Incorporating spatial configuration priorities into conservation incentive programs

Andrew F. Reeson, Kristen J. Williams and Stuart M. Whitten

CSIRO Ecosystem Sciences, GPO Box 1700, Canberra, Australia

Abstract Enhanced landscape function, such as connectivity between conserved sites or complementarity of habitat types, is desirable in most conservation programs. This paper presents a framework for identifying when and how to incorporate landscape function through spatial configuration into the prioritization methods used in conservation incentive programs. To achieve the best results, the ecological benefits of proposed conservation projects should be valued at the landscape-scale, rather than on a site-by-site basis. This can be done either by assessing the marginal contribution of individual projects to the landscape, or by assessing the total ecological value of the landscape under alternative packages of projects. If spatial configuration is valued in the project assessment process, landholders have an incentive to provide it within their projects, for example by offering adjoining parcels of land for a wildlife corridor. The complexity of the landscape-focused assessment approach can be readily tailored to fit a wide variety of purposes and budgets; at its most basic it can employ virtually identical input data to that employed by standard projectfocused approaches. Taking the landscape-scale approach offers an opportunity to more directly target ecological function in incentive programs, increasing the cost effectiveness of public investment and better protecting species and ecological communities in an era of rapid social and environmental change. Keywords: Payments for ecosystem services; connectivity; complementarity; environmental economics; biodiversity conservation; incentives; landscape scale; whole landscape; ecological metric

1. Introduction Incentive payments are a policy tool used to promote the conservation of biodiversity (or the provision of other ecosystem services) on private land. Payments can be fixed, offering all landholders a certain rate per unit of area conserved or service provided, or market-based, for example with landholders submitting bids for conservation funding through a competitive tender. Where multiple private landholders are seeking incentive payments, a program manager needs to discriminate between alternative projects in order to select those which offer the best value for money, thus ensuring the best possible outcomes are obtained with the limited funds available. However, there is often a gap between conservation planning, which takes a broad geographical perspective, and conservation incentive programs, which typically assess sites individually. While

Spatially configuring incentive payments private land is managed on a property (or smaller) scale, ecological communities function across a range of spatial and temporal scales. There is an opportunity to improve the ecological effectiveness of conservation tenders and other incentive programs by taking a more landscapeoriented approach to conservation investment (Schulte et al. 2008, Khanna and Ando 2009, Reeson et al. 2011, Williams et al. 2012). This paper considers options for targeting enhanced spatial configuration within incentive payment programs in order to deliver more resilient ecological populations and communities within functioning landscapes. Spatial configuration is important in biodiversity conservation. Flora and fauna cannot persist in isolation, but rather must be part of viable meta-populations and communities. Connectivity allows the movement of individuals and genes between habitat patches and populations within fragmented landscapes (e.g. Tewksbury et al. 2002). It is functionally important at a range of spatial and temporal scales (e.g. Lindenmayer et al. 2008). In the short term and at the local scale, connectivity may be necessary to ensure that individuals within a population can access sufficient resources for that population to remain viable (e.g. Williams and Kremen 2007). In the medium term local populations need to be part of a viable metapopulation, with the potential for dispersal between populations and colonisation of any patches that become vacant (Macdonald and Johnson 2001). In the longer term species and ecological communities will need to be able to adjust their ranges in response to climatic change and other disturbances (Krosby et al. 2010). Spatial configuration is also relevant to complementarity (Ferrier and Wintle 2009). Conserving complementary patches, for example patches at different altitudes, is likely to protect a greater range of biodiversity than conserving similar patches with the same types of biodiversity. Both connectivity and complementarity considerations mean that the marginal conservation value of a site depends on its spatial and functional relationships to other habitat patches across the landscape (Davis et al. 2006). These values cannot be captured if projects are assessed on a site by site basis. Where connectivity and/or complementarity are important, assessment should be done at the landscape-scale (Ferrier and Drielsma 2010). Projects offered in an incentive program represent management units, rather than ecological units. In contrast, ecological function needs to be assessed at the habitat patch and landscape scales, rather than individual property or field scales (Wiens et al. 1993). This paper outlines ecological principles and practical considerations for proactively targeting desired spatial configurations as a component of landscape function within conservation incentive programs. For this context, the most important spatial configuration is likely to be ecological connectivity, reflecting the interconnectedness of ecosystems and habitats at different scales, especially in fragmented landscapes (Bennett 2003). Achieving complementarity among species and ecosystems through adequate representation among conservation sites and protected areas is also likely to be important in some cases (Margules and Pressey 2000). A framework is presented for determining when, and how, to target spatial configuration in conservation incentive programs. 2. Valuing landscapes To make informed decisions in a conservation program it is necessary to consider the benefits and costs of alternative options. In a growing number of conservation incentive programs this process is formalised through the use of a metric which quantifies the benefits offered by proposed conservation projects into some form of biodiversity units (e.g. Parkes et al. 2003). This is a not a straightforward process, but it does allow the benefits of alternative projects to be directly

