13 200 2 2 400 2 3 200 Marxan uses two interdependent methods to generate reserves: (I) simulated annealing; and (2) -37-the total score calculation 1 0 (See Sections 1.5 and 2.7). Although both of these methods are not inherently intuitive, when explained simply and clearly, it is possible for people to grasp these procedures at a basic conceptual level. Smith (2004b) presents a particularly clear and comprehensible explanation of simulated annealing and the total score calculation. Also, the Reserve Design Game" developed by Rochester and Possingham at the University of Queensland illustrates the process of simulated annealing in a manner that is interactive and easy to understand. Marxan supports the generation of several output files. Like Marxan input files, the output files are reasonably intuitive. The Best Solution file (Table 3.4a) identifies the planning units that are included in the best reserve, the Best Missing Conservation Features file (Table 3.4b) describes the representation performance of the best reserve and the Summary Information file (Table 3.4c) contains summary information for each run. Table 3.4: Marxan Output Files Table 3.4a: Best Solution File Role Example Identifies the planning units that were included in the best reserve. Planning Solution This table displays sample records from a Best Solution file. It is 23766 1 composed of a list of the planning unit ID's which constitute the 23765 1 reserve with the lowest Marxan score. 23764 1 23763 1 23762 1 23761 1 Table 3.4b: Best Missing Conservation Features File Role Example Describes the representation performance of the best reserve. Amount Target Feature Name Target Held Met Wetlands 1.217 1.246 Yes Indefinite River 0.852 0.853 Yes Definite River 4.029 5.207 Yes Indefinite Lake 0.076 0.161 Yes Definite Lake 5.247 5.673 Yes Volcanic 29.427 29.428 Yes Ultramafic 0.089 0.086 No Sedimentary 28.773 31.934 Yes This table displays eight of 88 records in the Best Missing Conservation Feature file produced by Marxan. The columns signify the conservation feature name (from Table 3.2b), the representation target (from Table 3.2b), the amount of the conservation feature captured in the reserve and whether or not the target was met. 1 0 Marxan calls this the 'Objective Function.' " The Reserve Design Game is available online at: http://www.uq.edu.au/%7Euqwroche/resgame/ -38-Table 3.4c: Summary Information File Role Example Contains summary information for each run; the run with the lowest Marxan score is the "best" run. This table displays two records from the Summary Information file. The columns indicate: (1) the run number; (2) the total score for the reserve (see section 2.7); (3) the combined cost of the planning units (see Section 2.6.2); (4) the number of planning units in the reserve; (5) the cost of the reserve boundary; (6) the penalty for missing conservation targets; (7) the combined shortfall from the missing conservation features; and (8) the number of features with unmet targets. In this simulation, Run 55 had the best Marxan score. Run #- Score PU Cost # PU Boundary - Penalty ""'Shortfall Missing 55 1,259 704 2,974 555 0.774 0.0084 4 56 1,373 719 3,064 654 0.148 0.0024 1 Three factors combine to hinder people's ability to understand Marxan. First, most of the terminology used to describe Marxan is not familiar (e.g. boundary modifier, conservation feature penalty factor, simulated annealing). In some instances word usage can lead to confusion. For example, the planning unit 'cost' (Table 3.3a and 3.4c) does not necessarily reflect the familiar notion of the cost of land acquisition, but rather can be measured in different ways. (In this thesis it is a relative measure of naturalness — units that are more natural have a lower value (making them more desirable for Marxan to choose) than those that are more disturbed). Second, the format and sheer volume of some Marxan files make them difficult to assimilate and understand. In a number of cases, maps would be more appropriate forms of communication. For example, a map of the distribution of interior Douglas fir would be easier to understand than looking at over 23,000 values in the Distribution file (Table 3.3c). Third, as noted in Criterion I, Marxan's unsophisticated user interface does not facilitate exploration of Marxan inputs o r solutions. The foremost challenge in understanding Marxan's solutions is that it is difficult to grasp why a particular area is selected. It is not possible to query the solution maps and derive summary information on selected sets of areas using Marxan. In sum, Marxan is transparent in that all information that goes into Marxan and comes out of Marxan is available for examination. Where it fails is that the details of the decision-making are not easily understandable. For this reason, Marxan only partially satisfies Criterion 4. 3.2 Park Selection Cr i ter ia 3.2.1 Criterion 5: Marxan should derive solutions that meet coarse-filter representation targets. The premise here is that in order to protect the terrestrial and aquatic diversity of a region, a -39-reserve design application must derive reserve configurations that achieve coarse-filter representation targets. To evaluate Marxan's coarse-filter representation performance, I examined the information contained in the Best Missing Conservation Features files (Table 3.4b) produced by Marxan. For each of the 36 reserves, I investigated how many of the 61 coarse-filter conservation features met 100% of their targets, which features were underrepresented (these features met less than 100% of their targets) and the proportion of the target met (area reserved for feature/feature target). Table 3.5 summarizes the coarse-filter representation performance of the 36 solutions generated by Marxan. -40-Table 3.5: Coarse-filter Representation Performance for 36 Marxan Solutions Notes: The first three columns describe the Marxan simulation, the next two columns describe the number and percent of coarse-filter conservation targets that were achieved in each simulation and the final three columns describe the missing coarse-filter conservation features, the conservation feature penalty factor of missing features, and the proportion of the target met. BM is short for Boundary Modifier. N o . of T a r g e t s % Of F e a t u r e % of T a r g e t Scenar io B M A c h i e v e d (Out of 61) T a r g e t A c h i e v e d Missing F e a t u r e Pena l ty F a c t o r T a r g e t M e t 1 60 98.4% Relief: 452 3 99% 1 3 60 98.4% Rock: Ultramafic 1 99% 6 60 98.4% Rock: Ultramafic 1 99% 1 60 98.4% Rock: Ultramafic 1 83% 2 3 60 98.4% Relief: 532 1 99% 6 59 96.7% Relief: 421 3 99% Rock: Ultramafic 1 99% 1 61 100.0% Null Null Null 3 3 61 100.0% Null Null Null 20% 6 61 100.0% Null Null Null 1 61 100.0% Null Null Null 4 3 61 100.0% Null Null Null 6 61 100.0% Null Null Null 1 61 100.0% Null Null Null 5 3 61 100.0% Null Null Null 6 61 100.0% Null Null Null 1 60 98.4% Rock: Ultramafic 1 99% 3 61 100.0% Null Null Null 6 59 96.7% Relief: 122 3 99% Rock: Ultramafic 1 99% Relief: 452 3 99% 1 58 95.1% Relief: 532 3 99% 1 Rock: Ultramafic 1 95% 3 60 98.4% Rock: Ultramafic 1 97% 6 60 98.4% Rock: Ultramafic 1 93% Rock: Ultramafic 1 95% 1 57 93.4% Relief: 110 3 99% 2 1 Relief: 451 3 99% Relief: 541 3 99% 3 60 98.4% Rock: Ultramafic 1 93% 6 61 100.0% Null Null Null 1 61 100.0% Null Null Null 12% 3 3 61 100.0% Null Null Null 6 61 100.0% Null Null Null 1 61 100.0% Null Null Null 4 3 61 100.0% Null Null Null 6 61 100.0% Null Null Null 1 61 100.0% Null Null Null 5 3 61 100.0% Null Null Null 6 61 100.0% Null Null Null 1 59 96.7% Rock: Ultramafic 1 97% 6 1 Relief: 510 3 97% 3 60 98.4% Rock: Ultramafic 1 95% 6 60 98.4% Rock: Ultramafic 1 97% Overall Total 2172 2196 % 99 Average 97.1 -41-Overall, Marxan met or exceeded 2,172 of 2,196 (99%) coarse-filter representation targets in its 36 solutions. A l l 61 coarse-filter targets were met in 20 out of 36 solutions (55.6%). Eleven of 36 solutions (30.6%) were missing one coarse-filter conservation feature, three of 36 solutions (8.3%) were missing two coarse-filter conservation features and one of 36 (2.8%) solutions was missing three and four coarse-filter conservation features respectively. O f the 24 underrepresented features, 14 of 24 (58%) features met 99% of their target. Overall, the mean percentage of target met for underrepresented features was 97%. The range was 83 to 99%. Marxan