Spatial analysis to inform the mitigation hierarchy
Abstract
Human activities such as urbanization, infrastructure and agriculture are driving global biodiversity declines. In an attempt to balance economic development goals with biodiversity conservation, governments and industry apply a decision-making framework known as the mitigation hierarchy, with a goal of achieving no net loss or net gain outcomes for biodiversity. Successful application of the mitigation hierarchy requires biodiversity assessments and spatial planning to inform the design of mitigation policies, identify priority areas for biodiversity conservation and impact avoidance, assess the biodiversity impacts of developments, and identify appropriate mitigation measures including offsetting residual impacts. However, guidance on the necessary data and assessment techniques is often lacking, especially in countries where formal mitigation policies do not exist or are in their infancy. Here, we discuss and demonstrate analyses that can help answer some key questions for formulating effective mitigation policies and applying the mitigation hierarchy. We focus on data and analyses that can inform the avoidance and offset steps in particular, and demonstrate these techniques using a case study in Mozambique. While these analyses will not replace field-based assessments for projects, they offer rapid, low-cost approaches to support scoping and development of mitigation policy, planning and decision-making, especially in relatively data-poor regions.
1 INTRODUCTION
Much of Earth's surface is now impacted by human land uses that result in significant biophysical disturbance to natural habitats, such as large-scale land conversion, industrial activity, or infrastructure development (Watson et al., 2016). These changes are driving unprecedented losses of biodiversity (Ceballos et al., 2017; Díaz et al., 2019; Dirzo et al., 2014) and compromising the healthy functioning of ecosystems (Hooper et al., 2012). Driven primarily by land use change and overexploitation, this biodiversity loss is underlain by continued economic expansion through both formal developments undertaken by private and public actors, and urban and rural expansion that is not usually subject to environmental licensing (e.g., slash-and-burn agriculture, unregulated building, non-commercial mining). Many jurisdictions have introduced regulations covering the formal development sectors and their impacts on biodiversity, and over 100 countries either have, or are developing biodiversity compensation/offset policies (GIBOP, 2019).
In an attempt to limit the negative impacts on biodiversity caused by developments such as mining, oil and gas, infrastructure and agriculture projects, most biodiversity compensation policies are based around a decision-making framework known as the mitigation hierarchy (BBOP, 2012). This approach is designed to address impacts on biodiversity through first seeking to avoid impacts wherever possible, then minimizing impacts and restoring damaged biodiversity, and finally by offsetting any residual impacts with the aim of achieving no net loss or a net gain of biodiversity (BBOP, 2012; Ekstrom et al., 2015). Policies explicitly requiring or enabling the mitigation hierarchy are increasingly commonplace (Bull & Strange, 2018), and compliance with such regulations is often the strongest driver for industry to effectively reduce and compensate for their biodiversity impacts (Hanson et al., 2012). Sustainability standards by many financial institutions also require adherence to the mitigation hierarchy and incorporate no net loss and net gain goals (e.g., Equator Principles, 2020; Performance Standard 6; IFC, 2019a). Similar approaches are being used to develop corporate biodiversity strategies (de Silva et al., 2019; SBTN, 2020), and some sectoral bodies like the Roundtable for Sustainable Palm Oil are following suit (RSPO, 2015).
While mitigation policies are increasingly widespread, in practice the implementation and design of actions across the mitigation hierarchy is often problematic. For example, avoidance measures—the most important step of the hierarchy—are often not fully considered or implemented in development projects (Phalan et al., 2018) and 77% of compensation policies do not explicitly state offsets as a last resort (GIBOP, 2019). Offsets face an array of technical and practical challenges that make them difficult to implement in practice (Maron et al., 2016) with those implemented having mixed success in reaching NNL targets (zu Ermgassen et al., 2019). There is therefore a need for cost-effective tools and analyses that can be used to help improve the design and implementation of measures across the hierarchy, using available datasets of biodiversity values and threats.
When developing mitigation policies, or when applying existing policies, spatial analyses and assessments of biodiversity and development impacts play a vital role. They can provide information that helps define the rules and requirements of offset systems, and inform project design so that impacts are avoided at the earliest stage of project planning. In nations like Australia or South Africa, investment in the production and assessment of high-quality spatial data over many years, along with comparatively well-funded environmental departments or programs, allow for a sophisticated, spatially explicit understanding of the state and trends of biodiversity (Jackson et al., 2016; Skowno et al., 2019). However, many development frontiers are in developing countries that face the combined problem of rapid development, weaker application of policy, and limited data for effective mitigation planning (Oakleaf et al., 2015). In these regions, simple analyses using readily available data are useful to support development of mitigation policies, help governments and business improve application of policy, maintain the intact nature of the most important sites for biodiversity, and achieve jurisdictional biodiversity goals.
