WPS8718 Policy Research Working Paper 8718 Reducing Environmental Risks from Belt and Road Initiative Investments in Transportation Infrastructure Elizabeth Losos Alexander Pfaff Lydia Olander Sara Mason Seth Morgan Macroeconomics, Trade and Investment Global Practice January 2019 Policy Research Working Paper 8718 Abstract The Belt and Road Initiative, due to its diverse and extensive transportation projects across the globe. This paper studies infrastructure investments, poses a wide range of environ- environmental risks—direct and indirect—from Belt and mental risks. Some projects have easily identifiable and Road Initiative transportation projects and the mitigation measurable impacts, such as energy projects’ greenhouse strategies and policies to address them. The paper concludes gas emissions. Others, such as transportation infrastructure, with a recommendation on how to take advantage of the due to their vast geographic reach, generate more complex scale of the Belt and Road Initiative to address these con- and potentially more extensive environmental risks. The cerns in a way not typically available to stand-alone projects. proposed Belt and Road Initiative rail and road invest- In short, this scale motivates and permits early integrated ments have stimulated concerns because of the history of development and conservation planning. significant negative environmental impacts from large-scale This paper is a product of the Macroeconomics, Trade and Investment Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/research. The authors may be contacted at elizabeth.losos@duke.edu. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Reducing Environmental Risks from   Belt and Road Initiative Investments in Transportation  Infrastructure1    Elizabeth Losos, Alexander Pfaff, Lydia Olander, Sara Mason, and Seth Morgan  Duke University                                                                       1  We thank Michael Toman, Glenn Morgan, Sven‐Uwe Mueller, Graham Watkins, Daniel Taras, Bekele Debele,  Michele Ruta, and seminar participants at the World Bank for their helpful suggestions. Tristan Reed and Sasha  Trubetskoy were especially generous in sharing their transport infrastructure data and knowledge. All errors  remain our own.  INTRODUCTION  The location of Roman roads built almost two millennia ago still shapes modern landscapes and local  economies (Dalgaard et al. 2018). That scholars can still detect the effect of this transportation footprint  is testament to the transformational and enduring influence of transportation networks on people, the  economy, and the environment. Environmentally speaking, networks of roads and railways often have  been identified as one of the most significant anthropogenic interventions of the earth's ecosystems (W.  Laurance 2017; W. F. Laurance et al. 2015; W. F. Laurance and Arrea 2017; Popp and Boyle 2017).   Historically, large‐scale transport infrastructure projects have had significant negative environmental  impacts across the globe. Given the enormity of the Belt and Road Initiative (BRI), what environmental  risks do the proposed BRI transportation projects pose? In Part 1 of this working paper, we review many  of the most significant risks facing BRI transportation projects. We consider both the direct effects from  roads and rail – typically immediate and localized – and indirect effects from opening up new landscape  frontiers, shifting human populations and markets, altering demands for transportation, and balancing  energy efficiency with induced demands. Next, in Part 2, we review potential environmental mitigation  strategies that could improve the environmental performance of BRI transportation projects. Finally, in  Part 3 we will describe policies that can shape the implementation and enforcement of mitigation.   Though not addressed in this paper, BRI projects in other sectors also have environmental impacts. In  particular, BRI energy projects such as fossil fuel and renewable energy projects can have a considerable  effect on greenhouse gas emissions (GHG) and other pollutants. Serious concerns have been raised that  the promotion of BRI fossil fuel investments (especially coal plants) could lock host countries into fossil  fuel dependency for the coming decades and hamper them from reaching their nationally determined  contribution carbon targets as established under the Paris Agreement on Climate Change (Gallagher and  Qi 2018; Sausmikat et al. 2017; Zhou et al. 2018). While BRI transport projects – the focus of this paper –  have less direct impact on GHG emissions, the vast geographic reach of road and rail networks results in  a more complex set of environmental risks, which can include significant impacts upon GHG emissions.   Transportation investments also pose serious social risks, many closely paralleling environmental risks.  Social risks can include economic and physical displacement from taking of land and assets; impacts on  disadvantaged, vulnerable, or groups with special rights; impacts on quality of life; degradation of labor  and working conditions; and community health, safety and security. These can be profound yet – like  environmental risks – they vary by context such as settlement density, working conditions, labor force,  worker protections, planning provisions, and other conditions. Social issues are not addressed here yet it  should be noted that there is a complementarity between social and environmental risks and mitigation  strategies. For instance, our recommendations are equally viable for addressing many of the social risks.   Throughout, we provide examples drawn from the BRI transportation projects within the BRI Corridors,  illustrated in the appendices.  Appendix 1 is an overview of environmental risks at the scale of the entire  BRI.  Appendices 2‐9 focus on a number of aspects from one specific economic corridor, the China‐ Indochina Peninsula Economic Corridor (CICPEC) – specifically the three alternative transportation  routes between Kunming, China and Bangkok, Thailand – to illustrate more general points made  throughout the paper.   2    In sum, our working paper identifies serious environmental risks from the BRI transportation projects  and the potential for mitigation via Corridor SEA planning done as early as possible. The CICPEC example  will provide a model of how a BRI Corridor Strategic Environmental Assessment could meet these needs.    A.  Data Basis  Our working paper and especially the CICPEC examples draw on the World Bank data set from  Reed and Trubetskoy (2018) of BRI railway and road projects. These authors have geolocated almost  30,000 km of new or upgraded railways and roads that have been constructed or are in the process of  being constructed since 2013 and almost 15,000 km more are currently in the planning stages (see Table  1; Reed and Trubetskoy 2018). The majority of these investments have been or will be in rails, not roads:  more than 60% (by distance) of those recently constructed or under construction are rail projects and  90% of the planned BRI projects are railways (Reed and Trubetskoy 2018).2 When possible, we will  distinguish between transport categories (new versus upgraded infrastructure; train versus rail; high  capacity versus conventional rail; divided highway versus undivided highway). Where research is lacking,  we must speculate based on prior similarities and differences among these categories in other locations.   Table 1. BRI road and rail projects, completed, on‐going, and planned  Km under  Additional km  BRI Projects  construction or   planned,   Km total  already built  not built     km  %  km  %  km  %          New Undivided Highways  5,547   19%  378   3%              5,926   14%          New Divided Highways  809   3%  ‐    0%                  809   2%          Upgrade to Divided Highways  4,723   16%  841   6%              5,564   13%    Total Road Improvement  11,079    1,220    12,299      New Conventional Railways  8,649  30%  4,030   27%  12,680  29%          New High Capacity Railways  2,835   10%  3,975   27%              6,809   16%          Upgrade to High Capacity Railways  6,228   22%  5,768   38%            11,997   27%    Total Rail Improvements  17,712    13,774    31,486                                                                 2  Because there is no official list of BRI projects published by the Government of China, analysts have assembled  their own lists of transport. We utilize the BRI projects identified in Reed and Trubetskoy (2018). One of the criteria  used for inclusion is that the endpoints of the road or rail investments in question are cities with at least 300,000  inhabitants, consistent with the BRI vision of trade corridors that would connect major hubs throughout Eurasia.  Not surprisingly, then, the projects in the Reed and Trubetskoy (2018) data set are mostly large highways and high‐ speed rail corridors. Other analysts have chosen broader definitions concerning “BRI transportation projects” (see,  for instance, analyses in Reconnecting Asia of the Center from Strategic and International Studies (Hillman 2018) or  by Aiddata (BenYishay et al. 2016)). Those broader BRI classifications tend to capture more smaller projects.  3            Total  28,791   100%  14,993   100%            43,785   100%     Source: Reed, T. and S. Trubetskoy (2018) The Belt and Road Initiative and the Value of Urban Land. World Bank mimeo.  B.  Road & Rail Typologies  We subdivide each type of transport infrastructure following Reed and Trubetskoy (2018) to consider:  undivided roads, divided roads, conventional rail, and high‐capacity rail. The four categories are  described below with particular reference to proximities of both population centers and intact frontier  landscapes – two factors that are especially important for understanding investments’ environmental  risks.    High Capacity Rail: High‐capacity rail is dual track and/or electrified rail that transport freight and/or  passengers, including high‐speed rail (HSR). HSR are defined as high capacity and frequency railway  services achieving an average speed of over 200 km/hr (Givoni 2006). The HSR systems have  restricted geographic access because passengers and/or freight are limited to entry/exit at very few  stations. Further, those HSR stations are usually sited either in or adjacent to pre‐existing urban  centers, thus limiting new access to frontier landscapes.3     Conventional Rail: Conventional rail travel is at speeds less than 200 km/hr and is typically fueled by  diesel or other fossil fuels. Conventional rail stops at many more stations than the high‐speed rail,  thus creating many more entry points to access frontier landscapes. Smaller stations are sited in less  populated areas, potentially increasing access to intact frontier landscapes.   Divided Highway: These “high‐quality” roads, classified as four‐lane highways or larger, tend to be  sited between large urban areas, though they may pass through less developed areas. They are  restricted to high speed traffic. The divided highways often have limited or controlled access. The  degree of access depends on the frequency of off‐ramps. Access may also be restricted through  tolls.   Undivided Highway: These are “low‐quality” roads, i.e., roads which have no more than two lanes  but are serviceable for transporting freight. They  connect population centers of varied sizes and are  less likely to feature restrictions on access through ramps and/or tolls. BRI undivided highways may  also be constructed to connect to other transport hubs (for example, to larger BRI projects) such as  dryports, airports, or train stations. Because undivided highways typically have unrestricted access  for entry and exit, new roads may expand gateways into undeveloped frontiers.  There are several additional categories of minor roads – roads smaller than undivided highways – that  also have environmental risks but typically are not a focus of BRI projects. These can range from smaller  two‐lane highways to unpaved seasonal or logging roads. Minor roads could be included within the BRI  as a form of access to primary BRI projects. Additionally, unplanned spur roads that diverge from larger  roads (typically from undivided highways, due to their unrestricted access) may spontaneously result  from planned BRI projects. Spur roads might penetrate into frontier landscapes, yielding, for instance, a  grid of "fishbone development" as has been extensively documented in the Amazon (Barber et al. 2014).                                                               3  HSR is often flanked by roads for maintenance and emergencies that may provide some access between stations.  4    Part 1: Environmental Risks from BRI Investments in Transport   Transportation infrastructure can pose many risks to the environment. In this section we  distinguish direct effects – abiotic, ecosystems, wildlife – from the indirect effects induced by changes in  transport costs and land use. While the former tend to be localized and unambiguously linked to a road  or rail project, the latter often have more complex connections but potentially more pervasive impacts.  The magnitudes of various effects differ by context. As illustrated in Appendix 1 (Forest Cover Change  Across the Entire Belt and Road), deforestation – a relatively straightforward proxy for environmental  risks – has shown heterogenous responses to transportation investments, depending upon both the  ecological and development settings in which investments occur.  Cumulative risks of multiple factors  are even more challenging to assess and predict, as these factors may evolve and also may interact.   Environmental risks are typically divided between those incurred in the design and construction phases  of any infrastructure project and those experienced during the operations.  This working paper will only  review the latter because the former tend to be extremely site specific and can be confidently identified  during a project feasibility study or environmental impact assessment process (World Bank 2012).   I. DIRECT EFFECTS  A.  Abiotic Effects   i. Edaphic, Topographic, and Hydrologic Impacts  Once operational, roads modify their physical environment by alterations of hydrologic systems,  sediment erosion dynamics, and debris deposition dynamics. The extent and intensity of these abiotic  effects can vary as a result of the position of the road relative to slope, nearby waterways and drainage,  surrounding land cover, and prevailing winds (Coffin 2007; Forman 1998; Sidle, Ghestem, and Stokes  2014). Most abiotic factors have localized effects immediately around the road and roadside. These  impacts can be severe, as in the case of destructive landslides that damage nearby communities and  ecosystems. Some abiotic effects, however, permeate across a wider landscape. For example, poorly  constructed roads on mountainous terrain can generate increased sedimentation in rivers and streams,  creating long‐term effects on downstream aquatic communities and producing widespread flooding  risks (Forman 1998; Renaud, Sudmeier‐Rieux, and Estrella 2013; Sidle, Ghestem, and Stokes 2014).  