achieved better representation of conservation features with high feature penalty factors (100%), than those with medium (99.38%) or low penalty factors (92.3%). Al l high priority conservation features (biogeoclimatic zones and aquatic features) were fully represented in all solutions. By comparison, medium priority features (physical relief classes) achieved conservation targets 99.4% of the time, and low priority features (rock types) achieved conservation targets 92.3% of the time. Medium priority features achieved a higher percentage of their targets (mean = 99%), than low priority features (mean = 96%). Notably, 14 of 36 Marxan solutions (39%) underrepresented the presence of ultramafic rocks (igneous rocks with very low silica content (less than 45%) dominated by the minerals olivine and pyroxene), a rare but important rock at the Earth's surface which is sometimes home to endemic species adapted to the metallic soils. In total, ultramafic rocks comprised 58% of all missing conservation features. Scenarios where one or more protected areas were locked in (3, 4 and 5) achieved better representation (100%) than scenarios where protected areas were not locked in (I, 2 and 6) (97.8%). However, in Scenarios where protected areas were locked in Marxan reserved a larger portion of the study area. In sum, apart from a small percent of features that were slightly underrepresented, Marxan succeeded in meeting coarse-filter terrestrial and aquatic representation targets and therefore satisfies Criteria 5. 3.2.2 Criterion 6: Marxan should derive solutions that meet fine-filter representation targets. The premise here is that in order to protect exceptional natural phenomena and vulnerable, threatened or endangered habitats, a reserve design application should derive solutions that meet fine-filter representation targets. To assess how well Marxan met fine-filter representation targets, I applied the same methodology used in Criterion 5, except this time I investigated the performance of the 27 fine--42-filter conservation features rather than the 61 coarse-filter conservation features. Table 3.6 summarizes the fine-filter representation performance of the 36 solutions generated by Marxan. Table 3.6: Fine-filter Representation Performance for 36 Marxan Solutions Notes: The first three columns describe the Marxan simulation, the next two columns describe the number and percent of fine-filter conservation targets that were achieved in each simulation and the final three columns describe the missing fine-filter conservation features (see Appendix I for explanation of feature codes), the conservation feature penalty factor of missing features, and the proportion of the target met. T a r g e t Scenar io B M N o . of T a r g e t s A c h i e v e d (Out of 27) % of T a r g e t s A c h i e v e d Missing Features F e a t u r e Pena l ty F a c t o r % of T a r g e t M e t 1 27 100% Null Null Null 1 3 26 96% P.H: W F 2 99% 6 27 100% Null Null Null 1 27 100% Null Null Null 2 3 27 100% Null Null Null 6 27 100% Null Null Null 1 27 100% Null Null Null 3 3 27 100% Null Null Null 6 27 100% Null Null Null 1 27 100% Null Null Null 4 3 27 100% Null Null Null 6 27 100% Null Null Null 20% 1 27 100% Null Null Null 5 3 27 100% Null Null Null 6 27 100% Null Null Null P.H: C D 2 99% 1 23 85% P.H: SN 2 85% 1 P.H: A N 2 84% P.H: PA 2 76% P.H: SN 2 84% 6 3 24 89% P.H:AN 2 84% P.H: PA 2 76% P.H: C D 2 99% 6 23 85% P.H: SN 2 84% P.H: A N 2 84% P.H: PA 2 76% -43-Table 3.6 (Continued): Fine-filter Representation Performance for 36 Marxan Solutions T a r g e t Scenar io B M N o . of T a r g e t s A c h i e v e d (Out of 27) % of T a r g e t s A c h i e v e d Miss ing Fea tures Fea ture Pena l ty Fac tor ?o Of T a r g e t M e t i 1 26 96% TE: W H W O 2 99% i 3 27 100% Null Null Null 6 26 96% PH: SB 2 99% 1 27 100% Null Null . Null 2 3 27 100% Null Null Null 6 27 100% Null Null Null 1 27 100% Null Null Null 3 3 27 100% Null Null Null 12% 6 27 100% Null Null Null 1 27 100% Null Null Null 4 3 27 100% Null Null Null 6 27 100% Null Null Null 1 27 100% Null Null Null 5 3 27 100% Null Null Null 6 27 100% Null Null Null 1 27 100% Null Null Null 6 3 26 96% P.H: PA 2 99% 6 26 96% P.H: WF 2 99% Overall Total 956/972 * % ' &• 98% - -a Average 89% Overall, Marxan achieved 956 of 972 (98%) fine-filter representation targets in its 36 solutions. A l l 27 fine-filter targets were met in 28 out of 36 solutions (77.8%). Five of 36 solutions (13.9%) were missing one fine-filter conservation feature, one of 36 solutions (2.8%) was missing three fine-filter conservation features and two of 36 (5.6%) solutions were missing four fine-filter conservation features. O f the 16 underrepresented features, seven of 16 (44%) features met 99% of their target. Overall, the mean percentage of target met for underrepresented features was 89%. The range was 76 to 99%. Although both priority habitats and threatened, endangered and focal species habitats have the same conservation feature penalty factor (low), Marxan achieved better representation of threatened, endangered and focal species habitats (99.8%) than priority habitats (96.8%). Threatened, rare and endangered species habitats also achieved a higher percentage of their targets (mean = 99%) than priority habitats (mean = 88.5%). The priority habitat Ponderosa Pine - antelope brush - coarse-textured soils comprised 25% of all missing conservation features, Antelope Brush - needle and thread grass - coarse textured soils comprised 18.8% of missing conservation features, as did Big Sagebrush --44-needle-and-thread grass coarse-textured soils. Scenarios where one or more protected areas were locked in (3, 4 and 5) and Scenario 2 achieved better representation (100%) compared to Scenario I (98.1%) and Scenario 6 (92%). However, in Scenario 6 private land was not available for selection in the final reserve. In sum, apart from a small percent of features that were slightly underrepresented, Marxan succeeded in meeting fine-filter representation targets and therefore satisfies Criteria 6. 3.2.3 Criterion 7: Marxan should derive park boundaries with a size and configuration that support the persistence of native species and ecological processes. Marxan provides four mechanisms to incorporate ecological viability into reserve design. The first mechanism, the boundary modifier (see Section 2.6.4.), controls the compactness of the reserve. As documented in Section 3.1.3, increasing the boundary modifier encourages Marxan to select large contiguous areas rather than small fragmented areas. Recall that on average, changing the boundary modifier from I to 3 resulted in a 21% decrease in the total perimeter of the reserve, a 28% drop in the compactness ratio and a 38% decrease in the number of patches. Large, spatially cohesiveness reserves are generally more desirable because they reduce edge effects, facilitate the dispersal and recolonization of empty habitats and allow natural processes to operate at their characteristic temporal and spatial scale (Cabeza and Moilanen 2001; McDonnell et al. 2002). The second mechanism, planning unit status, provides an effective means of locking undesirable areas out of reserves and locking highly desirable areas in. I used this feature to lock urban areas out of all simulations since these areas are not likely to be restored to a natural state. I also used this parameter in Scenario 3 and 5 to lock in Snowy Mountain Protected Area — a large core wilderness area, with a wide range of vegetation and wildlife, a low compactness ratio and critical winter range for California bighorn sheep. With more scientific information, the planning unit status could be used to lock in other areas that are vital to the persistence of native species and ecological processes (such as migration corridors, nesting/breading habitat, wildlife staging areas, pristine watersheds, critical water bodies, etc.). Likewise, areas that threaten the persistence of native species and ecological processes (such as paved roads, transmission corridors, dams, areas of known invasive species, etc.) could be locked out. The third mechanism, incorporating naturalness into the 'cost' of planning units, encourages Marxan to -45-select more pristine areas. For example, in this thesis, planning unit 'cost' was based on the naturalness of the watershed. Units that were more natural were given a lower value, making them