Here, we highlight how spatial data and analyses can guide policy framing and help set requirements for the mitigation hierarchy. We focus specifically on how spatial analyses can inform the actions taken at the avoidance and offset steps of the mitigation hierarchy, through geographical site selection for development projects, and informing high-level scoping of potential offset sites. We focus on the avoidance and offset planning steps in the mitigation hierarchy, as these can be informed by broad scale spatial analyses, while minimization and restoration are more typically project level (and specific) decisions during the development phase (Ekstrom et al., 2015). We use a case study in Mozambique to demonstrate how such analyses can inform biodiversity impact mitigation in practice. Figure 1 displays a generalized overview of how spatial analysis can inform the avoidance and offset phases of the mitigation hierarchy, and these steps are described in more detail in the following sections.
2 IDENTIFYING POTENTIAL AVOIDANCE AREAS
The first step of the mitigation hierarchy, impact avoidance, requires developers to “anticipate and prevent adverse impacts on biodiversity before actions or decisions are taken that could lead to such impacts” (Ekstrom et al., 2015). Avoidance is generally considered the most important stage of the mitigation hierarchy, as it is the most effective stage in reducing impacts of development on biodiversity (Phalan et al., 2018). It can side-step many of the difficulties and limitations associated with ecological restoration and offsets, such as uncertain or inadequate outcomes for biodiversity, insufficient financing and resourcing for offset activities, ineffective offset actions in the face of complex drivers of biodiversity loss (Maron et al., 2010), restoration time lags (Bendor, 2009) and limits to what can be offset (Maron et al., 2010; Sonter et al., 2020).
Spatial planning to inform impact avoidance can be particularly effective if analysis occurs in two phases: first, strategic landscape-level planning can identify broad priorities for avoidance (including no go zones) and, conversely, key regions for development (Figure 1a). This is particularly useful when developing mitigation policies that must set rules around avoidance and limits to what can be offset, and also to inform alternatives for siting projects, thus helping reduce cumulative impacts from multiple projects (Heiner et al., 2019; Kiesecker et al., 2010). Second, more detailed, project-level analysis can inform direct application of the mitigation hierarchy at a finer scale (e.g., improve project design, comparison of project layouts; Figure 1b). This process can improve biodiversity outcomes and may well save costs, for example by preventing unforeseen project delays or reputational risks associated with higher biodiversity impacts. In particular, spatial analyses can help plan for avoidance by (i) identifying discrete areas with biodiversity of high conservation concern where development should be avoided, and (ii) using measures of biodiversity condition to inform avoidance options.
2.1 Identifying priority avoidance areas
The first step in identifying priority areas for avoidance should be to check for areas that have already been identified as having high biodiversity value, and thus where development is unlikely to be allowed or where more stringent regulations could be applied to discourage development projects (Figure 1a; Maron et al., 2016). This will involve conducting a data gap analysis for the region of interest, including a comprehensive search of published and gray literature, in order to identify and collate existing assessments or datasets that identify areas of high biodiversity value. The nature of these avoidance areas may be set out in existing legislation, including in mitigation policy, or in applicable (but often voluntary) safeguard standards by overarching industry bodies or financial institutions. For example, the International Finance Corporation requires developers to identify the environmental and social risks and impacts of a project (IFC, 2012a), and categorize land near the project into Modified, Natural, or Critical Habitat, in which different conditions for project finance must be met (e.g., requiring offsets to achieve net-gain for any impacts occurring in critical habitat; IFC, 2012a, 2012b, 2019b). In many cases, it is preferable for developers to simply avoid impacting Natural or Critical Habitat, removing the need to meet more stringent conditions for financing and the associated risks and costs. Global data identifying the location of likely Critical Habitat are now available, for both the terrestrial and marine realms (Brauneder et al., 2018; Martin et al., 2015), although these maps are limited by the availability and quality of the underlying data. As such, they are primarily useful as screening layers used to inform more detailed subsequent assessments, and are useful to help with avoidance through broad site selection (Figure 1a).
In most jurisdictions, datasets identifying areas of high biodiversity value will already exist, providing a useful starting point for mapping avoidance priorities (Stephenson & Stengel, 2020 provide a useful inventory of biodiversity data sources, but we summarize some key ones here). For example, some mitigation policies do not allow development within existing protected areas, and data on existing protected areas are often available from Protected Areas agencies. These data have been collated at https://www.protectedplanet.net, although these data should be double-checked against official sources (e.g., jurisdictional government maps) to account for updates or missing information. The World Database of Key Biodiversity Areas (KBAs) should also be consulted (http://www.keybiodiversityareas.org/home), as KBAs are defined through a set of criteria as sites contributing significantly to the global persistence of biodiversity and are thus likely to be areas where development should be avoided. The IUCN Red List of Threatened Species (https://www.iucnredlist.org/) also provides spatial information on threatened species habitat, and in some cases development is unlikely to be allowed where species above a certain threat status are present (Pilgrim et al., 2013), or where species of special anthropological significance occur (e.g., great apes; IFC, 2019a, 2019b). These sources provide globally available data on potential avoidance areas, but jurisdictional governments or other relevant bodies are likely to have additional mapped biodiversity priority areas which can be incorporated into avoidance planning (e.g., threatened ecosystems; Botts et al., 2020). Online tools such as the Integrated Biodiversity Assessment Tool (https://www.ibat-alliance.org/) or Tremarctos (http://www.tremarctoscolombia.org/) can also be used to quickly produce reports about whether a development site overlaps with areas of high biodiversity value, although the cost of subscriptions can be prohibitive for small organizations.
