Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Global congruence of carbon storage and biodiversity in terrestrial ecosystems

Global congruence of carbon storage and biodiversity in terrestrial ecosystems Introduction Conversion of natural ecosystems is the second largest source of human‐induced climate change, accounting for 17–20% of anthropogenic greenhouse gas emissions ( Gullison 2007 ; Metz 2007 ), and is the single most important driver of species extinctions ( Baillie 2004 ). Yet, despite major global scientific ( Baillie 2004 ; MEA 2005 ; Gullison 2007 ; Metz 2007 ) and political ( United Nations 1992, 1993 ) backing for the importance of forests and other ecosystems for global climate regulation and biodiversity conservation, deforestation rates remain unabated ( FAO 2006 ). Conversion is now concentrated in the most carbon‐rich and biodiverse biome on Earth, the tropical forests ( FAO 2006 ). Recent research suggests that the role these forests play in global climate regulation might be greater than previously thought ( Stephens 2007 ; Bonan 2008 ; Lewis 2009 ) and that some of its richest biodiversity hotspots are still poorly explored ( Carnaval 2009 ). New studies have also highlighted that climate change, deforestation, carbon storage in biomass and biodiversity are closely interlinked. For instance, in addition to contributing directly to global warming, deforestation also makes forests more susceptible to the effects of climate change ( Malhi 2009 ; Phillips 2009 ). Climate change will be a major driver of future species extinctions ( Thomas 2004 ) and one of its possible feedbacks is a further depletion of forest carbon stocks ( Malhi 2009 ). Biodiversity, on the other hand, might alleviate some of these effects by making ecosystems more resilient ( Reusch 2005 ; Tilman 2006 ) and, possibly, more productive ( Flombaum & Sala 2008 ). A landmark global deal being discussed under the United Nations Framework Convention on Climate Change is likely to include an international mechanism of financial incentives to reduce emissions from deforestation and forest degradation (REDD) ( Gullison 2007 ; UNFCCC 2007 ; Strassburg 2009 ). As the estimated magnitude of the financial incentives being discussed ( Kindermann 2008 ; Strassburg 2009 ) dwarfs current conservation expenditures in developing countries ( James 1999 ), REDD could trigger the biggest paradigm shift in conservation history. While it is generally assumed that REDD would have positive impacts for biodiversity conservation, this assumption has not been rigorously tested. Attention to the design of such a mechanism could enable increased gains for biodiversity, while reducing the risks, e.g., for those ecosystems that fall outside the scope of REDD ( Miles & Kapos 2008 ). Better understanding of the congruence between carbon storage and biodiversity conservation would help to maximize the gains and better address the risks. Here we use high‐resolution data to analyze for the first time the global congruence between carbon storage in biomass and biodiversity in order to investigate the synergies between climate‐ and biodiversity‐oriented conservation. Methods and data Our analysis is based on three global data sets of the distribution of mammal ( Schipper 2008 ), amphibian ( Stuart 2004 ), and bird ( Orme 2005 ) species, totaling 20,697 species, and on a new global carbon data set ( Ruesch & Gibbs 2008 ). The carbon data set presents estimates of above‐ and below‐ground (root) biomass, and is the first to apply consistently the IPCC Good Practice Guidance ( Penman 2003 ) to the whole terrestrial surface. We investigated the congruence between carbon and each of three biodiversity indices: richness (number of species per cell), threat (number of threatened species per cell), and range‐size rarity (number of species per cell whose ranges are in the lowest quartile for their class—see Supporting Information for details). All data (described and discussed in the Supporting Information) were analyzed on a geodesic discrete global grid system, defined on an icosahedron and projected to the sphere using the Inverse Snyder Equal Area (ISEA) Projection ( Sahr 2003 ). This corresponds to a hexagonal grid composed of individual units (cells) that retain their shape and area (∼12,500 km 2 , similar to a 1 × 1 degree rectangle at the equator) throughout the globe. These are more suitable for a range of ecological applications than the most commonly used rectangular grids ( Birch 2007 ). The analysis was conducted on a subset of this global grid: all 15,018 cells containing any land. The range of each species was converted to the hexagonal grid for analysis; a species was assumed to be present in a given cell if any part of the cell overlapped the species’ mapped range. For each cell: a Richness Index was calculated as the total number of species mapped in the cell (Figure S1A); a Threat Index was calculated as the number of threatened species per cell (Figure S1B); and a Restricted‐Range Index was calculated as the number of restricted‐range species (the 25% species in each taxon with the smallest range size) per cell (Figure S1C). Carbon data were also converted to the hexagonal grid, with the value in each cell corresponding to mean carbon density for the land area (Figure S2). Congruence between carbon and biodiversity was investigated visually (through maps) and analytically (through correlations). Global maps of the relationship between carbon and each of the biodiversity indices ( Figure 1 ) were