Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Climate change is leading to the development of land-based mitigation and adaptation strategies that are likely to have substantial impacts on global biodiversity. Of these, approaches to maintain carbon within existing natural ecosystems could have particularly large benefits for biodiversity. However, the geographical distributions of terrestrial carbon stocks and biodiversity differ. Using conservation planning analyses for the New World and Britain, we conclude that a carbon-only strategy would not be effective at conserving biodiversity, as have previous studies. Nonetheless, we find that a combined carbon-biodiversity strategy could simultaneously protect 90% of carbon stocks (relative to a carbon-only conservation strategy) and > 90% of the biodiversity (relative to a biodiversity-only strategy) in both regions. This combined approach encapsulates the principle of complementarity, whereby locations that contain different sets of species are prioritised, and hence disproportionately safeguard localised species that are not protected effectively by carbon-only strategies. It is efficient because localised species are concentrated into small parts of the terrestrial land surface, whereas carbon is somewhat more evenly distributed; and carbon stocks protected in one location are equivalent to those protected elsewhere. Efficient compromises can only be achieved when biodiversity and carbon are incorporated together within a spatial planning process.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Anthropogenic climate change represents a major threat to biodiversity (Sala et al. 2000; Thomas et al. 2004; Malcolm et al. 2006; Parry et al. 2007) as well as to human well-being. Humanity's response is to (attempt to) develop and implement mitigation strategies that minimise the speed and eventual level of climate change, and to establish adaptation strategies that allow humans to cope with the climate change that does take place. These mitigation and adaptation strategies are themselves set to become major environmental drivers. For example, growth in biofuel production may increase the total global land area covered by low-biodiversity agricultural systems, whereas sequestering carbon in ecosystems may benefit biodiversity by maintaining larger intact areas than would otherwise have been the case (Laurance 2007; Betts et al. 2008; Coomes et al. 2008; Ebeling & Yasue 2008; Miles & Kapos 2008; Naidoo et al. 2008; UNEP-WCMC 2008; Tilman et al. 2009; Venter et al. 2009; van der Werf et al. 2009; Harvey et al. 2010; Scharlemann et al. 2010; Busch et al. 2011; Persson 2012; Strassburg et al. 2012; Turner et al. 2012). Prominent among these is the scheme known as Reducing Emissions from Deforestation and Degradation, REDD+, which focusses on 44 (mainly) tropical countries (UN-REDD 2012). However, it is likely that land-based carbon stocks and fluxes will eventually be audited for most if not all countries and, hence, carbon conservation strategies are also relevant outside the tropics. The biodiversity consequences of any carbon conservation strategy should ideally be built explicitly into the plans to maximise the co-benefits of such changes, targeting locations that are important for biodiversity conservation as well as for the retention of carbon stocks (UNEP-WCMC 2008; Harvey et al. 2010). Failure to achieve this targeting would represent a massive lost opportunity for biodiversity conservation or, worse, could cause direct harm to biodiversity by deflecting conservation attention and resources away from sites of higher biodiversity importance. Protection of high-carbon ecosystems may result in increased demand for timber production and alternative land uses, and hence increase deforestation pressures in unprotected regions. This may result in increased deforestation elsewhere, a phenomenon known as leakage (Gan & McCarl 2007; Ebeling & Yasue 2008; Strassburg et al. 2010), and some of these locations may be more important for biodiversity than those protected. Hence, the implementation of strategies to maintain ecosystem carbon stocks will determine how and whether they also benefit biodiversity.