Because so many of the BRI Economic Corridors pass through steep terrain, BRI transportation projects  are especially vulnerable to such risks. The consequences can be severe: Twenty‐five million inhabitants  living downslope from two proposed BRI road projects in Myanmar, for example, are potentially at risk  due to vulnerability to increased sedimentation and flooding (Helsingen et al. 2018).     There is little environmental research (beyond engineering feasibility plans) on the effects of railways on  hydrologic systems, sediment erosion, and debris deposition. However, it is expected that they will be  similar to roads, with impacts depending on topography, hydrology, and climate. It is possible that risks  related to uneven topography and steep slopes may be more severe for rail projects (especially HSRs),  which for engineering purposes are constrained to relatively straight trajectories. Less able to adjust to  the terrain, HSR may face greater complications than roads, and may require more tunnels and bridges.  5    ii. Pollution  In the context of the Belt and Road Initiative, new or upgraded highways or rail will be built primarily  with the intention of increasing freight traffic, with the secondary effect of potentially increasing  passenger traffic.  In terms of roads, it is expected that increased high‐speed freight truck traffic on new  and upgraded BRI highways will generate significant noise and air pollution including greenhouse gases  (GHGs). Water pollution will also be created from metal corrosion, tire wear, exhaust emission, trash  and other pollutants that are washed into waterways, especially during heavy rainstorms or floods when  drainage capabilities of ditches and soil are exceeded (Economic Commission for Europe 1975).  Measuring pollution requires data on volume and composition of traffic as well as the weight, speed,  and efficiency of the cars and trucks (Janic and Jovanović 2012). Assessing the increase in pollution from  the new or improved BRI is more complicated, however, as it also requires identifying induced demand  for highway travel and transport, displacement of traffic from other routes, and substitution between  different modes of transportation. These issues will be discussed below in the Indirect Effects section.    With regards to pollution from BRI rail projects, conventional rail typically has higher emissions per  passenger or per ton freight than high‐speed rail because the latter is predominantly powered by  electricity rather than diesel. HSR emissions are directly related to the sources used to generate  electricity and their energy consumption. If significant coal is used as the source of the electricity, then  HRS can contribute significantly to local air pollution around the coal plants through the production of  SO2.  However, in general both conventional and HSR contribute significantly fewer GHGs per passenger  or freight than other forms of transportation such as road, air transport, and waterways (Givoni 2006; X.  Li, Fan, and Wu 2017; National Rail 2009; Viana and Cenamo 2008). Conventional freight rail has two to  five times higher fuel efficiency than truck transport (Dorsey, Olsson, and Rew 2015), as well as uses  21% less energy per passenger mile (Federal Transit Administration 2009). In Europe along the Mega‐ TransEuropean transport corridor, the substitution from trucks to freight trains resulted in 64‐70%  savings in GHGs (Janic and Jovanović 2012). Train are even more efficient than airplanes. As compared  to air transport, HSR have lower emissions of CO2 and NOx. Moreover, NOx emissions at high altitude  affect climate change much more than ground‐level emissions, making air transport even more  damaging (Givoni 2006). As with roads, the ultimate pollution impact from a BRI rail project will depend  not only on the relative fuel efficiency and pollution controls, but also the induced traffic demand that  could increase overall pollution, which will be addressed in the Indirect Effects section below.    Noise pollution from trains (freight and passenger) has also been recognized as harmful to the health of  humans and wildlife populations. For freight trains, the longer, heavier, and faster trains generate higher  level of noise but exposures are shorter due to higher speeds (Givoni 2006; Janic and Jovanović 2012).     B. Ecosystem Effects   i. Habitat Loss  The most basic environmental impact from a road or rail is the destruction of habitat to create a  transportation corridor. The direct loss in habitat due to the footprint of roads and roadside is typically  1‐2% of the land cover of most countries, with a greater proportion near urban centers and the lowest  percentage in the less developed landscape frontier (Forman 1998). Habitat loss is closely correlated  with many accompanying environmental risks that are related to loss of ecosystem services such as  biodiversity habitats, carbon storage and sequestration, water provision and quality, soil stabilization  6    and erosion protection (A. Dobson et al. 2006; Song, Huang, and Townshend 2014). (See Appendix 3:  Environmental Indicator: Ecosystem Effects as Measured by Forest Cover Change in CICPEC.)    Roads are considered one of the leading proximate causes of habitat destruction, especially tropical  deforestation.4 For example, a meta‐analysis by Geist and Lambin (2001) of the existing literature on the  causal factors for tropical deforestation found that the extension of road infrastructure was a proximate  cause of deforestation in 61% of the cases studied (93 of 152) while rail expansion was a proximate  cause in 11% (17 or 152). Such evidence has led to the development of a relatively new discipline, “road  ecology,” which addresses solely the impacts of road construction and operations and their mitigation  (Coffin 2007; Forman 1998; Forman et al. 2003; van der Ree, Smith, and Grilo 2015). More recently,  “railway ecology” has also been proposed (Popp and Boyle 2017; Wingard et al. 2014), although the  general assumption has been that the environmental impacts from rail, while similar in nature, are less  severe in intensity to those from roads. Perhaps for this reason, the great majority of the research on  environmental impacts has focused on roads rather than railways. For example, a survey of transport‐ related impacts on wildlife found that 94% of peer‐reviewed articles were about impacts of roads, with  only 6% on rail impacts (Popp and Boyle 2017).   ii. Edge Effects  The "road effect" (also known as the “edge effect”) is the area over which the ecological effects  of a road and traffic extend into the adjacent landscape due to habitat disturbance effects (van der Ree,  Smith, and Grilo 2015). This is a more subtle but pervasive ecosystem effect of a road that extends well  beyond the initial loss in habitat from the transportation corridor.  The newly created edges along a  roadside allow penetration of light, wind, and chemical pollution and modify microclimatic conditions.  Such alterations affect the distribution and abundance of plant and animal species. Specialized “habitat‐ interior” species of plants and animals are often outcompeted by “edge‐adapted” generalist or weedy  species (Bruschi et al. 2015). The edge effect can also lead to cascading abiotic effects. For example,  because the edges are desiccated from light and wind exposure, they are especially prone to forest fires.  In the Amazon, edges are particularly vulnerable to fires initiated in adjacent burned pastures and/or  selective logged forests with their built‐up fuel load. Once the forest fire enters along the roadside,  forest fires can burn deep into the interior of the forest (Cochrane and Laurance 2002).     The penetration of the road‐effect zone into the adjacent landscape is determined by the characteristics  of the (i) road (divided versus undivided, paved versus unpaved, elevation relative to adjacent  landscape); (ii) traffic (vehicle type, volume, speed); (iii) adjacent landscape (topography, hydrography,  vegetation type, habitat quality); (iv) prevailing wind speed and direction; and (v) species traits and their  sensitivity to the impact (R. van der Ree, Smith, and Grilo 2015).  The road effect zone can be felt as far  as 1500 m from a highway (Bruschi et al. 2015). It is estimated that 15‐20% of the earth’s landcover is  influenced by this road effect (Forman 1998). Given that most BRI road projects are relatively large, we  can estimate that the “road effect” will extend on average at least a kilometer into the adjacent  roadside habitat (Benítez‐López, Alkemade, and Verweij 2010; Ibisch et al. 2016). The China‐Indochina  Peninsula Economic Corridor in tropical Southeast Asia is especially vulnerable to edge effects because  research has shown that this effect is especially pronounced in tropical ecosystems (Goosem 2015).                                                                4  It should be noted that much of the research focused on roads and deforestation in Latin America, especially the  Amazon, and that more recent rigorous research in Asia also very strongly supports that the contexts are critical.  7      Often railways are lumped into the same category as roads and other linear infrastructure for their  contribution to transforming landscapes (Bruschi et al. 2015). The magnitudes of edge effects from  railway clearings are not well understood because the field of railway ecology is still nascent (Popp and  Boyle 2017). However, there is strong reason to believe that the ecosystems effects for BRI rail projects  would be, on average, less than those from BRI highway projects. The edge effect – that is, the area over  which the ecological effects of rail extend into the adjacent landscape – is likely to be less severe for  several reasons:  First, the width of clearing is usually narrower for railways. The loss of habitat from the  rail right‐of‐way corridor – typically 15 m on either side of the tracks – is on average smaller than the  corridor destroyed for roads and roadsides (and substantially smaller than that of divided four‐lane  highways) (Dorsey, Olsson, and Rew 2015). More narrow corridors lead to less change in light exposure,  wind patterns, and microclimates, and thus less opportunity for invasion by exotic and weedy species.  Second, train traffic is less frequent than car and truck traffic, thus creating less frequent though louder  behavioral disruptions of wildlife.      iii. Habitat Fragmentation  The edge effect is exacerbated by another attribute of the road network: fragmentation of the  landscape. Roads often fragment large habitat expanses into smaller patches, leading to dramatic  landscape transformation and loss of the ability to support healthy ecosystems, populations of plants  and animals, and other ecosystem services (Bruschi et al. 2015; Ibisch et al. 2016; Potapov et al. 2017). A  review of almost four decades of fragmentation experiments around the world found that habitat  fragmentation reduces biodiversity by 13 to 75%, decreases biomass and carbon storage, and alters  nutrient cycles. These impacts were most severe in the smallest and most isolated fragments and  continued to grow in magnitude over time (Haddad et al. 2015).     The edge effect – described above – further exacerbates the impact of habitat fragmentation.  The  smallest and most irregularly‐shaped patches have the largest ratio of perimeter to volume and  consequently the greatest edge effects.     From a BRI perspective, it is important to distinguish the impact of fragmentation for different classes of  roads, since approximately half the BRI road projects are large divided highways and most of the rest are  relatively large, paved two‐lane roads. Many of these road or rail projects represent upgrades rather  than new transportation corridors, so may not be fragmenting additional habitat. By comparison, much  of the fragmentation research reported in the literature has focused on smaller roads – often built for  logging access – that abut intact frontier landscapes and areas with little prior development. Most but  not all of this research has taken place in Latin America (Chomitz and Gray 1996; Pfaff 1999; Deininger  and Minten 2002; Barber et al. 2014). The few studies of fragmentation that compare different size  classes of roads have found that the smaller roads are responsible for most of the fragmentation. For  example, a study of the impacts of the expansion of road network from 1970 to 2008 in Xishuangbanna  in Yunnan Province, China found that increases in road density for the minor roads resulted in  significantly greater levels of landscape fragmentation than for the larger roads and highways (S. Liu et  al. 2011). While this would imply that the large BRI highways do not create as much concern about  forest fragmentation, in the Indirect Effects section below we will discuss the conditions under which  8    BRI highway projects may lead to access to undeveloped landscapes and further generation of additional  roads and thus fragmentation.    It is likely that rail lines result in less habitat fragmentation due to the fact that, for engineering  purposes, rail lines are constrained to follow straighter pathways – especially for high‐speed rail – and  thus create fewer irregularly‐shaped patches. Perhaps most significantly, railways allow greater control  of access to the adjacent land by limiting the construction of new secondary roads, which drive  fragmentation (Viana and Cenamo 2008), as will be discussed in the Indirect Effects section below.      iv. Intact Frontier Landscapes  A complementary concern is that, as more of the landscape is partitioned into smaller and smaller  fragments by the road network, fewer large, undisturbed roadless tracts remain.  As these large tracts  diminish in number and size, the ecological importance of the remaining tracts increases because they  create critical refugia for many plant and animal species that cannot survive in smaller fragments. They  also provide significant ecosystem services such as carbon storage and sequestration, climate  stabilization, water provision, indigenous culture, and the maintenance of human health (Ibisch et al.  2016; Selva et al. 2015; Watson et al. 2018). These large areas of undisturbed wilderness or frontier  landscapes can be measured as “intact forest landscapes” (IFL), that is, connected mosaics of forest and  naturally treeless ecosystems with no remotely detected signs of human activity and a minimum area of  500 km2. As of 2013, ILFs represented less than 21% of once forested land cover globally. Tropical South  America and the boreal regions contained the greatest percentage of the IFLs. (See Figure 1.) In  Southeast Asia only 8% of IFLs still remain while the northern Eurasia/temperate and southern boreal  region contains just 9% (Figure 1; Potapov et al. 2017). Because of the outsized role that IFLs serve in  protecting biodiversity and generating ecosystem services, encroachment on IFL areas by transportation  projects creates straightforward risks to these environmentally critical areas.  