Spatially configuring incentive payments compared, enabling a decision-maker to preferentially select those projects which offer the best returns per dollar invested. While no metric can be perfect, it is essential to have one which can adequately discriminate between alternative projects; if it does not a program is unlikely to succeed in efficiently allocating scarce conservation resources. In order to value spatial configuration it is necessary to incorporate it into a metric. This will inevitably add to the already complex nature of metric design. However, if it is not valued it is unlikely to be delivered. Biodiversity metrics may take a single species or community approach. Ideally a metric should focus on the desired outcome of enhanced ecological function. Where that is not possible (as is likely in most cases), a good proxy is required. For example, while structural (as seen on a map) connectivity, is the most amenable to measurement, consideration should be given each time as to whether this is the most appropriate connectivity measure for the particular context. For well known species or systems there may be greater potential to consider functional (ecological) connectivity (e.g., as demonstrated byDancose et al. 2011 for bison). The nature of the matrix (the areas between habitat patches, often under agriculture), and the characteristics of the species or communities in question determine functional connectivity. Most species can cross the matrix, to a greater or lesser extent. Management actions to make the matrix more permeable, for example by establishing or maintaining trees within farmland, can increase connectivity and hence the ecological function of a landscape. Matrix management can also enhance the condition or effective area of a habitat patch, for example by mitigating threatening processes and reducing edge effects, effectively increasing matrix permeability. Conservation programs should therefore not confine themselves to management actions within existing habitat patches. Assessing the functional value of a landscape for its connectivity or complementarity will require good data. Conservation programs which assess proposed projects on a site by site basis can rely on information provided by landholders (either within their proposals or through a site inspection). To assess the contribution of a site to the landscape, it is necessary to also have a reasonable understanding of the location and condition of other habitat patches, as well as any salient features of the surrounding matrix. For example, to assess complementarity, a metric will need to consider the beta-diversity of the landscape, rather than just the alpha-diversity of the patch (for definitions of alpha and beta diversity see, for example, Jost 2007). In some situations the availability of such data and its level of detail may in practice limit attempts to estimate fine-scaled landscape connectivity or complementarity values. 2.1. How much detail to include in a metric? The ultimate metric would assemble all aspects of landscape ecological function that contribute to biodiversity persistence into a single numeric measure. Of course in many cases this is simply not possible or desirable, given the need for accurate scientific information. Conservation programs could spend their entire budgets developing a metric without fully resolving all the issues. It is therefore always necessary to consider the benefits of developing an improved metric to better allocate funds among alternative projects, against the costs, which may reduce funds available for on-ground management. In addition to the financial costs, there are also methodological concerns with increasing the complexity of metrics to account for more features of the project in question (Hajkowicz et al. 2007, Hajkowicz et al. 2008). While simple metrics make assumptions about the system being assessed, more complex metrics are prone to error in parameters or system

Spatially configuring incentive payments representation. Complex metrics also become less transparent, which makes them harder to communicate to landholders and field officers and increases the risk of errors in implementation. The purpose of a metric is to combine various ecological attributes into a single numeric score, so trade-offs between different metric attributes will always be inherent. If the ecological metric is designed empirically rather than mechanistically, additional attributes will inevitably dilute the weighting of existing attributes. For example, consider a very simple metric which only incorporates habitat extent (i.e. area) and a separate condition score for each site. If an additional term is added for connectivity (e.g. a link score), and all scores are equally weighted, extent and condition will each now carry less weight overall. A project which offers less extent and/or condition may be ranked ahead of another by virtue of a higher connectivity score. The program can only deliver increased connectivity (as measured by the metric) at the expense of extent and/or condition of target habitat. (If connectivity is functionally correlated with extent and/or condition, for example through landscape permeability, then they need not be traded off against each other, but there is also little to be gained from then also incorporating connectivity into the metric.) A parsimony principle should be applied to metric design. Metrics will always be imperfect, both because the required scientific knowledge is often lacking or untested in mechanistic models, and because they are attempting to describe a complex ecological system with a single value. Some attributes, such as habitat extent, can be measured with certainty, while others, such as ecological condition, can only be estimated or calibrated using information from site assessments. Connectivity or complementarity should only be included within an empirical metric if it is considered sufficiently important to sacrifice these other attributes (extent and condition) to some degree. Alternatively, these can be included together in a mechanistically-designed metric in which extent and condition are inherent measures of effective habitat area used in the calculation of connectivity and complementarity values (e.g. Williams et al. 2012). Such attributes of landscape function should also only be included if their relative value can be reasonably estimated and the costs of doing so (i.e. developing ecological models, obtaining spatial data, etc.) are outweighed by the benefits. The substitutability between habitat extent (or number of patches) and connectivity will depend in part on the current status of the landscape (in relation to the degree of habitat loss and fragmentation) and the dispersal abilities of the species or assemblage of species in question. For more mobile species, increased connectivity can compensate to some degree for habitat loss, while for less mobile species it is more important to retain existing patches (Drechsler 2011, Johst et al. 2011). The habitat patch is the fundamental unit of landscape ecology. Even where connectivity and complementarity are of great importance, it is likely that the first priority for a metric will still be the viability of biota within an individual patch. The characteristics of the habitat patchsuch as size, shape, condition and isolationhave a direct relationship with viable populations or communities (Fahrig 2003, Williams and Kremen 2007, Magrach et al. 2011). What constitutes a viable patch, in terms of extent and connectivity, will vary between species and communities (e.g. as in the case of dispersal limited birds, Castellon and Sieving 2007). If a program has a small budget relative to the geographical area it covers, it is unlikely to be able to contribute significantly to connectivity and so will do better to focus on protecting key environmental assets. Larger programs or those focussed on particular areas are better placed to address enhanced landscape function through spatial configuration.

Spatially configuring incentive payments 2.2. Patches as ecological units, projects as management units The patch is an ecological unit which provides a basis for assessing conservation outcomes at both the local and landscape scale. A patch is not the same as the area covered by a proposed project. Conservation projects are typically constrained by property boundaries, while a single patch may cover multiple properties (see Schulte et al. 2008). Therefore in any conservation metric, and particularly those considering connectivity and complementarity, it is essential to distinguish between patches (as an ecological unit) and projects (as a management unit). A patch may be impacted by more than one project, and similarly (though less likely) a project may cover more than one patch. The value of a patch of habitat to overall biodiversity at the landscape scale is likely to increase in a non-linear manner with area (Jackson and Fahrig 2012). Larger patches are more likely to support viable populations. The smallest patches are likely to contribute little to population persistence (Fahrig 2003). Beyond a certain minimum size, value increases rapidly according to the species-area relationship as a patch becomes more biodiverse. It will continue to increase in value, but at a decreasing rate, as patch size gets larger. Effective patch size is determined by the total area of vegetation and the functional connectivity within it (Shanahan et al. 2011). Edgeeffects mean that the shape of a patch is also important, along with the characteristics of the surrounding matrix (Lidicker 1999, Laurance et al. 2007, Koh et al. 2010, Prevedello and Vieira 2010, Magrach et al. 2011). For species or communities which are negatively impacted by edge effects, the effective patch size will be smaller than the actual patch size. Such edge effects may be mitigated by creating a buffer around the patch, so as to increase the core area of the patch to some minimum viable level for a focal species. The effects of patch size will of course vary between species and taxa (Jackson and Fahrig 2012) what is small for a mammal may not be for an invertebrate (both because they require fewer resources and the available habitat is greater due to fractal geometry). Patch condition or habitat quality may be impacted by on-site or off-site threatening processes (e.g. Minor and Gardner 2011, Brudvig et al. 2012). For example, off-site processes reducing fertiliser runoff or erosion from surrounding (or upstream) fields may enhance the condition of a wetland through positive effects on water quality. Any project will have an impact on a patch, either directly (by altering extent) and/or indirectly (by increasing effective area or enhancing habitat condition). The value of these off-site management actions can be measured in terms of their marginal contribution to the condition or extent of the patch or directly on the populations occupying the patch (Baeza and Estades 2010). This conceptual framework recognises the contributions to biodiversity persistence of off-site actions (including those on neighbouring properties) alongside on-site actions. It provides a more ecologically realistic approach to valuing and selecting among alternative projects. While a patch is defined in landscape ecology as a relatively homogeneous area of habitat, more mobile species are not bound by patches. An area of functionally, but not structurally, connected habitat may be considered as a meta-patch. Linking patches by making the matrix more permeable may be a suitable management action to increase the effective area of a meta-patch in order to link otherwise isolated populations of a species in a meta-population. In other cases the contribution to a meta-patch may be counted proportionately. For example, isolated trees or small areas of remnant vegetation between habitat patches may be counted proportionately in the meta-patch area (e.g. 10% cover of scattered trees equated to habitat area). Exactly what constitutes a meta-patch