After collating data on existing areas of high biodiversity value, an important next step is to assess the location and status of ecosystems across the jurisdiction. An ecosystem classification and map commonly forms the basis for the application of the mitigation hierarchy, as ecosystems are often used as the unit with which to quantify impacts and offset requirements (Brownlie et al., 2017). Importantly, ecosystems can be mapped consistently across an entire jurisdiction, unlike species data that are often patchy and biased toward areas with high sampling levels. Understanding the threat status of ecosystems is crucial, because developments may not be permitted or advisable in threatened ecosystems, and under some policies offset requirements are scaled to be higher in more threatened ecosystems (Brownlie et al., 2017). For example, endangered ecosystems may qualify as Critical Habitat under the International Finance Corporation's Performance Standard 6, triggering much more stringent biodiversity requirements of clients and projects that would receive project finance (IFC, 2019a). Relatively detailed maps of ecosystem types are already available in many countries (e.g., Mozambique, South Africa, Myanmar), and some have also assessed ecosystem threat status (Botts et al., 2020; Murray et al., 2020). Many other countries have begun mapping and assessing ecosystems, following a global ecosystem typology being developed by the IUCN (Bland et al., 2017; Keith et al., 2020). These data can be used to inform avoidance measures through broad site selection (e.g., identifying several alternatives and selecting a project area where few or no threatened ecosystems would be impacted; Figure 1a) and subsequently through detailed project design (e.g., rerouting a road around a threatened ecosystem; Figure 1b).
For areas where national-scale ecosystem data are not already available, there are numerous free datasets available, ranging from global to regional scales (Table 1). However, it is important to verify the accuracy and suitability of such data, as they can often misclassify ecosystems, especially when analyzed at a fine scale. While not a trivial task, the rise of cloud analysis platforms—such as Google Earth Engine—is also making it increasingly feasible to quickly generate ecosystem classifications, if appropriate training data is available. Further, online tools such as Remap (https://remap-app.org/) can produce detailed ecosystem classifications reasonably quickly (Murray et al., 2018), without requiring any specialized GIS or scripting skills. Because tools like Remap and Google Earth Engine have access to the publicly available LANDSAT satellite archive, it is possible to access data and generate maps as far back as 1984 at a 30-m resolution (Woodcock et al., 2008). This makes it possible to assess changes in ecosystem distribution over a relatively long period using freely available data alone.
| Dataset | Geographical coverage | Source | |
|---|---|---|---|
| Ecosystem maps | Ecological Land Units | Global | Sayre et al. (2014) |
| Potential Natural Vegetation | Global | https://explorer.naturemap.earth/map | |
| Global Ecoregions | Global | Dinerstein et al. (2017) | |
| Global Wetlands | Global | https://www.cifor.org/global-wetlands/ | |
| Terrestrial Ecosystems of Africa | Africa | Sayre et al. (2013) | |
| Potential Natural Vegetation of Eastern Africa | East Africa | http://vegetationmap4africa.org/ | |
| Congo Basin Forest Ecosystems | Congo Basin | Shapiro et al. (2021) | |
| Flora Zambesiaca | SE Africa | Wild and Barbosa (1967) | |
| Terrestrial Ecosystems of South America | South America | Sayre et al. (2008) | |
| Ecosystems of Temperate & Tropical Americas | Temperate and Tropical Americas | Comer et al. (2020) | |
| Ecoregions of North America | Canada, USA, Mexico | Omernik and Griffith (2014) | |
Ecological condition/human pressure Data |
Human Footprint | Global | Venter et al. (2016) |
| Biodiversity Intactness Index | Global | Newbold et al. (2016) | |
| Ecoregion Intactness Metric | Global | Beyer et al. (2020) | |
| GEDI Ecosystem Structure | Global | Dubayah et al. (2020) | |
| Forest Landscape Integrity Index | Forest Ecosystems | Grantham et al. (2020) | |
| Forest Structural Integrity Index | Humid Tropical Forests | Hansen et al. (2019) |
- Note: there are many other nationally available datasets that are not listed here.
While there are numerous methods for assessing ecosystem threat status, one of the most widely accepted is the IUCN Red List of Ecosystems (RLE). The RLE is a global standard for assessing the status of ecosystems, applicable at local, national, regional and global levels (Bland et al., 2017). Like the Red List of Threatened Species, this framework contains five criteria which can be used to assign ecosystems to risk categories (e.g., critically endangered, endangered, vulnerable). These criteria act as guides for evidence-based scientific assessments of the risk of ecosystem collapse, as measured by reductions in geographical distribution or degradation of the key processes and components of ecosystems. The results of RLE assessments can be used to inform development of mitigation policy (e.g., set rules around avoidance or offset requirements), and in project level planning (e.g., ensure development footprint does not overlap with threatened ecosystems).