more desirable for Marxan to choose. Using this approach, highly disturbed areas are not likely to be selected unless they contain conservation features that cannot be found in other areas. Finally, similar to the approach above, the boundary 'cost' can be used to incorporate a relative measure of neighbor compatibility. Thus, Marxan is encouraged to select planning units with compatible neighbors (e.g. existing protected areas) over those with incompatible neighbors (e.g. urban areas). Not surprisingly, Marxan like other reserve design software, falls short in its attempt to model the persistence of biodiversity. For example, it does not ensure connectivity between individual sites, a critical consideration for the persistence of species. Nor does it deal explicitly with the spatial dynamics of species populations. Moreover, Marxan does not permit the user to enter a minimum patch size for each cluster — a parameter that would help maintain fire and vegetation disturbance regimes, as well as species persistence. Also, using Marxan there is no simple way to constrain solutions to ecologically meaningful boundaries, such as watersheds (unless watersheds are used as the planning unit). Lastly, using Marxan it is difficult to target complete features (such as an entire body of water or a patch of habitat). Despite the above limitations, Marxan's flexible framework provides users with a number of effective mechanisms with which to exercise control over the size, shape and habitat quality of proposed reserves. Used correctly, these mechanisms will help support the persistence of species and ecological processes. In all, Marxan partially fulfills criteria 7. -46-4 . D I S C U S S I O N 4.1 Marxan's Overall Performance In this thesis, I used four usability criteria and three park selection criteria to assess whether Marxan could be a useful decision-support tool for Parks Canada to use when selecting potential park areas. Overall, the evaluation shows that Marxan performed fairly well. Marxan fully satisfied two of the park selection criterion and partially satisfied one. W i t h respect to usability, Marxan fully met two criterion, partially met one and failed to meet one. Fortunately, Marxan's one failure — that it is not easy to use — can be largely resolved (see Section 4.2). Although Marxan did not fully satisfy all six criteria, this research clearly demonstrates that it would be an effective tool for Parks Canada to use in future feasibility studies. 4.2 Usabil ity - Problems and Solutions Marxan is a complex program, and as discussed in section 3.1.1, it is not easy to use — input files are difficult to create, data cannot be easily queried, commands are not automated and output products are not easy to visualize. Fortunately, these limitations can be substantially resolved by using software that is freely available on the Internet. Both Smith (2004) and Riolo (2005) have developed front-ends to M A R X A N called C L U Z 1 2 and P . A . N . D . A 1 3 respectively. Both of these applications interface with GIS software ( C L U Z is compatible with ArcView 3.x and P . A . N . D . A is compatible with ArcGIS 8.x and 9.x). C L U Z provides a user-friendly interface for Marxan and offers tools to create, display, explore, and interactively modify Marxan data. I used C L U Z in this thesis to help automate the creation and population of several input files required by Marxan and to import Marxan solutions into ArcView. C L U Z makes it easy to query the solution maps and derive summary information for a selected set of planning units. A final benefit of C L U Z is that it allows reserve configurations to be interactively modified — users can select one or more planning units and lock these in or out of the reserves. This is extremely useful when fine-tuning the final reserve boundary. Overall, C L U Z resolves many of the interface and usability limitations of Marxan and makes using and understanding Marxan significantly 1 2 http://www.mosaic-conservation.org/cluz/ 1 3 http://www.mappamondogis.it/panda_en.htm -47-easier, especially for non-technical users. That said, C L U Z does not provide a full suite of tools — some additional spatial analysis, map production and report generation will likely be required. Regardless of what additional software program is used to run Marxan, an in-depth technical understanding of Marxan is mandatory. Unfortunately, there is still an outstanding need for a more comprehensive and user-friendly Marxan training manual and tutorial. The existence of these training materials would make Marxan easier to learn and use. It would also make the software more accessible to a wider audience. 4.3 Flexibility — Benefits and Drawbacks The selection and design of protected areas requires tools that are flexible. The results of criterion 2 demonstrate that Marxan is a very flexible reserve design tool. Marxan is able to integrate an unlimited number of layers of data at different spatial scales and levels of biological organization (from species to ecosystems). Moreover, the designer is given control over the input parameters which control the basic rules of the reserve, including: (I) planning unit size and shape; (2) planning unit status; (3) planning unit 'cost'; (4) conservation features; (5) conservation feature targets; (6) conservation feature penalties; (7) boundary modifier; and (8) boundary 'cost'. This flexibility is essential to planners because it facilitates the consideration of different goals and management options. Additionally, it allows design teams to compare and evaluate a range of reserve configurations which meet representation targets. However, proper use of these different parameters requires considerable technical and ecological expertise. Wi thout this expertise, input parameters can be used incorrectly, leading to meaningless or misleading results. Also, inputting a vast array of input data and parameters results in more opaque solutions — it can be difficult to understand why a particular area is chosen. 4.4 Sensitivity Analysis A SENSITIVITY A N A L Y S I S involves varying the input parameters (Table 3.1) in a given model to assess the level of change in the model outputs. A good model is robust, producing consistent and predictable outputs in response to reasonable changes in input parameters. In this study, different input parameters influenced the distribution, area, and compactness of Marxan outputs. Marxan responded predictably to changes in most input parameters. Marxan's response to increases in representation targets was fairly predictable. As expected, the area -48-of the reserve increased almost linearly as the representation target increased (i.e. double the targets and the area of the reserve will double). Interestingly, the relationship was slightly less than linear. This behavior can likely be attributed to an 'economy of scale' — as the reserve grows larger there is more spatial correlation (overlap) of conservation features. Marxan is not particularly sensitive to the boundary modifier. Tripling the boundary modifier from I to 3 reduced the number of clusters by 43% (less than half). Since this is an areal relationship, this is as would be expected. However, Marxan did not respond to changes in the boundary modifier consistently. Between 3 and 6, the influence of the boundary modifier significantly diminished — reserves did not become appreciably more clustered although they did tend to get slightly larger and significantly more 'costly'. In this study, a boundary modifier of 3 achieved a good balance between reserve area and compactness. The influence of the boundary modifier was more unpredictable when the simulations were constrained by a fragmented landscape in Scenario 6. These results demonstrate the importance of thoroughly testing this parameter prior to using Marxan. Marxan performed as would be expected with respect to the conservation feature penalty factor, prioritizing the achievement of higher priority features over lower features. It also responded predictably to incorporating neighbor compatibility into boundary costs and naturalness into planning unit costs. As expected, the reserves tend to cluster around protected areas. However, because I incorporated both of these relative cost values in all simulations it is impossible to know the degree to which each individual factor influenced the results. In the future, it would be useful to investigate the influence of these parameters independent of each other. It would also be useful to investigate the influence of other input parameters including: (I) different values of planning unit 'cost' (e.g. incorporating naturalness versus using area alone); (2) different values of boundary 'cost' (e.g. incorporating neighbor compatibility versus using boundary length); (3) different conservation features (e.g. coarse-filter conservation features versus fine-filter conservation features); and (4) different planning unit size and shape. 