Spatially configuring incentive payments will vary between species and communities, as some species move across open areas between small parcels of remnant vegetation, while others require contiguous understorey. Some mobile bird species utilise very small areas of remnant vegetation (Fischer and Lindenmayer 2002, Collard et al. 2009). For the birds these patches may be considered as part of a larger functionally, though not structurally, connected meta-patch. Their value will depend on being part of a linked mosaic of patches, and so can only be meaningfully valued within a landscape context (Anand et al. 2008). Similarly, potential for pollen and seed movement means that individuals within plant communities can be functionally connected even in structurally fragmented landscapes. 2.3. Patches and landscapes Landscapes are relevant to the meta-population and meta-community scale of conservation assessment. Landscapes comprise a series of more-or-less interconnected patches, and must support viable populations or communities in the medium term (i.e. decades). The value of a landscape may be considered as the sum of the biodiversity complementarity values of its component patches along with a measure of connectivity within and among patches. Connectivity will be a function of distance and permeability between patches (either nearest neighbours or all pairs), weighted by the relative condition of patches incorporating edge-effects associated with the surrounding land use matrix. Any such metric must be constructed in such a way that large, nonfragmented patches score more highly for biodiversity persistence than a series of fragmented patches with similar area (Saura and Pascual-Hortal 2007). Landscapes are likely to have thresholds, below which populations of some species are unviable (Dykstra 2004). Individual management actions, both within and between patches, will impact even if marginally on the overall value of the landscape for the maintenance of its constituent biodiversity. 3. Incorporating landscape values into metrics The relative importance of spatial configuration will vary between landscapes, depending on the degree of fragmentation and the dispersal attributes of the species constituting the ecological communities in question. Where it is considered desirable and feasible to include in a metric, the next question is how to do so. 3.1. Project-focussed metrics To date, most ecological metrics have been based around attributes of individual conservation projects (e.g. Juutinen et al. 2009, Khanna and Ando 2009). These project-focussed metrics typically generate scores by combining values for a number of different attributes of the site, which are then used to rank alternative projects in terms of value for money (i.e. their environmental benefits offered per dollar invested). The most obvious way to include scores for spatial configurations such as connectivity and complementarity is through an additional attribute within a project-focussed metric (e.g. Parkes et al. 2003). Including additional scores for connectivity and/or complementarity alongside other attributes of a project does not require any major changes to the existing structure of the metrics used in conservation programs. A key question is how much weight the connectivity score should have relative to other aspects of the metric. This is a complex issue which does not have a definitive answer, so is likely to be based on expert judgement or policy considerations as relevant to the purpose of the conservation program. For example, if the connectivity weighting is relatively small, then it will only serve to discriminate between projects with similar overall scores from other attributes of the metric. The higher the connectivity weighting, the more likely it is that bids which are well connected, but 6

Spatially configuring incentive payments otherwise poor with respect to other metric attribute values, will be selected. Similarly for complementarity, projects may be assigned a bonus score based on their relative scarcity in the region or how well they are represented among other protected areas. Again, the weighting of the bonus score should depend on its relative importance compared to other factors. The more attributes that are additively considered within a metric to represent the objectives of an incentive program, the less the impact any single one is likely to have. 3.2. Landscape-focussed metrics While expert judgement and policy objectives can be used to guide the relative weights assigned to the different components of a project-focussed metric, it is a largely subjective decision. As discussed above, the habitat units in which ecological values are best expressed (i.e. patches and landscapes) do not necessarily correspond to the management units in which conservation projects are carried out by landholders (i.e. properties or parts of properties). Project-focussed metrics assess individual sites as though they were distinct patches. While this makes for relatively simple and tractable metrics, it does not adequately represent landscape function and is largely incapable of accommodating off-site management actions. Ecological units will often cross property boundaries, while project management units are more likely to be restricted to individual properties or parts of properties, except where coordination across properties is an project explicit goal (Windle et al. 2009). Connectivity (beyond the patch scale) and complementarity are not properties of individual projects. Attempting to value them on a site by site basis will therefore never be fully satisfactory this can be done far better with landscape-focussed metrics. A key test of a metric is how it copes if individual projects are split or combined. Flexibility in project design is often encouraged through multiple bids over different habitat patches or management actions representing different dollar values (Windle et al. 2009). A project-focussed metric will value part of a patch inconsistently with the whole of the patch. A landscape-focussed metric should give consistent scores regardless of whether projects are split or lumped. By distinguishing between management projects and habitat patches it becomes possible to value conservation activities beyond the site boundary and over time; for example by increasing connectivity or reducing threats to habitat patches. Such matrix management does tend to be neglected in existing conservation programs, despite its importance in a range of ecological processes (e.g. Attwood et al. 2009). Moving from project-focussed to landscape-focussed metrics can allow a far greater range of valuable management actions to be incentivised. It recognises that management actions do not occur at the same scale as ecological outcomes and allows each to be assessed at the most appropriate scale. This also provides a better framework for evaluating and reporting the effectiveness of programs. Landscape-focussed approaches deal properly with non-additive interactions between management actions (due, for example, to complementarity and connectivity) in calculating the overall benefits, whereas project-focussed approaches incorrectly assume individual scores are additive. Another advantage of this landscape-scale approach is that beneficial management actions on one property can be recognised for their contributions to habitat patches on neighbouring properties. There are at least two methods for selecting among projects based on total landscape values. One method considers the marginal contribution of a project to the total landscape value. An alternative method is to select the package of projects with the highest overall landscape value.

