Systematic conservation assessments: before you start

A) Define your planning region

The planning region is your study area and its boundaries are usually based on political or biological factors. Political factors often play an important role because conservation plans are generally implemented by national, provincial or local agencies, which have responsibility for specific regions.

These boundaries, however, rarely coincide with ecological regions and this can be problematic. For example, there may be little point in conserving an area in one country based on its role in an important ecological process, if this is dependent on a threatened area in the neighbouring country. Similarly, problems can arise when setting targets for a particular conservation feature without considering its abundance in neighbouring regions (Erasmus et al., 1999).

Therefore, the ideal situation is to develop transregional conservation plans wherever ecologically coherent regions cross political boundaries, although identifying the boundaries of these ecoregions can be problematic (McDonald et al., 2005).

If this is not possible then compromises can be made by extending the planning region boundaries to include neighbouring areas, to help ensure that potential conservation areas match with existing areas on the other side of the border.

Your first step should be to produce an ArcView shapefile file that defines your planning region boundary based on feedback from your stakeholders and other expert groups.

B) Select which conservation features should be included in the system

Before explaining how to select which conservation features should be included, it is worth stressing that you do not need large amounts of biodiversity data to undertake a conservation assessment. Anyone can carry out a preliminary assessment of their planning region using data that is freely available from the Internet or maps produced through expert review. In addition, assessments based on cost-effective landcover maps can produce very useful results (Smith et al., 2006).

In a perfect world though, your conservation assessment should include data to ensure that all biodiversity is represented. However, mapping every population, species and habitat types, as well as their associated interactions, is impossible, so you will need to identify conservation features that act as surrogates for this broader biodiversity.

The best way of choosing these surrogates is still contentious (Cowling et al., 2004) and is likely to vary between planning regions because of differences in ecology, data availability, funding and capacity. However, work from a number of projects suggests that one good approach is to base the assessment on some type of environmental surrogate map and then supplement this with suitable species, ecological process and other relevant data. More details on these different types of data are listed below:

Environmental surrogates
Environmental surrogate data attempts to capture information that relates to the habitats and ecosystems found in the planning region. This can range from very basic systems, such as using an elevation map to identify broad ecosystem types, to detailed landcover data. These data types are useful because they tend to be relatively unaffected by sampling bias and describe the whole planning region. However, caution is needed to ensure that the classification system has ecological relevance. In addition, it is important to combine data on potential habitats with maps showing where natural areas have been transformed by urbanisation or agriculture. Environmental surrogate maps include:

• Elevation and slope
• Geology types
• Soil types
• Potential vegetation types
• Landcover

Species
Species data are generally used for two distinct types of conservation assessment. The first type uses coarse-scale data from biodiversity atlases to map the distribution of a large number of species belonging to relatively well known taxonomic groups, such as birds and mammals. These assessments are usually based solely on species data and are used to identify large priority areas, within which finer-scale assessments can be undertaken.

The second type of assessments uses finer-scale data to design conservation landscapes. Using species data in these assessments can be a lot more problematic because most fine-scale data are affected by sampling bias (although some excellent datasets do exist). Thus, it is common for fine-scale assessments to select a smaller number of species than are used in coarse-scale exercises and to model their distribution based on species records and expert opinion.

It is also common for fine-scale assessments to include habitat data, so the selected species are often chosen to complement this broader measure of biodiversity and to address representation gaps. Selected species could include:

• Threatened species - globally or regionally threatened
• Range-restricted species
• Keystone species
• Wide-ranging species - eg, large carnivores
• Important sites for key species - eg, roosting or breeding sites
• Economically important species - eg, important for tourism or trophy hunting
• Culturally important species
• Species that have been mapped by previous projects

Ecological processes
It is vital for the long term persistence of biodiversity that ecological processes are maintained, but representing these phenomena spatially can be difficult. There are, however, a range of processes that can be incorporated by using expert mapping or using appropriate surrogates (Pressey et al., 2003), and these include:

• Corridors - eg, for gene dispersal or response to climate change
• Sand dune systems
• Hydrological systems
• Speciation hotspots - eg edaphic interfaces
• Areas large enough for natural fire regimes
• Areas large enough for natural herbivore/carnivore interactions

Other data
Conservation assessments can include data on a range of other important features. Many of these features are associated with particular sites and there may be some situations where only a proportion of these sites can be protected. Examples of such features are:

• Sacred sites - eg, sacred groves and burial sites
• Sites containing archaeological artefacts - eg, cave paintings
• View points - eg, sites with views of attractive scenery.