4.5 Representation Assessment The results of the coarse and fine-filter representation assessment (section 3.2.1 and 3.2.2) prove Marxan's ability to generate a range of representative reserve configurations that meet stated conservation targets. Apart from a very small percent of features that were slightly underrepresented, Marxan captured the variability of the biological and physical features of the study area and acquired the -49-habitats of rare and endangered species. One of Marxan's greatest assets is that it provides an effective and defensible method to identify a range of representative reserve options — a very useful feature for Parks Canada where regional representation is the cornerstone of park policy. 4.6 Ecological Integrity As mentioned in section 1.6, representation alone will not ensure the long-term success of new national park reserves. Effective reserve design tools must also promote the persistence of native species and ecological and evolutionary processes. Designing ecologically viable reserves is a very challenging task. Fortunately, Marxan provides a number of mechanisms with which to exert control over the ecological integrity of proposed reserves. In this thesis, explicit measures were taken to: (I) protect ecologically intact areas; (2) preserve sites with compatible adjacent land use; (3) create compact reserve configurations; (4) lock in areas that were ecologically desirable; and (5) lock out areas that were undesirable. The SOLS study area has been heavily modified by urbanization, agricultural development, and other human activities. Consequently, it is essential to ensure that Marxan protects areas that are less modified by human activities. Like Lieberknecht et al. (2004), I found that incorporating naturalness into planning unit 'cost' encouraged Marxan to choose more ecologically intact areas to achieve its representation goals. Similarly, incorporating neighbor compatibility into boundary 'cost' encouraged Marxan to choose areas with compatible neighboring land use. These mechanisms had an added benefit — Marxan tended to avoid areas that were important from an economic perspective (such as agricultural lands). In future, it would be useful to investigate the individual influence of each of these mechanisms. Like other researchers (e.g. McDonnell et al. 2002, Leslie et al. 2003, Leiberknecht et al. 2004), I found the boundary modifier to be an effective means to incorporate spatial compactness into reserve design. Representation can be achieved with reserves that have varying levels of fragmentation. However solutions derived with too low a boundary modifier are of limited potential use from an ecological and management perspective, and those with too high a boundary modifier tend to be less efficient in terms of size. In this study, a boundary modifier of 3 appeared to achieve a good balance between reserve area and compactness. Modifying the planning unit status proved to be an effective means to strategically lock undesirable areas -50-out of reserves and highly desirable areas in. I used it to lock out areas that were not likely to be restored to a natural state, and to lock in core wilderness areas. As previously mentioned, with more scientific information, the planning unit status could be used to lock in areas that are vital to the persistence of native species and ecological and evolutionary processes and lock out areas that threaten the persistence of native species and ecological processes. Cowling et al. (2003) explicitly targeted a number of spatial components associated with ecological and evolutionary processes (e.g. gradients, corridors) and forced the inclusion of these areas in the reserve. 4.6. / Using Other Tools in Conjunction with Marxan As previously mentioned, Marxan like other reserve design software, falls short in its attempt to model the persistence of species and ecological and evolutionary processes. One way to overcome this limitation is to use other ecological tools and techniques in conjunction with Marxan. In this study, GIS-based wildlife habitat capability and suitability predictions (Warman et al. 1998, Warman and Hodges (in progress)) were used to target suitable habitats for each of the threatened, endangered and focal species. P O P U L A T I O N VIABILITY A N A L Y S I S (PVA) has been used in conjunction with reserve selection algorithms in order to produce reserves that are more biologically adequate (Noss et al. 2002; Carrol l et al. 2003). Noss et al. (2002) combined P V A models created using P A T C H (Program to Assist in Tracking Critical Habitat) with reserve design software in their multicriteria assessment of sites in the Greater Yellowstone Ecosystem. P A T C H projects temporal changes in populations of terrestrial vertebrate species using habitat maps for an individual population, specifications for habitat use (such as territory size), vital rates (survival and reproduction) and descriptions of a species' movement ability (USEPA 2004). GIS-enabled metapopulation models can also be used in conjunction with reserve design software. Metapopulation models calculate the number of individuals living in a patch, the carrying capacity of a patch, the distance between patches and rates of dispersal and recolonization among patches. This information has been used to prioritize the conservation of specific areas, to evaluate connectivity and other factors necessary for the long-term survival of species and to rank different designs for nature reserves or corr idor networks (Carroll 2003; Applied Biomathematics 2004). Obviously, dynamic spatial modeling tools such as P V A and metapopulation modeling require species-specific population information. This type of data rarely exists for all species of interest. Incorporating ecological and evolutionary processes into reserve design is very difficult in practice because species and processes operate at different spatial and temporal scales (Margules and Pressey 2000). As Cabeza & Moilanen (2001) explain, the explicit consideration of persistence and spatiotemporal dynamics is likely to remain a major future challenge. 4.7 Analysis of Results Although this study was designed as a trial exercise, an analysis of the results yields some useful findings which could be used to inform Parks Canada's concurrent feasibility study being conducted in the SOLS. 4.7.1 General Recommendations Each of the 36 reserve options presented in this study represents the physical and biological diversity of the SOLS study area and the region's special elements. As discussed is Section 3.2.1 and 3.2.2, the majority of these solutions meet virtually all 88 coarse and fine-filter representation targets. A visual scan of the 36 reserve configurations reveals that three distinct core areas — Snowy Protected Area, South Okanagan Grasslands Protected Area and White Lake Grasslands Protected Area — are selected in each and every Marxan solution. A national park reserve that includes these three areas could meet many of Parks Canada's objectives for representation and ecological integrity. These three areas could be considered the most important for immediate protection. 