Spatially configuring incentive payments 3.3. Marginal contribution to total landscape value The most straightforward approach to implement a program with a landscape-scale metric is to consider the marginal contribution of each project to landscape ecological function. Changes in some patches will have a greater impact on landscape function and metapopulation viability than others, dependent on factors such as the size and condition of the patch, and its proximity to other patches (Home et al. 2009). Rather than considering each site as a distinct patch, this approach can value the contribution of a project to the overall landscape. For example, where a site represents part of a functional patch (e.g. because the patch extends onto neighbouring properties), it would be assessed on its contribution to that patch, and the landscape beyond, rather than being assessed as though the area covered by the project were an island in an ocean. This approach requires the estimation of total landscape ecological values with and without the project in question (Seddon et al. 2010). Subtracting one from the other provides the marginal value of that project to the landscape. This could then be divided by the asking price in order to rank projects in the same way as is done for conventional site-focussed metrics (e.g. Parkes et al. 2003). Project assessment could be done on a stepwise basis, with the best value (greatest marginal benefit to existing landscape/asking price) accepted first. This site could then be added to the existing landscape in order to work out the marginal value of other projects. (Barton et al. 2009) describe an algorithm (TARGET) for iteratively selecting conservation locations which provide the highest level of complementarity to an existing set of conserved sites. TARGET selects sites if their landscape benefits exceed the opportunity cost of conserving the site, though this requires the trade-off between cost and regional biodiversity complementarity (effectively a price for biodiversity outcomes) to be defined (Faith and Walker 1996). This approach could be modified to select the most cost effective project, in terms of the ratio of contribution to landscape biodiversity value to asking price. Landscape-scale ecological benefits can be estimated using one of a number of existing theoretical frameworks. (Ovaskainen and Hanski 2003) derive measures for assessing the value of an individual habitat fragment for the dynamics and persistence of a metapopulation living in a network of many fragments. The value of a fragment depends both on the properties of the landscape and the species (e.g. the fine scale movement by forest birds, Gillies et al. 2011). Metapopulation capacity characterises the ability of a fragmented landscape to support long-term persistence of a species (Hanski and Ovaskainen 2000). It looks at patch size (carrying capacity, which may also refer to quality) and connectivity. Metapopulation capacity can be used to rank different landscapes in terms of their capacity to support viable metapopulations (Hanski and Ovaskainen 2000). It can also calculate how the metapopulation capacity is changed by removing habitat fragments from or adding new ones into specific spatial locations, or by changing their areas (Drielsma and Ferrier 2009). A potential extension to this is landscape equivalency analysis, which measures the services provided by patches in fragmented habitats from a metapopulation perspective, taking into account genetic variability as well as population distribution and abundance (Bruggeman et al. 2005). Spatial configuration impacts on genetic variance within a metapopulation; landscapes which promote gene flow (e.g. through connectivity) are more likely to support viable populations as inbreeding is reduced. Landscape equivalency analysis expresses the conservation value of a patch

Spatially configuring incentive payments (or project) as its marginal contribution to metapopulation sustainability (Bruggeman et al. 2005). Changes to a landscape which result in rates of recruitment and migration that more closely reflect those that would occur in an unfragmented landscape are considered beneficial (Bruggeman et al. 2009). The probability of connectivity index (PC) (Saura and Pascual-Hortal 2007) is based on the probability that two individuals within a landscape occupy habitat patches that are interconnected. It also considers the carrying capacity of the patch (although this may be omitted). Total landscape connectivity can be calculated. Landscape elements (i.e. patches) can then be ranked according to their percentage contribution to total connectivity. This approach combines patch area (/quality) with connectivity. If permeability varies according to the nature of the intervening matrix, the full PC index may be worthwhile, otherwise a simpler binary version (Pascual-Hortal and Saura 2006) would suffice. This index measures connectivity at the landscape level rather than at the patch level, which is the more appropriate measure. It can demonstrate the marginal impact of a patch (along with the impact of the loss of some or all of a patch). 3.4. Total landscape package value Estimating the marginal value of a site to the landscape provides an excellent way of valuing a single project. It may become somewhat less satisfactory where multiple projects are to be valued simultaneously. The value of a patch and its landscape may depend on one or more projects, whose effects may not be additive. To value the patch it may be necessary to consider the various possible combinations of projects. In this case the marginal value of any one project will depend on which others go ahead. Rather than valuing and selecting individual projects, a more rigorous approach is to select the package of projects which together offer the best overall outcome within a budget constraint (e.g. Williams et al. 2012). This allows connectivity and complementarity to be valued holistically for a package of projects within their landscape context rather than on a site by site basis. The package approach is consistent with the objective of selecting the best value set of projects for conservation management (Faith et al. 1996). However, in such a metric, individual projects are not scored or ranked against one another. Rather the program will select the best possible package of projects which fits within a given budget; it is effectively selecting an overall configuration of projects, rather than a series of individual projects. It is also possible to identify the optimal packages which can fit under a range of different overall budgets, to create a cost curve for total biodiversity benefits against total budget. This could be used to guide policy makers in deciding how much to invest in that particular program. The package value approach provides a comprehensive solution to addressing spatial configuration at landscape and regional scales. It still requires appropriate levels of information and understanding in order to value alternative landscape configurations. It is also computationally challenging. Ideally it requires calculating a biodiversity value for each possible package of projects within the budget constraint. The number of combinations rises rapidly with the number of projects. For 30 projects there are one billion possible combinations; for 40 there are one trillion. As the number of projects increases, the number of calculations to value every possible package rises beyond the capability of even the most powerful computers. There are heuristics (computational short-cuts) such as genetic algorithms (e.g. Moilanen and Cabeza 2002) and simulated annealing (e.g. McDonnell et al. 2002) which can be applied to make these problems