Strategies to maintain carbon in intact ecosystems need to operate on multi-decadal and, preferably, centennial time scales, over which periods ecosystems will change and species shift their distributions. Ideally, we would be able to predict the future distributions of all species and account for these changes when determining priority areas, but conservation decisions must be made before this is possible – if it ever is. Nonetheless, well-established biogeographical principles provide guidance. The majority of terrestrial species have small geographical range sizes (endemics) where they either evolved or to which they retreated during the warm Holocene period (Gaston 2003). Not only do most species individually have small geographical ranges, but they are also concentrated into small areas of the Earth's land surface (Gaston 2000; Myers et al. 2000). These endemic-containing regions typically possess climates that are relatively unusual (compared to surrounding regions) and they are often relatively cool and heterogeneous, associated with mountain ranges and other sources of environmental heterogeneity (Ohlemüller et al. 2008). As a first-order approximation, they represent a ‘bioassay’ of the locations where species have in the past survived periods of varying climatic conditions, and specifically the Holocene warming (Sandel et al. 2011). We surmise that most of these endemic species have no prospect of undertaking major range shifts to different regions, so their survival will depend on the existence of suitable habitats within the immediate areas where they currently occur (e.g. at higher elevations or along watercourses). Hence, protecting regions containing narrowly distributed species remains a robust conservation strategy. These regions may also become the future refugia for species that are currently more widespread, for example as they retreat from the surrounding lowlands.
This does not mean that all of the species currently present in these regions will necessarily survive, but minimising other harms represents their best chance of survival. Species that are dwindling in areas of declining climatic suitability are expected to do so more rapidly if they face habitat degradation and other threats in the same regions: the quantity and quality of habitats remains paramount to successful biodiversity conservation (Hodgson et al. 2009, 2011; Canale et al. 2012). Although there has been much debate about strategies to conserve biodiversity under climate change (Heller & Zavaleta 2009), almost all rely on the existence of high-quality breeding habitats where species might be expected to survive (in large undisturbed and heterogeneous areas, climatic refugia, etc.) or to which they might move (e.g. via habitat corridors, stepping-stones, or through deliberate translocation). As species shift their ranges to new locations, they disproportionately colonise the habitats that exist within protected areas (Thomas et al. 2012). Maintaining intact ecosystems within regions where narrow-range species occur is, therefore, likely to benefit species shifting their distributions in response to climate change as well as those already in situ.
Trade-offs between the conservation of biodiversity and carbon stocks arise because not all of the regions where species are concentrated on Earth represent the highest-carbon environments, and reserves are often placed in remote locations where they are most convenient (Gaston et al. 2008; Joppa & Pfaff 2009), rather than where they provide greatest benefits to biodiversity or to the maintenance of carbon in ecosystems. The task is to find acceptable compromises where both carbon and biodiversity conservation can be achieved. Biodiversity is generally positively (but rather weakly) associated with ecosystem carbon, but the association is geographically variable (Strassburg et al. 2010), and even reverses in some regions (Anderson et al. 2009). Hence, ‘win–win’ conservation options may be available, but carbon-only conservation policies would not be guaranteed to deliver additional biodiversity benefits in all parts of the world (Chan et al. 2006; Turner et al. 2007, 2012; Naidoo et al. 2008; UNEP-WCMC 2008; Strassburg et al. 2010, 2012; Siikamäki & Newbold 2012). More efficient conservation solutions may be found by considering the distributions of both biodiversity and carbon within a single analysis (Venter et al. 2009; Moilanen et al. 2011a). However, the strength and sign of relationships between biodiversity and carbon stocks vary (Anderson et al. 2009), so further analyses are required before general conclusions can be drawn. Thus, the first major issue we address relates to the trade-off, in terms of conservation priorities, between selecting locations to protect carbon stocks in ecosystems and biodiversity. How much biodiversity can be gained for a specified sacrifice of carbon?