4.7.1 Scenario Based Recommendations The six scenarios that were used in this study reveal how different conservation and management options influence the configuration of reserves in the SOLS. A number of these simulations help inform the current feasibility study. Recall that in Scenario I, Marxan was allowed to consider all planning units. This proved to be an effective means of identifying priority areas for key conservation partnerships at the greater ecosystem level. In all six simulations involving Scenario I, Marxan proposed solutions that crossed Lower Similkameen Indian Band lands connecting Snowy Protected Area with South Okanagan Grasslands Protected Area. This clearly demonstrates the strategic importance of these lands to the conservation -52-of the greater ecosystem. These lands could be considered priority areas for conservation partnerships. In Scenario 3, 4 and 5 different combinations of protected areas were locked into the reserve and First Nations reserves were locked out. These scenarios demonstrate the ecological benefits of building on existing protected areas. O f the 36 simulations considered in this study, Scenario 3 and 5 (Fig 2.4 and 2.5) with 20% targets and boundary modifiers of 3 and 6 most meaningfully address spatial reserve design variables such as size, shape, connectivity, neighbor compatibility and appropriate alignment of boundaries. Each of these four configurations reserves all of Snowy Protected Area, all or the majority of South Okanagan Grasslands Protected Area and a large proportion of Whi te Lake Grasslands Protected Area. These reserve configurations are characterized by: (I) a large, spatially compact, relatively undisturbed core (Snowy Protected Area); (2) a wide range of vegetation and wildlife; (3) a large number of rare, endangered, and focal species; and (4) reasonable compactness ratios (3.49 to 3.77). The Snowy core area is particularly valuable from an ecological integrity perspective because it is fairly well aligned with complete sub-watersheds, and, for the most part, has suitable neighbors with complementary management objectives. However, these configurations do have several ecological shortfalls: (I) the core areas are small in size in relation to large-scale ecological processes; (2) the three core areas are not connected; and (3) several large highways act as barriers to animal migration. In Scenario 6, private property and First Nations reserves were locked out of the final reserve. The results of these simulations imply that when private land is locked out, reserve configurations are highly fragmented. This suggests that extensive land acquisition on a 'willing buyer, willing seller' basis will be necessary to achieve representation and ecological integrity goals. Future research should involve running more scenarios involving the removal of private land. For example, it would be interesting to run Scenario 6 with a boundary modifier of 9 or 12 to see if a higher boundary modifier would encourage Marxan to derive more spatially cohesive configurations for this scenario. It would also be informative to run a new scenario which combines scenario 5 (Snowy Mountain Protected Area and South Okanagan Grasslands Protected Area are locked in) and Scenario 6 (lock out private property). In this way some of the ecological benefits of Scenario 5 could be evaluated in relation to available land. Together, the preliminary reserve configurations act as a good starting point for informing the design of a national park reserve in the SOLS. However, further processing is required to develop a final refined -53-and defensible candidate reserve. 4 .8 Next Steps A number of refinements and steps would be required to fine-tune the reserve configurations proposed in this thesis and derive results that could more meaningfully inform the feasibility study being conducted by Parks Canada. These steps are outlined below. Step I: Have a team of experts (ideally local scientists from both within and outside of Parks Canada) evaluate the reserve configurations generated in this study (you could forward the best 6-10 for evaluation). Solicit feedback on the different reserve configurations (e.g the strengths, weaknesses, opportunities, threats). Gather information on what areas were omitted from the reserves that should be there and what areas are included in the reserves that should not be there. Also, have the experts suggest potential future simulations or refinements of existing simulations. Step 2: Use the input from Step I and from public consultations to refine and re-run more iterations of Marxan. Run entirely new scenarios if need be (e.g. it would be useful to run a new Scenario involving a combination of the constraints in Scenario 5 and Scenario 6). Consider adding more criteria or constraints to existing simulations that are yielding results that are promising (e.g. lock in areas that are desirable or lock out areas that are undesirable). Step 3: Have a team of experts evaluate the results from the second round of Marxan. If necessary, run more Marxan iterations and keep repeating the above steps. When you are satisfied that you have one or more good reserve options, move to Step 4. Step 4: Have the team of experts choose the best reserve configuration from all iterations. Step 5: Fine-tune the boundaries of the reserve using information derived from experts and from the public. The boundaries can be fine-tuned interactively using C L U Z . C L U Z allows individual or groups of planning units to be added or removed from the reserve. Use C L U Z to ensure the park boundaries closely mirror complete watersheds, to increase the compactness and connectivity of the reserve, to exclude sites that are known threats and to exclude undesirable lots of private land. The advantage of using C L U Z is that information on how well the reserve meets representation targets will be interactively updated as the final editing occurs. Fine-tune the reserve until you have a candidate -54-boundary that best meets ecological integrity and representative objectives, and the needs of the public. Step 6: Present the candidate reserve to the public and obtain feedback. Based on the results of the public consultations, further refine the boundaries of the candidate reserve using C L U Z . 4.9 Operational Limitations In doing this study I had a number of operational criticisms of Marxan. These limitations are discussed in this section. There are aspects of Parks Canada's ecological targets that could not be directly or easily incorporated into Marxan. For example, Marxan does not allow users to enter a minimum patch size for each cluster of planning units. This parameter would significantly help in modeling ecological processes (such as fire and vegetation disturbance regimes) and viable species habitats. Marxan does not ensure connectivity between different patches, a consideration that is critical to the long-term persistence of species. Also, using Marxan it is difficult to protect complete ecological features (such as a linear body of water or a patch of habitat). Finally, using Marxan there is no simple way to constrain solutions to ecologically meaningful boundaries, such as watersheds (unless watersheds are used as the planning unit). Marxan is a program that takes considerable time to operationalize. It takes time to understand the software program and the language it employs, to learn how to use the different input parameters appropriately, to figure out how to create the different input files and to interpret the results. The time it takes to get Marxan up and running is exacerbated by the fact that the Marxan user manual does not adequately explain the progression of methods required to use the software to design a reserve. N o r does it include a tutorial or tips and tricks for using the application. The lack of a Marxan tutorial is a significant barrier to learning how to use the software. A great deal of time is required to prepare data for Marxan. A number of the data preparation steps are repetitive and tedious even when using add-on applications such as C L U Z . Some of the input files (e.g. the Distribution file) are not easy to update when newer or more complete data becomes available. Another operational criticism of Marxan is that it is not interfaced with GIS software. GIS software provides the tools needed to prepare data for Marxan, analyze Marxan results, display Marxan outputs as maps and explore and understand Marxan solutions. Because Marxan is not interfaced with GIS -55-software, users must either develop the GIS support infrastructure themselves (as was done in both the re-zoning of the Great Barrier Reef Marine Park and in the Channel Islands) or download software that interfaces Marxan with GIS such as C L U Z or P . A . N . D . A . C L U Z significantly improves the usability of Marxan and resolves many of the interface limitations (see Section 4.2). Unfortunately, C L U Z is based on an older version of ArcView (3.x) and I had to continually switch between the old and new version of ArcView (3.x and 8 or 9.x) as I worked. When Marxan is being executed, it reports information on the progress of the algorithm as text on the screen. This text is very difficult to decipher. It would be easier to understand the methods and parameters that drive Marxan if this progress was interactively displayed on a map. This map could use different colors to represent planning units status. Finally, Marxan produces very crude reports. More sophisticated and intuitive summary reports would make interpreting and presenting the results much easier. 