produced by plotting each hexagon with a color defined on a two‐dimensional scale ( Williams & Gaston 1998 ), with intensity on the blue axis defined using the carbon density and intensity on the red axis defined from the biodiversity index. The color scales are based on statistical distribution of the data: a linear scale is used for values falling between minus three standard deviations and plus three standard deviations. Any cells with values outside this range are grouped into single color bands and the ends of the scale. Maps were plotted using GMT software ( Wessel & Smith 1998 ). Spearman's rank ( r S ) correlation coefficients were calculated for the relationship between carbon and each of the biodiversity indices ( Table 1 ), and their tests accounted for possible spatial autocorrelation (see Supporting Information). 1 Global congruence between biomass carbon and overall species richness (A), threatened species richness (B), and restricted‐range species richness (C). The two‐dimensional color scale used displays both the concentration of biomass carbon and biodiversity and the congruence between them. The intensity on the vertical blue axis represents biomass carbon density (tons of C per hectare) and the intensity on the horizontal red axis the richness of the respective biodiversity index (number of species per cell). Darker shadings correspond to higher concentrations. Colors along the 45° diagonal (white‐to‐green shading) indicate high congruence between biomass carbon and biodiversity. Areas with low congruence are represented by colors along the white‐to‐blue (higher carbon, lower biodiversity) or white‐to‐red (higher biodiversity, lower carbon) axes. The color scales are based on statistical distribution of the data: a linear scale is used for values falling between minus three standard deviations and plus three standard deviations. Any cells with values outside this range are grouped into single color bands at the ends of the scale. 1 Correlations between carbon biomass and biodiversity for each of the biodiversity indices considered (Richness, Threat, and Restricted Range) and for either all species or for each of the taxa (amphibians, birds, or mammals) separately. r S is Spearman rank correlation coefficients of all cells, EDF is Dutillieul‐corrected degrees of freedom for the first r S coefficients; P ‐values are the corresponding max P ‐values; r S nonautocorr present the mean and the 95% confidence intervals of Spearman rank correlation coefficients for replicates of subsets of cells 1,400 km apart from each other for which spatial autocorrelation is negligible (see supporting information) Richness index Threat index Restricted range index r S EDF P ‐values r S non‐autocorr r S EDF P ‐values r S non‐autocorr r S EDF P ‐values r S non‐autocorr All species 0.82 53.8 <0.0001 0.73 (0.59–0.83) 0.55 68.4 <0.001 0.39 (0.21–0.54) 0.42 149.3 <0.001 0.36 (0.17–0.52) Amphibians 0.84 51.0 <0.0001 0.77 (0.68–0.83) 0.37 250.6 <0.0001 0.35 (0.19–0.48) 0.24 357.5 <0.01 0.22 (0.06–0.35) Birds 0.80 57.6 <0.0001 71 (0.58–0.81) 0.43 73.8 <0.01 0.28 (0.09–0.47) 0.38 149.5 <0.001 0.32 (0.10–0.51) Mammals 0.79 54.3 <0.0001 0.69 (0.56–0.78) 0.52 64.7 <0.001 0.38 (0.22–0.49) 0.31 266.2 <0.001 0.29 (0.11–0.46) Results and discussion Overall, our results support the expectation that mechanisms for conserving biomass carbon (such as REDD) would have substantial cobenefits for biodiversity, and vice versa. Indeed, we found high congruence between species richness and biomass carbon at the global level ( r S = 0.82, Table 1 ; Figure 2A ). This result is not unexpected, given that on the one hand biodiversity‐poor ecosystems such as most deserts and polar regions are also carbon‐poor (shades of white and pale green in Figure 1A ), and on the other hand tropical forests (particularly the Amazon) are rich in both carbon and species (dark tones in Figure 1A ). 2 Global scale relationship between biomass carbon and overall species richness (A), threatened species richness (B), and restricted‐range species richness (C). In addition to all 15,018 individual data points, each panel presents an envelope containing 95% of the cells. To construct the envelope, cells were grouped based on their carbon content (each group being 10 tons of carbon wide). For each group the extremes were then excluded, with the cells with the highest or lowest 2.5% number of species falling outside the upper (blue line) or lower (red line) bounds, respectively. The black line represents the mean number of species of each group. There are also synergies between conservation of threatened or restricted‐range species and carbon‐based conservation ( Table 1 ), but these are less striking (at least at the global scale). Indeed, we found moderate congruence between biomass carbon and threatened ( r S = 0.55, Figure 2B ) and restricted‐range ( r S = 0.42, Figure 2C ) species. This is probably at least partially explained by the fact that although the two largest expanses of tropical forest (the Amazon and the Congo Basin) have simultaneously high carbon densities and species richness (hence high congruence in Figure 1A ), they have relatively few restricted‐range or threatened species. The coarse scale of our analysis is also likely to reduce the congruence between carbon and rare and threatened species by masking finer scale relationships in some areas (see below). In addition, prioritizing regions under threat of conversion (as a carbon mechanism would likely do) would likely substantially increase the correlation