A related issue is to protect as many species as possible, a goal of almost all conservation priority analyses but less commonly incorporated explicitly in studies of biodiversity as an ecosystem service. The conservation of species has been and is likely to remain a key goal of biodiversity conservation, even if we re-badge our motivation in terms of cultural and supporting ecosystem services. It is useful to consider, therefore, how well carbon-only, biodiversity-only and compromise strategies protect species (Venter et al. 2009), rather than simply using an overall metric for biodiversity. Overall metrics may suggest that a particular compromise solution is attractive but are likely to miss some species entirely. Hence, it is important to adopt a complementarity approach, prioritising places that support different sets of species to ensure that the largest possible number of species is protected across all sites and to make sure that localised species are represented; and the related concept of irreplaceability, locations that contain unique species that cannot be protected elsewhere (Pressey et al. 1997; Margules & Pressey 2000; Moilanen et al. 2009). Protecting two locations that have equally high local biodiversity is, by this logic, more valuable if they contain different species than if the species are shared, whereas the carbon stocks of the two locations are additive. Complementarity needs to be incorporated in analyses that consider ecosystem services and species together; in the present context to ensure that localised species are retained even when they occur in relatively low-carbon environments (Moilanen et al. 2011a). One could argue that protecting localised species should remain the preserve of biodiversity-only conservation strategies and organisations, which could identify and then protect the species and ecosystems that have been overlooked by carbon conservationists. However, operating separate priorities is likely to be inefficient as well as institutionally and intellectually undesirable: the potential co-benefits for biodiversity and human well-being (Turner et al. 2012) are major attractions of the REDD+ and other carbon-based conservation approaches. Addressing carbon and biodiversity conservation together seems appropriate. Thus, the second major issue we consider here is how the benefits of different possible conservation strategies are distributed across species, to ensure that certain species are not missed in a stampede to establish conservation strategies to maintain ecosystem carbon stocks.
Species and carbon priorities
We first consider the performance of ‘biodiversity-only’ and ‘carbon-only’ conservation strategies to evaluate how effective these would be at protecting ‘each other’: that is, how much would a carbon-only strategy benefit biodiversity if biodiversity was not deliberately prioritised? It is then possible to consider the consequences of adjusting the relative priority (weighting) given to biodiversity vs. carbon in a series of combined carbon-biodiversity analyses.
We consider these trade-offs for the New World and Great Britain to illustrate continental-scale and national-scale trade-offs and priorities. We include extra-tropical regions in the expectation that land-based carbon accounting will eventually be deployed in many if not all countries, and because a tonne of ecosystem carbon oxidised inside and outside the tropics makes an equivalent contribution to greenhouse gasses. Although REDD+ is the current focus of international policy development, any analysis of current UN-REDD Programme Partner Countries would be incomplete, given that additional countries may join (e.g. Venter et al. 2009 analysed a larger set of countries than those that were partner countries, without knowing which will eventually join) and that the distributions of species we might want to protect do not necessarily respect the national boundaries of REDD countries. The New World analysis for birds is relevant to, but not confined to, the UN-REDD Programme Partner Countries in that region. The analysis for Britain shows higher resolution trade-offs for an individual temperate zone country to illustrate the trade-offs in a country where the government increasingly prioritises the conservation of ecosystem services (Defra 2011). By analysing such different geographical systems, it is also possible to assess whether it may be possible to generalise from the results.
The analyses were carried out using Zonation (Moilanen et al. 2005). Zonation is a spatial conservation planning tool that produces a priority ranking across the entire landscape, which has been applied to large-scale high-resolution prioritisation of biodiversity elsewhere (Kremen et al. 2008; Moilanen et al. 2008, 2011b, 2012; Franco et al. 2009) and of ecosystem services (Moilanen et al. 2011a). The principle of Zonation is that, applied to a gridded landscape, it successively discards cells of the lowest value (of cells still remaining), thereby producing a ranking of the contribution of each cell. For the biodiversity work, we used the core-area Zonation analysis variant (Moilanen 2007), in which the value of each cell is based on the proportion of the range of each species contained within that cell; making it harder to discard a cell containing small-range species than cells containing only widespread species. For the carbon-only analysis, the ranking is simply based on the estimated carbon stock in each cell.