4.10 Advantages of Marxan Marxan supports a systematic and scientifically defensible approach to reserve design. It operates as part of a broad, multi-objective decision-support system and has the computational capacity to solve complex reserve design problems involving large amounts of data in a timely manner. One of the most significant benefits of Marxan is its ability to generate reserve configurations that meet stated conservation targets. In this study, I used Marxan to design and explore 36 reserve options. The majority of these options meet stated targets for coarse and fine-filter representation. For this reason, each of these configurations is defensible with respect to achieving representation goals. Designing representative reserves using manual GIS or pencils and maps would be very difficult. There are simply too many conservation features and parcels of land for the human mind to consider. Marxan provides a flexible environment in which to design protected areas. Marxan can consider simultaneously a broad set of conservation targets at multiple levels of biological organization. Users can also experiment with different conservation and management options (e.g. to include an existing protected area or to exclude native reserves). Marxan has the capacity to incorporate spatial considerations into the reserve design process (e.g. compactness) and includes a number of mechanisms to generate ecologically intact reserves. -56-Marxan produces a range of reserve configurations that meet conservation objectives. With a range of options, it is more likely that a solution will be found that maximizes conservation interests while minimizing negative economic, social or cultural impacts. Marxan also has the flexibility to support participatory planning processes and to help negotiate acceptable outcomes amongst multiple stakeholders. 4.11 Recommendations Marxan is a flexible, defensible and technically proficient reserve design tool. Overall, I believe that Marxan's benefits outweigh its limitations. Parks Canada should consider adopting Marxan as a conservation planning tool. Marxan would allow Parks Canada to design and explore a range of representative, spatially cohesive, and scientifically defensible reserves. This evaluation of Marxan has lead to a series of recommendations, which Parks Canada should consider if they choose to adopt Marxan as a conservation planning tool. 4.1 I.I Invest in training To use Marxan effectively requires an in-depth conceptual and methodological understanding of both Marxan and GIS. Sufficient time should be provided for a staff member(s) with proficiency in GIS to gain familiarity with Marxan terminology, concepts and methods. Because there is currently no single comprehensive source of information about Marxan, it is critical that this person draw on a wide range of sources to learn about Marxan, including the Marxan website14, the CLUZ website15, published journals, and unpublished reports. Liaising with other Marxan users, attending workshops and participating in training programs are also valuable means to acquire knowledge. 4.11.2 Set aside ample time for data compilation and preparation Data compilation, management and preparation are typically the most time-consuming aspects of designing a reserve using Marxan. Ample time is needed to build a comprehensive and inclusive database, assess data quality, identify data gaps, manage the data and prepare the data for Marxan. 1 4 http://www.ecology.uq.edu.au/index.html?page=27710 1 5 http://www.mosaic-conservation.org/cluz/marxan_intro.html -57-Consider using a user-friendly front-end to Marxan, such as CLUZ to help automate and speed up the creation of input files. 4.11.3 Assess the quality of the data Like other reserve design tools, the results Marxan generates are only as good as the data it is asked to consider. It is critical to be cognizant of the quality, completeness and limitations of input data and to be aware of how data limitations can influence outputs. More complete, accurate and high quality data will lead to more sound conservation decisions. 4.11.4 Conduct a thorough sensitivity analysis Understanding how Marxan responds to a range of key parameters (e.g. the boundary modifier, planning unit 'cost', boundary 'cost') takes time and experimentation. Users should conduct a thorough sensitivity analysis to test the influence of input parameters (see Table 3.1) on Marxan outputs. A thorough sensitivity analysis is important because each reserve design problem involves a different study area, and no two problems are alike. In other words, parameters that work for one study will not necessarily translate to another study. 4.11.5 Interface Marxan with freely available software Because Marxan is not interfaced with GIS software, users must either develop the GIS support infrastructure themselves or download software that interfaces Marxan with GIS. Consider using freely available software such as CLUZ or P.A.N.D.A to create, display, query, explore and interactively modify Marxan data. 4.11.6 Use Marxan's flexible architecture to maximize ecological integrity Marxan includes a number of mechanisms to generate ecologically intact reserves. It is critical to take full advantage of these mechanisms. Users should use the boundary modifier to generate more spatially cohesive reserves. The 'cost' of planning units that are more heavily modified by human activities should be increased so that Marxan will, when possible, favor more natural areas. The 'cost' of boundaries that are adjacent to incompatible neighbors should be increased so that Marxan will, when possible, select more desirable neighbors. Finally, the planning unit status should be used to lock in particularly pristine or desirable areas, and to lock out undesirable areas. -58-4.11.7 Use Marxan as a decision-support tool Marxan operates as part of a planning process, and is not designed to act as a stand-alone reserve design solution. Its effectiveness is dependent upon the involvement of people, the adoption of sound ecological principles, the establishment of scientifically defensible conservation goals and targets and the development of useable, accurate and precise datasets. Marxan should be used in collaboration with expert knowledge and other decision-support tools. These other forms of knowledge are essential to the refinement of Marxan inputs, the interpretation of Marxan outcomes and the refinement of final reserve boundaries. 4.12 Conclusion The planning and design of terrestrial national park reserves is becoming increasingly challenging due to competing land use, diminishing budgets and the inherent complexity of ecological systems. Effective and scientifically defensible tools are needed to establish representative and ecologically viable reserves that are socially and economically acceptable. In this thesis, I used four usability criteria and three park selection criteria to evaluate whether Marxan could be a useful decision-support tool for Parks Canada to use when selecting potential park areas. Although Marxan did not fully satisfy all seven criteria, this research clearly demonstrates that it would be a very effective tool for Parks Canada to use in future feasibility studies. Marxan provides a useful means to design and explore a range of representative, spatially cohesive, and scientifically defensible reserves. However, it is critical to stress that using Marxan is not easy; the effective use of this software requires technical and ecological expertise, a comprehensive GIS infrastructure, good data and time. Today, Marxan's momentum in the conservation community is growing. Since the research for this thesis began in the summer of 2004, numerous peer-reviewed articles have been published and the Marxan user base has significantly increased — over 1000 people from 95 countries have downloaded Marxan in the past two years 1 6 . The sophistication and usability of the software itself is also growing. Currently, a new version of Marxan called Marzone is being developed at the University of Queensland. This software is scheduled for release in 2007, and will overcome some of the limitations mentioned in this thesis. These developments are further testaments to Marxan's usefulness and promise. 