This is the accepted version of the following article: Di Fulvio, F. et al. Spatially explicit LCA analysis of biodiversity losses due to different bioenergy policies in the European Union in Science of the total environment (Ed. Elsevier), vol. 651, part 1 (Feb. 2019), p. 1505-1516. Which has been published in final form at DOI 10.1016/j.scitotenv.2018.08.419 © 2019. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ 1 Spatially explicit LCA analysis of biodiversity losses due to different bioenergy policies 2 in the European Union 3 4 Fulvio Di Fulvio1*, Nicklas Forsell1, Anu Korosuo1, Michael Obersteiner1, Stefanie 5 Hellweg2 6 7 *Corresponding Author difulvi@iiasa.ac.at 8 1 9 Ecosystems Services and Management Program (ESM), International Institute for Applied 10 Systems Analysis (IIASA), Laxenburg, Austria 2 11 Ecological Systems Design, Institute of Environmental Engineering (IfU), ETH, Zurich 12 13 14 15 16 17 18 19 20 21 1 22 Spatially explicit LCA analysis of biodiversity losses due to different bioenergy policies 23 in the European Union 24 25 26 Abstract 27 In this study, the potential global loss of species directly associated with land use in the EU and 28 due to trade with other regions is computed over time, in order to reveal differences in impacts 29 between the considered alternatives of plausible bioenergy policies development in the EU. 30 The spatially explicit study combines a life cycle analysis (LCA) for biodiversity impact 31 assessment with a global high resolution economic land use model. Both impacts of domestic 32 land use and impacts through imports were included for estimating the biodiversity footprint of 33 the member states of the (EU28). The analyzed scenarios assumed similar biomass demand 34 until 2020 but differed thereafter, from keeping the growth of demand for bioenergy constant 35 (CONST), to a strong increase of bioenergy in line with the EU target of decreasing greenhouse 36 gas (GHG) emissions by 80% by 2050 (EMIRED) and with the baseline (BASE) scenario falling 37 between the other two. 38 As a general trend, the increasing demand for biomass was found to have substantial impact on 39 biodiversity in all scenarios, while the differences between the scenarios were found to be 40 modest. The share caused by imports was 15% of the overall biodiversity impacts detected in 41 this study in the year 2000, and progressively increased to 24% to-26% in 2050, depending on 42 the scenario. The most prominent future change in domestic land use in all scenarios was the 43 expansion of perennial cultivations for energy. In the EMIRED scenario, there is a larger 44 expansion of perennial cultivations and a smaller expansion of cropland in the EU than in the 45 other two scenarios. As the biodiversity damage is smaller for land used for perennial 46 cultivations than for cropland, this development decreases the internal biodiversity damage per 47 unit of land. At the same time, however, the EMIRED scenario also features the largest 2 48 outsourcing of damage, due to increased import of cropland products from outside the EU for 49 satisfying the EU food demand. These two opposite effects even out each other, resulting in the 50 total biodiversity damage for the EMIRED scenario being only slightly higher than the other two 51 scenarios. 52 The results of this study indicate that increasing cultivation of perennials for bioenergy and the 53 consequent decrease in the availability of cropland for food production in the EU may lead to 54 outsourcing of agricultural products supply to other regions. This development is associated with 55 a leakage of biodiversity damages to species-rich and vulnerable regions outside the EU. 56 In the case of a future increase in bioenergy demand, the combination of biomass supply from 57 sustainable forest management in the EU, combined with imported wood pellets and cultivation 58 of perennial energy crops, appears to be less detrimental to biodiversity than expansion of 59 energy crops in the EU. 60 61 Keywords: biodiversity damage, bioenergy, land use, perennial energy crops, forestry, EU 62 footprint, trade. 63 64 65 66 67 68 69 70 71 72 3 73 1. Introduction 74 75 The EU recently updated its targets for bioenergy use in order to reach a 40% reduction of 76 greenhouse gas emissions by 2030 compared to 1990 levels (European Commission, 2016; 77 European Parliament and Council of the European Union, 2016). Within the refined target, 27% 78 of the total energy consumption is expected to be provided through renewable resources by 79 2030 (European Parliament and Council of the European Union, 2016). Bioenergy currently 80 provides 59% of the renewable energy consumed in the EU (Eurostat 2016). In addition to the 81 increased renewable energy consumption targets, awareness of the sustainability of bioenergy 82 supply is also on the rise. 83 An increase in the demand for woody biomass in Europe is expected to lead to an increased 84 harvest level in currently managed forests through elevated tree part utilization, expanding 85 forest area, and short rotation coppice plantations, as well as increasing wood imports from 86 other regions, and/or increasing wood supply from outside the forests (Mantau et al. 2010, Lauri 87 et al. 2014, Forsell et al. 2016, Schelhaas et al. 2006). 88 Depending on the different point of demand, the biomass can assume different shapes, for 89 example, solid wood fuels such as wood pellets or it can be converted into biofuels. 90 In this context, many European environmental non-governmental organizations (NGOs) argue 91 that without appropriate sustainability criteria for most biofuel production, the increased use of 92 woody biomass may lead to negative environmental impacts (Obersteiner et al. 2018). 93 Therefore, increased use of woody biomass to replace fossil fuels is likely not a side-effect free 94 solution to climate change problems. Increased biofuel production could lead to increased loss 95 in biodiversity and may also indirectly impact food security through possible increases in food 96 prices or further competition for land use (Söderberg & Eckerberg 2013). Liquid biofuels can be 97 divided into two categories: first-generation biofuels made from the sugars and vegetable oils of 4 98 arable crops, and second-generation biofuels made from ligno-cellulosic biomass, such as 99 woody biomass. The EU has reported that the negative impacts of first-generation biofuels, 100 such as deforestation, competition with food production, and indirect land use change, provide 101 motivation for a preference for second-generation biofuels from ligno-cellulosic biomass 102 (European Parliament and Council of the European Union, 2015). Accordingly, both the EU and 103 the USA have actively been promoting a revision of their policies with a shift away from first- 104 generation biofuel crops such as corn, sugarcane, and oilseeds towards cellulosic biofuels that 105 utilize the woody or fibrous parts of plants (Baumber 2017). 106 107 One of the global criteria for sustainable development (i.e., Sustainable Development Goal 15) 108 is to ‘Protect, restore, and promote the sustainable use of terrestrial ecosystems, sustainably 109 manage forests, combat desertification, halt and reverse land degradation, and halt biodiversity 110 loss’ (UN 2015). The EU has also recognized the importance of biodiversity explicitly, and 111 adopted a strategy to halt biodiversity loss by 2020 (European Commission 2011). This strategy 112 includes, among others, targets to improve the conservation status of habitats and species, and 113 to improve and restore ecosystems and ecosystem services wherever possible. Land use and 114 its changes are considered the main drivers for biodiversity loss in terrestrial ecosystems 115 (Pereira et al. 2010), and the general consensus is that more land protection is required to 116 preserve global biodiversity (Heller and Zavaleta 2009). With this connection, it is evident that 117 the impacts of future bioenergy policies need to be assessed in light of their impacts on the land 118 use, land-use change, and forestry (LULUCF) sector, as well as their related impacts on 119 biodiversity. 