Carbon-only rankings are illustrated for the New World and Great Britain (Fig. 1a and c). New World total organic carbon layers combining vegetation and soil at 50 km resolution were derived using data from the International Soil Reference and Information Centre (ISRIC-WISE 2000) and the Center for Sustainability and the Global Environment (SAGE, Gibbs 2006; Gibbs et al. 2007). The global ISRIC-WISE (2000) data contain soil carbon densities estimates (kg C m−2) to a depth of 1 m, and the SAGE data are based on satellite-derived land cover estimates of live vegetation carbon (mean Mg C ha-1). For Britain, 2 × 2 km grid resolution vegetation carbon data were based on NERC Centre for Ecology and Hydrology data, and soil carbon density (to 1 m depth) estimated using soil parameter, land use and soil series data from the National Soil Resources Institute (NSRI 2003); giving total carbon per 2 × 2 km grid cell. For the New World, carbon is concentrated in high-latitude northern peatlands and in tropical forests; in Britain, carbon is also concentrated in high latitude and altitude peat soils (Fig. 1a and c).
To illustrate the application of Zonation to biodiversity data, the biodiversity-only Zonation analysis was used to select cells designed to maximise species representation of the New World's birds and conservation-priority species in Great Britain. This approach identifies locations where many small-range species exist as priority areas for conservation (Fig. 1c and f). We converted GIS polygon shapefiles of the distributions of 3988 New World bird species (Ridgley et al. 2005) to presence/absence data at 50 × 50 km grid resolution on a Behrmann projection equal area grid using ArcGIS 9.3 (ESRI, Redlands, California, USA). The British species represent those prioritised for conservation by the UK government, based on distributions, abundances and distribution and population trends (http://www.ukbap.org.uk/NewPriorityList.aspx). Species included are all terrestrial UK Biodiversity Action Plan species for which data were available. The species included were 18 mammal species, 41 birds, 10 herptiles, 22 butterflies, 207 plants and 102 bryophytes. Distributions of these 400 species were derived at tetrad (2 × 2 km) resolution using data from the Centre for Ecology and Hydrology Biological Records Centre, Butterfly Conservation and the British Trust for Ornithology. Based on these data, priority areas are identified as mainly in montane areas of South and Central America, Caribbean islands and the Atlantic forests of the New World (Fig. 1c). In Britain, they are concentrated in the south and in northern mountains (Fig. 1f). These two examples illustrate that the approach can be employed at different spatial resolutions and over different spatial extents.
Comparisons of Fig. 1a and c, for the New World, and Fig. 1d and f, for Britain show that the priority areas for biodiversity and for carbon are quite different. For the New World, conserving the 30% of land (using 50 × 50 km grid resolution) with the highest predicted ecosystem carbon would maintain 47.4% of the estimated continents’ ecosystem carbon stocks, but only 34.1% of the biodiversity value for birds. Biodiversity value is measured here as the proportion of the range of each species contained within priority areas, averaged across species.
Alternatively, the biodiversity-only Zonation analysis for the New World would maintain as much as 71.7% of the biodiversity value for birds within 30% of the land (Fig. 1c), twice that obtained by the carbon-only analysis. However, the biodiversity-only solution contained only 30.8% of terrestrial ecosystem carbon, much lower than the best possible choice of locations for carbon conservation. Consequently, maintaining natural ecosystems for biodiversity alone would not be particularly effective at safeguarding carbon stocks – at this continental scale. Likewise, maintaining carbon stocks alone would be an inefficient way of delivering biodiversity conservation.
The results were qualitatively similar for Britain. The largest carbon stocks are held predominantly in peat soils in the north (Fig. 1d), whereas high biodiversity values are found predominantly in uncultivated habitats in the south (Fig. 1f). In Britain, the best 30% of cells for carbon (Fig. 1d) contain 58.7% of the carbon stocks, but only 25.4% of the biodiversity conservation value for the species considered (the UK Biodiversity Action Plan priority species) – a selection of cells that is worse than random, given that a random selection of 30% of cells will on average deliver 30% of the biodiversity value. In contrast, using Zonation to target biodiversity conservation (ignoring carbon), the best 30% of cells for these species (Fig. 1f) contain 92.3% of the biodiversity value, yet only 25.4% of the ecosystem carbon – worse than a random selection of sites for carbon.