Almost none of the BRI projects identified by Reed and Trubetskoy (2018) cross or even approach intact  frontier landscapes, except for a proposed road and a rail projects in the northern reaches of the China‐ Figure 1. Intact Frontier Landscape extent for the year 2013, IFL area reduction from  2000 to 2013, and regional boundaries. From Potapov et al. (2017)  9    Mongolia Russia Economic Corridor (CMREC) and several segments of projects within CICPEC Southeast  Asia pass. This may not be surprising, as BRI projects are designed to connect population centers.  Given  how few IFLs remain in the world, it is especially critical to safeguard the integrity of the few remaining  IFLs in those two BRI Economic Corridors.  (See Appendix 4: Environmental Indicator: Ecosystem Effects  as Measured by Intact Frontier Landscapes.) It should be noted that the Ice Silk Road in the arctic region,  however, traverses a region with a much greater proportion of large tracts within which IFLs are still  relatively extensive in northern Eurasia and the Northern boreal region (45%) (Potapov et al. 2017).   C. Wildlife Effects   i. Habitat and Behavioral Disruption  Since the 1970s, researchers have focused on effects of roads on wildlife more than any other risks.   Extensive research has been carried out concerning the risk of roads to mammals, birds, herpetofauna,  pollinators, non‐pollinating insects and other invertebrates (Ahmed et al. 2014; Coffin 2007; Popp and  Boyle 2017; R. van der Ree, Smith, and Grilo 2015). For some animal species, road‐related mortality  ranks as one of the causes, if not the most significant cause, of species endangerment (Seiler and Helldin  2006).     Wildlife impacts are mediated through a range of mechanisms, the first of which was covered above,  i.e., risks due to habitat loss, fragmentation, and the edge effects. A meta‐analysis of fragmentation  investigations shows that reduced fragment area and increased fragment isolation has significantly  reduced the abundance of mammals, birds, and insects (Haddad et al. 2015).  Understory birds in  Amazonia are particularly vulnerable to edge effects, road‐induced fragmentation, fire, selective logging,  hunting, and traffic disturbance (Ahmed et al. 2014). While populations of many habitat‐interior species  may be diminished in the roadside buffer, other animal species are attracted to the open canopy and  thick understory. These tend to be edge‐adapted species – secondary forest species with life history  traits to maximize growth rates, often referred to as “weedy species.” Many of these weedy species are  not native to the area (Bruschi et al. 2015). Because edge‐adapted animal species are generalists and  have very large ranges and high reproductive rates, they contribute little to the biological diversity levels  of the broader region. Meanwhile, the loss of species adapted for the intact habitats diminishes the  species diversity and ecosystem resilience. (See Appendix 5: Environmental Indicator: Wildlife Effects as  Measured by Biodiversity hotspots and Umbrella Species in CICPEC.)    Wildlife are vulnerable to a range of additional road‐related conditions beyond ecosystem effects.  Wildlife‐vehicle collision (WVC) is perhaps the most visible and well‐studied. Vehicles are inherently  dangerous to wildlife.  High‐volume and high‐speed traffic – which typifies BRI road projects –  contribute to the highest rates of WVC (Gunson, Mountrakis, and Quackenbush 2011; Litvaitis and Tash  2008). The situation is exacerbated for wildlife that are attracted to open flyways and paths, further  increasing their risk of WVC (Bruschi et al. 2015; Coffin 2007; Morelli et al. 2014; Myczko et al. 2017).  This phenomenon has been documented especially for predators and large herbivores and explains the  high level of collision mortality among Florida panthers (Felis concolor), Iberian lynx (Felis pardina), wolf  (Canis lupus), and brown bear (Ursus arctos) (Seiler and Helldin 2006). The decreased wildlife abundance  due to vehicle collisions has been shown in some studies to pose a greater risk than the negative effects  of habitat fragmentation (Jackson and Fahrig 2011).    10    Within railway ecology, environmental research has also focused primarily on wildlife effects. Wildlife‐ train collisions (WTC) may be similar to that from roads, but their magnitude of the impact is uncertain  (Dorsey, Olsson, and Rew 2015; Forman et al. 2003; Popp and Boyle 2017; Seiler and Helldin 2006;  Wingard et al. 2014). Where railways pass through high‐quality wildlife habitat, collisions with ungulates  and carnivores are common. Some characteristics of rail as compared to roads make them more prone  to collisions, such as dense vegetation (habitat) near tracks that obscures visibility. Railway corridors –  lacking the hard asphalt pavement of roads – are also more likely to serve as corridors for animals to  travel on.  For species attracted by the rail “corridor effect,” WTC can be especially dangerous. In area of  Alaska, for example, moose populations declined by 70% due to WTC (Dorsey, Olsson, and Rew 2015).     For most BRI road and rail projects – which tend to be large – the more serious risk is not the attraction  of open corridors but rather the barriers created by the linear infrastructure project.  Large divided  highways may serve as barriers to all species except those that are habituated to open‐air flight. These  large roads are not only wide but most have additional barriers such as medians in the middle and  fencing on either side of the road (Bruschi et al. 2015). In addition to physical obstacles, frequent and  loud traffic can create behavioral modifications that lead some wildlife species such as elk during the  hunting season to avoid areas with roads (Paton et al. 2017). At the other extreme, narrow roads with  low traffic volume and some regrowth create far less of a barrier, though even in these cases the road  can inhibit sensitive species that tend to avoid both edge‐affected habitat and the road clearing itself  (W. F. Laurance et al. 2014). Like roads, rail lines can create a barrier for some species. Rail lines,  especially HSR, often are lined by fences that create added barriers to both resident and migratory  wildlife species (Bruschi et al. 2015).     The barriers created by roads and rail endanger animal populations  by disrupting migration, splitting  populations, and thus reducing genetic variability (Sawyer et al. 2013; Wingard et al. 2014). Roads are  responsible for decreased movement and genetic variability among populations of understory birds in  the Amazon (S. G. W. Laurance, Stouffer, and Laurance 2004). For migratory and nomadic mammal  species, these barriers can create an even more serious obstacle. The Qinghai‐Tibet Railway, for  example, created a barrier to the migration of the endangered Przewalski’s Gazelle which has reduced  the genetic viability of that species (Yu et al. 2017).     Railway creates some unique hazards to wildlife not found in roads. For example, train embankments,  when high, can present a significant barrier (Wingard et al. 2014). Similarly, wildlife can be electrocuted  by the third rail of HSR. Small animals, especially turtles, can become trapped between the two rails  (Dorsey, Olsson, and Rew 2015; Seiler and Helldin 2006).     ii. Wildlife Poaching and Illegal Trade  Roads and railways not only create conduits for wildlife travel, but they also often increase  access for hunters. Roads, and in particular minor roads that penetrating frontier landscapes, can  increase both legal and illegal hunting of wildlife (Clements et al. 2014; Gray et al. 2018; W. F. Laurance  2015; Wingard et al. 2014).  This is likely a greater threat from roads than railways, due to the limited  human access points along rail lines. Train stations are built in human settlements, so rarely offer new  access for poaching in frontier landscapes.  That said, as noted earlier some HRS lines have access roads  flanking the rail line for maintenance and emergencies which could be used by poachers.    11       Transportation infrastructure – both roads and rail – poses a second and perhaps more serious threat to  wildlife trade: Improved transportation networks can facilitate the movement of illegal wildlife traffic.  This is an especially serious concern for the CICPEC  in Southeast Asia, one of the world’s most active  centers for illegal wildlife trafficking (BenYishay et al. 2016; Broussard 2017; Felbab‐Brown 2013a,  2013b). Ultimately, it is this improved transportation efficiency – ironically, a key goal of the BRI  Economic Corridors –that presents the greatest threat for some of wildlife species highly prized in East  Asian markets. (See Appendix 5: Illegal logging and wildlife trade in CICPEC.)  II. INDIRECT EFFECTS  Once road and rail projects begin operating, they set into motion additional, unintended  changes that can affect the environment even more profoundly than direct effects.  Two particularly  notable indirect effects are the changes in forest cover and GHG emissions that often result from new  frontier access and changes within markets. In this section, we will review the conditions under which  the land use and GHG change are most prevalent and the implications for BRI road and rail projects.    A. Land‐Use Change & Deforestation  Deforestation caused by land‐use changes following changes in transport costs can dramatically  exacerbate environmental risks. Road building opens up frontiers to settlement, while increasing market  access for farmers and ranchers to forests. Studies of deforestation in Brazil (A. Pfaff et al. 2007), Mexico  (G. C. Nelson and Hellerstein 1997), Belize (Chomitz and Gray 1996), and Democratic Republic of Congo  (Man Li et al. 2014) indicate some close relationships between road building and deforestation. More  generally, in fact, road building has been correlated with deforestation rates − noting that the research  has been heavily focused on the impacts within many settings in Latin America in the 1980s and 1990s  (Chomitz and Gray 1996; Cropper, Griffiths, and Mani 1999; Deininger and Minten 2002; Ferretti‐Gallon  and Busch 2014; Geist and Lambin 2002; G. C. Nelson and Hellerstein 1997; A. S. P. Pfaff 1999). Nearly  95% of all deforestation in the Amazon, occurs within 5.5 km of roads or 1 km of rivers (Barber et al.  2014). This can generate a spiraling loss of forest and increasing damage to biodiversity and ecosystems  from forest loss, habitat fragmentation, and edge effects discussed in the Direct Effects section above.  However, spatially‐explicit studies with attention to diverse contexts reveal significant heterogeneity in  the deforestation generated by roads − by the locations of roads (and perhaps also the types, although  evidence is limited). The degree to which roads lead to deforestation vary with the topography (Freitas,  Hawbaker, and Metzger 2010; Hoyos, Cabido, and Cingolani 2018) and agricultural suitability (Ruan, Qiu,  and Dyck 2016), local wealth levels (Deininger and Minten 2002; M. Li, Wu, and Deng 2013), protection  enforcement (BenYishay et al. 2016; Bhattarai, Conway, and Yousef 2009; Hargrave and Kis‐Katos 2013),  and tenure arrangements (Geist and Lambin 2001; Mena, Bilsborrow, and McClain 2006). Crucially for  forest dynamics, it has been found that prior land clearing and prior development affect roads’ impacts  (Andersen 1996; Cropper, Puri, and Griffiths 2001; A. Pfaff et al. 2018). In sum, for a number of reasons  the deforestation that a policy maker should expect due to a new road varies greatly by the setting.   i. Evidence on Road Impacts Given Prior Development and Deforestation  Chomitz (2007) delineates three main types of forest settings: forest‐agriculture mosaicland  where population densities are relatively high, markets are near, and deforestation rate is high for the  sparse remaining natural forests; frontier and disputed areas where pressures for deforestation and  12    degradation are high or increasing, and control is often insecure and in conflict; and areas beyond the  agricultural frontier where substantial forest remains with few and largely indigenous inhabitants, and  some pressure on timber resources.  These categories accord with those used by Pfaff et al. (2018) in  looking at heterogeneous impacts of new roads, distinguishing between settings with high, medium, and  low levels of prior roads and deforestation. For those settings with medium prior deforestation, new  roads are expected to lead to significant deforestation, as those areas tend to be where the frontier of  development is located and expanding. By comparison, more developed forest‐agriculture mosaiclands  may have already experienced more deforestation given that they already had relatively low transport  costs and higher economic activity. Such forest‐agriculture mosaicland areas may even experience net  reforestation as a result of new road construction if the further improvements in transport cost lead to  outmigration or a transition away from agriculture as the primary economic activity. Finally, for regions  beyond the agricultural frontier with low prior deforestation it is important to distinguish between the  short‐run and long‐run impacts. In the short run, inputs for rapid expansion of production, such as labor  and capital, are limited. Thus, deforestation responses can be minimal. However, if new roads generate  profit opportunities, inputs and labor will enter the area, increasing production—and deforestation—in  the long run. That can lead, in turn, to spatially path‐dependent dynamics in which early arrivals create  the conditions for others, including by lobbying for other infrastructure, so that the frontier expands.  The implications of this relationship between prior development and deforestation for environmental  impacts are manifold. New road access to forest with medium prior development as compared to new  access to areas with high prior development, on average, leads to greater absolute deforestation and its  consequent habitat loss, fragmentation, edge effects, and biodiversity loss. New road access to intact  frontier landscapes (areas with low prior development) could do either little or a great deal to forests,  depending on the time period being considered as well as any complementary mitigation policies such  as protected areas that could flank investments (see Part 2). If such access sets in motion a significant  economic development dynamic, the impact on forests can be highest of all. This impact is further  compounded by the fact that habitat loss and degradation to intact frontier landscapes is relatively  more harmful than to secondary or disturbed habitats because of the ecosystem functions of large  intact areas of forest, including higher levels of biodiversity, high levels of ecosystem services such as  carbon sequestration, and role as a refugia, as described above in the Direct Effects section on IFLs.    ii. Theory about Shifts in Such Risks as Economic Development Proceeds  Two well‐known theoretical frameworks – not mutually exclusive − provide “macro‐scale” rationales for  expecting heterogeneities across sites in the impacts of investments in transportation infrastructure. In  considering average impacts of new roads, given other factors, and marginal impact for each condition,  both Environmental Kuznets Curve (EKC) and Forest Transition (FT) frameworks consider evolutions of  whole economies. They are potentially relevant if BRI investments unleash significant economic shifts.  