120 121 Managing policy trade-offs connected to the LULUCF sector and biodiversity is complicated by 122 the interconnected nature of global energy, food, feed, and fiber markets. While some impacts 123 of increased bioenergy production are direct (observed in the areas where biomass is 5 124 produced), others are indirect, affecting land use change and the supply of food, feed and fiber 125 in other distant locations (Berndes et al., 2011). Through indirect land use change, the impacts 126 of EU policies are also transferred to highly vulnerable habitats in other regions such as Asia or 127 South America (e.g., Rivas Casado et al. 2014, Britz & Hertel 2011). That is, the interaction 128 between bioenergy supply and larger global systems leads to indirect consequences on the 129 globe beyond the direct effects connected to the bioenergy production chains (Elbersen et al. 130 2013). The rapidly increasing demand for biofuels, driven in part by EU policies, is a clear 131 example of this due to the global nature of biofuel markets. In this case, the reported effects 132 were damages to biodiversity and ecosystem services provision through both direct and indirect 133 land use changes (Holland et al., 2015). 134 135 Previous studies have measured the impacts of the LULUCF sector on biodiversity in the EU 136 under different bioenergy policy scenarios (Eggers et al. 2009, Rivas Casado et al. 2014, 137 Schulze et al. 2016). These approaches assess the suitability of different land uses as habitat 138 for different species. However, as these studies only measure change in biodiversity related to 139 change in land use within the area directly impacted by a policy, they fail to account for changes 140 in biodiversity related to two other vital processes namely: i) changes in biodiversity related to 141 the intensity of land use and forest activities, and ii) changes in biodiversity in areas that are 142 only indirectly affected by the policy, for example the impacts on biodiversity outside the EU as 143 a result of market effects and international trade of food, feed and fiber commodities. 144 145 Recently, global databases containing responses of species to different land uses and 146 intensities of management have been made available (Hudson et al 2014; Schipper et al. 2016). 147 These databases have allowed for a regionalized quantification of biodiversity losses consistent 148 with a global framework (Newbold et al. 2015, Chaudhary et al. 2015). 149 6 150 The development of spatially explicit biodiversity indicators within LCA has progressed 151 substantially in the last years (for a review see Curran et al., 2016). Among the most notable 152 developments are the methods of de Baan et al. (2012, 2013), which were the first to quantify 153 local, regional, and permanent biodiversity loss on a global scale. Following the suggestion of 154 Verones et al. (2013), Chaudhary et al. (2015) developed these approaches further by including 155 more data and weighing regional species loss with a factor combining the rarity and threat level 156 of species. Their work provides impact factors that measure biodiversity loss in units of global 157 species extinctions at a steady state, that is, the number or fraction of species that are 158 committed to extinction in the long term as a consequence of land use for six land use classes 159 and 804 ecoregions. The joint Life Cycle Initiative (2016) between the United Nations 160 Environment Programme (UNEP) and the Society of Environmental Toxicology and Chemistry 161 (SETAC) tentatively recommended the method of Chaudhary et al. (2015) as best practice for 162 the assessment of land-related impacts on biodiversity loss. This method has been used to 163 assess the biodiversity impacts of global agriculture and forestry (Chaudhary et al. 2016a) and 164 also for global trade (Chaudhary & Kastner 2016). However, none of the existing studies have 165 assessed biodiversity loss of prospective land use scenarios under different policies. 166 In this paper, we set out a global framework that is able to jointly assess and analyze the 167 biodiversity implications of policies related to: direct land use change, changes in intensity in 168 land use and forestry, and in-direct land use effects. 169 We build on the recent development of biodiversity indicators within LCA, and provide a spatially 170 explicit analysis of LULUCF driven biodiversity loss from different European policies in the 171 bioenergy sector. In our analysis, the biodiversity loss factors are coupled with the results of the 172 Global Biosphere Management Model (GLOBIOM) – a high resolution economic model 173 providing prospective land use scenarios that also allow us to analyze impacts on other regions 174 and on international trade (Havlık et al. 2011, Havlik et al, 2014). Three alternative bioenergy 175 policy scenarios are considered in the EU during the period from 2000 to 2050. The potential 7 176 global loss of species directly associated with land use in the EU and due to trade with other 177 regions is computed over time, in order to reveal differences in impacts between the considered 178 alternatives of plausible policy development in the EU. 179 180 181 182 2. Material and methods 183 184 2.1 Approach for assessing land-use related impacts of biodiversity loss 185 For the assessment of biodiversity loss from LULUCF, the life cycle impact assessment method 186 “potential loss of global species (PSLglo)” was used. The approach quantifies the percentage of 187 global species lost at a steady state, thereby providing an indicator of global extinctions that will 188 result as a consequence of LULUCF. Species loss is quantified using the countryside species 189 area-relationships (SARs). In contrast to the original SARs, it takes into consideration that some 190 species will also survive in anthropogenically transformed land, depending on their affinity. 191 Regional species loss is further weighted with the total range and threat level of species to 192 provide an indicator of global species extinctions (i.e., global species equivalents lost per m2). 193 The PSLglo method provides characterization factors (CFs i.e., the factors indicating the 194 biodiversity damage caused by the unit area of a particular land use in a particular region) for 195 six land use types (annual crops, permanent crops, pasture, urban areas, extensive forestry, 196 intensive forestry), four vertebrate taxa (mammals, birds, amphibians, and reptiles), vascular 197 plants, and 804 ecoregions. Ecoregions are chosen as spatial units containing distinct 198 communities of species, and their boundaries approximate the original extent of natural 199 ecosystems prior to major land use change (Olson et al. 2001). Taxa aggregated CFs for each 200 land use type per ecoregion are also available in the unit of potentially disappeared fraction of 8 201 global species (PDF/m2) (UNEP-SETAC 2016, Chaudhary et al. 2015). To get from the unit 202 “global species equivalents lost” to PDFs, the former is divided by the total number of existing 203 species on earth, for each taxonomic group, thus denoting the fraction of global species that is 204 projected to go extinct. PDFs of various taxonomic groups are then aggregated by calculating a 205 weighted average, following the procedure documented in UNEP-SETAC (2016). In this paper, 206 we used the “marginal” characterization factors reported in UNEP-SETAC (2016). 207 208 2.2 Biodiversity impact of future land-use scenarios 209 The regionally specific CFs were combined with land use maps of annual crops, permanent 210 crops (i.e., miscanthus and short rotation energy plantations), pasture, and managed forests in 211 the EU (EU28) computed from the GLOBIOM model under different bioenergy policy scenarios. 212 213 The GLOBIOM model is an economic partial equilibrium model of the global forest, agriculture, 214 and biomass sectors with a bottom-up representation of agricultural and forestry management 215 practices (Havlik et al. 2011, Havlik et al, 2014). In this study, the GLOBIOM model was run 216 recursively for 10-year time steps (i.e., the years 2000, 2010, 2020, 2030, 2040, and 2050) for 217 three different bioenergy policy scenarios. The results from the GLOBIOM model were analyzed 218 at the resolution of 246 European administrative units (NUTS2) (supplementary information (SI) 219 1) and they are presented as land-use maps for the assessment of biodiversity impacts. 