The biodiversity-only analysis for the New World picks out centres of endemism that contain species not found elsewhere in the world, a desirable feature of conservation priority planning. The British-scale analysis identifies particular locations that contain habitats for species that are particularly localised or threatened within Britain. Again, this is potentially a desirable attribute from a national perspective, in the absence of many endemic species within Britain. However, neither strategy would be particularly effective at conserving carbon.
It is becoming clear that these conflicts between carbon and biodiversity priority areas are likely to be qualitatively robust across regions, spatial scales and analytical approaches. For example, Siikamäki & Newbold (2012) found, in a global analysis, that the number of endemic mammal + bird + amphibian species in an ecoregion was positively correlated with the above-ground carbon density, but the correlation coefficient was 0.1, thus explaining only 1% of the variation, suggesting that carbon conservation would perform little better for biodiversity than randomly located ecosystem protection. Similarly, Venter et al. (2009) carried out a country-level analysis of the potential allocation of REDD funds to forest-losing countries, and concluded that ‘if REDD focusses solely on cost-effectively reducing carbon emissions, its benefits for biodiversity are low, protecting only slightly more vertebrate species than if funds were allocated at random’. These and our results imply that ‘biodiversity-blind’ carbon planning would represent a missed opportunity and, at worst, could have some negative consequences for biodiversity if protecting some areas to maintain carbon stocks could result in a degree of redirection of land use changes (i.e. ‘leakage’; Ebeling & Yasue 2008; Gan & McCarl 2007; Strassburg et al. 2010), potentially increasing the pressures in areas of higher biodiversity value. Similarly, ‘carbon-blind’ biodiversity prioritisation could result in increased atmospheric CO2 levels if this resulted in additional losses (e.g. conversion to agriculture) of unprotected high-carbon ecosystems outside biodiversity priority regions. Biodiversity conservation does not automatically follow from the conservation of carbon, or vice versa.
Conservation planning tools that identify efficient multi-species conservation priorities using the principles of complementarity can also prioritise other ecosystem services – here carbon stocks – at the same time as species (Moilanen et al. 2011a). To achieve this aim, we successively discard locations of lowest value for carbon stocks and biodiversity (species), varying the relative weightings of carbon vs. biodiversity in different analyses. Effectively, carbon is treated as equivalent to an additional species in the analysis. For standard biodiversity analyses within Zonation, each species can be given a weighting (e.g. to prioritise certain species), which makes it harder to discard cells that contain species with higher weights. In the carbon-only analysis described above, each species was weighted 0 (and hence ignored) and carbon 1, and in the biodiversity-only analysis the weighting of each species was set to 1 and carbon to 0 (and hence ignored). In the combined carbon-biodiversity analyses, we set each real species to a weight of 1, and varied the weight of carbon: a very high weight on carbon leads to a solution that converges on selecting locations by carbon only, whereas the solution converges on that for biodiversity alone when a high weight is given to biodiversity. The weighting ascribed to carbon was defined in units of n; where n was the total number of biological species in the analysis (n =3988 bird species for the New World; n =400 priority species in Britain). We varied the weighting assigned to carbon in the following steps: 64n (mainly carbon), 32n, 16n, 8n, 4n, 2n, n (carbon weight equivalent to the weighting ascribed to all species together), 0.5n, 0.25n, 0.125n, 0.0625n, 0.0312n, 0.0155n, 0.00781n, 0.00390n (mainly biodiversity). These are displayed in Fig. 2 as a Relative Weighting [RW] Index, with values ranging from 2−6 corresponding to a carbon weighting of 64n (i.e. species are weighted as 2−6 of the value of carbon), through an RW Index of 20 when the carbon weighting is n (i.e. all species together are equivalent to carbon), to 28 for a carbon weighting value of 0.00390n (i.e. all species together carry a weight 28 that of carbon). Only the relative weight is important to the analysis, so the same results would have been obtained had the weight given to carbon been held constant, and those for species varied. The relative weights could be set by the relative financial value of carbon and species, if the value of species could be agreed, or by politically based (social) values. Alternatively, the set of potential results for all relative weights could be assessed for their social and political acceptability, after the analysis is completed.