Spatially configuring incentive payments manageable. Simulated annealing, for example, has been extensively used with an objective function incorporating achievement of biodiversity targets and spatial compactness while minimising socio-economic costs to prioritise areas for conservation planning (see review by (Sarkar et al. 2006). It is feasible to develop and implement ecological metrics which assess packages of projects at the landscape scale. However, the process is inevitably more complex and costly than uncoordinated alternatives. With further research these metrics should become better understood and easier to apply. In the meantime this package approach is likely to be required only where landscape ecological function expressed through spatial configuration, for example, is of paramount importance to environmental outcomes, or where the values of individual projects are highly interdependent. It will be particularly appropriate where there is potential for individual projects to have a major impact on the outcomes of biodiversity persistence at the landscape or regional scale. A drawback of the package approach is that the process may be compromised if one or more projects drop out of the process before contracting or implementation. If some projects in the preferred package are lost, its overall value may be significantly compromised. A form of sensitivity analysis could be carried out to identify any individual projects whose loss would significantly reduce the overall value of the selected package these landholders could be contracted first, allowing the selection process to be re-run if one or more drop out. Alternatively a package of projects could be identified which is more robust to the loss of any particular elements. A resilient landscape will require a series of large patches and/or metapopulations, hence dependency on individual sites should be minimised across the landscape as a whole. Another drawback with the package approach is that the transparency of the project selection process is reduced. Once projects are assessed collectively, it is no longer so clear why some are in the preferred package and others are not. Some successful projects may appear to offer poorer value than some unsuccessful projects if considered on their own. This issue would need to be addressed through follow-up communication with landholders. A package metric can demonstrate cumulative landscape-scale outcomes from conservation investment where the whole is greater than the sum of the parts. It can do the same for the cumulative impacts of habitat degradation. Such an approach may also be considered to avoid the death by a thousand cuts which can result from incremental, uncoordinated planning decisions. 4. Metric selection Figure 1 shows how these different metrics would work in practice. A framework is needed to evaluate what sort of metric should be applied to any particular policy program. This depends on the policy objective, available knowledge and data, and the relative importance of spatial configuration in the context of species, assemblages and landscapes. As discussed above, the inclusion of spatial features such as connectivity and complementarity to represent landscape function should only be done if it adds to the policy objective, since the more attributes that are considered in an empirical metric the less relative weight each one will have. This is less of a problem for metrics that are designed with mechanistic process of biodiversity persistence in mind (e.g. Ferrier and Drielsma 2010). Table 1 compares the different approaches.

10

Spatially configuring incentive payments Figure 1: Metric options

H A B G C I
A hypothetical fragmented landscape across nine properties (grid boundaries). Here we assume condition and biodiversity composition are the same in all remnant habitat patches (shaded areas). Letters represent locations of potential projects. In each case, values are based on the potential outcome for biodiversity persistence. 1. Project-focussed, no spatial attributes. Focus on site attributes only. Biodiversity value = area x condition Site D would have the highest ecological value as the largest site, followed by A, B, C and F. Small remnants have little value. Actions in the matrix are not valued. 2. Project -focussed, with spatial attributes. Connectivity is represented through a landscape value proxy such as percent vegetation cover in the surrounding area (a specified neighbourhood), and/or direct links to protected areas. Biodiversity value = area x condition x landscape value Sites A, B and C are likely to be favoured over the more isolated F. Sites G and H have more value than I because they are closer to other sites. 3. Landscape-focussed, marginal value of project. Total value of landscape with and without each project estimated. Extent (area), condition and connectivity are all considered. Marginal biodiversity value = landscape with project (area, condition, connectivity) landscape without project (area, condition, connectivity) Sites A and B would be valued higher than C as each contributes to a larger patch. Paddock trees at J, K are valued for their contribution to connectivity. Small remnants G, H are valued for their proximity to a larger patch. G is also valued for contributing to connectivity. Potential for additional matrix management actions (e.g. buffering of remnant D at E, or establishing paddock trees) are valued for their contribution to effective patch area and connectivity. 4. Landscape-focussed, optimal package selected. Total value of landscape for each possible combination of projects that fits within a budget constraint. Biodiversity value = the package of projects that maximises benefit within budget. Applies same ecological model as (3), but also considers interdependencies between projects. For example the value of J may be limited if K is not protected. Similarly the value of G may depend on B and C both being protected. Matrix actions are still valued, but their dependencies on other bids are recognised. E is worth little without D.

J K

D E

11

Spatially configuring incentive payments

Table 1: Criteria for evaluating alternative metric options using spatial configuration as an example (outlined in Figure 1). Landscape alteration terms follow McIntyre and Hobbs (1999). What it does 1 Focuses on site attributes only. Minimises costs of designing and running program. 2 Incorporates spatial attributes (e.g. proximity to other projects or protected areas). What it does not do No consideration of spatial configuration, complementarity or matrix management actions. No consideration of matrix/off-patch actions. Consideration of landscape context may be limited. When to use Small budget, only a few sites likely to be purchased. Landscape alteration is relictual in nature, with numerous options of similar value. Spatial configuration considered important, but more rigorous ecological models unavailable. Landscape alteration is relictual to fragmented in nature with few options for incorporating spatial context. Spatial configuration and/or matrix/off-patch management actions relatively important. Ecological model and data available. Landscape alteration is fragmented to variegated in nature with multiple options for incorporating spatial context. Spatial configuration and/or matrix/off-site management actions relatively important. Ecological model and data available. Landscape alteration is fragmented to variegated in nature with multiple options for incorporating spatial context. Value of projects conditional on others. Impact of program likely to be high relative to existing protected areas.

Considers value of project to landscape. Incorporates greater range of on- and off-patch management actions within metric framework.

Does not consider conditional or package values. For example, individual components of corridor each have relatively low marginal values and may not be selected, even though together they are highly valuable.

Selects optimal overall outcome from available proposals.

Computationally complex. Values are not expressed for individual projects. Projects dropping out can be problematic.