The literature considering EKCs focuses on changes in per capita income and their implications for the  change in environmental quality. As incomes rise for any given population, consumption tends to rise  over time with it, while environmental quality tends to fall due to the effects of both production and  consumption. As a consequence, eventually the marginal utility of consumption tends to diminish while,  in contrast, rising scarcity of environmental quality raises the marginal utility of preserving ecosystem  services and environmental public goods. This could motivate changes in individual behaviors and could  lead to support for public environmental regulation that would increase environmental quality even if at  13    the expense of consumption. Empirical evidence for an EKC in terms of deforestation remains mixed  (Cropper and Griffiths 1994; Foster and Rosenzweig 2003; Koop and Tole 1999). Nevertheless, the point  holds that a wealthier society might find it worthwhile to use more costly routes to lower environmental  damage or put protected areas alongside roads to lower impacts. (See Part 2 on mitigation strategies.)  More recent studies suggest that EKC patterns could be more N‐shaped, with a return to high levels of  deforestation as wealth increases demand for land‐intensive products (Joshi and Beck 2017). DeFries  and colleagues (2010) provide evidence that concords with the proposed N‐shape: in a time series from  2000‐2005 they find that urbanization and trade are associated with increased demand for agricultural  products, which leads to increased deforestation. It seems therefore that any posited EKC is likely to be  a context‐dependent depiction of the relationship between demand for different commodities and their  corresponding land uses. At low levels of income, the most important factors may be local demand for  local agricultural goods vs. local demand for forest products. At higher levels of income, a country’s  forested regions are likely to be more integrated into the urban economy and global trade networks, so  the relevant variables will include demand for exportable agricultural commodities versus the demand  for sustainably harvested timber and other commodities or amenities provided by forest cover.  This competition between land uses appears to be significantly related to trade openness and  comparative advantage. Leblois and colleagues (2017) find that trade in agricultural commodities and  trade competitiveness are associated with deforestation. However, this effect is less pronounced where  agricultural land is already extensive (Leblois, Damette, and Wolfersberger 2017). It appears that regions  where agriculture is still expanding into forest areas will experience more deforestation pressure as  development progresses, while regions whose comparative advantage lies in sustainable timber or  agroforestry production systems may see net reforestation. This pattern is borne out by the experience  of the American South, where land use has transitioned from agriculture toward forestry in the 20th  century (Carter, Kellison, and Wallinger 2015). Similarly, Kenya has seen significant secondary forest  regeneration as a result of post‐colonial demand for tree products on small‐holding farms (Holmgren,  Masakha, and Sjöholm 1994), while in Indonesia comparative advantage in annual agriculture versus in  tree plantations is significantly linked to deforestation outcomes (Baylis, Fullerton, and Shah 2016).  Forest Transition (FT) theory, which has focused upon forests in particular, posits that at least some  countries may undergo a large‐scale transition from net deforestation to net reforestation. FT theory  depicts a progression over time that starts from widespread intact forest landscape, moves through a  period of high deforestation, toward a state of stabilized forest cover and forest‐agriculture mosaics,  and eventually generates net reforestation (noting that historically this has included rising plantations).  Shifts from net forest losses to net forest gains have been observed in Europe  and the Eastern United  States (Rudel, Schneider, and Uriarte 2010), as well as in some developing nations such as Vietnam  (Meyfroidt and Lambin 2009), Costa Rica, Puerto Rico, and regions of Indonesia and India. However,  there are multiple possible mechanisms for such “forest transitions” and not all of them clearly involve  net forest gains when looking more broadly across space, for example, at the global level. Trade clearly  can allow for one country to rise in forest by simply displacing deforestation elsewhere via inputs of  timber and agricultural products (Meyfroidt, Rudel, and Lambin 2010; A. Pfaff and Walker 2010).   Another mechanism that could support a forest transition via movements within one country might be  labor reallocation from agricultural to urban areas − where forest impact per person may be lower. If  labor is scarce in an area, and the non‐agricultural economy expands, then agricultural land use could  14    decrease and forests could increase. This accords with patterns of structural transformation and could  be caused by improved transportation infrastructure (Asher and Novosad 2015; D. J. Kaczan 2016). This  can happen when the manufacturing sector expands and may continue as the services sectors expand.  Thinking at the regional level, employing a general equilibrium perspective, transport improvements  might themselves increase the pace of urbanization, facilitating migration to urban areas. This can also  be part of increasing the relative productivity of manufacturing and services in urban areas versus  agriculture, so land and population exit agricultural production in rural areas (Deng et al. 2011).  Moving to general equilibrium effects for the economy as a whole, more transportation infrastructure  could increase the rate of economic growth. Country‐wide growth could lead to increased demand for  agricultural commodities or timber, raising deforestation. Alternatively, growth could contribute to an  EKC pattern, and a possible net reforestation, since the marginal impact of increased income is not the  same at all points of economic development but instead appears to vary with food and natural scarcity.   Which of these mechanisms dominates in a particular setting and time period is an empirical question.  With more information about a setting, we can speculate about the likely impacts of various forms of  investments in transport infrastructure, including new transportation corridors and various upgrades.  iii. Empirical Examples  The empirical literature on roads and forest cover on the whole, then, provides the opportunity  to match cases to theoretical predictions. The following summaries review empirical studies by region,  then raise the question (not sufficiently explored in the empirical literature) of the type of investment.     Brazil: The Brazilian Amazon contains a significant share of the world’s primary forest and has  attracted a high degree of attention from conservationists, including due to its extensive intact frontier  landscape (Potapov et al. 2017). The literature about roads and deforestation in Latin America indicates  that road building tends to raise deforestation in such settings and higher road density if anything raises  land clearing, particularly for cattle production. As studies have suggested that deforestation impacts  differ by pre‐existing clearing patterns (Pfaff et al. 2018), and road‐linked deforestation in some areas  was lower than some expected (Chomitz and Thomas 2003), avoiding deforestation from investments  seems likely to require both targeting of particular settings and complementary mitigation policies.  India: Relative to Brazil, much of India is already characterized by forest‐agriculture mosaicland,  with virtually no remaining intact forest landscape (Potapov et al. 2017). Studies of the Prime Minister’s  Rural Road project have concluded that in this type of setting increasing the road network in rural India  has resulted, on average, in roughly zero net deforestation (Asher, Garg, and Novosad 2017; D. J. Kaczan  2016). Heterogeneity analysis suggests net reforestation for densely populated agricultural areas (D. J.  Kaczan 2016) with net deforestation for more distant and heavily forested regions. One mechanism that  seems to at least in part explain the result for densely populated areas is labor reallocation, since road  building is found to be linked to an increase in non‐agricultural employment (Asher and Novosad 2015).  China: China also has a significant rural population living in what might be characterized as  forest‐agriculture mosaicland and little intact forest landscape (Potapov et al. 2017). Deng et al. (2011)  find that road connections in rural China, controlling for economic variables at the watershed level, are  associated with net reforestation. The mechanism posited is labor migration, as roads in China tend to  allow migration into swiftly growing industrial centers of production. Another element here is the high  15    degree of economic growth. In addition to causing rural‐urban migration, a growing economy creates  more demand for commodities which may be produced at the expense of forests. General equilibrium  effects like these are more difficult to empirically verify but may have significant impact in reality.     Cross‐country Comparisons: Dasgupta and Wheeler (2016) and Danyo, Dasgupta and Wheeler  (2018) provide spatially explicit estimates of forest loss associated with road upgrading in Cameroon,  Bolivia, Myanmar and the Lao People’s Democratic Republic. Their estimates indicate a footprint  between 5 and 10 km where forest loss is likely following road improvements. But within their estimates  significant heterogeneity supports the framework presented in this report. For Lao PDR, the most  significant forest loss is expected in the northern region where low pre‐existing clearing coincides with  relatively high expected returns from agriculture. Similarly, the less‐developed northern highlands of  Cameroon experience significant forest loss associated with road improvements. Myanmar’s central and  east‐central forests and Bolivia’s Yungas region are also seriously threatened. Protected areas appear to  have a positive effect on forest cover in all countries covered by these studies, pointing toward a need  for protection in areas with both low prior clearing and high potential for agricultural profitability. This  can complement roads and rails.  iv. Application to BRI  BRI road projects are proposed for a wide range of settings, so it is not helpful to offer a single  view concerning the indirect deforestation effects from new construction and upgrading of BRI roads.   Yet several factors lead us to believe that for large fractions of potentially affected areas, the patterns  identified in India and China to date may on average be more applicable than those documented for the  Brazilian Amazon − although we hasten to emphasize that even within regions of India and China it  appears that the heterogeneity of impacts can include falling versus rising forest area. (Appendix 7:  Environmental Indicator: Indirect Effects as Measured by Forest Cover Change in CICPEC explores in more  detail indirect effect trends in Southeast Asia and their relevance to three BRI transportation routes.)   The single clearest reason for the applicability of some more than other past results for roads’ impacts is  that most BRI road projects – including even the new construction projects – are built or intended to be  built either on or very near to the existing footprints of smaller roads. Thus, they are not providing any  entirely new access to landscape frontiers even if they are improving access (perhaps to critical levels).   Most such planned investments are also connecting population centers and, as such, not creating access  to frontiers at their endpoints − though potentially creating access to intact frontier forests in the areas  between the population centers (it is within those latter types of areas that the type of infrastructure  and complementary mitigation policies are likely to matter a lot for forest outcomes). For BRI divided  highway projects with four lanes or more, access to intact forest may be restricted by exit ramps or tolls.  In such cases, a limited‐access highway that replaces a smaller, unrestricted road could actually reduce  access to frontier forests, at least in principle. Within the already more densely developed areas, effects  from increased commerce and expansion of new manufacturing or services, urbanization, changes in  land prices, and other equilibrium effects are likely to dominate.  Finally, while most BRI transportation  projects have not been sited or planned near to Intact Frontier Landscapes – remote and undeveloped  wilderness areas would be disturbed by any deforestation from new road construction – a few planned  projects already have been. (See map of IFLs in Figure 1 and the example of a Thai project in Appendix 4:  Environmental Indicator: Ecosystem Effects as Measured by Intact Frontier Landscapes.)   16    v. Roads versus Rails        Most BRI projects are rail not road projects, yet the literature on indirect effects from rails on  land use is far less extensive. Donaldson (2010) finds that rail connections in India decrease trade costs  and price gaps between regions, integrating markets and transmitting prices across the sub‐continent. In  the US context, Vessali (1996) finds that rapid transit infrastructure tends to increase urban density near  stations, if urban zoning regulation is aligned with this outcome. Thus, it appears that rail investments  exhibit similar economic impacts as do roads, creating opportunities for the expansion of agricultural  commodities while also potentially increasing urbanization and outmigration (de Soyres et al. 2018).  