220 221 The GLOBIOM model covers the following six main land use categories: unused forests, 222 managed forests, cropland (both annual and permanent), pastures, other natural vegetation, 223 and urban areas. However, for the assessment of the biodiversity implications, only changes in 224 managed forests, cropland, and grassland are used. Unmanaged forests and other natural 225 vegetation were considered as the reference ecosystems in each ecoregion, assuming impacts 9 226 from human modifications were negligible, while urban areas were considered to be out of the 227 scope of analyses in our scenarios. 228 229 Managed forests are forests used over a certain period to meet wood demand. These forests 230 are managed for woody biomass production, which implies a certain rotation time, thinning 231 events, and final harvest. The unmanaged forests do not currently contribute to wood supply, 232 based on economic decision rules in the model. However, they may still be a source for 233 collection and production of non-wood goods (e.g., food, wild game, or ornamental plants). 234 235 The land allocated to “managed forests” in GLOBIOM was divided between “intensive” and 236 “extensive” management. Area shares of intensively and extensively managed forest in each 237 NUTS2 unit in the EU were calculated according to a European forest management suitability 238 map from Hengeveld et al. (2012). For this purpose, the “combined objective” forests in 239 Hengeveld et al. (2012) were considered to be “extensive forests”, while the “even aged 240 forests”, and “short rotation forests” were classified as “intensive forestry”. The forest land used 241 outside the EU was divided between “intensive” and “extensive” forest according to the shares 242 of roundwood from plantations reported in Jürgensen et al. (2014) for five regions (i.e., South 243 America, Oceania, Asia, Africa, and North and Central America) in the period 2000 to 2010. The 244 projection of expansion rates for plantations from 2010 to 2050 was based on the trends 245 predicted in ABARE & Pöyry (1999) and Jürgensen et al. (2014). 246 247 The matching between the land-use spatial units of NUTS2 with the ecoregion-specific 248 characterization factors was done according to equation 1: 249 250 𝐶𝐹𝑖,𝑗 = ∑𝑛𝑔 =1 𝐶𝐹𝑔,𝑖 × 𝑝𝑔,𝑗 (Eq.1) 10 251 252 Where CFg,i is the characterization factor for the land use types i (cropland, permanent crops, 253 extensive forestry, intensive forestry, and pasture), j is an index for NUTS2 units, g is an index 254 for ecoregion, and pg,j is the share of area occupied in the NUTS2 region j by each ecoregion g. 255 256 Biodiversity damage BDi,j (species eq. lost) impact due to the different land uses in each NUTS2 257 was calculated by multiplying CFi,j by the area (Ai,j) occupied by the different land use types in 258 each of the NUTS2 (in m2), thus assuming a steady state change in biodiversity as: 259 260 𝐵𝐷𝑖,𝑗 = 𝐶𝐹𝑖,𝑗 × 𝐴𝑖,𝑗 (Eq. 2) 261 262 The sum of the BDi,j’s from different land uses i provided the NUTS2 level biodiversity damage: 263 264 𝐵𝐷𝑗 = ∑𝑖 𝐵𝐷𝑖,𝑗 (Eq. 3) 265 266 We assessed the impacts using the taxa aggregated CFi,j’s for each land use type in the 267 NUTS2. This provides the biodiversity impacts in the units potentially disappeared fraction per 268 m2 of land use (PDF/m2). The impacts due to land use from the NUTS2 were then also provided 269 on the country level. 270 271 𝐵𝐷𝑐 = ∑𝑗 𝐵𝐷𝑗 ; 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑗 𝑙𝑜𝑐𝑎𝑡𝑒𝑑 𝑤𝑖𝑡ℎ𝑖𝑛 𝑐𝑜𝑢𝑛𝑡𝑟𝑦 𝑐 (Eq. 4) 272 273 The PDFs due to forest land use in the EU were divided by the roundwood production from 274 each NUTS2 unit, and a map of impacts as global PDF/m3 roundwood was obtained. The same 275 indicator (PDF/m3) was also calculated at the country and EU levels. 11 276 277 Similarly, for permanent crops including willow and poplar short rotation coppices (SRC) and 278 miscanthus, the PDFs due to land use were divided by their respective production (solid m3) and 279 the PDF/m3 perennials were obtained at the NUTS2, country and EU levels. For miscanthus, the 280 conversion to solid m3 was obtained by calculating the amount of biomass (oven dry tonnes) 281 required for achieving the same energy as 1 m3 of woody biomass. 282 283 2.3 Impacts from trade 284 In the GLOBIOM model, trade is modeled between 30 global trade regions (i.e., 29 regions and 285 the EU28) (SI 2). The model provides the amount of goods traded by the EU28 countries with 286 the other 29 regions on the globe in each scenario and year. In addition, the model computes 287 the amount of goods traded by each EU country with other European countries. 288 289 A trade balance was created for the aggregated EU28, as the difference between the import 290 and export of each product (i.e., net import or, in the case of negative values, net export). 291 Similarly, the trade balance was also created for each country within the EU28. The net import 292 to the EU28 was allocated to the member states proportionally, based on the magnitude of their 293 respective net imports for each of the products. 294 295 Agricultural products imported from or exported to the EU28 (i.e., barley, dry beans, cassava, 296 chick pea, corn, cotton, groundnuts, millet, palm oil, potatoes, rapeseed, rice, soybeans, 297 sorghum, sugar cane, sunflower, sweet potatoes, and wheat) were classified as “annual crops” 298 for this specific assessment and were all converted to fresh tonnes of biomass. The amounts 299 were divided by the average yields (tonne/ha, SI 3) in each trading region to obtain the average 300 area of crop used to produce the amounts being traded. 12 301 The forest products imports/exports accounted for in this assessment included pulp logs, 302 sawlogs, woodchips, and wood pellets. Their imports/exports were all converted to solid m3. The 303 traded amounts were divided by the average forest increment (m3/ha, SI 4) in each trading 304 region to obtain the amount of intensive and extensive forestland used. 305 306 For the calculation of impacts BDimport,a,EU due to net imports from trade region a into the EU28 307 region, the characterization factors 𝐶𝐹𝑔,𝑖 for ecoregion g for land use types i were first multiplied 308 by their area share 𝑝𝑔,𝑎 in region a. The resulting characterization factors for each trade region 309 of origin a 𝐶𝐹𝑎,𝑖 were then multiplied by the area demand Aa,i,r of land use type i for producing 310 one unit of product r and the net amount of product t exported from trade region a to the EU28 311 region 𝑡𝑎,𝐸𝑈,𝑟 . Afterwards, the total biodiversity damage created outside the EU due to net 312 imports of biomass into the EU 𝐵𝐷𝑖𝑚𝑝𝑜𝑟𝑡,𝑟,𝐸𝑈 was calculated by summing the damages through 313 imports from all regions of origin (Eq. 7). 314 315 𝐶𝐹𝑎,𝑖 = ∑𝑔 𝐶𝐹𝑔,𝑖 × 𝑝𝑔,𝑎 (Eq. 5) 316 317 𝐵𝐷𝑖𝑚𝑝𝑜𝑟𝑡,𝑎,𝑟,𝐸𝑈 = ∑𝑖 𝐴𝑎,𝑖,𝑟 × 𝑡𝑎,𝐸𝑈,𝑟 × 𝐶𝐹𝑎,𝑖 (Eq. 6) 318 319 𝐵𝐷𝑖𝑚𝑝𝑜𝑟𝑡,𝑟,𝐸𝑈 = ∑𝑎 𝐵𝐷𝑖𝑚𝑝𝑜𝑟𝑡,𝑎,𝑟,𝐸𝑈 (Eq. 7) 320 321 The allocation of 𝐵𝐷𝑖𝑚𝑝𝑜𝑟𝑡,𝑟,𝐸𝑈 to single EU member states was obtained by multiplying the 322 damage by the share of net import for each country and product. 323 13 324 In case of a negative net import of products (i.e., a net export) from the EU region to the other 325 regions, the amount t of product r exported from the EU was allocated to the member states c 326 according to their share of net export for each product (𝑡𝑐,𝑟 ). The country specific 327 characterization factors were obtained from the 𝐶𝐹𝑔,𝑖 multiplied by their area share 𝑝𝑔,𝑐 as 328 occupied by each ecoregion in the country c (Eq. 8). The exported amount 𝑡𝑐,𝑟 was multiplied by 329 the area demand Ac,i,r of land use type i for producing one unit of product r and by the country 330 specific characterization factors 𝐶𝐹𝑐,𝑖 to obtain the biodiversity damage due to net export 331 𝐵𝐷𝑒𝑥𝑝𝑜𝑟𝑡,𝑐,𝑟 (Eq. 9). 332 333 𝐶𝐹𝑐,𝑖 = ∑𝑔 𝐶𝐹𝑔,𝑖 × 𝑝𝑔,𝑐 (Eq. 8) 334 335 𝐵𝐷𝑒𝑥𝑝𝑜𝑟𝑡,𝑐,𝑟 = ∑𝑖 𝐴𝑐,𝑖,𝑟 × 𝑡𝑐,𝑟 × 𝐶𝐹𝑐,𝑖 (Eq. 9) 336 337 For each EU member state, the biodiversity damage due to net exports was deducted from the 338 other damages in the computation of the EU biodiversity footprint. 339 340 The biodiversity damage due to internal trade within the EU28 region was also considered. For 341 each member state, the net export/import amount for each product was converted into BD as in 342 Eq. 6-9 and added to, or deducted from the impacts due to land use in each country. 343 344 2.4 Policy scenarios 345 In order to analyze the implications of increasing bioenergy consumption, three prospective 346 scenarios were considered. The scenarios were developed to depict different pathways for the 347 future development of the EU bioenergy sector (Forsell et al. 2016). 