with threatened species The regional patterns identified in Figure 1 show a wide variation in synergies among carbon and biodiversity conservation, with correspondingly varied policy implications. Some areas would provide a high degree of cobenefits for the goals of conserving biomass carbon and biodiversity, including areas of high congruence between carbon and overall species (e.g., the Amazon; Figure 1A ), threatened species (e.g., Indonesia; Figure 1B ), and endemic species (e.g., New Guinea; Figure 1C ). Some areas that are of high priority for one of the conservation objectives but have moderate importance for the other could still benefit from complementary financing from this second one. For example, Mainland Southeast Asia, a priority for biodiversity conservation, could benefit from complementary funding from carbon conservation. Finally, regions that are of high priority for one of the goals but of low importance for the other deserve careful consideration to ensure that they are not neglected. In particular, if a large‐scale carbon‐focused REDD mechanism is implemented, biodiversity‐rich and relatively carbon‐poor regions could suffer from a double conservation jeopardy, with conservation investment diverted away from them, and human pressure redirected toward them, as carbon‐rich areas become the focus of conservation efforts. Areas potentially at risk include some that are widely recognized as global biodiversity conservation priorities such as the Brazilian Cerrado, the Cape Floristic province, and the Succulent Karoo ( Mittermeier 2004 ). In some regions, the lack of congruence between carbon stocks and biodiversity might be an artifact of the coarse scale of our spatial analysis ( Figure 3 ). Extensive deforestation often leaves remaining natural habitats concentrated in sparse forest fragments. These regions have low carbon per 1° cell, but each remaining forest fragment can be carbon‐rich on a per‐forest area basis. Habitat loss means many of these forests’ species will be threatened, particularly if their ranges are small. Hence these regions might emerge in our coarse‐scale analysis as having low carbon per cell, yet high counts of threatened or range‐restricted species. If mapped at a finer scale, they might instead be identified as small fragments of high value for both carbon and biodiversity, surrounded by a matrix of low carbon and low biodiversity. This is almost certainly the case of parts of the Atlantic Forest of South America, the Tropical Andes, and Southeast Asia ( Figure 1B ). Often, those remaining fragments are also highly threatened from future deforestation, and so their conservation would provide high benefits in terms of both avoided emissions and avoided extinctions. Future analyses at finer scale would be able to better capture these relationships. This refinement would be especially important to investigate the synergies between carbon‐oriented conservation and threatened and rare species, and in particular for species that are both rare and threatened, maybe the most valuable group for biodiversity conservation ( Ricketts 2005 ). 3 An illustration of the effect of scale on the congruence between carbon and biodiversity in fragmented areas. At the spatial resolution employed in our analyses (cells ∼12,500 km 2 ), the Atlantic Forest of South America (part of which is outlined by a box) emerges as an area of low congruence between carbon and biodiversity, appearing to contain relatively low carbon (A; shades of green correspond to carbon per cell) but high biodiversity value, for example, in terms of numbers of threatened species (B; shades of red correspond to number of threatened species per cell). At finer scales, though, the congruence is much higher, as most of the species that give this region its exceptional biodiversity value are restricted to the remaining fragments, corresponding to 7% of the original forest cover ( Carnaval 2009 ), which are carbon‐rich. The endemic and Endangered Red‐billed Curassow ( Crax blumenbachii ), for example, (C; photo by Alberto Teleuko, Wikimedia Commons), is only known from a few forest patches (D, inset; stars—species’ records; green—forest fragments) ( Bird Life International 2008 ). We believe that these fragmented areas provide some of the most valuable opportunities for “win–win” carbon and biodiversity conservation. We have also calculated correlation coefficients between carbon and biodiversity indices (richness, threat, and restricted range) separately for each of the taxonomic groups (amphibians, birds, mammals) ( Table 1 ). We found that the results are consistent with those found for the overall biodiversity indices, demonstrating that they are not driven by any particular taxon among those analyzed, and suggesting that they are robust to the choice of taxa as surrogates of biodiversity. However, some areas have very high value for biodiversity conservation despite having lower vertebrate diversity (see Supporting Information). We investigated the effect that spatial autocorrelation has on our results through three different approaches (Tables S1–S3 and Supporting Information). Although most correlation coefficients go down when redoing the analyses for sets of non‐autocorrelated samples ( Table 1 and Table S2), the decreases are relatively small (e.g., for Species Richness, nonautocorrelated r S is 0.73, down from 0.82 for total cells). Most importantly, all coefficients have P ‐values < 0.01 even when controlling for spatial autocorrelation and none of the conclusions