1 6 Hurley, Karen, pers. communication. -59-R E F E R E N C E S Airame, S., J. E. Dugan, K. D . Lafferty, H . Leslie, D . A . McArdle, and R. Warner . 2003. Applying ecological criteria to marine reserve design: A case study from the California Channel Islands. Ecological Applications 13:S 170-S184. Allison, G . W . , S. D . Gaines, J. Lubchenco, and H . P. Possingham. 2003. Ensuring persistence of marine reserves: Catastrophes require adopting an insurance factor. Ecological Applications 13:S8-S24. Applied Biomathematics. 2004. R A M A S Ecological Software. Available from www.ramas.com/ (accessed September 2004). Ardron , J. A. , J. Lash, and D . Haggarty. 2002. Modeling a network of marine protected areas for the central coast of British Columbia. Ver. 3.1. Living Oceans Society, Sointula, British Columbia, Canada. Ball, I. R., and H . P Possingham. 2000. Marxan (VI.8.2): Marine Reserve Design Using Spatially Explicit Annealing, A manual prepared for the Great Barrier Reef Marine Park Authority. Available from www.ecology.uq.edu.au/marxan.htm (accessed August 2004). B C Ministry of Environment (MOE). 2006. Endangered Species and Ecosystems. Available from www.env.gov.bc.ca/atrisk/red-blue.htm (accessed October 2005). B C Ministry of Forests (MOF). 2005. Glossary of Forestry Terms. Available from www.for.gov.bc.ca/hfd/library/documents/glossary/index.htm (accessed January 2006). Cabeza, M . and A . Moilanen. 2001. Design of reserve networks and the persistence of biodiversity. T R E N D S in Ecology and Evolution 16:242-248. Cabeza, M . 2003. Spatial Population Dynamics in Reserve-Network Design. Academic Dissertation. Department of Ecology and Systematic, University of Helsinki, Finland. Canada Department of Justice. 2000. Canada National Parks Act . Available from http://lois.justice.gc.ca/en/N-14.01 /19271 .html (accessed January 2005). Cannings, R. J., and E. Durance. 1998. Human use of natural resources in the South Okanagan and Lower Similkameen valleys in I. M . Smith, and G.G.E. Scudder, editors. Assessment of species diversity in the Montane Cordillera Ecozone. Burlington: Ecological Monitoring and Assessment Network. Carrol l , C , R. F. Noss, P. C . Paquet, and N . H . Schumaker. 2003. Use of population viability analysis and reserve selection algorithms in regional conservation plans. Ecological Applications 13:S 1773-S1789. Christensen, J. 2004. Second thoughts for a designer of software that aids conservation. The N e w Y o r k Times, September 21, 2004: F2. Committee on the Status of Endangered Wildlife in Canada ( C O S E W I C ) . 2005. Available from www.cosewic.gc.ca/eng/sct5/index_e.cfm (accessed January 2005). Cowling, R. M. , R. L. Pressey, M . Rouget, and A . T. Lombard. 2003. A conservation plan for a global diversity hotspot-the Cape Floristic Region, South Africa. Biological Conservation I 12:191 -216. -60-Environment Canada. 2000. Canada's Endangered Desert Country. The Science and Environment Bulletin. November/December 2000. Available from www.ec.gc.ca/science/sandenov00/article2_e.html (accessed August 2004). Gonzales, E. K., P. Arcese, R. Schulz, and F. L. Bunnell. 2003. Strategic reserve design in the central coast of British Columbia: integrating ecological and industrial goals. Canadian Journal of Forestry Research 33:2129-2140. Groves, C . R., D . B. Jensen, L L. Valutis, K. H . Redford, M . L. Shaffer, J. M . Scott, J. V. Baumgartner, J. V . Higgins, M . W . Beck, and M . G . Anderson. 2002. Planning for Biodiversity Conservation: Putting Conservation Science into Practice. Bioscience 52:499-512. Iverson, K., A . Haney, and M . Sarell. 2005. Updated Antelope-brush Mapping for the South Okanagan Valley. Kirkpatrick, J. B. 1983. A n iterative method for establishing priorities for selection of nature reserves: an example from Tasmania. Biological Conservation 25:127-134. Lea, E. C , R. E. Maxwell, and W . L. Harper. 1991. Biophysical habitat mapping of the South Okanagan study area. Wildlife Work ing Report 17 (draft) B.C. Ministry of Environment, Victoria, British Columbia. Leslie, H. , M . Ruckelshaus, I. R. Ball, S. Andelman, and H . P. Possingham. 2003. Using Siting Algorithms in the Design of Marine Reserve Networks. Ecological Applications I 3:S 185-S198. Leslie, H . 2005. A synthesis of Marine Conservation Planning approaches. Conservation Biology 19:1701-1713. Lewis, A. , S. Siegers, D . Lowe, L. Muller, L. Fernandes, and J. Day. 2003. Use of Spatial Analysis and GIS techniques to Re-Zone the Great Barrier Reef Marine Park. Coastal GIS Workshop, July 7-8 2003, University of Wollongong, Australia. Lieberknecht, L. M. , J. Carwardine, D . W . Connor, M . A . Vincent, S. M . Atkins, and C . M . Lumb. 2004. The Irish Sea Pilot - Report on the identification of nationally important marine areas in the Irish Sea. J N C C report no. 347. Available from www.jncc.gov.uk/irishseapilot (accessed February 2006). Lomolino, M . V. 1994. A n evaluation of alternative strategies for building networks of nature reserves. Biological Conservation 69:243-249. Manseau, M. , F. Rennie, and C . Mondor. 2001. Determination of ecological boundaries for the establishment and management of Canadian National Parks in D . Harman, editor. Crossing boundaries in Park Management: Proceedings of the I Ith Conference on Research and Resource Management in Parks and on Public Lands. The George Wright Society, Hancock, Michigan. Margules, C R . and R. L. Pressey. 2000. Systematic Conservation Planning. Nature 405:243-253. McDonnell , M . D., H . P. Possingham, I. R. Ball, and A . Cousins. 2002. Mathematical methods for spatially cohesive reserve design. Environmental Modeling and Assessment 7:107-1 14. Meidinger, D. , and J. Pojar. compilers and editors. 1991. Ecosystems of British Columbia. British Columbia Ministry of Forests. Special Report Series N o . 6. 330p. -61-Noss, R. F. 1996. Ecosystems as conservation targets. Trends in Ecology and Evolution I 1:351. Noss, R. F. 2001. Beyond Kyoto: Forest Management in a Time of Rapid Climate Change. Conservation Biology 15:578-590. Noss, R .F., C . Carroll , K. Vance-Borland, and G . Wuerthner. 2002. A Multicriteria Assessment of the Irreplaceability and Vulnerability of Sites in the Greater Yellowstone Ecosystem. Conservation Biology 16:895-908. Parks Canada. 1994. Parks Canada Guiding Principles and Operational Policies. Available from www.pc.gc.ca/docs/pc/poli/princip/index_E.asp (accessed September 2004). Parks Canada. 1997. National Parks System Plan. Minister of Supply and Services Canada, Ottawa. Parks Canada. 2002. Parks Canada Agency Corporate Plan 2001 /02 - 2005/06. Department of Canadian Heritage - Parks Canada. Parks Canada. 2005. South Okanagan-Lower Similkameen National Park Feasibility Study: Building a Park Concept: Draft Planning Targets. Available from /www.pc.gc.ca/progs/np-pn/cnpn-cnnp/os-os/itm5-/feu-info/plan-plan/index_e.asp (accessed September 2005). Poiani, K, A. , B. C . Richter, M . G . Anderson, and H . E. Richter. 2000. Biodiversity Conservation at Multiple Scales: Functional Sites, Landscapes, and Networks. Bioscience 50:133-146. Pojar, J., K. Klinka, and D. V. Meidinger. 1987. Biogeoclimatic ecosystem classification in British Columbia. For. Ecol. Manage. 22:119-154. Possingham, H . P., I. R. Ball, and S. Andelman. 2000. Mathematical Methods for Identifying Representative Reserve Networks. Pages 291-305 in S. Ferson and M . Burgman, editors. Quantitative Methods for Conservation Biology. Springer-Verlag, N e w York. Pressey, R. L., and A . O . Nichols. 1989. Efficiency in conservation evaluation: scoring versus iterative approaches. Biological Conservation 50:199-218. Pressey, R. L. 1994. A d Hoc Reservations: Forward or Backward Steps in Developing Representative Reserve Systems? Conservation Biology 8:662-668. Pressey, R. L., H . P. Possingham, and C . R. Margules. 