2.2 Measuring biodiversity condition to inform avoidance planning
Beyond identifying discrete priority areas where development impacts are best avoided (e.g., protected areas, endangered ecosystems), avoidance planning can also be informed by measures of biodiversity condition. Here, we focus on measuring ecosystem condition, as ecosystems are commonly used as the units for which impacts and offset requirements are calculated (Bezombes et al., 2018). Landscape-level measurements of ecosystem condition can be useful to set the rules and regulations of mitigation policies, which are dependent on understanding the overall condition of biodiversity across a jurisdiction. For example, mitigation policies might restrict developments or require more stringent regulations in good condition areas, or encourage developments in lower condition areas depending on the ecosystem type. Such policy requirements can then inform high-level project planning, allowing for comparison of site alternatives and revision of development plans. For example, spatial analyses can be used to target development toward already degraded areas, or follow existing infrastructure, which can improve outcomes for industry and biodiversity (Runge et al., 2017; Figure 1b).
At the landscape-level, most methods to measure ecosystem condition focus on the human activities known to degrade ecosystem function, rather than measuring condition directly (Heiner et al., 2019; Venter et al., 2016). Often referred to as human pressures, activities and land-uses such as agriculture, urbanization, roads, and railways have been directly linked to constraints on and declines in biodiversity (Newbold et al., 2015; Safi & Pettorelli, 2010; Tucker et al., 2018). As such, areas where these pressures are prevalent are likely to be of lower condition than areas where such pressures are absent. Some of the most well-known human pressure maps use an approach known as cumulative impact mapping, which combines data on multiple human activities to generate a single index measuring human pressure on the environment. When using human pressure data, it is important to be mindful of the traditional actions of indigenous cultures (e.g., hunting, harvesting tree resources) which are unlikely to be the major cause of pressure in a landscape. Additionally, there are some high biodiversity ecosystems that have evolved for thousands of years with human intervention, such as Mediterranean cork oak and holm oak savannahs and woodlands (Bugalho et al., 2011; Hernández-Agüero et al., 2022). Focus should instead be placed on the human land uses that result in significant biophysical disturbance to natural habitats, such as large-scale land conversion, industrial activity, or infrastructure development. Global human pressure data is currently available at 1 km2 resolution (Sanderson et al., 2002; Venter et al., 2016), and it is relatively straightforward to update for specific regions where better data are available (e.g., Karimi & Jones, 2020; Li et al., 2018). There are also more detailed, ecosystem specific condition metrics available for some ecosystems (e.g., forests), many of which combine measures of human pressure with remotely sensed data on vegetation extent, health or structure (Grantham et al., 2020; Hansen et al., 2019). Human pressure datasets can be classified to show likely intact versus degraded habitat, and there are also other global datasets available which combine human pressure data with maps of important biodiversity areas to separate natural and modified habitat (Gosling et al., 2020). Such data are potentially very useful for avoidance planning at a broad-scale, where developments can preferentially be targeted toward degraded areas where impacts on biodiversity will be lower (Figure 1a).
At the project-level, there are many existing metrics used to measure the impacts of developments, such as the Habitat Hectares concept used in Australia (Parkes et al., 2003), or Biodiversity Metric 3.0 used in the United Kingdom (Panks et al., 2021). These metrics often use a suite of field indicators, such as number of large trees, patch size, and weed presence, and aggregate them into a weighted score to compare against benchmark scores for a “healthy” ecosystem (McKenney & Kiesecker, 2010). While such approaches are useful and necessary for quantifying the condition of impacted ecosystems and determining offset requirements, they are relatively time and labor intensive, and their utility for quickly assessing probable ecosystem condition is limited, especially when analyzing the landscape at a broader scale. For early-stage avoidance planning, simple proxies for measuring ecosystem condition can be useful to help revise development plans and minimize impacts, without the need for field-based indicators.
3 AVOIDANCE PLANNING IN MOZAMBIQUE
To demonstrate how identifying priority avoidance areas and measuring biodiversity condition can inform avoidance planning and ultimately reduce impacts to biodiversity, we formulated a hypothetical road development in the Cabo Delgado province of Northern Mozambique. This road development stretches for approximately 190 km between the towns of Manrassi and Miengueliua, a region that has been targeted by industry over the past 10 years due to the abundance of high-value natural resources, such as gem stones, graphite, and natural gas. Because the mitigation hierarchy in Mozambique focuses primarily on ecosystems, we conducted a simple assessment of predicted project impact by considering the conservation status and condition of impacted ecosystems. We show how these assessments can inform avoidance planning by directing impacts toward ecosystems of lower conservation significance, and preferentially impacting degraded areas over those in good ecological condition.