or relative comparisons discussed is affected. Our analysis was focused on carbon stored in above‐ and below‐ground biomass only. The soil carbon pool is 4.5 times the biotic carbon pool ( Lal 2004 ). But unlike biomass carbon, there is no agreement on the magnitude or even the sign of the change in this pool following land‐use change ( Guo & Gifford 2002 ), which varies greatly depending on several factors (e.g., replacement land cover). Therefore it is not possible to assess the importance of soil carbon stocks for climate change mitigation without taking into consideration a dynamic land‐use context. For these reasons, we opted to focus our analysis on biomass carbon, the primary carbon pool affected by land cover change. Nonetheless, a sensitivity analysis whereby total carbon is measured by adding biomass carbon and 25% of the soil carbon stock (Figure S3 and Supporting Information) showed only a relatively small decrease in correlation between vulnerable carbon and species richness ( r S = 0.73, down from r S = 0.82). Our analysis is about carbon stock (that can be lost due to land conversion) and it fails to include the value of different ecosystems for carbon sequestration. There is now clear evidence that mature forests, both tropical ( Phillips 2008 ; Lewis 2009 ) and temperate ( Wirth 2009 ), sequester substantial amounts of carbon every year, and some of these are highly valuable for biodiversity. We have assessed the synergies between climate mitigation and biodiversity conservation based solely on stocks of carbon and species. An important aspect missing from our analysis is that the conservation of these stocks is more relevant where their rate of loss might be greater (i.e., areas under higher risk of deforestation). In this sense, a further step to improve this assessment would be to analyze these synergies in a dynamic land‐use change context. A mechanism focused purely on carbon retention will not be optimal at conserving biodiversity, but there is much scope for improving the gains for biodiversity conservation, with low or no losses for carbon mitigation. Indeed, we found high variability in species counts across cells with similar biomass carbon levels ( Figure 2 ). For instance, for total species richness there is approximately a five‐fold difference in species richness per cell between the lower (red line, Figure 2A ) and upper (blue line, Figure 2A ) bounds of a 95% envelope. Cells in the upper bound usually have twice as many species as those expected by chance (mean, black line, Figure 2A ). Taking biodiversity patterns into account when prioritizing areas for implementation could therefore substantially improve the contribution of REDD or similar mechanisms to global biodiversity conservation. This could be done by inserting a biodiversity premium for emissions from more biodiversity‐rich areas directly into the REDD mechanism, by setting aside a fraction of REDD financing for targeting biodiversity‐rich areas that would not be conserved for their carbon content alone ( Strassburg 2009 ), or by promoting cooperation between programs for REDD and conservation at a national to international scale, redirecting conservation funding to these areas ( Grainger 2009 ). Climate change and biodiversity loss are two crises of global magnitude, each posing individual as well as synergistic risks to human well‐being and curtailing humanity's future options ( MEA 2005 ; Metz 2007 ). Most of the world's governments have pledged both to reduce the rates of biodiversity loss ( United Nations 1993 ) and to prevent dangerous climate change ( United Nations 1992 ). Rather than addressing these two crises separately, it makes sense, where appropriate, to combine efforts and thereby achieve mutual gains. Our results confirm that mechanisms to reduce carbon emissions from deforestation and forest degradation will have side benefits in terms of biodiversity protection, as expected given that forest conversion is a major driver of both climate change and biodiversity loss. Nonetheless, a purely carbon‐focused mechanism does not necessarily focus on those forests where biodiversity conservation is most urgent, and attention should also be given to biodiverse areas outside the scope of proposed REDD mechanism that could become increasingly threatened. But our results suggest that if biodiversity distribution is taken into account, there might be substantial room for additional gains for biodiversity conservation without compromising the effectiveness of climate change mitigation. Our study confirms that the Parties to the United Nations Framework Convention on Climate Change have a singular opportunity to address two of the biggest challenges of our time, first and foremost by approving incentives for reducing carbon emissions from deforestation and forest degradation, and second by maximizing the potential of this mechanism to reduce rates of global biodiversity loss. Editor : Dr. Pablo Marquet Acknowledgments We are grateful to the thousands of experts who contributed to the IUCN Global Mammal Assessment, to the IUCN Global Amphibian Assessment, and to the Red List assessment of bird species (coordinated by BirdLife International). We thank two anonymous referees for comments. B.S. thanks CAPES/Brazilian Ministry of Education for the financial support received. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Conservation Letters Wiley

Global congruence of carbon storage and biodiversity in terrestrial ecosystems

Download PDF
8 pages

Loading next page...