Yet one might nonetheless conceptualize a continuum in terms of the degree of local access to transport  networks and consequent local environmental risks. Rail provides access to freight and passengers only  at stations. High‐speed rail lines service fewer stations than conventional railways. As such, rail allows  access at fewer points and thus they may open less access to frontiers. This has led some to advocate for  rail links as an alternative to roads in sensitive areas like the Amazon (Viana and Cenamo 2008). Risks  could still be high if stations are near forest frontiers or serve as new nodes for transportation network  growth which generates expansion into forest frontier areas. Moreover, many high‐speed rail lines are  often accompanied by minor access roads near the tracks to facilitate maintenance and emergency  repairs. Thus, BRI rail investments are likely to allow some access to frontier forests, even if perhaps in  general that is less than is created by roads. Limited‐access highways similarly provide access only at  designated exits, while the conventional road infrastructure might allow access at any point along its  length. Thus, it might be expected that both railroads and limited‐access highways cause less indirect  environmental risk on average than do highways which provide greater access.  Moreover, one could  envision the possibility that such investments in rail systems or limited‐access divided highways could  potentially reduce pressure on habitats if they draw traffic from smaller roads that offer greater access.   B. Carbon and Other GHG Emissions   The potential for indirect effects of land‐use change and deforestation from BRI road and rail  construction, as described above, could not only profoundly affect forest cover and ecosystem health  but also generate a significant impact on the global climate.  Deforestation, especially in the tropics,  represents a significant source of carbon emission, to the point of contributing close to one‐fifth of all  anthropogenically‐produced emissions (Harris et al. 2012). Deforestation within tropical IFLs could be  especially harmful for their impact on carbon emissions from dense forests (Potapov et al. 2017).     BRI projects can have indirect effects on GHG emissions through the secondary impact via shifts in trade  and resultant changes in vehicle traffic and land use (Maliszewska and van der Mensbrugghe in prep.).  An assumption has been that fuel‐efficient modes of transportation should reduce energy consumption  and GHG emissions, as above. High‐speed rail has higher energy efficiency and lower GHG emissions on  a per passenger or per ton freight basis than conventional rail, which has higher efficiency than freight  trucks and cars, which in turn have higher efficiency than air transport. However, whether the upgrade  to a more efficient mode translates into an absolute decrease in GHGs or air pollution emitted depends  on several factors including substitution (how many passengers or freight shift from air transport, cars,  or conventional rail) and traffic generation (how much new demand for transport is generated by the  construction). Emissions saved also depends on how freed capacity (on the road and runway) is used. If  newly available runway capacity is used for more energy‐intensive long‐haul flights, for instance, then  mode substitution could raise environmental impact (Givoni 2006). The results vary depending on the  17    setting and the competing modes of transportation as well as uncertainty in future transport demand,  patterns of urbanization, technology, and sources of energy production (e.g. Chai et al. 2016; National  Rail 2009; Westin and Kågeson 2012). For example, when comparing HSR and air transport using a  duopoly model, D'Alfonso and colleagues (2016) found HSR are more energy efficient per passenger/km  but because of the induced demand, the new railway system tends to increase energy consumption and  GHGs when competing with air transport. Similarly, modeling by Westin and Kågeson (2012) found that  substitution would have to be very large, and mostly from air transport, to balance GHG emissions from  new HRS construction and increased traffic.  A general equilibrium study by Chen and colleagues (2016)  of high‐speed rail in China used data from the last 15 years of HRS construction. The authors found that  HSR construction has resulted in increased demand for transport services, thereby resulting in more  GHG emissions than were displaced. Studies like these indicate that induced demand is an important  element of transportation infrastructure’s impact that may affect land use and deforestation as well.     Part 2: Mitigation of Environmental Risks from BRI Investments in  Transportation Infrastructure  While environmental risks generated by BRI road and rail projects can be significant and extensive, a  range of choices exist that could lower both the chances and the magnitudes of such potential harms.  Within Part 2, we present these activities in terms of the “mitigation hierarchy,” a sensible framework to  organize the consideration of ways to limit varied environmental risks from BRI road and rail projects to  biodiversity and ecosystem services. Such a hierarchy is commonly comprised of four types of actions –  avoid risks, reduce risks, restore, and offset damages – that often are used in combination (Figure 1).    For the BRI, such activities could balance  economic gains from transport investment  with gains from sustaining natural capital.  Both support human well‐being locally and  around the globe. Mitigation activities can  help to reduce the direct and the indirect  effects from BRI projects (described in Part  1). Such activities can help meet standards  or guidelines established by multilateral  development banks, Chinese banks, industry  associations, and various other financial  stakeholders. They can also help to meet  Figure 2. The Mitigation Hierarchy.   environmental policies and regulations set  www.thebiodiversityconsultancy.com/approaches/mitigation‐hierarchy/   by host country governments and by  Chinese government agencies overseeing BRI and other overseas investments. Environmental policies  including laws, regulations, standards, and guidelines will be discussed directly in Part 3. Taken together,  mitigation activities can reduce the risks, costs, and delays of BRI projects and improve stakeholder  support in host countries, in China, and internationally (Ekstrom, Bennun, and Mitchell 2015).   Appropriate mitigation actions for any BRI project may be identified through environmental assessment  processes if they are applied at sufficiently early planning stages and at sufficiently broad spatial scales.  For the most important mitigation step – avoidance – strategic environmental assessment (SEA) from  18    the very start of planning can be a critically useful tool. SEA are typically a distinct regional or sectoral  assessment conducted by government agencies at the stage of prioritization within plans, programs, or  policies prior to the development of any individual project (UN Environment 2018a). Environmental  assessments that are conducted at early stages in the planning process and with a broad spatial scope –  for example, across an entire “BRI Economic Corridor” – can help to avoid substantial environmental risk  and, further, allow the coordination of mitigation choices to broaden project benefits. This allows all  decision makers the most possible flexibility in placing new roads and rails where they are likely to cause  the least environmental and social harm, while maintaining desired economic and connectivity benefits.   For other steps within the mitigation hierarchy, well‐established environmental impact assessment (EIA)  processes can usually guide individual project planning and implementation, especially for the localized  direct effects generated during construction phases. Such mitigation strategies might include reducing  impact through altering project design, restoring impacted areas, or developing offsets to compensate  for unavoidable impacts and damages. Both SEAs and EIAs will be discussed in greater detail in Part 3.    Table 2. Potential mitigation actions for BRI transportation Infrastructure investments using mitigation hierarchy  Avoid  Integrated planning of route choice to avoid vulnerable environments and maximize net gains:[1]   Identify alternative route options that can simply avoid the environmentally and socially  most sensitive areas, i.e., where there are clearly high environmental or social damages;   Conduct an integrated cost‐benefit analysis (considering economic, environmental, and  social impacts) to guide selection between alternative routes, given that all of the benefits  and costs are heterogeneous across routes and some routes may be better on many counts.   This should be done at the micro level, for BRI corridors, and the BRI as a whole, to account  for the interdependency of locations, investment impacts, and affected environments.    Reduce  Mitigate impacts through environmentally‐conscious engineering and complementary policy:   Choices of techniques: Wildlife crossings (bridges and underpasses, with mechanisms to  ‘funnel’ wildlife to crossing), sound barriers, lights downwards to reduce light pollution,  retention of trees, timing construction to avoid important times for animal migration or  mating; tunnel‐bridge‐tunnel engineering in order to reduce landslide and erosion risks.   Choices of complementary policies: Regulations, their enforcement, and incentives to  reduce deforestation, poaching, and vulnerable species trade. This may include the creation  of protected areas (PAs) nearby to transport investments, addressing weaknesses in  enforcement capacity, and/or incentive/compensatory payments to landlords or local  governments in return for maintaining forests and ecosystems. PAs  should be coordinated  along the BRI, to ensure these reduce, rather than displace, harmful activity.   Apply integrated cost‐benefit analyses in selecting across transport options (road  categories, rail versus roads, electric versus standard rail) as well as regulation on vehicle  emissions and maintenance, etc. These considerations raise the case for (particularly high‐ speed electric) rail over roads – due to lower pollution, and reduced encroachment on  frontier landscapes due to fewer access points and their concentration in already dense  cities.                                                               [1]  Such as Intact Frontier Landscapes (IFLs), biodiversity endemism hotspots, protected areas, forests liable to  deforestation, landscapes with topographical or earthquake hazards, and other vulnerable landscapes.  19    Restore  Take remedial action to repair damage inflicted by the construction process: Stabilize damaged  slopes; replant vegetation; and repair waterways or wetlands disrupted by new investments.  Offset  Compensate for environmental damage that cannot be avoided, reduced, or restored by investing in  off‐site locations to achieve overall net neutral or net positive environmental outcomes:   carbon offsetting, or, following harm to biodiverse areas, the enhancement of alternative  comparable sites elsewhere (e.g. with similar endemic species or ecosystem functions).    mechanisms for achieving such impacts elsewhere could include protected areas (that  target pressure), “Payment for Ecosystem Services” (PES), biodiversity compensation funds  into which projects pay, biodiversity banks selling off‐setting credits, and more ad hoc  project‐by‐project solutions, all supported by national or local off‐setting laws.  Off‐setting is proposed, however, only as a ’last resort’, because at least in light of past evidence it  appears to suffer both efficiency (impact) and equity (distribution) challenges.  For the rest of Part 2 we describe in detail the four categories of the mitigation hierarchy in relation to  BRI transportation infrastructure. Table 2 describes examples of mitigation actions and Appendices 7  and 8 describe how the mitigation hierarchy framework can be used to develop a suite of mitigation  actions for the CICPEC in Southeast Asia.   I. AVOID: Selection of Projects and Routes    Avoidance is generally considered the most effective way of reducing potential negative impacts from  transportation investments (ICMM and IUCN 2013; McKenney and Wilkinson 2015; Pilgrim and Ekstrom  2014). Avoiding the risks to the environment from BRI transportation projects could entail relocating or  rethinking a project or prioritizing other projects when resources are limited. Such spatial planning and  prioritization across all possible BRI transportation projects (not just single projects) would ideally be  part of an early‐stage SEA‐type regional assessment that would consider entire or large sections of BRI  Economic Corridors. It is important to conduct such planning during the early stages of the planning  process when cost‐effective avoidance options are still available and  feasible. As a proposed project  moves toward approval it is more likely to just proceed as initially conceptualized, regardless of the  gains from alternative options (Ekstrom, Bennun, and Mitchell 2015; UN Environment 2018b). It is   important to conduct such early planning over a wide region, with all potential projects and locations  considered, so that individual projects are not simply displaced to other locations with similar risks.  Project planning for infrastructure investments often identifies environmentally sensitive areas that  should be avoided. At the level of an individual project, a feasibility study is conducted early in the  process to identify disqualifying risks such as high vulnerability to landslides, flooding, hydrologic  disruptions, or other risks that cannot be reasonably minimized and managed. Sensitive areas for  biodiversity and ecosystem services should also be addressed in project feasibility studies, though that is  not always the case (W. F. Laurance 2015). Protected areas (PAs) and large expanses of undeveloped  wilderness are often used to designate the environmentally sensitive areas that should be avoided.   As described in the Part 1 and in Appendix 4, PAs and intact frontier landscapes (IFLs) typically function  as important refugia for biodiversity and safeguard many ecosystem services. To be effective, though,  these protected areas should be established as “off limits” or avoidance areas for environmental  20    objectives early within a planning process (Bruschi et al 2015; Mahmoud et al. 2017).5 Appendix 8  (Mitigation Strategies: Protected Areas) illustrates the opportunities that might be available within  CICPEC for locating new protected areas or increasing the enforcement of existing PAs within relatively  undisturbed forested areas that are targeted for planned BRI highway projects. Ensuring adequate  enforcement is an important component of this conservation strategy, as is illustrated by the  encroachment in Snuol Wildlife Reserve, Cambodia, where enforcement was lacking (Appendix 8).    It is relatively easy to use PAs and IFLs for planning purposes – again leaving implementation as critical  for achieving actual impacts – because they are relatively easy to designate spatially. However, some of  the areas at risk that contain some of the richest, most unique and threatened biodiversity or ecosystem  services are outside protected areas.  These are particularly important to avoid when siting roads and  railways because of their lack of protected status.  