3.1 RLE assessment
To assess the conservation significance of ecosystems, we utilized the results of a recent RLE assessment conducted for Mozambique. This process is documented in Lötter et al. (2021), but we provide a brief summary here. The RLE framework provides five criteria which can be used to assess ecosystem threat status (Criterion A–E), with each criteria reflecting a different threatening process, and RLE assessments can apply as many criteria as available data allow (Bland et al., 2017). Mozambique's RLE assessment involved the development of a detailed historical ecosystem map, which was undertaken via a combination of machine-learning supervised classification algorithms and manual expert validation. Following development of this map, all ecosystems were assessed against Criterion A (reduction in geographic distribution) and Criterion B (restricted geographic distribution), and a subset of ecosystems were assessed against Criterion D (disruption of biotic processes). While RLE assessments can be very useful for avoidance planning, it is important to note that many assessments are not able to assess all RLE criteria, and thus these non-comprehensive assessments may underestimate the threat status of ecosystems. It is therefore crucial that mitigation policies emphasize the importance of impact avoidance and overall impact minimization, even in ecosystems of lower threat status or conservation concern.
In Mozambique's RLE assessment, Criterion A3 (historical change) was assessed by using 2016 land cover data (FNDS, 2019) to mask areas of the historical ecosystem map that have since been converted to human land uses (e.g., urban areas & agriculture; Lötter et al., 2021). By comparing the historical extent of ecosystems to their current extent, the area of loss for each ecosystem can be calculated, and ecosystems assigned to criterion A threat categories accordingly (Figure 2). For example, an ecosystem that has lost over 70% of its historical distribution, or lost over 50% of its distribution in any 50 year period, is classified as endangered (Bland et al., 2017). Criterion B was assessed by using the R package redlistr (Lee et al., 2019) to compute a minimum convex polygon that encompassed the entire distribution of each ecosystem within Mozambique (Criterion B1—Extent of occurrence), and the number of 10 × 10-km grid cells occupied by each ecosystem (Criterion B2—Area of occupancy; Figure 2). Criterion D was assessed by analyzing the extent and severity of biotic change in forest ecosystems using the Forest Landscape Integrity Index (Lötter et al., 2021).
3.2 Measuring ecosystem condition
Mapping ecosystem condition can help ensure that development will be prioritized to degraded areas rather than intact areas within the same ecosystem, resulting in less impact on biodiversity and reduced offset requirements. To assess ecosystem condition in Mozambique, we followed an approach known as cumulative impact mapping (Ban et al., 2010; Sanderson et al., 2002). This framework uses data on multiple human activities known to impact biodiversity, and combines them to generate a single index measuring human pressure on the environment, known as the human footprint. Specifically, we adapted the approach outlined in Venter et al. (2016), and considered the following human pressures: (1) the extent of built environments; (2) crop land; (3) human population density; (4) night-time lights; (5) railways; (6) roads; and (7) navigable waterways. Each pressure was placed on a 0–10 scale and then summed together to create an overall human footprint index for Mozambique (Figure 3a). To separate intact and degraded areas, we followed previous studies and defined degraded areas as those where human footprint scores were greater than four (Figure 3b; Di Marco et al., 2018; Jones, Venter, et al., 2018; Venter et al., 2016).
3.3 Avoidance planning
The first step in avoidance planning is to assess the overlap between the project's area of influence (the area predicted to be affected by development) and priority/important ecosystems. This analysis can inform decisions around project feasibility (e.g., are there unacceptable impacts that cannot be mitigated), and then attempt to find ways to reduce the spatial overlap between project footprint and priority ecosystems. For example, a pipeline could be: (i) rerouted to follow existing roads or railways, or to avoid priority ecosystems (avoidance through site selection); (ii) buried to remove a barrier to movement (avoidance through project design; Ekstrom et al., 2015); or (iii) canceled if no feasible route can be identified. Regardless of the threat status or condition of ecosystems likely to be affected by a project, the overall area impacted should always be reduced as much as is feasibly possible. This should occur in the pre-project planning phase, as project plans are less likely to be revised for avoidance once construction has begun (Ekstrom et al., 2015). While real-world avoidance planning is an extremely complex process with a multitude of factors to be considered, one of the key steps involves generating and comparing a series of alternative project layouts in order to decide on the most appropriate design for a given situation (Figure 1b).
In our case study, we conducted a formal comparison of project alternatives by assessing the total area affected, the ecosystems impacted and ecological condition of land impacted by each potential project design (Table 2). We found that a revision of road plans can prevent impacts in an endangered ecosystem, and reduce impacts in a vulnerable ecosystem (Figure 4a; Table 2). This kind of avoidance planning is not only beneficial for biodiversity, but is also likely to benefit project proponents, as many mitigation policies place higher offset requirements on threatened ecosystems. Whilst there would be higher costs associated with constructing the revised road layout, the reduced offset requirements (and associated costs), as well as the lower reputational (i.e., public scrutiny) and financial risks (e.g., loss of access to IFC PS6 linked financing) in the revised scenario may benefit the project proponent. There may therefore be a substantial disincentive to development in threatened ecosystems, although developments may still occur in cases where the predicted economic or social benefits are so high that they make meeting these strict offset requirements worthwhile.
| Original road | Revised road | |
|---|---|---|
| Total area impacted (km2) | 20.71 | 30.24 |
| Area of endangered ecosystems impacted (km2) | 15.53 | 0 |
| Area of vulnerable ecosystems impacted (km2) | 3.11 | 2.46 |
| Area of intact land impacted (km2) | 2.69 | 0.86 |
- Note: Spatial arrangement of roads can be seen in Figure 4.