 
/lp/wiley/global-congruence-of-carbon-storage-and-biodiversity-in-terrestrial-WwUJKtqzNN

References (39)

Publisher
Wiley
Copyright
"Copyright © 2010 Wiley Subscription Services, Inc., A Wiley Company"
eISSN
1755-263X
DOI
10.1111/j.1755-263X.2009.00092.x
Publisher site
See Article on Publisher Site

Abstract

Introduction Conversion of natural ecosystems is the second largest source of human‐induced climate change, accounting for 17–20% of anthropogenic greenhouse gas emissions ( Gullison 2007 ; Metz 2007 ), and is the single most important driver of species extinctions ( Baillie 2004 ). Yet, despite major global scientific ( Baillie 2004 ; MEA 2005 ; Gullison 2007 ; Metz 2007 ) and political ( United Nations 1992, 1993 ) backing for the importance of forests and other ecosystems for global climate regulation and biodiversity conservation, deforestation rates remain unabated ( FAO 2006 ). Conversion is now concentrated in the most carbon‐rich and biodiverse biome on Earth, the tropical forests ( FAO 2006 ). Recent research suggests that the role these forests play in global climate regulation might be greater than previously thought ( Stephens 2007 ; Bonan 2008 ; Lewis 2009 ) and that some of its richest biodiversity hotspots are still poorly explored ( Carnaval 2009 ). New studies have also highlighted that climate change, deforestation, carbon storage in biomass and biodiversity are closely interlinked. For instance, in addition to contributing directly to global warming, deforestation also makes forests more susceptible to the effects of climate change ( Malhi 2009 ; Phillips 2009 ). Climate change will be a major driver of future species extinctions ( Thomas 2004 ) and one of its possible feedbacks is a further depletion of forest carbon stocks ( Malhi 2009 ). Biodiversity, on the other hand, might alleviate some of these effects by making ecosystems more resilient ( Reusch 2005 ; Tilman 2006 ) and, possibly, more productive ( Flombaum & Sala 2008 ). A landmark global deal being discussed under the United Nations Framework Convention on Climate Change is likely to include an international mechanism of financial incentives to reduce emissions from deforestation and forest degradation (REDD) ( Gullison 2007 ; UNFCCC 2007 ; Strassburg 2009 ). As the estimated magnitude of the financial incentives being discussed ( Kindermann 2008 ; Strassburg 2009 ) dwarfs current conservation expenditures in developing countries ( James 1999 ), REDD could trigger the biggest paradigm shift in conservation history. While it is generally assumed that REDD would have positive impacts for biodiversity conservation, this assumption has not been rigorously tested. Attention to the design of such a mechanism could enable increased gains for biodiversity, while reducing the risks, e.g., for those ecosystems that fall outside the scope of REDD ( Miles & Kapos 2008 ). Better understanding of the congruence between carbon storage and biodiversity conservation would help to maximize the gains and better address the risks. Here we use high‐resolution data to analyze for the first time the global congruence between carbon storage in biomass and biodiversity in order to investigate the synergies between climate‐ and biodiversity‐oriented conservation. Methods and data Our analysis is based on three global data sets of the distribution of mammal ( Schipper 2008 ), amphibian ( Stuart 2004 ), and bird ( Orme 2005 ) species, totaling 20,697 species, and on a new global carbon data set ( Ruesch & Gibbs 2008 ). The carbon data set presents estimates of above‐ and below‐ground (root) biomass, and is the first to apply consistently the IPCC Good Practice Guidance ( Penman 2003 ) to the whole terrestrial surface. We investigated the congruence between carbon and each of three biodiversity indices: richness (number of species per cell), threat (number of threatened species per cell), and range‐size rarity (number of species per cell whose ranges are in the lowest quartile for their class—see Supporting Information for details). All data (described and discussed in the Supporting Information) were analyzed on a geodesic discrete global grid system, defined on an icosahedron and projected to the sphere using the Inverse Snyder Equal Area (ISEA) Projection ( Sahr 2003 ). This corresponds to a hexagonal grid composed of individual units (cells) that retain their shape and area (∼12,500 km 2 , similar to a 1 × 1 degree rectangle at the equator) throughout the globe. These are more suitable for a range of ecological applications than the most commonly used rectangular grids ( Birch 2007 ). The analysis was conducted on a subset of this global grid: all 15,018 cells containing any land. The range of each species was converted to the hexagonal grid for analysis; a species was assumed to be present in a given cell if any part of the cell overlapped the species’ mapped range. For each cell: a Richness Index was calculated as the total number of species mapped in the cell (Figure S1A); a Threat Index was calculated as the number of threatened species per cell (Figure S1B); and a Restricted‐Range Index was calculated as the number of restricted‐range species (the 25% species in each taxon with the smallest range size) per cell (Figure S1C). Carbon data were also converted to the hexagonal grid, with the value in each cell corresponding to mean carbon density for the land area (Figure S2). Congruence between carbon and biodiversity was investigated visually (through maps) and analytically (through correlations). Global maps