Appendix 5 (Environmental Indicator: Wildlife Effects  as Measured by Biodiversity Hotspots and Umbrella Species) provides an example of hotspots for  biodiversity which considers species richness, endemism, and vulnerability within CICPEC. Identifying  and mapping such hotspots for biodiversity, water provision, carbon storage and sequestration, and  other ecosystem services is more challenging than mapping PAs and IFLs but attention to these natural  assets early in the planning process can preclude costly actions at a later stage of project development.    Many examples exist in which geospatial modeling and data analysis have been used to assist with the  mapping of avoidance areas for linear infrastructure. Most focus on road networks, such as studies of  alternative routes in Myanmar, Nigeria, and Tanzania (Caro et al. 2014; Dobson et al. 2010; Fyumagwa  et al. 2013; Helsingen et al. 2018; Mahmoud et al. 2017; but see Fyumagwa et al. 2013), showing how  route choice can avoid damage to protected areas, ecosystem services, and wildlife migration patterns  while still generating high socioeconomic benefits by connecting agricultural centers with markets.  Dasgupta and Wheeler (2016) and Danyo, Dasgupta and Wheeler (2018) spatially explicitly estimate the  losses that could be avoided within upgrading of roads in Cameroon, Bolivia, Myanmar and Lao PDR.     Similar balancing of objectives could be done for rail projects to optimize environmental, engineering,  economic, and other factors. For example, Dong and colleagues (2018) developed an integrated risk  evaluation model using ecological, social, and economic data from the high‐speed railway proposed for  the China‐Mongolia‐Russia Economic Corridor. Using data on biodiversity, forest cover, protected areas,  proneness to fire and earthquake disasters, the authors identified regions within Heilongjiang, Republic  of Buryatia, Irkutsk Oblast, Zabaykalsky Krai, and the Lake Baikal area that face high ecological risk from  HSR. To avoid these ecologically sensitive areas yet allow social and economic benefits, they proposed   alternate routes and policies to mitigate construction risks. In Kenya, public attention to environmental  risks from the Standard Gauge Railway led to the modeling and mapping of risks and alternative routes  between Nairobi and both Mombasa and Malaba. The goals of mapping and modeling included avoiding  wildlife‐train collisions, pollution, barriers to wildebeest migration and more (Ambani 2017).  Appendix 9  (Mitigation Strategies: Using Umbrella Species for Mitigation Planning in CICPC) provides some examples  of mapping and planning in the China‐Indochina Peninsula Corridor to identify key areas for mitigation.  China has experiences with environmental safeguards within the development of its rail network − over  30 years − working in close coordination with international institutions such as the World Bank (World  Bank Technical Assistance Program 2009). China routinely does national and project‐level railway EIAs,                                                               5  Establishing PAs or IFLs as off limits to road and rail development can be politically challenging because, in some  cases, the very fact that they have no preexisting development could make them attractive for linear infrastructure  since a government would not have to purchase or expropriate private land and/or resettle landowners.  21    which include analyses of alternative routes to avoid environmentally sensitive sites. For example, in its  planning of the recently completed HSR Guiyang‐Guangzhou Railway Project, the Chinese government  identified 47 environmentally sensitive sites that included protected areas, forest parks, scenic areas,  watersheds, and cultural relics sites. In the project feasibility study, planners created many alternative  routes and the route selected avoided 40 of those sensitive areas (Wang, Yang, and Quintero 2012).  II.  REDUCE: After Projects/Routes Selected    If avoiding risks is impossible, then reducing6 environmental risks should be considered, as suggested by  Ekstrom and colleagues (2015): “Measures taken to reduce the duration, intensity, significance and / or  extent of impacts (including direct, indirect and cumulative impacts, as appropriate) that cannot be  completely avoided, as far as is practically feasible.” Reductions in the environmental risks of a project  below thresholds of harm may be possible. If not, actions may reduce the scale and the expense of the  remediative measures, such as restoration or offsets, that may be required later for unmitigated loss.     Risk reduction via project and route selection has been introduced above, within the Avoidance section.  For some infrastructure projects, such as the HSR Guiyang‐Guangzhou Railway Project, China, described  above, there can be an overlap between decisions to fully avoid impacts versus partially avoid impacts,  i.e., reduce impacts. In that example, railway planning completely avoided 40 sensitive sites yet broadly  it would also be right to claim risk reduction by 85% (40 of the 47 sensitive sites). For this case, actions  could be considered “avoidance” or “reduction” but we will not repeat the avoidance discussion here.     Instead, for the rest of this section on reducing risks, we will consider several other reduction decisions  for BRI transportation projects. First is the selection of techniques during infrastructure construction and  operations. Second is the use of complementary policies that could change the impacts of investments.    Just as for avoidance, reductions should be proposed during early project feasibility studies (additional  options will likely be identified through EIAs). There are many strategies to reduce direct environmental  impact during construction and, to a lesser extent, operations of roads and railways. Many concentrate  on adjusting the design of the infrastructure, but others focus on location and timing. Here we list just a  few examples that address reducing the impacts from a range of different environmental direct effects.  An exhaustive review of strategies to reduce impacts is found in van der Ree et al. (2015). Prescriptive  good engineering design strategies are also detailed in voluntary sustainable infrastructure standards  developed by organizations such as SuRe®, CEEQUAL, Envision®, and Greenroads® (see Part 3 and  Appendix 15).  A.  Local Strategies for Reducing Direct Effects: Good Engineering Design  Strategies to reduce local direct effects such as abiotic and wildlife impacts from investments in  transport often have been focused upon changes to the design and the construction of infrastructure.  Good engineering design solutions at both the project and landscape level are among the most frequent  strategies to address hydrologic, sedimentation, and erosion dynamics. For example, tunnel‐bridge‐ tunnel engineering schemes can be instituted to avoid dangerous landslides or erosion along steep                                                               6  The term “minimize” also is often used alongside or instead of “reduce,” for the second element of this mitigation hierarchy.  As it is hard to minimize more than one objective, and as truly minimizing means avoiding, we use reduce as our default.  22    terrain (Wang et al. 2012). Similarly, engineers can install wildlife crossing structures such as bridges and  underpasses  to reduce wildlife collisions and barriers to migration while increasing habitat connectivity  (Iuell et al. 2003; Litvaitis and Tash 2008). Underpasses include tunnels, culverts, and viaducts beneath  raised roadways while overpasses can range from small canopy rope crossings for arboreal primates to  large vegetated landscape bridges up to 50m wide (Iuell et al. 2003; van der Ree, Smith, and Grilo 2015).  Such structures must be coordinated with the funneling of animals to the appropriate crossing points.  The most common funneling technique is a fence (Iuell et al. 2003; R. van der Ree, Gagnoy, and Smith  2015; D. J. Smith, van der Ree, and Rosell 2015) preventing wildlife from entering a road or rail line  anywhere other than at safe crossing points. This is especially important for reducing wildlife access to  “roadkill hot spots” along roads or rails – where some physical, biological, or topographical factor  increases the likelihood of wildlife collisions (Litvaitis and Tash 2008).  A range of design options using behavioral modifications exist to reduce negative impacts of transport  corridors. Focusing artificial lighting downward reduces bird attraction (Blackwell, DeVault, and Seamens  2015) and constructing sound barriers reduces stress or interference with animal communication (Parris  2015). Retention of trees retains habitat features (Soanes and van der Ree 2015; Weller 2015), while  specialized noise and light spectra can be used to deter certain targeted wildlife species from a road or  railway to avoid collisions (D’Angelo and van der Ree 2015). Recordings of natural animal warning calls  triggered by oncoming trains have been shown to elicit faster reaction times and increased escape by  certain wildlife species (Babińska‐Werka et al. 2015). In contrast, commonly used roadside reflectors  have had limited success in deterring wildlife from entering roads (Angelo et al. 2006; Iuell et al. 2003).     Timing of construction can also help reduce impacts. For instance, Construction could be scheduled to  avoid times of the year when certain species are especially mobile. Amphibians are especially vulnerable  when they move across the landscape during their breeding season and birds when they migrate for  wintering and nesting (Andrews, Langen, and Struijk 2015; Hamer, Langton, and Lesbarreres 2015).  Wildlife‐vehicle collision rates have been shown to be higher during migrations (Litvaitis and Tash 2008).  Timing construction or traffic to avoid times of high wildlife migration can reduce collision incidence.    Some direct impacts result from human behavior and are best reduced through environmental policies,  regulations, and enforcement. For example, roads and rail provide access for illegal logging and wildlife  poaching and provide transportation for illegal trafficking of this contraband. Environmental policies and  enforcement to address these activities can be implemented in tandem with the BRI investment.  Environmental policies may be more effective influencing impacts from some transportation modes  versus others. For instance, airborne particulate pollution and GHG emissions from increased freight  traffic are relatively easy to reduce in rail projects – via high efficiency engines, efficient scheduling, and  appropriate maintenance which limits emissions – while emissions from road projects are less easily  regulated because of the dispersed nature of related pollution sources. EIAs typically call for national  measures regarding, for instance, regulations on fuel type and purification of exhaust gases, vehicle  emission regulations and standards, and the local enforcement of laws concerning vehicle condition.    B.  Broader Strategies for Reducing Indirect Effects: Complementary Policy  Reducing indirect impacts from transportation projects – such as induced land use changes – is less  straight forward than the infrastructure design modification strategies employed to reduce direct  impacts. Complementary policies are the preferred strategy for modifying human decision making.   23    i.   Complementary Policies Interact with Transportation Investments  Much as the heterogeneity of impacts from investments in transportation infrastructure already leads to  gains from route choice, the targeting of policies also is raised by interactions between types of policies  and, in particular, interactions between development investments in transport and conservation policy.  The optimal policy blend can generate a ‘win‐win’ outcome from integrated early planning as it can raise  the conservation impacts of conservation policy resources, while lowering the environmental risks from  development investments and still preserving their raison d’être, i.e., benefits of economic connectivity.   Consider that protected areas or payments for ecosystem services often have not effectively addressed  environmental threats and thus have low impact (see the literature discussion within IV. Offsets below).  In both cases, the problem is that implementation avoids threats. For protected areas, this is common  when economic interests deflect intended protection, so that protected areas end up far from pressure,  where even perfect enforcement has zero impact. For payment for ecosystem services, this occurs when  private actors are paid to maintain forest on unprofitable lands they would have kept in forest anyway.  In contrast to those scenarios, with good integrated early planning public actors may consciously place  investments in transport and conservation near each other, e.g., protected areas near to roads or rails.  For the longer run, that has impact even if the road or rail investments are traversing relatively pristine  areas (e.g., between distant big cities) not previously or currently facing high threat. The future clearly  could hold higher threats, as investments create profitable opportunities that lure labor and capital as  well as public investments as in schools or health.  As such, previously low‐pressure frontiers may have  high environmental value which could be vulnerable to impact after a rail or road line is established.   ii. Public and Private Actions, Interactions, and Further Responses  While any such policy interactions (above with explicit joint targeting) always involve some public roles,  here we also want to highlight the value of considering possible private responses to public policies.  Those private reactions may also, in turn, affect the optimal follow‐on public responses. Generally, more  such interactions again suggest that there are gains from integration of development and conservation.   The logic above concerning protected areas, for instance, could be applied to payments to suggest siting  payments for ecosystem services alongside investments in roads or rails. However, for local landowners  living alongside the proposed BRI road or track to voluntarily forgo cutting down forests once a new rail  or road is in place, we must ask whether there are sufficient private incentives to join any such program.  For instance, lands that might have been readily offered for conservation even for a low payment before  a road or rail investment occurs may no longer be volunteered after the investment (as is suggested by  the past purchasing of lands as speculative investments when possible plans for roads are made public).  Yet, nonetheless, there might exist productive combinations of transport investments and protection  and payments, a triple integration, if the placing of protected areas immediately around new transport  lowers the economic expectations for those parcels nearby which are private and not being protected.  