Figure 2 shows how the expected amount of carbon and biodiversity conservation that could be achieved varies with these relative weightings, for 30% of the land protected. The shapes of the two curves are important: the biodiversity value curves in Fig. 2 show that adding information about species to the analysis results in a rapid increase in the biodiversity value of the priority areas. The benefit to biodiversity, measured as the average proportion of each species’ range protected, is achieved with little initial reduction in carbon protected because the carbon stock curve is more sigmoid. This is true for both sets of data, despite the variation in extents, resolutions and taxa considered.
By adjusting the relative carbon-to-species weighting, it is possible to evaluate the extra biodiversity value gained for a given amount of carbon foregone. To illustrate this, we consider how much extra biodiversity could be obtained if we were prepared to forego 10% of the maximum carbon that we could protect. We ran iterations of Zonation with different relative weightings until we obtained a value for carbon corresponding to this 10% sacrifice. The Zonation solutions that correspond to foregoing 10% of the maximum carbon value (that would be achieved for 30% land protection) are shown in Fig. 1b and e. The combined carbon-species solutions that maintain 90% of the maximum possible carbon also maintain 91.4% of the maximum possible biodiversity value for the Americas, and 90.9% of the biodiversity value for Britain (compared with the biodiversity-only solutions, in each case). Thus, the solution that foregoes 10% of the carbon represents biodiversity in conservation areas nearly as well as the biodiversity-only solution. This qualitative conclusion is not sensitive to the percentage of land included within protected areas, or the exact percentage of carbon foregone (until the solutions are heavily weighted towards biodiversity). Policies to achieve this might include assigning different monetary values to the carbon protected in different locations based on the ranking of locations within the analysis.
The combined priority areas are biased towards known centres of endemism in the New World analysis (Fig. 1b) and locations containing highly localised species in Britain (Fig. 1e), indicating that satisfactory target areas have been identified. Of course, the higher the quality and spatial resolution of data used, the ‘better’ the solution is likely to be, and data quality may be a constraint in some circumstances. Nonetheless, global data are increasingly good for vertebrates, at least, and the challenges in improving the quality of data on ecosystem carbon stocks (and then fluxes) are no different for this Zonation approach than for any other implementation of REDD+ strategies.
Not all ecosystems and regions are equally threatened with conversion to low-carbon, low-biodiversity land uses (Lambin et al. 2003; Venter et al. 2009). Northern peatlands do not fall within REDD Programme Partner countries (UN-REDD 2012), for example, although degrading upland and northern peatlands are among the highest priorities in the United Kingdom (IUCN 2012) and the UK government is developing a more ecosystem-based approach to conservation (Defra 2011). However, we would not advocate carrying out analyses on priority regions or target countries in isolation: such analyses generate artefacts whereby priority areas are biased towards national and other planning boundaries (Moilanen et al. 2012). It appears to be more robust to carry out one analysis for a larger area (as here for New World birds) and then to ‘cut out’ the rankings for cells within a focal country, or across REDD Programme Partner countries, if one wishes to define priorities within a specified sub-region. This also avoids priority areas changing every time a new country joins a scheme.