348 14 349 Baseline (BASE) 350 The Baseline scenario (BASE) was specified as close as possible to that of the EU Reference 351 Scenario 2013 published by the European Commission. The Baseline scenario depicts the 352 development of biomass use under bioenergy policies that aim at a 20% reduction of GHG 353 emissions in the EU28 by 2020, but where the EU climate-energy targets for 2030 are not 354 considered. The results show that increased demand for bioenergy will lead to a considerable 355 increase in the EU domestic production of woody biomass (an increase by as much as 10% by 356 2030 in comparison to 2010 levels), as well as increased EU reliance on imported biomass 357 feedstock, in particular wood pellet imports (an increase of 90% by 2030 in comparison to 2010 358 levels). From 2030 to 2050, the EU domestic production of biomass stabilizes as a result of 359 slower development of EU bioenergy demand. The largest changes in the EU28 production of 360 biomass feedstocks for bioenergy are seen in the development of SRC, which together with the 361 EU import of wood pellets are expected to increase considerably in the future. In addition, there 362 is an intensification in the use of EU forests, as well as an increase in the EU net import of 363 roundwood. The increase in EU forest harvesting is driven by both the increasing demand for 364 bioenergy, and the expected increase in demand of woody materials. 365 366 Constant demand 367 The Constant EU Bioenergy Demand scenario (CONST) investigates the effects of policies that 368 increase the EU bioenergy demand similarly to the BASE scenario until 2020, but stay constant 369 thereafter. There are only small differences between this scenario and BASE on the overall 370 aggregate material production sector. However, compared to the BASE scenario, this scenario 371 has more particleboard and less sawn wood production, driven by decreased demand for 372 industrial by-products from sawmills (wood chips and sawdust) for bioenergy production. A clear 373 difference is also seen in the composition of feedstocks used for energy production. Most 15 374 importantly, the sourcing of domestically produced SRC and import of pellets is smaller than in 375 the BASE scenario. Pellet imports increase until 2020, but remain almost constant thereafter. 376 377 Greenhouse gas emission reduction 378 The development seen in the BASE scenario is found to be accentuated in the EU Emission 379 Reduction scenario (EMIRED), which builds on the policy target of decreasing GHG emissions 380 by 80% by 2050 in the EU. In this scenario, the development of biomass use follows that of the 381 BASE scenario until 2030. Thereafter, the results show a considerable increase in the EU 382 import of wood pellets and domestic production of SRC. The increasing production of SRC in 383 the EU after 2030 leads to some reductions in cropland and grazing land areas as compared to 384 the BASE scenario, which in turn affects food and feed production. Additionally, we also see 385 large quantities of roundwood directly used for bioenergy production in small and large-scale 386 conversion facilities, especially by 2050. In other words, the bioenergy demand increases to an 387 extent where stemwood that is of industrial roundwood quality and could be used for material 388 purposes by the forest-based sector is instead being used directly for energy production. The 389 increased use of biomass for energy has a direct impact on forest harvests, which are almost 390 9% higher than in the BASE results in 2050. 391 392 3. Results 393 394 3.1 Land use impacts occurring on EU territory 395 The area for the land uses considered in the EU from the year 2000 to 2050 increases by 12%, 396 10%, and 15% in the BASE, CONST, and EMIRED scenarios respectively (Figure 1 and SI 6 397 Table 1), as an effect of increasing bioenergy and food demand in the future. 16 398 The most relevant increase is the land area for wood extraction (forests and permanent crops), 399 with a maximum in the EMIRED scenario, followed by the BASE and the CONST scenarios (for 400 figures see SI 6 Table 1). Cropland expansion is relevant in the CONST and BASE scenarios, 401 whereas it is less noticeable in EMIRED. This difference is due to the higher rate of conversion 402 from cropland into perennial cultivations in EMIRED compared to the other two scenarios, 403 especially after the year 2030 (Figure 1). Pasture land appears to be the category that is the 404 least sensitive to the different bioenergy scenarios (Figure 1, SI 6 Table 1). 405 406 17 407 Figure 1. The graphs on the left show land use difference relative to the base year (2000) for the 408 EU28 in the three different bioenergy scenariosas land uses within the EU, and net imported 409 land uses. The graphs on the right depicts differences in biodiversity impacts relative to the 410 base year (2000) due to land use within the EU and to net imports. The stacked columns 411 represent the differences for each land use category compared to the year 2000, while the red 412 dots represent the arithmetic sum of differences due to different land uses for each year. 413 The aggregated biodiversity impact due to land use in the EU28 from year 2000 to 2050 causes 414 0.08% of the global species extinction (7.63 × 10-4 PDF) in the BASE scenario. Cropland and 415 grassland reduce their shares over time from respectively 78% and 16% (year 2000) to 76% 416 and 15% (year 2050) of impacts. In the meantime, perennials reach 3.6% of land use impacts in 417 the year 2050, while the share from used forests remains almost constant over time (6.0% to 418 5.9%). 419 The difference between the three scenarios increases after the year 2020: In the year 2050, the 420 EMIRED scenario produces 0.4% more impacts than the BASE scenario, while the CONST 421 scenario produces 1.5% less than the BASE scenario (SI 6 Table 2). 422 The impacts due to land use are amplified in South Europe, where the ecoregions are hosting 423 more species richness than in the North (Figure 2-5). South western European countries show a 424 total impact due to land use in the order of 0.1% of global species loss (10-3 PDF), while in the 425 rest of EU countries the impacts are in the order of 0.00001% to 0.01% of global species loss 426 (10-4 to 10-7 PDF). This spatial difference is magnified if considering the impacts per hectare of 427 land (as PDF · ha, Figure 2). 18 428 429 Figure 2. Biodiversity impacts in the units PDF·ha-1 for the BASE scenario in 2050 in the 430 EU28 NUTS2 administrative units due to land use (i.e., excluding trade). 431 19 432 433 Figure 3. Biodiversity impacts in the units PDF for the BASE scenario in 2050 in the 434 EU28 NUTS2 administrative units due to land use (i.e., excluding trade). 435 436 20 437 438 Figure 4. Biodiversity impacts in the units PDF·ha-1 for the BASE scenario as difference 439 between the years 2050 and 2000 in the EU28 NUTS2 administrative units due to land 440 use (i.e., excluding trade). 441 21 442 443 Figure 5. Biodiversity impacts in the units PDF for the BASE scenario as difference 444 between the years 2050 and 2000 in the EU28 NUTS2 administrative units due to land 445 use (i.e., excluding trade). 446 The biodiversity impacts from utilized forests increase in the future in all three scenarios due to 447 a growth in roundwood extraction, which is expected to increase by 26% in the BASE scenario 448 from the year 2000 to 2050. The corresponding numbers in the CONST and EMIRED scenarios 449 are 20% and 41% respectively (SI 6 Table 3). Meanwhile, the surface of utilized forests 450 increases by 19% in the BASE scenario, 15% in the CONST scenario, and 29% in the EMIRED 451 scenario. The corresponding potential biodiversity damage due to forest management increases 22 452 by 8.9%, 10.0%, and 18.5% over time in the CONST, BASE, and EMIRED scenarios 453 respectively (SI 6 Table 3). 454 For all the scenarios, there is generally a reduction of impacts per unit of roundwood extracted 455 over time, however the difference between the scenarios and over time is of a relatively small 456 magnitude. 