Beyond using ecosystem threat status to inform avoidance planning, it is also possible to use measures of ecosystem condition to plan developments such that impacts preferentially occur in areas that are already degraded. Within ecosystems of any threat status, there will likely be spatial variation in degradation levels, and it is preferable to impact degraded areas over good condition areas. In our case study, we are able to change the location of a machinery storage area to impact a degraded area rather than an intact area (Figure 4b). Examining ecosystem condition also shows that the alternative road, designed to avoid impacting an endangered ecosystem, shifts a small section of road from degraded land into an intact area (albeit in a least concern ecosystem). Given that new roads in intact ecosystems can facilitate further development and land speculation (Laurance et al., 2009; Suárez et al., 2009), it may be that a small impact to a threatened ecosystem (original road plan) is preferable to a road which cuts through an intact area (alternative road plan). This would depend on a large number of factors, and real-world avoidance planning requires assessing development plans against relevant mitigation policy, and considering environmental, social, technical, and financial criteria. Because the rapid assessments we present here are based on freely available data that is often coarse in scale, and our intention is to broadly illustrate the application of such data, verification, and validation of project designs and impacts to biodiversity must always be verified through more detailed analysis and fieldwork.
4 IDENTIFYING POTENTIAL OFFSET SITES
Strategic landscape-level planning is crucial for determining the most rewarding and cost-effective opportunities for securing biodiversity gains, and hence determine the most appropriate set of offset activities and locations (Figure 1c). Offset design is generally dependent on the principle of equivalency, which aims to ensure that the biodiversity gain from offsets is a fair (adequate, commensurate) exchange for what was lost due to a development (Quétier & Lavorel, 2011). Although the exact rules may change between mitigation policies, the equivalence of an offset site generally depends on the type of biodiversity that occurs there, the condition or state of that biodiversity, and the potential of that site to generate the gain of biodiversity that is required (BBOP, 2012; Quétier & Lavorel, 2011).
The first step in designing potential offsets will be to assess the distribution of biodiversity similar to that which will be impacted by a development. In some mitigation policies offsets are only permitted within the same ecosystem as was impacted, whereas others allow impacts to be offset in different ecosystems as long as pre-determined “exchange-rules” are followed. These rules often involve “trading up” to conserve biodiversity that is of greater conservation concern than that impacted (Gardner et al., 2013), and spatial assessments of ecosystem status (Figure 3) can be useful to determine appropriate exchange-rules. Exchanges will likely be most useful in a jurisdiction with some widespread ecosystems and some extremely threatened ecosystems, such that losses in widespread ecosystems are acceptable if they are balanced by gains in threatened ecosystems (trading up). In such cases, clear guidelines are essential to prevent the exchange of threatened or irreplaceable ecosystems for those of lower importance (Pilgrim et al., 2013; Walker et al., 2009). If allowed by offset legislation, the use of spatial prioritization tools to design offset programs has been shown to deliver substantially improved outcomes for biodiversity when compared to strict like-for-like policies (Kujala et al., 2015). It is also important to consider whether mitigation planning and compliance is legislated to occur at the project level or at a wider scale (e.g., landscape or national level), as this can have substantial impacts on project decisions and the overall success of mitigation policies (Kennedy et al., 2016).
As well as simply holding similar biodiversity values to what was impacted (or other biodiversity values deemed acceptable by exchange rules), potential offset sites should be able to provide the necessary biodiversity gains to reach the applicable mitigation target (BBOP, 2012). As such, landscape-level assessments of ecosystem condition or restoration potential are useful for determining the kinds of offsets that should be prioritized in mitigation policies (e.g., Worthington & Spalding, 2020). For example, in a jurisdiction where most good condition habitat is already effectively protected, restoration of degraded habitat may be the most effective way to deliver biodiversity gains to offset development impacts. Alternatively, in places where good condition habitats are relatively widespread but being rapidly lost (e.g., Cabo Delgado in Northern Mozambique), effective protection of existing good condition habitat may be the best way to avert loss and deliver biodiversity gains (Devenish et al., 2022). As described in Section 2, there are multiple spatial datasets available to monitor ecosystem condition at the landscape-level. For example, products like the Forest Landscape Integrity Index (Grantham et al., 2020) or the Forest Structural Integrity Index (Hansen et al., 2019) combine measures of human pressure, forest disturbance, and/or forest structure to monitor forest condition. Alternatively, measures of human pressure such as the human footprint (Venter et al., 2016) can also be used as proxies for ecosystem condition, as demonstrated here. Beyond being useful for informing mitigation policy at a broad scale, these approaches may also be used at the project-level as “first-pass” filters for identifying good condition areas for possible protection, or degraded areas which may be suitable for restoration (see Section 4).