Species representation in conservation solutions
The above analyses score biodiversity value as the average representation (proportion of initial range) per species. Variation in representation among species is also important because species that are largely or entirely excluded from a protected area network may be at risk of extinction if habitats outside protected areas are destroyed, particularly if excluded species have small geographical ranges. Carbon-only solutions for 30% of the land protected resulted in low representation (taken as < 5% of the range area of a species falling within a priority area) of 404 of 3988 (10.1%) of New World bird species and 179 out of 400 (44.7%) British priority species (Fig. 2; histograms labelled ‘A’). In both cases, carbon-conservation priorities ‘miss’ substantial numbers of species. The percentage missed appears much larger in the British analysis because these have already been identified as priority species, and often have small ranges, whereas the New World analysis includes all bird species, including those that are widespread (see below). In contrast, the biodiversity-only solutions contained low representation of only 137 (3.4%) of the New World bird species and 0 (0%) of the British priority species (Fig. 2; histograms labelled ‘E’).
When biodiversity and carbon are included within the same analysis, all species are deliberately retained through the Zonation ranking as a result of its built-in complementarity. The process also maintains a higher fraction of the ranges of small-range species: Zonation algorithms retain cells if they contain a high fraction of the remaining distribution of any species. Hence, there is a rapid decline in numbers of species with low representation, even when the biodiversity weighting is very low. The combined analyses that forwent 10% of the maximum possible carbon (for 30% of land area) encompass the distributions of most species; showing low representation of only 111 (2.8%) New World bird species and none of the British species (Fig. 2; histograms labelled ‘D’). This 10% sacrifice analysis is as effective as the biodiversity-only analysis in representing the British species, and the New World 10% sacrifice analysis appears to support an extra 26 species, compared to the biodiversity-only analysis (although these extra 26 species are not particularly threatened).
As described in the introduction, conservation strategies should ideally encompass the distributions of small-range species. This is not achieved in the carbon-only analyses, for which small-range species are little or no better represented in conservation solutions than are more widespread species (Spearman correlations between the range size of a species and the proportion of its range included in top 30% priority areas: New World birds Rs = −0.052, P < 0.001, n = 3988; British priority species Rs = −0.014, P = 0.787, n = 400). Much of the apparent biodiversity value of the carbon-only solutions arises through the conservation of widespread species, many of which do not need a great deal of protection, whereas it is hit and miss for small-range species. For carbon-only solutions, large blocks of land are prioritised in ‘high-carbon ecosystems’, so some small-range species are entirely encompassed within the priority areas, whereas others are completely missed – universally distributed species are always protected to a degree.
In contrast, the combined analyses (30% area, 10% carbon foregone) disproportionately protect localised species, with strong negative correlations between range size and the proportion of the range protected (New World birds Rs = −0.641, P < 10−10, n = 3988; British priority species Rs = −0.811, P < 10−10, n = 400). Indeed, the 2.8% of New World bird species that appeared to be poorly represented (i.e. < 5% of their ranges fall within the 30% of land prioritised) in the combined analysis were widespread species (that mainly live in lower carbon ecosystems), for which specific conservation measures are unlikely to be required. Note that it is possible to alter the weighting assigned to different types of species to ensure that threats other than range size are incorporated (Moilanen et al. 2005). Indeed, the British analysis shown here is based on conservation-priority species, effectively weighting non-threatened species as zero within the analysis.
The analyses provide a numerical argument that biodiversity conservation does not automatically follow from the conservation of carbon stocks, but that quantitative tools can deliver win–win solutions. However, relationships between carbon and biodiversity differ among regions (Fig. 1; Anderson et al. 2009), so specific analyses are required to ensure that biodiversity co-benefits follow from the conservation of carbon stocks within ecosystems.