457 The impacts due to perennials (miscanthus and short rotation energy plantations) increase 458 significantly over time in all scenarios; in the CONST scenario they stabilize after the year 2020, 459 while in the BASE and EMIRED scenarios they continue to grow until the year 2050. In the year 460 2050, the potential biodiversity impact in the EMIRED scenario is almost doubled compared to 461 the BASE scenario (i.e., a 95% increase) (Figure 6). The impacts due to the expansion of 462 perennials is more relevant in the regions of South West and Central West Europe, representing 463 46% and 24% respectively of total damage in the EMIRED scenario in 2050 (Figure 6). 464 In the BASE scenario, the PDF per m3 of perennials increases by 82% from 2010 to 2050, the 465 corresponding increases are 22% and 48% in the CONST and EMIRED scenarios respectively 466 (SI 6 Table 4). The impacts per m3 in the CONST and EMIRED scenarios are similar to those of 467 the BASE scenario from 2010 to 2030. After the year 2030, in these scenarios the impacts per 468 m3 are 11% to 33% lower than in the BASE scenario. This could be due to different reasons: in 469 the CONST scenario, the demand for perennials is lower than in the BASE scenario, therefore 470 the expansion of perennials is limited to natural vegetation and pasture land with relatively high 471 yields compared to the land occupied by perennials in the BASE scenario. In the EMIRED 472 scenario, the demand for perennials is higher than in the BASE scenario, which causes a further 473 expansion of perennials in relatively high fertility croplands. However, in most of the regions, the 474 increase of demand in the EMIRED scenario compared to the BASE scenario did not 475 correspond to a significant expansion of perennials in croplands, leading to higher impacts per 476 m3 than in the BASE scenario. 23 477 478 479 480 Figure 6: Development of biodiversity impacts due to land use in the units PDF from the 481 year 2000 to 2050 due to perennial land use for the different regions of the EU (for a list 482 of countries in each region see SI 5). 483 484 3.2 Impacts from trade 485 The net import of woody biomass from forestland to the EU28 progressively increases from the 486 year 2000 to 2050. In the meantime, the EU28 increases the net import of cropland until 2030, 487 and it then either stabilizes or continues to increase, depending on the scenario (Figure 1, SI 6 488 Table 5)In terms of traded product mass, the most important partners to the EU with regard to 489 the net import of crops (i.e., from cropland) are Brazil, Australia, New Zealand, the Pacific 490 Islands, Turkey, and Ukraine. From these regions, the most relevant imported crops are 491 sugarcane, soy, rapeseed, sunflower, and cassava. Meanwhile, the most important net export 492 regions are Africa and the Middle East. Our results show that over time, Canada and the former 493 USSR also become relevant export regions. The most important net exported crops are wheat, 494 corn, barley, and potatoes. 24 495 Pulp logs represent the largest share of net imports in terms of mass within the forest sector in 496 the BASE and CONST scenarios, while in the EMIRED scenario, pellets achieve the same 497 mass as pulp logs in 2040 and then in 2050 exceed pulp logs. The largest shares of pulp logs 498 are imported from the former USSR and Malaysia, while for pellets the leading exporters are 499 Canada, the former USSR, and the US (SI 6 Table 5). 500 The total biodiversity damage caused by net imports is in the order of 0.1% to 0.2% global 501 species loss (1-2 × 10-3 PDF). Cropland causes 99% of this impact, and the remainder is mostly 502 due to pulp logs (0.4-0.8%) and pellets (0.1-0.5 %) (Table 6). The impacts in 2050 are 2.2, 2.1, 503 and 2.0 fold the ones observed in 2000 for the EMIRED, BASE, and CONST scenarios 504 respectively. The differences between scenarios are amplified after the year 2030: In 2050, the 505 impacts for the EMIRED scenario are 6% higher than for the BASE scenario, while for the 506 CONST scenario they are 2% lower than in the BASE scenario (cf. SI 6 Table 1 and 6). 507 At the regional level, the largest shares of impacts due to net imports for the BASE scenario in 508 the year 2000 are caused by Central West (54% of impact in the EU28) and South West Europe 509 (39% of impact in the EU28). Over time there is a progressive increase of net imports for 510 Central West Europe relative to the other regions (Figure 7). Consequently, Central West 511 Europe causes 67% of EU28 damage due to net import in the BASE 2050 scenario. 512 513 514 515 516 517 25 518 519 520 521 522 523 524 Figure 7: The development of biodiversity impacts due to trade in the units PDF from 525 the year 2000 to 2050 expressed as the total of impacts due to external trade for the 526 different regions within the EU regions. The red dots represent the arithmetic sum. 527 Negative values denote net exports. 528 529 3.3 The biodiversity footprint of Europe 530 The total biodiversity damage, here referred to as the “EU footprint” was calculated as the sum 531 of impacts due to domestic land uses in the EU28 summed to the impacts due to imports and 532 decreased by the exports. 26 533 The EU footprint is in the order of 0.7% to 0.9% of global species loss (7-9 ×10-03 PDF). The 534 impact of the BASE scenario increases by 26.1% from the year 2000 to 2050. The 535 corresponding growths in the CONST and EMIRED scenarios are 24.1% and 28.6% 536 respectively. The difference between scenarios is less than 1% until 2030. This increases over 537 time and in the year 2050 impacts for the EMIRED scenario are 1.9% larger than in the BASE 538 scenario. In the CONST scenario, they are 1.7% lower than in the BASE scenario (Table 1). 539 In all scenarios there is a growth over time in the share of impacts due to imports compared to 540 land use, starting from 15% in the year 2000 and reaching 24% to 26% in 2050 (Figure 1). 541 After correcting for internal trade in the EU, the results show that countries in the Central West 542 EU that are strongly dependent on imports (i.e., the UK) reach a total footprint (i.e., sum of land 543 use and import) comparable in magnitude to the ones in the South West EU (cf. Fig. 2 and Fig. 544 4). The compensatory effect of imports is already evident in the year 2000 and increases over 545 time. 546 In the BASE scenario, the countries with the largest share of the total footprint in the year 2000 547 are countries in the southwestern region. These countries represent 57% of EU impacts in the 548 year 2000, and their share decreases to 54% by 2050 in the BASE scenario. Countries in 549 Central West Europe, which are generally more dependent on net import, enlarge their share of 550 the total EU footprint from 20% in 2000 to 26% in 2050. Similar tendencies are observed across 551 all scenarios (Figure 8). 552 Table 1. Total biodiversity footprint from the EU (PDF), as the sum of impacts due to 553 land use and net imports to the EU 28 in the three different bioenergy scenarios. Year 2000 2010 2020 2030 2040 2050 Baseline (PDF) 7.50×10-03 7.84×10-03 8.23×10-03 8.66×10-03 9.02×10-03 9.46×10-03 Constant demand 0 0 -3.55×10-07 -8.19×10-05 -1.33×10-04 -1.56×10-04 27 (PDF Δ to baseline) Emission reduction (PDF Δ to baseline) 0 0 -7.08×10-06 -3.32×10-05 5.67×10-05 1.79 ×10-04 554 555 556 557 Figure 8: The development of the total biodiversity footprint from the EU in units PDF 558 from the year 2000 to 2050 for the different EU regions. 559 560 4. Discussion 561 In this paper, we set out a global framework that is able to jointly assess and analyze the 562 biodiversity implications of policies related to direct land use change, changes in intensity in 563 land use and forestry, and in-direct land use effects. Utilizing this framework, we analyzed three 564 different bioenergy policies in the EU28 and their effects on biodiversity, focusing on the 565 expected changes in domestic land use and the possible damage on global biodiversity through 566 trade. 567 In all scenarios, we observed a significant increase of biodiversity damage over time. In the long 568 term (by 2050), the potential species loss due to the EU footprint was found to increase from 28 569 0.75% in 2000 to almost 1% of global species in 2050. Previous assessments suggest that ca. 570 10% of species globally could potentially have disappeared by 2050, compared to the year 2000 571 (CDB 2014). Given this background, the dynamics we analyzed for the EU28 have considerable 572 impact on a global scale. 