Beyond data such as the human footprint or forest landscape integrity index, which are coarse resolution, global-scale datasets with a number of limitations, there are numerous other approaches to scope areas that could benefit from actions to improve condition and lend themselves as offset sites. For example, vegetation productivity indices (e.g., NDVI) can be used to monitor ecosystem degradation (Easdale et al., 2018), and there are now user-friendly GUIs for quickly assessing such indices using cloud computing (Langner et al., 2018). Alternatively, the IUCN Restoration Opportunities Assessment Methodology provides a flexible and affordable framework to rapidly identify suitable restoration areas, but it is crucial that the indicators used in such approaches are linked to the objectives of restoration efforts (IUCN and World Resources Institute, 2014; Worthington & Spalding, 2020). Regardless of the method used to assess degradation or ecological condition, such analyses are a first step and it will always be necessary to validate potential offset sites through fieldwork. This will include verifying the ecological suitability of a site, for example by ensuring it holds or once held appropriate biodiversity, that it has the potential to be enhanced, and that it can be effectively protected.
Assessing the social acceptability, feasibility and environmental justice of implementing offsets in a given location is another key component of the site selection process and application of the mitigation hierarchy as a whole. It is crucial that best practice guidelines on ensuring no net loss for people and biodiversity are followed (BBOP, 2009; Bull et al., 2018). As a priority, offset developers need to understand how different actors (e.g., communities, government institutions, etc.) use the potential offset site and how these actors should be considered and engaged in biodiversity management (BBOP, 2012; Bull et al., 2018). It is also crucial to recognize that the aspects of biodiversity written into offset legislation or policy, such as broad ecosystems or species ranges, may not reflect what is important to people at local scales. Many of the elements of biodiversity or ecosystem services that are important to people at a local scale are not well captured by spatial data, especially the coarse-scale data highlighted here. As such, comprehensive consultation with affected communities is essential to measure changes to people's wellbeing caused by a development and its offsets, and to design mitigation measures to ensure people are no worse (or are better) off than before the project (BBOP, 2009). Measuring and adequately addressing human well-being is a complex participatory process, but generally involves identifying a suite of wellbeing indicators and using these to measure project impacts and design mitigation/compensation programs. Given the complexity of these issues, industry best-practice guidelines recommend close consultation with social scientists throughout the project lifetime (Bull et al., 2018).
Designing and implementing offsets is often complex, time-consuming, and costly, so offsets should ideally be placed in areas that are already identified as priorities for biodiversity conservation. This has a double advantage, as it (i) allows for offsets to contribute to the achievement of national biodiversity targets; and (ii) may reduce the negotiation period and transaction costs, ease stakeholder discussions, and may ultimately improve outcomes (Ekstrom et al., 2015). Whilst local and context-specific knowledge is vital for guiding the design and location of these sites, broad spatial analyses can help with initial identification and prioritization. Potential sites can also be specified in mitigation policies, identified through existing national conservation plans and strategic environmental assessments, or using priorities identified through spatial biodiversity assessments and planning or mechanisms such as IUCN RLE (i.e., target offsets toward the most endangered ecosystems), or Key Biodiversity Areas (IUCN, 2016; Smith et al., 2019).
Finally, with the world on track for at least 1.5° of warming (Rogelj et al., 2016), it is important to consider the impacts of climate change in offset design and implementation, and there are numerous ways to do so. Offsets can be preferentially targeted toward conservation priorities that are recognized as climate refugia, that is, those places that are predicted to undergo less severe change due to global temperature increases (Ashcroft, 2010; Jones et al., 2016), such that biodiversity in the offset site will be less impacted by climate change compared to other areas. Spatial data on climate refugia are now available in many locations, and a general framework for refugia identification is available in Keppel et al. (2012). Similarly, offsets could be targeted toward areas that increase connectivity between isolated habitat patches to facilitate the movement of species in the face of climate change (Littlefield et al., 2019; Morelli et al., 2020). Alternatively, modeling techniques (e.g., species distribution modeling, environmental niche modeling) can be used to identify areas of potentially suitable habitat under future conditions, allowing identification of places that will remain suitable for biodiversity offsets into the future (Jones et al., 2016). Beyond impacts on biodiversity directly, it is important to consider the human response to climate change, and assess how this may impact biodiversity and human well-being in and around offset sites. While it is difficult to predict how any individual will react to a changing climate, human responses to climate change can be addressed using actions that either resist human responses (e.g., anti-poaching patrols), accommodate change (e.g., restoration of degraded forests), or provide dual-benefits (e.g., ecosystem-based adaptation; Maxwell et al., 2015). In many cases, biodiversity offsets can be planned such that they either accommodate human responses to climate change or provide dual-benefits. For example, restoration of mangroves as a biodiversity offset may also help reduce the impacts of storm surges and sea level rise driven by climate change (Menéndez et al., 2020).