By combining both carbon and species within a single conservation planning framework, large increases in biodiversity conservation could be achieved for only a small reduction in the amount of carbon stocks maintained within the protected ecosystems. These gains could be defined in terms of the average representation across all species, the inclusion of (almost) all species in priority areas, and increased representation of localised species. Our expectation is that these gains will prove to be quite general. Most species have very small geographical ranges (Gaston 1996, 2003), so the capacity to protect a high proportion of species within a small area is general. Some of these areas were ‘missed’ in carbon-only conservation strategies. However, this feature of biodiversity enables it to be incorporated within an overall strategy at relatively low cost (Venter et al. 2009). Different species are found in different ecosystems (in different environments/regions), whereas any two locations within the same ecosystem are likely to share many species (Nekola & White 1999; Qian et al. 2005; Gaston et al. 2007). Foregoing the conservation of a relatively small area of one carbon-rich ecosystem has little impact on the total number of species protected within that ecosystem, provided that much of it will be protected anyway. In contrast, adding a corresponding small area of a different ecosystem that would not be prioritised for carbon alone rapidly increases species representation by including a different set of species associated with the new ecosystem. The gains are strongest when the extra areas to be protected are located in biodiversity hotspots where many small-range species are present.
Species can only survive if their habitats survive. The locations prioritised in the combined analyses appear to be appropriate strategies under climate change as well as under current conditions, in as far as this is possible to judge. The priority areas identified here include the ‘ends of’ most temperature and moisture gradients, including in the Andes and Central American mountains, the Atlantic forest of Brazil, along the course of the Amazon river, and high latitude areas, such as the rainforests of Chile and the western USA (Fig. 1). These are unusual local climates and topographies (in some cases geologies), relative to those in the surrounding regions, and hence contain many small-range species (Ohlemüller et al. 2008). Although the climates of these regions will change over the coming century, topographies, geologies and proximities to coastlines remain. Hence, most of these regions are likely to continue to support regionally unusual conditions – and so support localised species. Furthermore, these regions typically possess high local environmental heterogeneity, apparently enabling species within then to survive shifting climatic conditions (Ohlemüller et al. 2008; Sandel et al. 2011), an emerging principle of climate change conservation (Hodgson et al. 2009).
In conclusion, the shapes of the trade-off curves (Fig. 2) when selecting areas to conserve carbon and biodiversity enable extremely large biodiversity gains to be achieved for relatively modest level reductions in the amount of carbon protected. Incorporated properly within economic incentives and policies, the conservation of carbon stocks could revolutionise humanity's capacity to protect biodiversity. Whether this is achieved will depend on how carbon conservation strategies are incentivised and deployed.
The UK Population Biology Network (Natural Environment Research Council and Natural England) funded the research. UK soil data were supplied by the National Soil Research Institute (DEFRA funded project SP0511; as part of a licence LC0072/00 for the UK Centre for Terrestrial Carbon Dynamics). UK vegetation data were provided by the Centre for Ecology and Hydrology (CEH-Edinburgh); global soil and vegetation carbon data by the Oak Ridge National Laboratory data centre (http://daac.ornl.gov/). New World bird data were provided by NatureServe in collaboration with Robert Ridgely, James Zook, The Nature Conservancy Migratory Bird Programme, Conservation International Center for Applied Biodiversity Science, World Wildlife Fund USA, and Environment Canada WILDSPACE. We thank all those responsible for the original bird records. The British Trust for Ornithology (BTO) and CEH provided UK biodiversity data. We thank the national schemes and societies and associated body of volunteer recorders for collecting and collating the UK biodiversity data used in this article: the Botanical Society of the British Isles, the British Bryological Society, BTO, Butterfly Conservation, the Mammal Society and the National Amphibian & Reptile Recording Scheme. UK analyses are based on data provided through EDINA UKBORDERS with support of ESRC and JISC, and uses boundary material copyright of the Crown. K. J. G. holds a Royal Society-Wolfson Research Merit Award. A. M. was supported by the Academy of Finland, Finnish Centre of Excellence Programme 2006–2011, grants 213457 and 129636; A.M. thanks the ERC-StG grant #260393 (GEDA) for support.
All authors contributed to the development of the project; AH, TQ, DBR, SG contributed data; BJA, AM & FE carried out analyses; CDT wrote the first draft of the manuscript and all authors contributed to revisions.