573 The increase of the biodiversity footprint over time is due to both an expansion of domestic land 574 use and, especially, to land use imported through agricultural products into the EU. The 575 international character of the problem is emphasized over time: The damage due to imported 576 land use increased form 15% of total damage in the year 2000 to 24% to 26% in 2050, meaning 577 that the footprint is progressively outsourced. This overall increase in the share of footprint 578 caused by imports is mainly due to an increase of imported agricultural products to fulfill the 579 growing European food demand and the area needed for this production outside the EU. This 580 trend is reinforced by the conversion of cropland into perennials in the EU, which leads to 581 outsourcing some of the cropland production to outside the EU. The biodiversity damage is 582 magnified, as the imports of agricultural products include countries of origin located in tropical 583 regions, in areas particularly rich of biodiversity and vulnerable species (i.e., Brazil, Australia, 584 New Zealand, and the Pacific Islands). In these countries, the indirect damage per tonne of 585 product is 5.9 to 8.9 times larger than in the EU. This result is in line with the findings of 586 previous studies, which have found that the food consumption in industrialized countries drives 587 biodiversity loss in tropical developing countries through international trade (Chaudhary & 588 Kastner 2016, Lenzen et al. 2012). 589 Within the EU, agricultural production remains the largest domestic driver of land use related 590 biodiversity impacts in all scenarios. The increase of food demand in the whole EU is expected 591 to lead to a 1-8 Mha expansion of domestic cropland. However, the contribution of domestic 592 cropland to the total EU biodiversity footprint (including imported land) is expected to decrease 593 over time from 66% in 2000 to 54% to 59% in 2050. The most relevant future change of 29 594 domestic land use in all scenarios is the expansion of perennial cultivation for energy, which is 595 expected to increase to 2 to 14 Mha by 2050. The perennials are projected to increase 596 especially after the year 2020, although their contribution to the total biodiversity footprint 597 remains limited to 1% to 6% in 2050. Forests under active management in the EU expand over 598 time by 10 to 20 Mha. Nevertheless, the domestic forest management area continues to be of 599 minor relevance for the total biodiversity footprint (4% to 5% of the total footprint in 2050) 600 compared to damages due to other domestic land uses and imported land use. 601 The difference between the three scenarios was found to be small compared to the magnitude 602 of biodiversity damage increase over time. This finding is similar to the findings of Eggers et al. 603 (2009), who also observed that different biofuel targets in the EU had a much smaller effect on 604 biodiversity than the overall trend of biodiversity reduction observed over time from 2000 to 605 2030. 606 In our study, the scenario with the highest demand of bioenergy (EMIRED) created similar 607 damage than the other two scenarios in 2050 (the difference between the scenarios was only 608 1.9% to 3.6%). In the EMIRED scenario, there is a larger expansion of perennials and a smaller 609 expansion of cropland in the EU than in the other two scenarios. As the biodiversity damage is 610 smaller in perennials than for cropland, this development lowers the internal biodiversity 611 damage per unit of land occupied. In the meantime, in the EMIRED scenario there is also the 612 largest outsourcing of damage, due to increased import of cropland products from outside the 613 EU for satisfying the EU food demand. The two opposite effects even each other out, resulting 614 in the total biodiversity damage for the EMIRED scenario being similar to the other two 615 scenarios. 616 Over time, the growth in bioenergy demand also increases the import of wood pellets to the EU. 617 In 2050, imports of pellets in the EMIRED scenario are 2.1 to 2.7 times higher than in the 618 CONST or REF scenarios. However, the biodiversity damage created by wood pellet imports 30 619 has only marginal relevance compared to the import of agricultural products, given the relatively 620 lower characterization factors for managed forests compared to cropland and the main countries 621 of origin for wood exports (USSR, Canada, and the US). In these countries, the biodiversity 622 damage per unit of wood pellet is 4 to 24 smaller than the damage in the EU resulting from 623 perennial cultivation. 624 The internal distribution of the EU footprint is determined by the split of the area into different 625 land uses in each region in terms of the biodiversity richness in the different ecoregions, and 626 especially by the amount of net imports. For these reasons, the largest biodiversity footprints in 627 the EU were initially observed in the Mediterranean region, which is the region that hosts most 628 of the biodiversity in the EU. However, over time, the biodiversity footprint increases in central- 629 western EU countries that are particularly dependent on imports, due to the relatively more 630 severe damage per unit of land caused by imports compared to the damage caused by 631 domestic production of biomass (cf. Fig. 8). 632 We used only one indicator of species loss (potentially disappeared fraction of global species, 633 PDF), which was obtained by aggregating the richness of species across the different taxa. A 634 single indicator will not capture damages due to changes in species composition that take place 635 following disturbances. The same methodology can be repeated through the use of 636 characterization factors for the single taxa (cf. Chaudhary et al. 2015). Eggers et al. (2009) 637 investigated the suitability of different species and concluded that mammals and birds were the 638 most damaged by the expansion of biofuel crops. Therefore, to investigate the biodiversity 639 damage in more detail, an investigation into impacts across the different taxa could be a 640 possible extension of the current study. Furthermore, in the current study only global extinction- 641 equivalents were accounted for, thus neglecting regional extinctions. However, the latter may 642 also be important to warrant local ecosystem functioning and should be assessed in future 643 research. 31 644 The strength of the approach that we proposed for evaluation of the biodiversity damage is that 645 it is already consolidated in the literature (see UNEP/SETAC 2016). The biological functionality 646 of an ecosystem, such as functional diversity, could not be assessed using species richness as 647 an indicator. This is a topic for future studies, as there is currently a lack of available 648 approaches for combining and evaluating different indicators of functionality (Maia De Souza et 649 al. 2014). 650 651 In our calculations we assumed a steady state change in species extinctions and neglected the 652 temporal evolution of biodiversity loss. In reality, the species will not instantaneously go extinct 653 or return when land use change takes place. This delay in the species dynamics means that it is 654 likely that the changes assessed in this paper will happen more gradually than assumed. We 655 also did not consider aspects such as land fragmentation or ecological corridors, which are 656 important to biodiversity with regard to landscape level continuity. Landscape analyses could 657 help to understand the effects of different patterns of land use. 658 The economic model used in this study produced land use projections at the resolution of 659 NUTS2 administrative units. Trade was modeled between global trade regions, and internal 660 trade within the EU was modeled at the country level. The characterization factors were 661 originally available on an ecoregion scale, hence they were re-scaled to fit the different 662 resolutions (NUTS2, Country, trade region). Although some accuracy is lost through this re- 663 scaling, the geographical scale is still rather small and is considered to be a strength of the 664 study. 665 666 In our study we estimated the intensity of forest management in the EU using a suitability map, 667 which considered 28% of managed forests under intensive management. 