of the relationship between carbon and each of the biodiversity indices ( Figure 1 ) were produced by plotting each hexagon with a color defined on a two‐dimensional scale ( Williams & Gaston 1998 ), with intensity on the blue axis defined using the carbon density and intensity on the red axis defined from the biodiversity index. The color scales are based on statistical distribution of the data: a linear scale is used for values falling between minus three standard deviations and plus three standard deviations. Any cells with values outside this range are grouped into single color bands and the ends of the scale. Maps were plotted using GMT software ( Wessel & Smith 1998 ). Spearman's rank ( r S ) correlation coefficients were calculated for the relationship between carbon and each of the biodiversity indices ( Table 1 ), and their tests accounted for possible spatial autocorrelation (see Supporting Information). 1 Global congruence between biomass carbon and overall species richness (A), threatened species richness (B), and restricted‐range species richness (C). The two‐dimensional color scale used displays both the concentration of biomass carbon and biodiversity and the congruence between them. The intensity on the vertical blue axis represents biomass carbon density (tons of C per hectare) and the intensity on the horizontal red axis the richness of the respective biodiversity index (number of species per cell). Darker shadings correspond to higher concentrations. Colors along the 45° diagonal (white‐to‐green shading) indicate high congruence between biomass carbon and biodiversity. Areas with low congruence are represented by colors along the white‐to‐blue (higher carbon, lower biodiversity) or white‐to‐red (higher biodiversity, lower carbon) axes. The color scales are based on statistical distribution of the data: a linear scale is used for values falling between minus three standard deviations and plus three standard deviations. Any cells with values outside this range are grouped into single color bands at the ends of the scale. 1 Correlations between carbon biomass and biodiversity for each of the biodiversity indices considered (Richness, Threat, and Restricted Range) and for either all species or for each of the taxa (amphibians, birds, or mammals) separately. r S is Spearman rank correlation coefficients of all cells, EDF is Dutillieul‐corrected degrees of freedom for the first r S coefficients; P ‐values are the corresponding max P ‐values; r S nonautocorr present the mean and the 95% confidence intervals of Spearman rank correlation coefficients for replicates of subsets of cells 1,400 km apart from each other for which spatial autocorrelation is negligible (see supporting information) Richness index Threat index Restricted range index r S EDF P ‐values r S non‐autocorr r S EDF P ‐values r S non‐autocorr r S EDF P ‐values r S non‐autocorr All species 0.82 53.8 <0.0001 0.73 (0.59–0.83) 0.55 68.4 <0.001 0.39 (0.21–0.54) 0.42 149.3 <0.001 0.36 (0.17–0.52) Amphibians 0.84 51.0 <0.0001 0.77 (0.68–0.83) 0.37 250.6 <0.0001 0.35 (0.19–0.48) 0.24 357.5 <0.01 0.22 (0.06–0.35) Birds 0.80 57.6 <0.0001 71 (0.58–0.81) 0.43 73.8 <0.01 0.28 (0.09–0.47) 0.38 149.5 <0.001 0.32 (0.10–0.51) Mammals 0.79 54.3 <0.0001 0.69 (0.56–0.78) 0.52 64.7 <0.001 0.38 (0.22–0.49) 0.31 266.2 <0.001 0.29 (0.11–0.46) Results and discussion Overall, our results support the expectation that mechanisms for conserving biomass carbon (such as REDD) would have substantial cobenefits for biodiversity, and vice versa. Indeed, we found high congruence between species richness and biomass carbon at the global level ( r S = 0.82, Table 1 ; Figure 2A ). This result is not unexpected, given that on the one hand biodiversity‐poor ecosystems such as most deserts and polar regions are also carbon‐poor (shades of white and pale green in Figure 1A ), and on the other hand tropical forests (particularly the Amazon) are rich in both carbon and species (dark tones in Figure 1A ). 2 Global scale relationship between biomass carbon and overall species richness (A), threatened species richness (B), and restricted‐range species richness (C). In addition to all 15,018 individual data points, each panel presents an envelope containing 95% of the cells. To construct the envelope, cells were grouped based on their carbon content (each group being 10 tons of carbon wide). For each group the extremes were then excluded, with the cells with the highest or lowest 2.5% number of species falling outside the upper (blue line) or lower (red line) bounds, respectively. The black line represents the mean number of species of each group. There are also synergies between conservation of threatened or restricted‐range species and carbon‐based conservation ( Table 1 ), but these are less striking (at least at the global scale). Indeed, we found moderate congruence between biomass carbon and threatened ( r S = 0.55, Figure 2B ) and restricted‐range ( r S = 0.42, Figure 2C ) species. This is probably at least partially explained by the fact that although the two largest expanses of tropical forest (the Amazon and the Congo Basin) have simultaneously high carbon densities and species richness (hence high congruence in Figure 1A ), they have relatively few restricted‐range or threatened species. The coarse scale of our analysis is also likely to reduce the congruence between carbon and rare and threatened species by masking finer scale relationships in some areas (see below). In addition, prioritizing regions under threat of conversion (as