12

Spatially configuring incentive payments If landscape function values such as connectivity and complementarity are considered of relatively minor importance for conservation of a particular species or community it is likely that little will be gained from attempting to target them. This is often the case in highly fragmented landscapes where enhancing the condition of remnant habitat is the highest initial priority (e.g. Table 1, option 1). In addition to forcing trade-offs with other attributes incorporating connectivity and complementarity at this stage will also result in a more complex and administratively costly program. Similarly, for programs with small budgets the additional costs of considering spatial configuration are more likely to outweigh the benefits. In such cases a simple approach focussing just on habitat extent and condition is likely to be preferable in order to target key conservation assets. Off-patch actions may still be incorporated to a limited extent by considering their impact on the condition or effective area of a neighbouring patch. The greater the variability in patch quality, and the lower the dispersal ability of the species in question, the more connectivity is likely to vary between different habitat configurations, so the more there is to be gained from a landscape metric (Visconti and Elkin 2009) (i.e., Table 1, option 2). Landscape-scale metrics also do better when patches are aggregated rather than randomly dispersed, and where off-patch threatening processes have a greater impact. For example in the case of wetland communities, pollutant run-off from upstream agriculture is often a major conservation issue. The impact of threat mitigation actions will depend heavily on their location in the landscape and their relationship to high value patches and other mitigation activities valuing such proposals in isolation is unlikely to be satisfactory. It is also necessary to consider whether there is sufficient spatial information and ecological knowledge to reliably measure connectivity and complementarity at scales relevant to the landscape context of the tender. The level of detail required is likely to vary depending on the objectives of the scheme. Comparing connectivity at the landscape-scale may require detailed mapping or modelling of current land-uses and vegetation cover/conditions. For some species and communities coarser scale information may suffice. Some knowledge or reliable estimates of the relationships between connectivity and biodiversity is required. If the necessary information is not available and too costly to acquire, the potential for a landscape-scale metric is limited. The level of complexity of landscape metrics can be tailored both to the requirements of the program and the available resources in most cases they need not be considered too hard for timely and costeffective implementation. Valuing projects on their marginal contributions to landscapes means that some potentially high value combinations may be missed where the marginal value of individual projects is too low for any one of them to be successful. However in some cases it may prove more tractable than undertaking a full package optimisation (particularly for large numbers of projects as the computational issues increase rapidly). The marginal value approach is likely to work best where the projects funded in a conservation tender constitute a relatively small part of the overall ecological function of the landscape. It will also be more attractive if only a small number of projects are likely to be funded (e.g. due to budget constraints). The package approach will be relatively more attractive where the budget is large and there are many different options for the desired outcomes (e.g. many different ways of forming links across the landscape). Another consideration is how existing habitat patches should be incorporated, particularly those on private land which are not covered by a formal conservation agreement. While such sites may currently be in good condition, it could be risky to assume that this will always be the case. One 13

Spatially configuring incentive payments option is to only consider formal protected areas as contributing to habitat patches, although in many cases existing protected areas will not provide an accurate representation of effective landscape function. A better option would be to include all existing habitat patches, irrespective of current management, perhaps accounting for an expected trajectory of deterioration in unprotected patches by linking management regime to habitat condition outcomes. Current and predicted condition values of both protected and unprotected sites could be based on business as usual statetransition trajectories, which would show degradation, improvement or constancy depending on the land use and management regime (Drielsma and Ferrier 2006). 5. Incorporation into incentive programs By valuing a projects contribution to the overall landscape landholders have a financial incentive to submit bids which contribute to desired spatial configurations by considering what their neighbours are currently or planning to do also. However, where the details of individual projects relate to others in the landscape (such as aligning habitat corridors across properties), additional mechanisms may be required to incentivise management to enhance landscape function. One possibility is an agglomeration bonus in which landholders providing desired spatial configurations receive an additional payment (Parkhurst et al. 2002), which can be more cost effective than making spatially homogenous payments (Drechsler et al. 2010). Multi-round conservation tenders, in which landholders are shown the location within the landscape of offers from the previous round, also present a means of providing both the incentive and the necessary information for landholders to coordinate their offers to achieve landscape connectivity (Rolfe et al. 2009). The potential of multi-round tenders to deliver spatially linked conservation sites has been demonstrated both in the laboratory (Reeson et al. 2011) and in the field (Windle et al. 2009, Williams et al. 2012). The landscape-scale approach can also be applied to tradable credit schemes to accrue biodiversity benefit (Bruggeman et al. 2009). Connectivity is not a general property of landscapes, but rather varies depending on the attributes of the species or ecological communities in question. There can therefore be no single formula for managing for connectivity. The relative importance of connectivity, and how it should be measured and managed, will vary according to the objectives of each conservation program. Figure 2 summarises the steps to consider in determining the most appropriate form of an ecological metric. Where connectivity or complementarity are important, projects should be assessed at the landscape scale, rather than on a site by site basis, in order to better reflect ecological outcomes. However, this is only feasible for programs which have sufficient resources to make an impact at the landscape scale (and the greater the geographical scope of the program, the more money is likely to be required to have a significant impact), and where ecological models and data are available (or can be obtained at reasonable cost).

14

Spatially configuring incentive payments

Is spatial configuration important relative to other attributes? Yes Is total budget likely to be sufficient to impact spatial configuration? Yes Are data and models available to value spatial configuration? Yes Landscape-focussed metric

No

No

No

Patch-focussed metric

Are there significant interdependencies between sites? Yes Package landscapevalue metric

No

Marginal landscapevalue metric

Figure 2: Metric selection process Where these conditions are not met, a metric should focus on individual projects. Such metrics should be parsimonious, focussing on key attributes rather than attempting to incorporate everything. However, they may be extended to consider the impact of off-patch management actions on the extent or condition of neighbouring patches where these effects are important. The package approach is preferred where the total values of combinations of projects are significantly different to the sum of their individual marginal values. This is most likely where spatial requirements are particularly exacting. For example, if it is essential to have contiguous habitat, a single gap could make the difference between a viable and an unviable population. Landscape-scale metrics are feasible and tractable, and so are ready to be applied in conservation incentive programs. While in many cases this would represent a change from existing metrics, the landscape-scale approach does not necessarily have to be much more complex (see examples in 15

Spatially configuring incentive payments Ferrier and Drielsma 2010). In many cases the ecological data required and the on-ground site assessment process would be similar to current practices, the available information is simply configured differently for landscape-scale assessment (Seddon et al. 2010, Williams et al. 2012). Landscape-scale metrics can better describe ecological function, and so can enhance the capacity of incentive programs to deliver the best possible biodiversity outcomes, increasing the cost effectiveness of public investment and better protects species and ecological communities. Including the landscape in project assessments also means that landholders have a financial incentive to submit bids which contribute to desired landscape function outcomes by considering, also, their neighbours actions. As ecological models and economic incentive mechanisms become increasingly sophisticated, there is a growing opportunity for conservation incentive programs to move beyond the confines of project-based metrics to take on a more comprehensive landscapescale assessment process. More integrated, dynamic assessment processes, such as that outlined by (Ferrier and Drielsma 2010) will enable the marginal changes in landscape function to be evaluated as a result of changes in management (Saura et al. 2011) or as a consequence of rapid climate change (Krosby et al. 2010).