32 668 Intensifying the use of forest biomass could also affect forest management regimes, leading to a 669 reduction in rotation periods, a possible increase of monocultures, or collection of residual wood 670 debris. These could in turn negatively affect biological diversity and natural habitats, which could 671 lead to further reductions of local biodiversity (Lassauce et al. 2012, Söderberg & Eckerberg 672 2013). All these aspects could result in more substantial damage to the hosted biodiversity 673 compared to our analyses. 674 Using the growth rates for plantations predicted in ABARE & Pöyry (1999) and Jürgensen et al. 675 (2014), which assumes further intensification of European forest management than what 676 resulted from our scenarios, intensified managed forests in the EU could reach 39% of 677 managed forests in 2030 and 55% by 2040-2050. In this new condition, the potential species 678 loss in 2050 would increase and forests would cause 7.4% to 7.8% of the internal land use 679 damage. However, we considered only two classes of intensity, while Chaudhary et al. (2016) 680 provided response ratios for then different classes of forest management. They found significant 681 and different species losses produced by plantations, clear-cutting, and conventional selective 682 logging. This suggests that forest management intensity may have a larger effect than what is 683 shown in the current study. Our simplification was due to the scarcity of data regarding forest 684 management statistics, which did not allow us to distinguish globally among more than the two 685 classes. 686 The intensity of forest management outside the EU was based on the regional statistics of wood 687 supply from planted forests reported in Jürgensen et al. (2014) and projected according to the 688 long term growth rates estimated by ABARE & Pöyry (1999). Currently, there is a lack of data 689 for validating the area of forest plantations. The statistics that are globally available from the 690 Food and Agriculture Organization of the United Nations (FAO)’s Forest Resource Assessment 691 (FRA 2015), report the surface and change rate of “planted forests” per country without 692 specifying the different uses of the planted forests (production, protection, etc.). In some 33 693 regions, the current growth rates observed in FRA 2015 could deviate from the ones projected 694 in our study. However, it is not straightforward to distinguish planted forests from plantations 695 that currently contribute to wood supply in the different regions. The current EU import of forest 696 biomass is mostly sourced from forests and wood plantations assimilated to intensively 697 managed forests. For this reason, we assumed future wood imports to have also originated from 698 intensively managed forestland. If considering the most biodiversity adverse situation (the whole 699 EU import of wood pellets in 2050 will be sourced from perennial plantations), the damage due 700 to imports of wood products in 2050 would increase by a factor of 1.2 to 1.5. Under this 701 condition, the biodiversity damage per m3 of wood pellets imported is still 2 to 17 smaller than 702 the damage for perennials in the EU. However, more significant damage could be induced if 703 perennial plantations outside the EU would displace food production. If considering both the 704 domestic and external intensification in the supply of woody biomass, the potential species loss 705 in 2050 would increase by a factor of 1.01. 706 The GLOBIOM model used for projecting land uses is based on the economic convenience of 707 allocating land to different uses, or in practical terms, on demand and supply curves. The growth 708 of a bioeconomy in the EU could lead to an intensification of local demand points in some 709 regions where the industry could expand more easily, and this could reduce costs and increase 710 the profitability of supplying biomass locally (Hellman & Verburg 2011). This development could 711 significantly alter the land use allocated to biomass production within the different regions 712 compared to our results. Therefore, the results must be seen as representative of general 713 trends in the EU, and not as being an exhaustive description of development within each 714 administrative unit. 715 Agricultural and forestry yields were kept constant in our study, which could have led to an 716 overestimation of damages produced by future cropland expansion. Increasing agricultural 717 yields in regions with significant yield gaps could lead to intensification, which could further lead 34 718 to future sparing of land from agriculture and instead utilizing it for possible bioenergy use 719 (Lamb et al. 2016). Consequently, we were rather conservative from this point of view. 720 721 5. Conclusions 722 Our results show that policies promoting bioenergy in the EU may contribute to a further global 723 decline of biodiversity. While a strong expansion of perennial crops for bioenergy production 724 could be an interesting option for climate change mitigation, it could have negative impacts on 725 biodiversity through loss of species habitats. Further, our results indicate that through 726 international trade, an increase in bioenergy demand may result in a considerable leakage of 727 biodiversity damage to species-rich and vulnerable regions outside the EU. Therefore, in the 728 case of future increase in bioenergy demand, the combination of supply from sustainable forest 729 management in the EU and imported wood pellets combined with the cultivation of perennial 730 energy crops, appears to be less detrimental to biodiversity than only an expansion of energy 731 crops. 732 733 734 735 736 737 738 739 35 740 741 742 743 744 745 746 747 748 Acknowledgements 749 The “Imbalance P” project is acknowledged for financial support under the European Research 750 Council Synergy Grant 610028. 751 This project has received funding from the European Union’s Horizon 2020 research and 752 innovation programme under ALTERFOR project (Alternative models and robust decision- 753 making for future forest management), Grant Agreement No. 676754. 754 We would like to thank our colleague Dr. Pekka Lauri for professional advice. 755 We would like to thank Ms. Christie Walker and Ms.Hansa Heyl for the English proofreading. 756 757 758 759 36 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 37 778 779 780 781 782 783 784 785 786 References 787 788 ABARE/Pöyry, 1999. Global outlook for plantations. Australian Bureau of Agriculture and 789 Resource Economics (ABARE) and Jaakko Pöyry Consulting. ABARE Research Report 99.9. 790 Canberra. Available at: 791 http://143.188.17.20/data/warehouse/pe_abarebrs99000431/PC11463.pdf 792 793 Baumber A. 2017. Enhancing ecosystem services through targeted bioenergy support policies. 794 Ecosystem Services 26: 98–110. 795 796 Berndes, G., Fritsche, U., 2016. May we have some more land use change, please? Biofuels, 797 Bioprod. Biorefin. 10 (3), 195–197. 798 38 799 Britz W, Hertel TW. 2011. Impacts of EU biofuels directives on global markets and EU 800 environmental quality: an integrated PE, global CGE analysis. Agric Ecosyst Environ 801 2011;142:102–9. 802 803 CDB 2014. Global Biodiversity Outlook 4. Secretariat of the Convention on Biological Diversity, 804 Montréal, 155 pp. 805 806 Chaudhary, A., Burivalova, Z., Koh, L.P., Hellweg, S. 2016. Impact of Forest Management on 807 Species Richness: Global Meta-Analysis and Economic Trade-Offs. Scientific Reports, 6 808 (23954): 809 810 Chaudhary A., Pfister S., Hellweg S. 2016. Spatially Explicit Analysis of Biodiversity Loss Due to 811 Global Agriculture, Pasture and Forest Land Use from a Producer and Consumer Perspective. 812 Environmental Science & Technology 2016 50 (7), 3928-3936 813 814 Chaudhary, A., Kastner, T. 2016. Land use biodiversity impacts embodied in international food 815 trade. Global Environmental Change 38, pp. 195-204. 816 817 Chaudhary, A.; Verones, F.; de Baan, L.; Hellweg, S. 2015. Quantifying Land Use Impacts on 818 Biodiversity: Combining Species−Area Models and Vulnerability Indicators. Environ. Sci. 819 Technol. 2015, 49 (16), 9987−9995. 820 821 Curran M., de Souza DM., Antón A., Teixeira R, Michelsen O, Vidal-Legaz B, Sala S, Milà i 822 Canals L. 2016. How Well Does LCA Model Land Use Impacts on Biodiversity? A Comparison 823 with Approaches from Ecology and Conservation. Environ Sci Technol. 