a carbon mechanism would likely do) would likely substantially increase the correlation with threatened species The regional patterns identified in Figure 1 show a wide variation in synergies among carbon and biodiversity conservation, with correspondingly varied policy implications. Some areas would provide a high degree of cobenefits for the goals of conserving biomass carbon and biodiversity, including areas of high congruence between carbon and overall species (e.g., the Amazon; Figure 1A ), threatened species (e.g., Indonesia; Figure 1B ), and endemic species (e.g., New Guinea; Figure 1C ). Some areas that are of high priority for one of the conservation objectives but have moderate importance for the other could still benefit from complementary financing from this second one. For example, Mainland Southeast Asia, a priority for biodiversity conservation, could benefit from complementary funding from carbon conservation. Finally, regions that are of high priority for one of the goals but of low importance for the other deserve careful consideration to ensure that they are not neglected. In particular, if a large‐scale carbon‐focused REDD mechanism is implemented, biodiversity‐rich and relatively carbon‐poor regions could suffer from a double conservation jeopardy, with conservation investment diverted away from them, and human pressure redirected toward them, as carbon‐rich areas become the focus of conservation efforts. Areas potentially at risk include some that are widely recognized as global biodiversity conservation priorities such as the Brazilian Cerrado, the Cape Floristic province, and the Succulent Karoo ( Mittermeier 2004 ). In some regions, the lack of congruence between carbon stocks and biodiversity might be an artifact of the coarse scale of our spatial analysis ( Figure 3 ). Extensive deforestation often leaves remaining natural habitats concentrated in sparse forest fragments. These regions have low carbon per 1° cell, but each remaining forest fragment can be carbon‐rich on a per‐forest area basis. Habitat loss means many of these forests’ species will be threatened, particularly if their ranges are small. Hence these regions might emerge in our coarse‐scale analysis as having low carbon per cell, yet high counts of threatened or range‐restricted species. If mapped at a finer scale, they might instead be identified as small fragments of high value for both carbon and biodiversity, surrounded by a matrix of low carbon and low biodiversity. This is almost certainly the case of parts of the Atlantic Forest of South America, the Tropical Andes, and Southeast Asia ( Figure 1B ). Often, those remaining fragments are also highly threatened from future deforestation, and so their conservation would provide high benefits in terms of both avoided emissions and avoided extinctions. Future analyses at finer scale would be able to better capture these relationships. This refinement would be especially important to investigate the synergies between carbon‐oriented conservation and threatened and rare species, and in particular for species that are both rare and threatened, maybe the most valuable group for biodiversity conservation ( Ricketts 2005 ). 3 An illustration of the effect of scale on the congruence between carbon and biodiversity in fragmented areas. At the spatial resolution employed in our analyses (cells ∼12,500 km 2 ), the Atlantic Forest of South America (part of which is outlined by a box) emerges as an area of low congruence between carbon and biodiversity, appearing to contain relatively low carbon (A; shades of green correspond to carbon per cell) but high biodiversity value, for example, in terms of numbers of threatened species (B; shades of red correspond to number of threatened species per cell). At finer scales, though, the congruence is much higher, as most of the species that give this region its exceptional biodiversity value are restricted to the remaining fragments, corresponding to 7% of the original forest cover ( Carnaval 2009 ), which are carbon‐rich. The endemic and Endangered Red‐billed Curassow ( Crax blumenbachii ), for example, (C; photo by Alberto Teleuko, Wikimedia Commons), is only known from a few forest patches (D, inset; stars—species’ records; green—forest fragments) ( Bird Life International 2008 ). We believe that these fragmented areas provide some of the most valuable opportunities for “win–win” carbon and biodiversity conservation. We have also calculated correlation coefficients between carbon and biodiversity indices (richness, threat, and restricted range) separately for each of the taxonomic groups (amphibians, birds, mammals) ( Table 1 ). We found that the results are consistent with those found for the overall biodiversity indices, demonstrating that they are not driven by any particular taxon among those analyzed, and suggesting that they are robust to the choice of taxa as surrogates of biodiversity. However, some areas have very high value for biodiversity conservation despite having lower vertebrate diversity (see Supporting Information). We investigated the effect that spatial autocorrelation has on our results through three different approaches (Tables S1–S3 and Supporting Information). Although most correlation coefficients go down when redoing the analyses for sets of non‐autocorrelated samples ( Table 1 and Table S2), the decreases are relatively small (e.g., for Species Richness, nonautocorrelated r S is 0.73, down from 0.82 for total cells). Most importantly, all coefficients have P ‐values < 0.01 