We might similarly consider private incentives if protection is not strict but, instead, permits some use –  as in extractive reserves that explicitly allow some smallholder production (and resulting forest losses).  They clearly change the political economy by making it easier to have both protection and development.  That can make protection more politically feasible. It also could generate local incentive for monitoring.  Thus, we may consider what types of conservation policies are best for interactions with roads and rail.  Further, in principle much the same could be said for having private rights to logging within concessions.  24    While in cases such rights have led to rates of forest losses higher than in the background for the region,  rights create incentive for actors to defend forest assets – much like smallholders in extractive reserves.  For concessions, actors may be firms, who may have incentives and the resources to stop illegal logging.  Generally, combinations of development and conservation tools could interact usefully on the frontier.  Sometimes, however, development and conservation would not appear to work well near to each other.  For instance, if a strong federal government environmental agency establishes a strict PA nearby a road  or rail, it may well function as a signal that, in the future, more development investment is unlikely in  this area. Such conservation signals – perhaps federal, though local enforcement is very important – can  encourage net private outmigration decisions. In turn, those could discourage any further public actions.  For instance, when in‐migration slows, less payoff may be perceived from maintaining older small roads.  Thus, public actions can, via private responses, lead to complementary public actions (Herrera 2015) and  in this scenario, the public‐private‐public response dynamic serves to improve the local forest outcomes.  In other settings, local public actors such as state or municipal governments are focused more upon the  local economic gains from policies than the environmental outcomes. That can remove environmental  complementarity across public actions. Federal actions might constrain local agencies’ choices − which  effectively maximize local interests subject to federal policies. For instance, when a federal “blacklist” in  the Brazilian Amazon constrained high‐deforestation municipalities, more than one local program with  international funding (through The Amazon Fund, e.g.) did not achieve stated objectives to lower local  deforestation. That is unsurprising if local programs were essentially to help the locality in question to  get off of the costly federal blacklist, i.e., effectively the local programs were aiming simply to achieve  the federal requirements yet at a lower local cost to the economy  (Correa et al. 2018; Sills et al. 2015).  Stepping back, though, even that can help integrated planning by increasing local political feasibility.  III. RESTORE: Activities to Neutralize Net Transport Impacts Locally    Restoration is conducted at project sites to repair direct or indirect impacts to biodiversity or ecosystem  services. Restoration is only for when avoidance and reduction are not feasible (Ekstrom, Bennun, and  Mitchell 2015). Restoration for BRI road and rail projects would primarily occur during the construction  phase and actions would likely be identified during the initial feasibility study with restoration strategies  outlined in the EIA. If extensive landscape transformation occurs, for example, through digging tunnels,  stabilizing slopes, or building embankments, then vegetative restoration will likely be required.   Restoration may also be needed if roads or rail construction disrupt wetlands and waterways.   IV. OFFSET: Compensation Elsewhere for Unavoidable Impacts    Offsets have been defined as: “Measurable conservation outcomes, resulting from actions applied to  areas not impacted by the project, that compensate for significant, adverse project impacts that cannot  be avoided, minimized and/or rehabilitated/restored” (Ekstrom, Bennun, and Mitchell 2015). Offsets are  a last resort, after all other steps in the mitigation hierarchy. They involve restoration, rehabilitation, or  protection of biodiversity, carbon, water, or other ecosystem services comparable to the project losses.       Offsets may aim for no net loss. Recently, some net gain has been promoted to address global declines  in biodiversity and ecosystem services. In project finance, the International Finance Corporation (IFC)  specifies the need for a net gain in biodiversity in critical habitat as a requirement for lending (IFC 2012).  The new World Bank framework notes a preference for a net gain of biodiversity as well (The World  25    Bank 2017). Over 75 major financial institutions have expressed the same preferences via adoption of  The Equator Principles for project finance of over US$10 million (Rainey et al. 2015).    Yet there are good reasons to consider offsets a last resort. They can be complex and expensive and yet  of uncertain benefit (Bull et al. 2013; IUCN 2016; Pilgrim and Ekstrom 2014). Some argue that offsets  give developers a free pass to both destroy and degrade biodiversity (Ledec, Campos, and Reay 2016)  and, further, that implementation challenges may in fact prevent effective net mitigation. Another   significant issue is compensation of those who suffer loss. Ecosystem services such as watershed  services are local and thus offsets may be hard pressed to “make whole” the affected communities if  gains are being realized in other locations. Project losses followed by offset gains could involve trade‐ offs where some services are reduced then others are enhanced (e.g., by soil stabilization after timber  production or the relocation of water access) and, as a result, some portions of a community could gain  on net while others could still lose out on net.    Below we consider a number of approaches to generating offsets which could, in principle, balance out  the net losses that may occur as the result of a transportation infrastructure project – once the rail or  road project has been selected and routed and its damages reduced and restored as much as possible.     A.  Biodiversity and Habitat Offsets   When a new or upgraded rail or road project results in the destruction of or damages to  biodiversity, offsetting can be accomplished by restoring or enhancing comparable biodiversity  elsewhere. An offset should “achieve no net loss and preferably a net gain of biodiversity on the ground  with respect to species composition, habitat structure, ecosystem function and people’s use and cultural  values associated with biodiversity” (BBOP 2009). Biodiversity offsets have taken the form of new  protected areas, restoration of degraded or destroyed habitat, or even increased enforcement of  existing protected areas (G. Bennett et al. 2017; Ledec, Campos, and Reay 2016). Biodiversity and  habitat offsets may focus on particular species (e.g., endangered or threatened) or habitat types (e.g.,  those hosting particular species communities) or functions (e.g., wetlands/streams).   Biodiversity offsets are relatively unproven, thus it will be important to design such programs well to  ensure impact and monitor their performance. There is no best way to design a biodiversity offset  because they are so dependent on context, yet there are best practices (IUCN and ICMM 2013). For  instance, two interrelated elements that many believe will enhance success in offsets are: planning for  offsets within the landscape context (BBOP 2012; McKenney and Wilkinson 2015); and setting up  aggregated biodiversity offsets (Ekstrom, Bennun, and Mitchell 2015; ICMM and IUCN 2013; Ledec,  Campos, and Reay 2016). These principles suggest that a BRI Economic Corridor may allow for planning  at the appropriate scale, with broad and early stage environmental assessments the appropriate tool to  plan effective offsets. Aggregated offsets can reduce transaction costs and consider interactions  between projects to enhance benefits like landscape connectivity (Ekstrom, Bennun, and Mitchell 2015;  Ledec, Campos, and Reay 2016) (Pilgrim and Ekstrom 2014; BBOP 2009). One could envision a roadmap  for an offsets corridor, even if the BRI network passes through multiple countries. Few national  examples exist, however. Such roadmaps can be seen for Liberia (The World Bank Group 2015),  Mozambique (Bechtel and Nazerali 2016), and Mongolia (TNC 2016) though these are new enough that  their success cannot be determined. A major blockage seems to be the seed funding required (Ekstrom,  Bennun, and Mitchell 2015).  26    Financing institutions such as the World Bank and IFC have mandatory biodiversity offset polices. These  policies have been a “small but significant driver in offset demand.” As of 2017, 9 IFC‐financed projects  created biodiversity offsets, with 12 more projects planning to implement offsets in the future (G.  Bennett et al. 2017). If the World Bank or IFC were to get more involved in co‐financing BRI projects,  their biodiversity offsetting standards would be applied to any project to which they were lending.    Offsets can be only as effective as their design, implementation, and actual enforcement allows. Clear  and detailed policy guidance is required. A lack of oversight is a primary reason offsets fail to achieve  their goals (Bull et al. 2013; Pilgrim and Ekstrom 2014). Such effective guidance and oversight could, in  principle, be provided by host countries or by BRI oversight organizations in the Chinese government.  However, to the extent global finance is involved, there may be gains from including outside parties.  It is hard to generalize the success of mitigation for biodiversity and habitat because programs measure  success using different metrics, which make comparisons or the analysis of trends difficult. Analyses of  large, standardized programs like the US stream and wetland mitigation program have found that only  30% of offsets are meeting full project objectives (Matthews and Endress 2008; n =76 sites) and 74% of  offsets are achieving no net loss (Brown and Lant 1999; n= 68 banks). With this in mind, it is clear that  there is high uncertainty of the success of biodiversity offsets. Many projects simply define success as  establishing a protected area; as discussed elsewhere in Part 2, that has a wide variety of outcomes.  B.  Offset Instruments: Protected Areas (PAs)  The most common conservation policy in terms of area affected is the establishment and  maintenance of protected areas (PAs). They contain over 13% of the world’s lands and, in principle,  could generate offsets by preserving forest cover, biodiversity, and ecosystem services. While PAs tend  to prevent forest loss on average, however, their impact varies substantially by location (Andam et al.  2008; L. N. Joppa and Pfaff 2011; Miteva, Murray, and Pattanayak 2015; A. Pfaff et al. 2009; A. Pfaff,  Robalino, Herrera, et al. 2015). Further, a PA’s impact often is less than it seems, since sites are biased  towards low‐pressure locations.7 This matters in terms of appropriate quantitative estimates of offsets.  The offsets a PA can generate are limited by the deforestation that would have occurred without the PA.  Thus, even a perfectly implemented and enforced PA simply would not avoid deforestation if it is  located in an isolated area where no deforestation was expected to occur. Generalizing, many factors  affect the net benefits from non‐PA land uses, i.e., a PA’s opportunity costs (foregone profits), which  creates high variation in the counterfactuals or baseline land uses that the PA could block when creating  any offset.   Often the most beneficial non‐PA use of a frontier parcel is agricultural use, in which case the lands with  steeper slopes, poorer soils, and sites farther from the important markets have lower opportunity costs.                                                               7  For low‐pressure locations, it is easy to overestimate PA impact. For instance, if comparing PAs with randomly chosen controls  or average unprotected land, low‐pressure PAs are being compared to higher pressure. Andam et al. (2008) find for some Costa  Rican PAs, “apples to apples” (using observable land characteristics) comparisons reduce the estimated gains from 44% to 11% .  Globally, Joppa and Pfaff (2010) find controls for land characteristics reduce estimated impact on average by about one‐half.    27    We would expect the lobbying efforts against PAs in those locations to be lower.8 However, it is possible  that the most beneficial non‐PA use of the land is, instead, for hydropower – in which case steeper slope  could increase the opportunity costs of maintaining an undisturbed PA. Indeed, we observe many cases  in which creation of hydropower dams has led to the degazettement of existing PAs (Symes et al. 2015).   Thus, understanding the land‐use dynamics in a given setting is critical for good estimates of a baseline  or counterfactual land‐use trajectory that, compared with the outcomes within PAs, determines offsets  (in the sense of computing how much area that was going to be deforested was conserved by the PA).  Offset calculations are likely to also include computations concerning the density of a desired service –  e.g., a hectare of forest in a favored habitat may be weighted more highly than other forest hectares or  avoided deforestation that affects water quality upstream of large cities may be especially appreciated.   Distinguishing across PAs, the political calculus involved in site choice for new PAs, as well as choices on  both monitoring and enforcement, is likely to vary with the level of government taking the PA decisions.  Various studies indicate that federal decisions on the environment can be expected to differ from local  decisions, since local actors are expected to put more weight on local costs (A. Pfaff and Robalino 2012).  One can imagine that decisions would also differ among countries across a BRI corridor. There is limited  such evidence for PAs, in general, though for the Brazilian Amazon this point appears to be supported.   Further, the local and global benefits and costs for different relevant actors in offsets vary with PA type.  Strict PAs that do not permit any entry or production vary in their net benefits from multiple‐use PAs or  sustainable reserves that permit some smallholder production. Offering some local benefits could easily  influence lobbying positions. Consequently, multiple‐use PAs may be more likely to be in locations with  relatively high forest pressure (for global evidence, see Nelson and Chomitz (2011); for a particular case  in Brazil, see Pfaff et al. (2014); and for an intermediate regional result for Peru, see Rico et al. (2018)).  Whether that implies higher impacts for one type of PA varies, even if mixed‐use PAs are in areas which  are more threatened (Blackman 2015; Ferraro et al. 2013). Not surprisingly the dynamics vary (see, e.g.,  Pfaff et al. (2015) and Pfaff, Santiago‐Ávila and Joppa (2017)). Understanding land‐use dynamics is key.  