50(6):2782-95. 824 39 825 Dauber J., Jones M.B., Stout J.C., The impact of biomass crop cultivation on temperate 826 Biodiversity. GCB Bioenergy (2010) 2, 289–309 827 828 de Baan, L.; Mutel, C. L.; Curran, M.; Hellweg, S.; Koellner, T.2013. Land use in life cycle 829 assessment: global characterization factors based on regional and global potential species 830 extinction. Environ. Sci. Technol. 2013b, 47 (16), 9281−9290. 831 832 Dimitriou, I., Baum, C., Baum, S., Busch, G., Schulz, U., Köhn, J., Lamersdorf, N., Leinweber, 833 P., Aronsson, P., Weih, M., Berndes, G., Bolte, A., 2011. Quantifying environmental effects of 834 Short Rotation Coppice (SRC) on biodiversity, soil and water. IEA Bioenergy. Task 43. 835 836 Eggers, J.; Tröltzsch, K.; Falcucci, A.; Verburg, P.H.; Ozinga, W.A. 2009. Is biofuel policy 837 harming biodiversity in Europe? Global Change Biology Bioenergy 1:18 - 34. 838 839 Elbersen, B., Fritsche, U., Petersen, J.-E., Lesschen J. P., Böttcher, H., Overmars, K. 2013. 840 Assessing the effect of stricter sustainability criteria on EU biomass crop potential. Biofuels, 841 Bioproducts and Biorefining 7(2): 173-192. 842 843 European Commission, 2016. Proposal for a COUNCIL DECISION on the conclusion on behalf 844 of the European Union of the Paris Agreement adopted under the United Nations Framework 845 Convention on Climate Change. COM(2016) 395 final - 2016/0184. 846 847 European Commission, 2011. Our life insurance, our natural capital: an EU biodiversity strategy 848 to 2020. /* COM/2011/0244 final */ 849 40 850 European Parliament and Council of the European Union, 2016. Proposal for a DIRECTIVE OF 851 THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on the promotion of the use of 852 energy from renewable sources. COM/2016/0767 final/2 - 2016/0382 853 854 855 European Parliament, Council of the European Union, 2015. Directive (EU) 2015/1513. Official 856 Journal of the European Union L239, 1–29. 857 858 Forsell, N. et al. 2016: Study on impacts on resource efficiency of future EU demand for 859 bioenergy (ReceBio). Final report. Project: ENV.F.1/ETU/2013/0033. Luxembourg: Publications 860 Office of the European Union, 2016. 43 p. 861 862 FRA 2015. Food and Agriculture Organization of the United Nations. The Global Forest 863 Resources Assessment 2015. Main report. Rome, 2015. 244 pp. 864 865 Havlík P., Schneider U., Schmid E., Böttcher H. Fritz S., Skalsky R., Aoki K., De Cara S., 866 Kindermann G., Kraxner F., Leduc S., McCallum I., Mosnier A., Sauer T., Obersteiner M. 2011. 867 Global land-use implications of first and second generation biofuel targets. Energy Policy, 2011 868 vol: 39 (10) pp: 5690-5702 869 870 Havlík, P, Valin, H, Herrero, M, Obersteiner, M, Schmid, Ed, Rufino, MC, Mosnier, A, 871 Thornton, PK, Böttcher, H, Conant, RT, Frank, S, Fritz, S, Fuss, S, Kraxner, F, Notenbaert, 872 A, 2014. Climate change mitigation through livestock system transitions. Proceedings of the 873 National Academy of Sciences of the United States of America. Volume 111, Issue 10, 11 874 March 2014, Pages 3709-3714 875 41 876 Heller, N. E., & Zavaleta, E. S. (2009). Biodiversity management in the face of climate change: a 877 review of 22 years of recommendations. Biological conservation, 142(1), 14-32 878 879 Hellmann F, Verburg PH 2011. Spatially explicit modelling of biofuel crops in Europe. Biomass 880 and bioenergy 35: 2411-2424. 881 882 Hengeveld, G. M., G.-J. Nabuurs, M. Didion, I. Van den Wyngaert, A. P. P. M. Clerkx, and M.-J. 883 Schelhaas. 2012. A forest management map of European forests. Ecology and Society 17(4): 884 53. 885 886 Holland, R.A., Eigenbrod, F., Muggeridge, A., Brown, G., Clarke, D., Taylor, G., 2015. A 887 synthesis of the ecosystem services impact of second generation bioenergy crop production. 888 Renew. Sustain. Energy Rev. 46, 30–40. 889 890 Hudson LN, Newbold T, Contu S, Hill S, Lysenko I, De Palma A, et al. 2014. The PREDICTS 891 database: a global database of how local terrestrial biodiversity responds to human impacts. 892 Ecology and Evolution 4 (24), 4701-4735. 893 894 Jürgensen, C., Kollert, W. and Lebedys, A. 2014. Assessment of industrial roundwood 895 production from planted forests. FAO Planted Forests and Trees Working Paper FP/48/E. 896 Rome. 897 898 Koellner T., de Baan L., Beck T., Brandão M., Civit B., Margni M., i Canals L.M., Saad R., de 899 Souza D.M. Müller-Wenk R. 2013. UNEP-SETAC guideline on global land use impact 900 assessment on biodiversity and ecosystem services in LCA. Int J Life Cycle Assess, 18 (2013), 901 pp. 1188-1202 42 902 903 Lamb , A , Green , R , Bateman , I , Broadmeadow , M , Bruce , T , Burney , J , Carey , P , 904 Chadwick , D , Crane , E , Field , R , Goulding , K , Griffiths , H , Hastings , A , Kasoar , T , 905 Kindred , D , Phalan , B , Pickett , J , Smith , P , Wall , E , zu Ermgassen , E K H J & Balmford , 906 A 2016 , ' The potential for land sparing to offset greenhouse gas emissions from agriculture ' 907 Nature climate change , vol 6 , pp. 488-492. 908 909 Lassauce A, Lieutier F, Bouget C. 2012. Woodfuel harvesting and biodiversity conservation in 910 temperate forests: effects of logging residue characteristics on saproxylic beetle assemblages. 911 Biol Conserv 2012;147:204–12. 912 913 Lauri P, Havlik P, Kindermann G, Forsell N, Böttcher H, Obersteiner M. 2014. Woody biomass 914 energy potential in 2050. Energy Policy 66: 19-31. 915 916 Lenzen, M.; Moran, D.; Kanemoto, K.; Foran, B.; Lobefaro, L.; Geschke, A. 2012. International 917 trade drives biodiversity threats in developing nations. Nature 2012, 486: 109−112. 918 919 Mantau, U., et al., 2010. Final Report — Real Potential for Changes in Growth and Use of EU 920 Forests. EUwood Project: Call for Tenders. No. TREN/D2/491-2008. 921 922 Newbold T, Hudson LN, Arnell AP, Contu S, De Palma A, Ferrier S, et al. 2015. Science 353 923 (6296): 288-291. 924 925 Obersteiner, M., Bednar, J., Wagner, F., Gasser, T., Ciais, P., Forsell, N., Frank, S., Havlik, P., 926 Valin, 43 927 H., Janssens, I.A., Peñuelas, J., Schmidt-Traub, G. 2018. How to spend a dwindling 928 greenhouse gas budget. Nature Climate Change, 8 (1): 7-10. 929 930 Olson, D. M., Dinerstein, E., Wikramanayake, E. D., Burgess, N. D., Powell, G. V. N., 931 Underwood, E. C., D'Amico, J. A., Itoua, I., Strand, H. E., Morrison, J. C., Loucks, C. J., Allnutt, 932 T. F., Ricketts, T. H., Kura, Y., Lamoreux, J. F., Wettengel, W. W., Hedao, P., Kassem, K. R. 933 2001. Terrestrial ecoregions of the world: a new map of life on Earth. Bioscience 51(11):933- 934 938. 935 936 Pereira, H. M., Leadley, P. W., Proença, V., Alkemade, R., Scharlemann, J. P., Fernandez- 937 Manjarrés, J. F., & Chini, L. (2010). Scenarios for global biodiversity in the 21st century. 938 Science, 330(6010), 1496-1501. 939 940 Rivas Casado, M., Mead, A., Burgess, P.J., Howard, D.C., Butler, S.J. 2014. Predicting the 941 impacts of bioenergy production on farmland birds. Science of the Total Environment 476-477, 942 pp. 7-19 943 944 Schelhaas, M.J., van Brusselen, J., Pussinen, A., Pesonen, E., Schuck, A., Nabuurs, G.J. and 945 Sasse, V., 2006. Outlook for the Development of European Forest Resources. A study prepared 946 for the European Forest Sector Outlook Study (EFSOS). Geneva Timber and Forest Discussion 947 Paper. ECE/TIM/DP/41. UN-ECE, Geneva. 948 949 Schipper A., Bakkenes M., Meijer J., Alkemade R., Huijbregts M. 2016. The GLOBIO model. A 950 technical description of version 3.5. PBL Netherlands Environmental Assessment Agency The 951 Hague, 2016 PBL publication number: 2369 952 44 953 Schulze, J., K. Frank, J. A. Priess, and M. A. Meyer. 2016. Assessing Regional-Scale Impacts 954 of Short Rotation Coppices on Ecosystem Services by Modeling Land-Use Decisions. Plos One 955 11:21. 956 957 Söderberg, C., Eckerberg, K. 2013. Rising policy conflicts in Europe over bioenergy and 958 forestry. Forest Policy and Economics 33: 112-119. 959 960 Souza DM ,Teixeira RFM, Ostermann OP 2015. Assessing biodiversity loss due to land use 961 with Life Cycle Assessment: are we there yet? Global Change Biology 21: 32–47. 962 963 UN General Assembly, 2015. Transforming our world: the 2030 Agenda for Sustainable 964 Development, 21 October 2015, A/RES/70/1, available at: 965 http://www.refworld.org/docid/57b6e3e44.html [accessed 13 September 2017] 966 967 968 45