5 SCOPING FOR OFFSET SITES IN MOZAMBIQUE
For our case study in Mozambique, we did not attempt to determine specific offset requirements, but simply to identify a broad set of potential offset sites. Mozambique's legislation outlines three options for biodiversity offsets, all of which are focused on biodiversity enhancement. First, offsets can be used to restore parts of existing protected areas that have been degraded such that their conservation objectives may be compromised. Second, offsets can be used to extend existing protected areas to encompass and improve unprotected biodiversity or establish connectivity between existing protected areas. Third, offsets can be used to protect and improve degraded parts of priority areas for biodiversity such as KBAs, RAMSAR sites, threatened ecosystems, or other important areas identified through a systematic conservation planning exercise. While there are a number of very valid concerns related to the additionality of permitting offsets to occur within protected areas (Githiru et al., 2015; Pilgrim & Bennun, 2014), the Mozambican government has committed to only adequately resourcing half of its existing protected area estate until 2030. As such, biodiversity offsets within parts of the unfunded protected area network, or within currently unprotected KBAs or threatened ecosystems can be considered to meet the additionality principle (BBOP, 2012) for biodiversity offsets in Mozambique.
To scope potential offsets sites (based on biodiversity enhancement) for our hypothetical road development in Mozambique, we first identified the ecosystems impacted by taking a 1-km buffer around the area of influence of the planned road (Figure 5a). Placing buffers around roads has been used to quantify indirect impacts within the area of influence of a project which extend beyond the direct project footprint, such as edge effects or increased access for hunting (Barber et al., 2014; Benítez-López et al., 2017; Laurance et al., 2009; Suárez et al., 2009). Next, we mapped the total extent of impacted ecosystems in Mozambique, in order to generate an overall area of interest in which to map potential offset sites (Figure 5b). We then used human footprint data to identify degraded land within the area of interest (Figure 5c). Finally, we mapped potential offset sites by identifying degraded areas of each ecosystem impacted by the development (Figure 5d). We also overlaid protected areas and KBAs, as offsets in Mozambique are preferentially to occur within them where possible. The areas mapped in Figure 5d represent potential offset sites and can inform more detailed local-scale studies to finalize site selection. The linear appearance of potential offset sites is due to roads and their buffers as mapped in the human footprint data. While roads and their immediate surroundings (e.g., shoulders, verges, etc.) are unlikely to be removed or closed as part of an offset, ecosystem degradation is often most severe close to roads, as they facilitate access for hunting, charcoal collection, and many other activities which impact biodiversity (Laurance et al., 2009; Santos & Tabarelli, 2002; Suárez et al., 2009). As such, areas close to roads are often degraded, and some likely have potential as offset sites if human activities can be restricted or controlled.
6 CONCLUSION
Despite widespread recognition of the need to halt biodiversity loss (Convention on Biological Diversity, 2014), transformation of the natural world continues to occur at a rapid pace (Jones, Klein, et al., 2018; Venter et al., 2016). As governments pursue ambitious international development goals (United Nations General Assembly, 2015), and will soon adopt a post-2020 Global Biodiversity Framework likely to contain ambitious biodiversity conservation targets (e.g., protect 30% of Earth's land by 2030), the need to balance development and conservation goals is ever growing. Although it should not replace other strong actions to reduce drivers of biodiversity loss, the mitigation hierarchy has great potential to help reduce the overall biodiversity loss associated with development projects, helping countries improve the state of biodiversity and human well-being at a local scale as well as achieve national biodiversity targets and international commitments. This paper outlines some simple spatial analyses and approaches (based on existing standards and best practices) that can guide development of mitigation policies in regions where they do not exist, and assist with direct application of the mitigation hierarchy, particularly during the project scoping phase. Because offsets can be complex to implement (zu Ermgassen et al., 2019), it is crucial that any application of the mitigation hierarchy places avoidance of impacts as its primary focus, resorting to offsets only when absolutely necessary. Additionally, the analyses presented here are based on freely available, often coarse scale data, and so there will always be a need for verification and field validation. However, the approaches we present can provide low-cost, rapid preliminary assessments to support decision-making and spatial planning, especially in developing regions.
ACKNOWLEDGMENTS
The authors would like to thank Agence Française de Développement (AFD) and The French Facility for Global Environment (FFEM) for their support in funding the COMBO project, the Foundation for the Conservation of Biodiversity (BIOFUND) for their financial and technical support, as well as all COMBO staff for their helpful discussions and suggestions. They would also like to thank two anonymous reviewers whose suggestions helped improve an earlier version of this paper and the Ministry of Land and Environment of Mozambique, in particular the National Directorate of Environment, for all the support provided during the implementation of the COMBO project.
CONFLICT OF INTERESTS
The authors declare no conflicts of interest.
AUTHOR CONTRIBUTIONS
Kendall R. Jones, Amrei Von Hase, Hugo Rainey, Hugo M. Costa, and Hedley S. Grantham conceived and designed the paper. Kendall R. Jones performed the research and analysis. All authors developed recommendations and wrote the paper.
Open Research
DATA AVAILABILITY STATEMENT
The datasets generated for this study can be made available upon request to the corresponding author.