Acknowledgements: We are grateful to Simon Attwood, Stefan Baumgrtner, David Boland, Jenny Boshier, Veronica Doerr, Simon Ferrier, Russell Gorddard, Art Langston and Charlie Zammit for comments on earlier versions of this paper. This work was funded by the Australian Government Department of Sustainability, Environment Water, Population and Communities and the CSIRO Terrestrial Biodiversity research theme.

16

Spatially configuring incentive payments References Anand, M. O., J. Krishnaswamy, and A. Das. 2008. Proximity to forests drives bird conservation value of coffee plantations: Implications for certification. Ecological Applications 18:17541763. Attwood, S. J., S. E. Park, M. Maron, S. J. Collard, D. Robinson, K. M. Reardon-Smith, and G. Cockfield. 2009. Declining birds in Australian agricultural landscapes may benefit from aspects of the European agri-environment model. Biological Conservation 142:1981-1991. Baeza, A., and C. F. Estades. 2010. Effect of the landscape context on the density and persistence of a predator population in a protected area subject to environmental variability. Biological Conservation 143:94-101. Barton, D. N., D. P. Faith, G. M. Rusch, H. Acevedo, L. Paniagua, and M. Castro. 2009. Environmental service payments: Evaluating biodiversity conservation trade-offs and costefficiency in the Osa Conservation Area, Costa Rica. Journal of Environmental Management 90:901-911. Bennett, A. F. 2003. Linkages in the landscape. The role of corridors and connectivity in wildlife conservation. IUCN Gland, Switzerland and Cambridge, U.K. Brudvig, L. A., S. A. Wagner, and E. I. Damschen. 2012. Corridors promote fire via connectivity and edge effects. Ecological Applications 22:937-946. Bruggeman, D. J., M. L. Jones, F. Lupi, and K. T. Scribner. 2005. Landscape equivalency analysis: Methodology for estimating spatially explicit biodiversity credits. Environmental Management 36:518-534. Bruggeman, D. J., M. L. Jones, K. Scribner, and F. Lupi. 2009. Relating tradable credits for biodiversity to sustainability criteria in a dynamic landscape. Landscape Ecology 24:775790. Castellon, T. D., and K. E. Sieving. 2007. Patch network criteria for dispersal-limited endemic birds of South American temperate rain forest. Ecological Applications 17:2152-2163. Collard, S., A. Le Brocque, and C. Zammit. 2009. Bird assemblages in fragmented agricultural landscapes: The role of small brigalow remnants and adjoining land uses. Biodiversity and Conservation 18:1649-1670. Dancose, K., D. Fortin, and X. L. Guo. 2011. Mechanisms of functional connectivity: the case of free-ranging bison in a forest landscape. Ecological Applications 21:1871-1885. Davis, F. W., C. Costello, and D. Stoms. 2006. Efficient conservation in a utility-maximization framework. Ecology and Society 11. Drechsler, M. 2011. Trade-offs in the design of cost-effective habitat networks when conservation costs are variable in space and time. Biological Conservation 144:479-489. Drechsler, M., F. Watzold, K. Johst, and J. F. Shogren. 2010. An agglomeration payment for costeffective biodiversity conservation in spatially structured landscapes. Resource and Energy Economics 32:261-275. Drielsma, M., and S. Ferrier. 2006. Landscape scenario modelling of vegetation condition. Ecological Management & Restoration 7:S45-S52.

17

Spatially configuring incentive payments Drielsma, M., and S. Ferrier. 2009. Rapid evaluation of metapopulation persistence in highly variegated landscapes. Biological Conservation 142:529-540. Dykstra, P. R. 2004. Thresholds in Habitat Supply: A Review of the Literature. Wildlife Report No. R-27. B.C. Ministry of Sustainable Resource Management and B.C. Ministry of Water, Land and Air Protection, Victoria, Canada. Fahrig, L. 2003. Effects of habitat fragmentation on biodiversity. Annual Review of Ecology Evolution and Systematics 34:487-515. Faith, D. P., and P. A. Walker. 1996. Integrating conservation and development: Effective tradeoffs between biodiversity and cost in the selection of protected areas. Biodiversity and Conservation 5:431-446. Faith, D. P., P. A. Walker, J. R. Ive, and L. Belbin. 1996. Integrating conservation and forestry production: Exploring trade-offs between biodiversity and production in regional land-use assessment. Forest Ecology and Management 85:251-260. Ferrier, S., and M. Drielsma. 2010. Synthesis of pattern and process in biodiversity conservation assessment: a flexible whole-landscape modelling framework. Diversity and Distributions 16:386-402. Ferrier, S., and B. A. Wintle. 2009. Quantitative approaches to spatial conservation prioritization: matching the solution to the need. Pages 1-15 in A. Moilanen, K. A. Wilson, and H. Possingham, editors. Spatial Conservation Prioritization - Quantitative Methods and Computational Tools. Oxford University Press, Oxford, UK. Fischer, J., and D. B. Lindenmayer. 2002. Small patches can be valuable for biodiversity conservation: Two case studies on birds in southeastern Australia. Biological Conservation 106:129-136. Gillies, C. S., H. L. Beyer, and C. C. St Clair. 2011. Fine-scale movement decisions of tropical forest birds in a fragmented landscape. Ecological Applications 21:944-954. Hajkowicz, S., A. Higgins, C. Miller, and O. Marinoni. 2008. Targeting conservation payments to achieve multiple outcomes. Biological Conservation 141:2368-2375. Hajkowicz, S., A. Higgins, K. J. Williams, D. P. Faith, and M. Burton. 2007. Optimisation and the selection of conservation contracts. Australian Journal of Agricultural and Economic Review 51:39-56. Hanski, I., and O. Ovaskainen. 2000. The metapopulation capacity of a fragmented landscape. Nature 404:755-758. Home, R., C. Keller, P. Nagel, N. Bauer, and M. Hunziker. 2009. Selection criteria for flagship species by conservation organizations. Environmental Conservation 36:139-148. Jackson, H. B., and L. Fahrig. 2012. What size is a biologically relevant landscape? Landscape Ecology 27:929-941. Johst, K., et al. 2011. Biodiversity conservation in dynamic landscapes: trade-offs between number, connectivity and turnover of habitat patches. Journal of Applied Ecology 48:12271235. Jost, L. 2007. Partitioning diversity into independent alpha and beta components. Ecology 88:2427-2439. 18