even when controlling for spatial autocorrelation and none of the conclusions or relative comparisons discussed is affected. Our analysis was focused on carbon stored in above‐ and below‐ground biomass only. The soil carbon pool is 4.5 times the biotic carbon pool ( Lal 2004 ). But unlike biomass carbon, there is no agreement on the magnitude or even the sign of the change in this pool following land‐use change ( Guo & Gifford 2002 ), which varies greatly depending on several factors (e.g., replacement land cover). Therefore it is not possible to assess the importance of soil carbon stocks for climate change mitigation without taking into consideration a dynamic land‐use context. For these reasons, we opted to focus our analysis on biomass carbon, the primary carbon pool affected by land cover change. Nonetheless, a sensitivity analysis whereby total carbon is measured by adding biomass carbon and 25% of the soil carbon stock (Figure S3 and Supporting Information) showed only a relatively small decrease in correlation between vulnerable carbon and species richness ( r S = 0.73, down from r S = 0.82). Our analysis is about carbon stock (that can be lost due to land conversion) and it fails to include the value of different ecosystems for carbon sequestration. There is now clear evidence that mature forests, both tropical ( Phillips 2008 ; Lewis 2009 ) and temperate ( Wirth 2009 ), sequester substantial amounts of carbon every year, and some of these are highly valuable for biodiversity. We have assessed the synergies between climate mitigation and biodiversity conservation based solely on stocks of carbon and species. An important aspect missing from our analysis is that the conservation of these stocks is more relevant where their rate of loss might be greater (i.e., areas under higher risk of deforestation). In this sense, a further step to improve this assessment would be to analyze these synergies in a dynamic land‐use change context. A mechanism focused purely on carbon retention will not be optimal at conserving biodiversity, but there is much scope for improving the gains for biodiversity conservation, with low or no losses for carbon mitigation. Indeed, we found high variability in species counts across cells with similar biomass carbon levels ( Figure 2 ). For instance, for total species richness there is approximately a five‐fold difference in species richness per cell between the lower (red line, Figure 2A ) and upper (blue line, Figure 2A ) bounds of a 95% envelope. Cells in the upper bound usually have twice as many species as those expected by chance (mean, black line, Figure 2A ). Taking biodiversity patterns into account when prioritizing areas for implementation could therefore substantially improve the contribution of REDD or similar mechanisms to global biodiversity conservation. This could be done by inserting a biodiversity premium for emissions from more biodiversity‐rich areas directly into the REDD mechanism, by setting aside a fraction of REDD financing for targeting biodiversity‐rich areas that would not be conserved for their carbon content alone ( Strassburg 2009 ), or by promoting cooperation between programs for REDD and conservation at a national to international scale, redirecting conservation funding to these areas ( Grainger 2009 ). Climate change and biodiversity loss are two crises of global magnitude, each posing individual as well as synergistic risks to human well‐being and curtailing humanity's future options ( MEA 2005 ; Metz 2007 ). Most of the world's governments have pledged both to reduce the rates of biodiversity loss ( United Nations 1993 ) and to prevent dangerous climate change ( United Nations 1992 ). Rather than addressing these two crises separately, it makes sense, where appropriate, to combine efforts and thereby achieve mutual gains. Our results confirm that mechanisms to reduce carbon emissions from deforestation and forest degradation will have side benefits in terms of biodiversity protection, as expected given that forest conversion is a major driver of both climate change and biodiversity loss. Nonetheless, a purely carbon‐focused mechanism does not necessarily focus on those forests where biodiversity conservation is most urgent, and attention should also be given to biodiverse areas outside the scope of proposed REDD mechanism that could become increasingly threatened. But our results suggest that if biodiversity distribution is taken into account, there might be substantial room for additional gains for biodiversity conservation without compromising the effectiveness of climate change mitigation. Our study confirms that the Parties to the United Nations Framework Convention on Climate Change have a singular opportunity to address two of the biggest challenges of our time, first and foremost by approving incentives for reducing carbon emissions from deforestation and forest degradation, and second by maximizing the potential of this mechanism to reduce rates of global biodiversity loss. Editor : Dr. Pablo Marquet Acknowledgments We are grateful to the thousands of experts who contributed to the IUCN Global Mammal Assessment, to the IUCN Global Amphibian Assessment, and to the Red List assessment of bird species (coordinated by BirdLife International). We thank two anonymous referees for comments. B.S. thanks CAPES/Brazilian Ministry of Education for the financial support received.

Journal

Conservation Letters – Wiley

Published: Apr 1, 2010

Keywords: ; ; ; ; ; ; ; ; ;

There are no references for this article.