1996. Optimality in reserve selection algorithms: W h e n does it matter and how much. Biological Conservation 96:259-267. Pressey, R. L., H . P. Possingham, and J. R. Day. 1997. Effectiveness of alternative heuristic algorithms for identifying minimum requirements for conservation reserves. Biological Conservation 80:207-219. Prendergast, J., R. M . Quinn, J. H . Lawton, B. C . Eversham and D. W . Gibbons. 1993. Rare species, the coincidence of diversity hotspots and conservation strategies. Nature 365:335-337. Prendergast, J. R., R. M . Quinn, and J. H . Lawton. 1999. The Gaps between Theory and Practice in Selecting Nature Reserves. Conservation Biology 13:484-492. Reed, T. M . 1983. The role of species-area relationships in reserve choice: A British example. -62-Biological Conservation 25:263-271. Resources Inventory Committee (RIC). 2000. Standard for Terrestrial Ecosystem Mapping (TEM) -Digital Data Capture in British Columbia. Ecological Data Committee Ecosystems Work ing Group/Terrestrial Ecosystems Task Force For: Resources Inventory Committee (RIC), Victoria, B.C. Riolo, F. 2005. P . A . N . D . A Protected Areas Network Design Application for ArcGIS. Available from www.mappamondogis.it/panda_en.htm (accessed January 2006). Round River Conservation Studies. 2003. A Conservation Area Design for the Coastal Forest and Mountains of Southeast Alaska and British Columbia (Draft) Available from www.roundr iver .org /CFM_CAD_REPORTS.pdf (accessed July 2004). Schwartz, M . W . 1999. Choosing the Appropriate Scale of Reserves for Conservation. Annual Review of Ecology and Systematics. 30:83-108. Simberloff, D. 1997. Flagships, Umbrellas, and Keystones: Is single-species management passe in the Landscape Era? Biological Conservation. 83:247-257. Smith, R. J. 2004. Conservation Land-Use Zoning ( C L U Z ) . Available from www.mosaic-conservation.org/cluz (accessed January 2006). Smith, R. J. 2004b. Conservation Land-Use Zoning ( C L U Z ) . Available from www.mosaic-conservation.org/cluz/marxanl.html> (accessed January 2006). Stewart, R. R., and Possingham, H . P. 2002. A Framework for Systematic Marine Reserve Design in South Australia: A Case Study. Available from www.ecology.uq.edu.au/links/Stewart_poss_2002_A_framework.pdf (accessed August 2004). Stewart, R. R., T. Noyce and H . P. Possingham. 2003. The opportunity cost of ad-hoc marine reserve design decisions - A n example from South Australia. Marine Ecology Progress Series 253:25-38. South Okanagan-Similkameen Conservation Program (SOSCP). 2000. Conserving Canada's Desert Country: A Prospectus. Environment Canada. South Okanagan-Similkameen Conservation Program (SOSCP). 2004. Recovery Strategy for Species at Risk in the South Okanagan and Lower Similkameen Valleys. Soule, M. , and R. F. Noss. 1998. Rewilding and Biodiversity: Complementary Goals for Continental Conservation. W i l d Earth Fall: 18-28. Soule, M. , and J. Terborgh. 1999. Conserving nature at regional and continental scales-a scientific program for Nor th America. Bioscience 49:809-817. Sullivan, T., M . Gitten, and P. Moskowitz. 1997. Evaluation of Selected Environmental Decision Support software. Available from www.epa.gov/tio/download/char/doedss.pdf (accessed February 2006). United States Environmental Protection Agency (USEPA), Western Ecology Division. 2004. Program to Assist in Tracking Critical Habitat ( P A T C H ) . Available from www.epa.gov/wed/pages/models/patch/patchmain.htm (Accessed September 2004). -63-Warman, L , S. Robertson, A . Haney, and M . Sarell, 1998. Habitat Capability and Suitability Models for 34 Wildlife Species. Ministry of Environment, Lands and Parks, Penticton, B C . Warman, L. D . 2001 Identifying Priority Conservation Areas Using Systematic Reserve Selection and GIS at a Fine Spatial Scale: A Test Case Using Threatened Vertebrate Species in the Okanagan, British Columbia. Academic Thesis. Department of Zoology, University of British Columbia. Warman, L.D., R. E. Sinclair, G . G . E. Scudder, B. Klinkenberg, and R. L. Pressey. 2004 Sensitivity of Systematic Reserve Selection to Decisions about Scale, Biological Data, and Targets: Case Study from Southern British Columbia. Conservation Biology 18:655-666. Warman, L. D . and K.E. Hodges, (in progress) A P P E N D I X I: Conservation Features, Targets and Penalties C a t e g o r y Feature C o d e Feature Penal ty 20% T a r g e t ( k m 1 ) 12% T a r g e t (km*) Biogeoclimatic AT Alpine Tundra 5 13.05 7.83 Representation BG Bunch Grass 5 89.76 53.86 ESSF Engelmann Spruce Sub-alpine Fir 5 65.66 39.40 IDF Interior Douglas Fir 5 167.23 100.34 MS Montane Spruce 5 83.58 50.15 PP Ponderosa Pine 5 75.93 45.56 Geological Intrusive Intrusive 1 223.08 133.85 Representation Metamorphic Metamorphic 1 172.32 103.39 Sedimentary Sedimentary 1 47.95 28.77 Ultramafic Ultramafic 1 0.15 0.09 Volcanic Volcanic 1 49.04 29.43 Aquatic Definite Lake Definite Lake 5 8.75 5.25 Representation Indefinite Lake Indefinite Lake 5 0.13 0.08 Definite River Definite River 5 6.72 4.03 Indefinite River Indefinite River 5 1.42 0.85 Wetlands Wetlands 5 2.03 1.22 Physical Relief 1 10 274-600, Level 3 37.47 22.48 Representation 121 274-600, Gentle, Cool 3 19.48 1 1.69 122 274-600, Gentle, Warm 3 16.78 10.07 131 274-600, Moderate, Cool 3 8.27 4.96 132 274-600, Moderate, Cool 3 6.57 3.94 141 274-600, Steep, Cool 3 3.67 2.20 142 274-600, Steep, Cool 3 2.81 1.68 151 274-600, V. Steep, Cool 3 1.78 1.07 152 274-600, V. Steep, Cool 3 1.25 0.75 210 600-1200, Level 3 5.16 3.10 221 600-1200, Gentle, Cool 3 25.82 15.49 222 600-1200, Gentle, Warm 3 24.99 14.99 231 600-1200, Moderate, Cool 3 25.73 15.44 232 600-1200, Moderate, Cool 3 19.71 11.83 241 1200-1800, Steep, Cool 3 15.19 9.12 242 1200-1800, Steep, Cool 3 11.72 7.03 251 1200-1800, V. Steep, Cool 3 9.72 5.83 252 1200-1800, V. Steep, Cool 3 7.32 4.39 310 1200-1800, Level 3 17.13 10.28 321 1200-1800, Gentle, Cool 3 56.23 33.74 322 1200-1800, Gentle, Warm 3 50.98 30.59 C a t e g o r y Fea ture C o d e Fea ture Penal ty 20% T a r g e t (km 1 ) 12%-'", T a r g e t ' ( km 2 ) Physical Relief Representation 331 1200-1800, Moderate, Cool 3 24.13 14.48 332 1200-1800, Moderate, Cool 3 16.20 9.72 341 1200-1800, Steep, Cool 3 10.64 6.39 342 1200-1800, Steep, Cool 3 6.70 4.02 351 1200-1800, V. Steep, Cool 3 4.62 2.77 352 1200-1800, V. Steep, Cool 3 3.31 1.99 410 1800-2400, Level 3 1.95 1.17 421 1800-2400, Gentle, Cool 3 17.46 10.48 422 1800-2400, Gentle, Warm 3 12.45 7.47 431 1800-2400, Moderate, Cool 3 10.62 6.37 432 1800-2400, Moderate, Cool 3 7.98 4.79 441 1800-2400, Steep, Cool 3 3.98 2.39 442 1800-2400, Steep, Cool 3 2.52 1.51 451 1800-2400, V. Steep, Cool 3 2.16 1.29 452 1800-2400, V. Steep, Cool 3 1.07 0.64 510 2400+, Level 3 0.03 0.02 521 2400+, Gentle, Cool 3 0.26 0.16 522 2400+, Gentle, Warm 3 0.18 0.1 1 531 2400+, Moderate, Cool 3 0.1 1 0.07 532 2400+, Moderate, Cool 3 0.12 0.07 541 2400+, Steep, Cool 3 0.07 0.04 542 2400+, Steep, Cool 3 0.07 0.04 551 2400+, V. Steep, Cool 3 0.11 0.07 552 2400+, V. Steep, Cool 3 0.03 0.02 Representation of A A M T I Tiger Salamander 2 22.24 13.34 Threatened A_SCIN Great Basin Spadefoot Toad 2 19.03 1 1.42 Endangered & Focal Species B_BUOW Burrowing Owl 2 1.59 0.96 B P E F A Peregrine Falcon 2 38.09 22.85 S_SATH Sage Thrasher 2 2.72 1.63 B _ W H W O White-Headed Woodpecker 2 38.84 23.31 B_WSOW Western Screech Owl 2 1.37 0.82 B_YBCH Yellow Breasted Chat 2 0.86 0.51 M_ANPA Pallid Bad 2 18.66 11.20 M_TATA American Badger 2 66.04 39.63 R_CRVI Western Rattlesnake 2 4.10 2.46 R H Y T O Night Snake 2 11.66 6.99 R_PIME Gopher Snake 2 5.37 3.22 M_OVCA California Bighorn Sheep 2 2.57 1.54 Representation of YS Yellow Pine - Saskatoon Fan 2 1.50 0.90 Priority Habitats PB Ponderosa Pine - water birch moist fan 2 0.02 0.01 Category Feature Code Feature Penalty 20% Target (km 2) 12% Target (km 1) Representation of Priority Habitats PA Ponderosa Pine - antelope brush - coarse-textured soil 2 3.63 2.18 WF Bluebunch Wheat-grass- Idaho fescue coarse-textured soil 2 0.87 0.52 AN Antelope Brush - needle and thread grass - coarse-textured soils 2 8.95 5.37 SN Big sagebrush - needle-and-thread grass coarse-textured soils 2 4.76 2.86 BA Barren 2 0.24 0.14 BD Water Birch - red osier dogwood swamp 2 1.03 0.62 CD Black Cottonwood — red osier dogwood floodplain 2 3.20 1.92 SB Silverweed - bulrush meadow 2 0.25 0.15 CT Common Cattail marsh 2 0.26 0.16 OW Shallow Open Water 2 0.55 0.33 PO Pond 2 0.37 0.22 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0092583/manifest

Comment

Related Items