Other considerations:

• Assessments should include features that are considered important by the stakeholders, as this increases local buy-in and makes it more likely that the final plans are adopted by the relevant implementation agencies.
  
  
• If you wish to re-introduce a species to your planning region, or wish to increase its present population size, then you should map its potential instead of its actual distribution. Similarly, you should map potential distributions for any habitat types that you plan to restore.
  
  
• It is often a good idea to divide large areas of the same habitat type or species range into different zones to make sure that you represent any biogeographic patterns more effectively.

C) Define your planning units

All conservation assessments involve dividing the planning region up into a number of planning units and identifying which planning units are needed to meet the representation targets. MARXAN and CLUZ have been designed so that you can divide your planning region into as many planning units as you want and these units can be any size or shape. Therefore, you should define your planning units based both on the underlying data and how the results will be used by conservation practitioners.

Planning unit shape
Two main approaches are used when deciding planning unit shape. The first approach divides the planning units into real-world units, such as watershed sub-catchments or cadastral boundaries. The second approach uses regular shaped polygons, such as hexagons or grid squares.

• Sub-catchments. Watershed sub-catchments are commonly used in freshwater conservation assessments, partly because these are the relevant management units and partly because freshwater data are often collected at this level.
  
  
• Cadastral and management boundaries. Many conservation plans involve buying land or working with landowners to ensure that their properties are managed to maintain their biodiversity value. In these cases, it can be helpful to base the planning units on cadastral data, so that each unit is a different land-owner's property (Pierce et al., 2005). In addition, some sectors, such as marine fisheries, use large management blocks and in this situation it is better to set the planning units as being the same as these management units (P. Eastwood, pers. comm.).

However, in some cases it is better not to use cadastral data because: a) identifying important properties in draft conservation assessments can cause a range of problems when dealing with land-owners (A. Driver, pers. comm.); b) it may be unnecessary to conserve all of a land-owners property and so using smaller sub-divisions of each property would be more appropriate, and; c) cadastral units are not helpful when dealing with areas under communal management.

• Hexagons and squares. Using hexagonal or square planning units has several advantages. First, the polygons do not correspond to real-world boundaries, making it easier to discuss draft conservation assessments without over-emphasising the importance of individual properties. Second, using planning units with equal areas makes it easier to combine several types of cost value (see below for more details).

The ideal shape of the planning units may depend on how the underlying data are created and how the resultant maps are displayed. Square planning units may be more accurate at representing grid-based data and lose less quality when converted into image files, which also consist of square-shaped pixels. Hexagons have a larger number of edges and so are better for identifying patches of planning units, so should probably used in preference to squares in most situations.

Planning unit size
Using real-world planning units means that the size of each unit is already defined. However, this is not the case for hexagons and squares, which can theoretically be any size. The main limit to using large numbers of smaller units is that MARXAN will take longer to run and CLUZ will take longer to display and process the results. In addition, you should never use planning units with a finer spatial resolution than your conservation feature and cost data.

The advantage of using smaller planning units is that the portfolios are less likely to include unsuitable areas that were only chosen because they fall within the same planning unit as some important conservation features. Moreover, small planning units can be used to divide up larger cadastral units, so that portfolios do not need to include all of a land-owners property. Smaller planning units are also much better at representing fragmented habitats accurately. In addition, maps based on smaller planning units are visually more appealing and easier for stakeholders to interpret.

In MARXAN, planning units are either entirely included or entirely excluded from a portfolio. This can be a problem when your PA and planning unit boundaries do not match, although any inaccuracies can be reduced by using smaller planning units (Araújo, 2004). Alternatively, you can combine the regular planning unit boundaries with the PA boundaries, thus ensuring the accurate measurement of the amount of each feature that is conserved.

D) Decide on what the planning unit cost values should represent

The cost of a planning unit can be based on a range of factors, but MARXAN will always act to minimise this cost in the final portfolio. So, if you set the planning unit costs as being equal to its area, then MARXAN will act to identify the portfolio with the smallest extent. Alternatively, if you set the cost as being equal to the planning unit's financial value then MARXAN will act to identify the portfolio that would be cheapest to buy.