Spatially configuring incentive payments Juutinen, A., M. Mnkknen, and A. L. Ylisirni. 2009. Does a voluntary conservation program result in a representative protected area network? The case of Finnish privately owned forests. Ecological Economics 68:2974-2984. Khanna, M., and A. W. Ando. 2009. Science, economics and the design of agricultural conservation programmes in the US. Journal of Environmental Planning and Management 52:575-592. Koh, L. P., T. M. Lee, N. S. Sodhi, and J. Ghazoul. 2010. An overhaul of the species-area approach for predicting biodiversity loss: incorporating matrix and edge effects. Journal of Applied Ecology 47:1063-1070. Krosby, M., J. Tewksbury, N. M. Haddad, and J. Hoekstra. 2010. Ecological connectivity for a changing climate. Conservation Biology 24:1686-1689. Laurance, W. F., H. E. M. Nascimento, S. G. Laurance, A. Andrade, R. M. Ewers, K. E. Harms, R. C. C. Luizao, and J. E. Ribeiro. 2007. Habitat fragmentation, variable edge effects, and the landscape-divergence hypothesis. PLoS ONE 2:e1017. Lidicker, W. Z. 1999. Responses of mammals to habitat edges: an overview. Landscape Ecology 14:333-343. Lindenmayer, D., et al. 2008. A checklist for ecological management of landscapes for conservation. Ecology Letters 11:7891. Macdonald, D. W., and D. D. P. Johnson. 2001. Dispersal in theory and practice: consequences for conservation biology. Pages 361-380 in J. Clobert, E. Danchin, A. A. Dhondt, and J. D. Nichols, editors. Dispersal. Oxford University Press, Oxford, UK. Magrach, A., A. R. Larrinaga, and L. Santamara. 2011. Changes in patch features may exacerbate or compensate for the effect of habitat loss on forest bird populations. PLoS ONE 6:e21596. 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 E.A. Cousins. 2002. Mathematical Methods for Spatially Cohesive Reserve Design. Environmental Modeling and Assessment 7:107-114. Minor, E. S., and R. H. Gardner. 2011. Landscape connectivity and seed dispersal characteristics inform the best management strategy for exotic plants. Ecological Applications 21:739749. Moilanen, A., and M. Cabeza. 2002. Single-species dynamic site selection. Ecological Applications 12:913-926. Ovaskainen, O., and I. Hanski. 2003. How much does an individual habitat fragment contribute to metapopulation dynamics and persistence? Theoretical Population Biology 64:481-495. Parkes, D., G. Newell, and D. Cheal. 2003. Assessing the quality of native vegetation: The 'habitat hectares' approach. Ecological Management & Restoration 4:S29-S38. Parkhurst, G. M., J. F. Shogren, C. Bastian, P. Kivi, J. Donner, and R. B. W. Smith. 2002. Agglomeration bonus: an incentive mechanism to reunite fragmented habitat for biodiversity conservation. Ecological Economics 41:305-328.

19

Spatially configuring incentive payments Pascual-Hortal, L., and S. Saura. 2006. Comparison and development of new graph-based landscape connectivity indices: towards the priorization of habitat patches and corridors for conservation. Landscape Ecology 21:959-967. Prevedello, J. A., and M. V. Vieira. 2010. Plantation rows as dispersal routes: A test with didelphid marsupials in the Atlantic Forest, Brazil. Biological Conservation 143:131-135. Reeson, A. F., L. C. Rodriguez, S. M. Whitten, K. Williams, K. Nolles, J. Windle, and J. Rolfe. 2011. Adapting auctions for the provision of ecosystem services at the landscape scale. Ecological Economics 70:1621-1627. Rolfe, J., J. Windle, and J. McCosker. 2009. Testing and Implementing the Use of Multiple Bidding Rounds in Conservation Auctions: A Case Study Application. Canadian Journal of Agricultural Economics 57:287-303. Sarkar, S., et al. 2006. Biodiversity conservation planning tools: present status and challenges for the future. Annual Review of Environment and Resources 31:123-159. Saura, S., C. Estreguil, C. Mouton, and M. Rodriguez-Freire. 2011. Network analysis to assess landscape connectivity trends: Application to European forests (1990-2000). Ecological Indicators 11:407-416. Saura, S., and L. Pascual-Hortal. 2007. A new habitat availability index to integrate connectivity in landscape conservation planning: Comparison with existing indices and application to a case study. Landscape and Urban Planning 83:91-103. Schulte, L., M. Rickenbach, and L. Merrick. 2008. Ecological and economic benefits of crossboundary coordination among private forest landowners. Landscape Ecology 23:481-496. Seddon, J. A., T. W. Barratt, J. Love, M. Drielsma, S. V. Briggs, P. Gibbons, and S. Ferrier. 2010. Linking site and regional scales of biodiversity assessment for delivery of conservation incentive payments. Conservation Letters 3:415424. Shanahan, D. F., C. Miller, H. P. Possingham, and R. A. Fuller. 2011. The influence of patch area and connectivity on avian communities in urban revegetation. Biological Conservation 144:722-729. Tewksbury, J. J., et al. 2002. Corridors affect plants, animals, and their interactions in fragmented landscapes. Proceedings of the National Academy of Sciences 99:12923-12926. Visconti, P., and C. Elkin. 2009. Using connectivity metrics in conservation planning - when does habitat quality matter? Diversity and Distributions 15:602-612. Wiens, J. A., N. C. Stenseth, B. V. Horne, and R. A. Ims. 1993. Ecological mechanisms and landscape ecology. Oikos 66:369-380. Williams, K. J., A. F. Reeson, M. J. Drielsma, and J. Love. 2012. Optimised whole-landscape ecological metrics for effective delivery of connectivity-focused conservation incentive payments. Ecological Economics 81:48-59. Williams, N. M., and C. Kremen. 2007. Resource distributions among habitats determine solitary bee offspring production in a mosaic landscape. Ecological Applications 17:910-921. Windle, J., J. Rolfe, J. McCosker, and A. Lingard. 2009. A conservation auction for landscape linkage in the southern Desert Uplands, Queensland. Rangeland Journal 31:127-135.

20