The following is a list of the different cost values that have been used in conservation assessments, or have the potential to be used with slight modification:

• Area
  
  
• Human population density (Luck et al., 2004)
  
  
• Financial value of the land (Pence et al., 2003)
  
  
• Opportunity costs (Stewart and Possingham, 2005)
  
  
• Predicted attitudes to conservation (Winter et al., 2005)
  
  
• Threat (Wilson et al., 2005)

As you can see from this list, a number of different cost values can be used but most people still use area as the measure of cost because it is easily calculated. Such an approach is understandable and perfectly acceptable for conservation assessments that aim to investigate methodological or theoretical issues. However, effective real-world planning exercises often need to include different types of threat and financial data (Wilson et al., 2006).

You might want to include two types of cost in your planning exercise, such as the financial value of each planning unit and its risk of being cleared for agriculture. Unfortunately, the current version of MARXAN only assigns one cost value to each planning unit, so you will need to combine the different cost measures into one metric. This can be done through a range of techniques, which decide how each of the different elements should be weighted and combined. In these situations it is also often better to use equal-sized planning units, as this avoids having to include planning unit area as another different factor that needs to be combined in the cost value.

E) Set your representation targets

Systematic conservation planning is a target driven process (Margules & Pressey, 2000), so MARXAN and CLUZ are designed to identify portfolios that meet all of your representation targets. This makes target setting a very important step but it is also one of the most difficult. The targets should ensure the long-term persistence of your conservation features, so it is important that each is tailored to the requirements of its associated feature. Unfortunately, it can be difficult to find information to help guide this process, so most assessments rely on the following four broad approaches:

1. Expert review. Most target setting exercises rely heavily on expert review. This is because little is known about most conservation features, so targets cannot be entirely data-driven. The best way to improve this situation is by ensuring that a number of experts are consulted and that the experts base their targets on the best available information.
   
   
2. Calculating Minimum Viable Populations. The best way to set targets for a species is to make calculations based on the theory of minimum viable populations, but this requires knowing a number of parameters about its biology and conservation (Cabeza & Moilanen, 2001). Therefore, most assessments only use this approach for a few well-studied species.
   
   
3. Using species-area curves for habitat types. This method has been pioneered in South Africa and uses field data to produce a species-area curve for each habitat type (Desmet & Cowling, 2004). You can then calculate the area of habitat that is needed to represent the required percentage of associated species.
   
   
4. Estimating the planning region's role in a metapopulation. Some wide-ranging species, such as large carnivores, may require an area larger than the planning region. In this situation it is better to view your planning region as part of a metapopulation and to set targets based on the total viable population and how much of that should be represented in your region. For example, you set your target for a vulture species as 40 pairs, based on the need to conserve 800 pairs worldwide and your planning region making up 5% of the vulture's range.

Other things to consider when setting targets :

• Conservation plans generally take years to be implemented and so knowledge about the conservation features is likely to improve over time. This means that their targets may change too, so you need to explain this to your stakeholders to avoid sending out conflicting messages.
  
  
• Target values can be very controversial when assessments identify important sites that have high value for other land-uses, such as agriculture or housing. Therefore, the target setting process has to be very transparent and it should be stressed that research funding may be needed to solve disputes over target values.
  
  
• It is common for stakeholders to want to include certain areas in the final assessment. If the reasons for this are justified then set targets to represent the relevant features or set the area as being conserved at the beginning of the process. Avoid selecting areas in a non-transparent way, eg by modifying the MARXAN parameters until the preferred planning units are included in the portfolio.
  
  
• There is no point in setting a target that is lower than is required to ensure the long-term persistence of a conservation feature. If it is impossible to meet the target for a feature then exclude the feature from the assessment.
  
  
• It can be a good idea to set higher targets for conservation features that are particularly threatened. For example, you might want to set higher targets for valuable medicinal plants, which are more likely to be affected by over-harvesting.
  
  
• In general, it is a bad idea to set targets based on the current distribution of a feature, eg 20% of current forest cover. This target-setting process under-represents features that have been affected by past habitat loss and these are generally the features that most need conserving (A. Lombard, pers. comm.).
  
  
• If you want to re-introduce or increase the range of a species, or restore a degraded habitat, then map their potential distribution and set targets based on the amount that you wish to represent in the future.