The same logic concerning local benefits and costs can also affect monitoring and enforcement. As was  raised by Albers (2010), local actors with incentives to keep the forests standing can contribute not only  through their own land‐use choices but also by contributing to monitoring of others. Thus, for example,  the multiple‐use PAs whose locations are highlighted above might also be better locally monitored. On  the theme of private incentives to keep forests standing, it is even possible that logging concessions –  whose rights create incentive to defend forest assets – could outperform PAs in terms of avoiding loss  should those PAs have fewer resources, less incentives, and poor enforcement (Panlasigui et al. 2017).  Consequently, on both the positive and the negative side, in estimating the offsets that are likely to be  generated by any given conservation investment of this type, the private incentives clearly do matter.  C.  Offset Instruments: Payment for Ecosystem Services  Payments for Ecosystem Services (PES) are a voluntary intervention for preserving forest cover  (Engel, Pagiola, and Wunder 2008; Ferraro and Simpson 2002; Wunder 2007). Their typical effectiveness  in terms of avoided forest loss has, to date, often been relatively low. Thus, at the least depending upon                                                               8  Given variation in the benefits and costs of PAs, and which groups in a society incur them, we expect that political economy  will affect the location as well as monitoring and enforcement of PAs. If we hold benefits constant, we predict that PAs should  go to the locations with the lowest opportunity costs. Globally, it has been confirmed that PAs are disproportionately “high,  steep and far” (L. N. Joppa and Pfaff 2009). More generally, characteristics that lower agricultural opportunity costs appear to  attract siting of PAs.  28    program targeting and implementation, one may not be able to count upon PES for significant offsets.  Payment for watershed services (PWS) – a specific type of PES –offset services like sediment retention,  pollutant filtration, and water retention (to reduce floods) that are affected by construction and  operation of roads and rail.   PWS or “water funds” offer payments from water users such as water  utilities to land owners or users in exchange for improving agricultural land management, including  riparian buffers to improve watershed management (Locatelli and Vignola 2009).      Even more than for PAs, additional forest beyond baseline is hard to guarantee when PES participation is  voluntary. Individuals who were not going to clear the forest on a parcel of land anyway would rationally  be the first in line to volunteer that parcel for a contract that pays money if the forest remains standing.  That can explain low impact estimates for PES (J. Robalino and Pfaff 2013; Sanchez‐Azofeifa et al. 2007).   Estimates of gains tend to be more encouraging if targeting is not entirely up to individuals but rather is  determined by other factors, e.g.: biological factors, as in the protection of Monarch butterfly habitats in  Mexico (Honey‐Rosés, Baylis, and Ramírez 2011); or by government actors or civil society, depending on  their formal or effective decision rules; or randomly, as in a trial in Uganda where the location featured  high deforestation for low opportunity costs and individuals had to put all their parcels into the PES  (Jayachandran et al. 2017). China’s Sloping Land Conversion Program is said to be impactful, although  not quite voluntary even if adequately covering local opportunity costs (yet distortions may interfere  with targeting, depending on incentives of local administrators (J. Xu et al. 2010; Z. Xu et al. 2004)).  Under this program, payments for removing sloped lands from agricultural production also resulted in  more laborers switching to non‐farm employment (M. T. Bennett et al. 2014; Kelly and Huo 2013).   Contract details offer other dimensions of PES design. PES contracts involve asymmetric information, as  land‐holders know their own opportunity costs and therefore the price at which they would be willing to  accept a PES contract. A government or NGO does not have this information (Ferraro 2008), which raises  cost and the chance that some of the funding does not lead to additional forest. That can discourage the  creation of new programs and could be addressed by auctions. PES designs that pay more to the farmers  whose previous willingness to contribute to conservation was low (suggesting high opportunity costs of  conservation and the need for incentives), however, can introduce perceptions of unfairness and lower  contribution by participants who were originally more willing to contribute (Alpízar et al. 2017b, 2017a).  Collective contracts show some promise, if designed well, using group enforcement and social pressure  to potentially increase compliance while lowering the costs of administration (D. Kaczan et al. 2017).  Information on baseline land use and behavior under the PES is lower cost at a higher, collective scale.  For instance, while not labeled as “PES”, the Brazilian Amazon “blacklist” mentioned above provided a  conditional incentive from the federal government at the collective scale of the municipality – some of  which are huge and all of which have pre‐existing political institutions – to be monitored by satellite.   Despite challenges, PES could offer an effective intervention, especially in concert with other policies.  Funds transfers within PES may be one element in bundles of interventions, helping to make elements  that are locally costly, such as strict PAs, more palatable. For generating offsets too, then, combinations  of conservation policies might usefully be considered at the level of the entire BRI Economic Corridor.  D.  Carbon Offsets   Carbon storage and sequestration provides an important climate‐stabilization ecosystem service  (The World Bank 2017). Loss of forests, wetlands, and other habitats due to construction and operation  29    of transportation infrastructure, though, can release carbon and reduce sequestration.  Thus, offsetting  here should include climate stabilization, i.e., compensating for the emissions and loss of sequestration.  To date, there is one example of offsets required for loss of GHG sinks or increases in emissions due to  infrastructure, a greenhouse gas regulatory regime in the State of California under state environmental  protection (CEQA), where one must assess, minimize and offset shifts in carbon sinks and  emissions.   However, it is not enough to do activities and claim offsets if activities did not actually lower emissions.  Studies of carbon offsets such as from projects in the Kyoto Protocol’s Clean Development Mechanism –  the world’s largest carbon offsetting program, with many energy and industrial types of offsets – show a  lack of impact (Haya 2010). Clearly, some methods for calculating carbon impacts and carbon offsets are  imperfect (Harangozo and Szigeti 2017; Johnson, Edwards, and Masera 2010). It is still the case, though,  that well‐designed and stringently enforced programs could lower emissions and even have co‐benefits  (e.g., some forest‐carbon efforts in the U.S. for the California market (Anderson, Field, and Mach 2017)).   E.  Spillovers and Net Impacts  As emphasized, it is a challenge to achieve impacts via interventions of any type and purpose –  avoiding, reducing, or offsetting – given that simply announcing a program does not change behaviors.  Simply creating a protected area does not guarantee there will be no illegal invasion and deforestation;  plus, even if the trees remain standing, it does not guarantee any impact if there were not any threats.  Simply creating an incentives program does not guarantee that anybody will sign up or will change any  behaviors. If one is paid to keep trees standing, one might simply volunteer agriculturally useless land.  However, sometimes interventions will in fact change behaviors on those lands within their boundaries.  Unfortunately, even that does not guarantee a net impact, since interventions might generate spillovers.  If a protected area does in fact lower clearing relative to what would have occurred without protection,  which many have done on average across the world, unfortunately it could still easily be the case that  the individuals who would have cleared inside the protected area just displace their deforestation  elsewhere. Within payments programs, even if one parcel that would have been cleared is not, due to  payments, there is often nothing preventing a landowner from just shifting the clearing to another such  parcel. Should such spillovers – in those cases “leakage” – occur, they can cancel out any positive  impacts.  Conceptual models of how market demands for land clearing play out across any landscape certainly  suggest that PAs could induce spillover deforestation into nearby locations (Armsworth et al. 2006).  Empirical estimates of forest spillovers from PAs find heterogeneities (A. Pfaff and Robalino 2017; J.  Robalino, Pfaff, and Villalobos 2015). That is in part because when lacking impact, PAs will not spill over,  e.g., if threats are not taken on then they cannot be blocked, for impact, but also will not be displaced.  Spillovers tend to be higher deforestation elsewhere if interventions are nearer to roads (J. Robalino,  Pfaff, and Villalobos 2017) – which is obviously relevant when considering integrated policies that place  protected areas near BRI projects − and if prices for commodities whose production is displaced are high  (Baylis, Fullerton, and Shah 2016). Spillovers may be less negative near PA entrances, where ecotourism  tends to concentrate and to raise the value of intact forest parcels  (J. Robalino, Pfaff, and Villalobos  2017), as well as when any displaced commodities’ prices are low (Baylis, Fullerton, and Shah 2016).   30    Spillover effects can even be positive, environmentally speaking, for instance in the sense that PAs lower  net forest losses nearby. For instance, if PES support introduction of trees into agriculture (agroforestry),  it may sometimes be the case that local profits are in fact enhanced (e.g., by shade for coffee and cows).  Upon seeing such an increase in profit, neighbors might voluntarily imitate those agroforestry practices.   The existing empirical results for deforestation spillovers from protected areas indicate that not only the  size (magnitude) but also even the sign (direction) of spillover effects will depend on comparative rents  (Angelsen 2007) and political signaling depending on which agencies take which actions (Herrera 2015).  For instance, the creation of a strict protected area by a strong federal agency that is known to currently  feature an environmental agenda can signal that the protected region will not be a development focus.  That, in turn, should influence private migration choices, public investments, and other public actions.  Spillovers outside interventions’ boundaries (distinct from multiple‐use PAs) can also be socioeconomic,  emphasizing again that they can vary even in sign by context and thus location. A number of studies find  PAs have weakly positive economic welfare impacts  (Alix‐Garcia et al. 2013; Sims 2010; Sims and Alix‐ Garcia 2015).  Sims (2010) finds that decreases in poverty may result from PAs – primarily those PAs  found at an intermediate distance from urban areas such that they are far enough that the PA is not  substituting for urban land use but close enough that these areas are accessible to paying tourists. In  other contexts, the creation of PAs has been tied to significant negative socioeconomic impacts, due for  instance to displacement of vulnerable populations (Geisler and de Sousa 2001) and reduced wages (J.  A. Robalino 2007) − and it is important to consider the distribution of the associated costs and benefits.       Part 3: Policies to Improve Environmental Net Impacts of BRI Corridors  Because transportation infrastructure dramatically transforms landscapes, policies have been developed  over decades to improve environmental and social performance. In Part 3, we review types of polices −  including laws, regulations, guidelines, agreements, standards, and safeguards − that can affect how BRI  transportation projects address the environment. To the extent possible, we will focus on environmental  policies that are distinctly relevant to BRI transportation projects. We focus first on national policies and  then on international policies, including Chinese policies related to direct foreign investments. Finally,  we discuss how strategic environmental assessments customized to BRI Economic Corridors are well  suited to effectively and efficiently address transboundary environmental impacts related to BRI  projects and could improve both environmental and social outcomes. While our focus is on  environmental policies, it should be noted that many of these policies – especially in recent years – also  address social issues.   I. NATIONAL AND PROJECT‐LEVEL ENVIRONMENTAL POLICIES   A.    Overview of National Environmental Policies  The first line of environmental protections is of course choices by the host countries. China has  pledged to follow host‐country standards and norms for all Belt and Road projects, so in principle host‐ country policies provide the minimal protections expected for every infrastructure project (Dollar 2018).  31    Of the roughly 80 countries involved in the BRI, each one has its own set of environmental policies that  must be understood to grasp the full regulatory landscape. There is a wide range of national, provincial,  and local regulatory policies that could have a bearing on  environmental risks from transportation  infrastructure projects including: environmental assessment laws; environmental protection laws for  public lands; environmental protection laws from private lands including easement laws; primary  protected areas legislation; criminal codes addressing violation of environmental protection laws and  related obligation; wildlife laws; construction permitting; wetlands protection and mitigation; vehicle  emissions standards; damages law for hazardous materials spills and leaks; indigenous peoples rights  laws/territory rights; zoning laws; imminent domain laws; and national procurement rules. In addition to  national legislation, countries may have policies, guidelines, and strategies for addressing many of these  issues. It is also important to track various complementary laws, regulations, and zoning ordinances  from provincial or local governments, since these can sometimes have as much if not more influence  over BRI projects than a central government (H. Liu and Lim 2018). There is a growing body of resources  that document, archive, and analyze environmental laws. The Green Growth Knowledge Platforms  (GGKP) provides an excellent meta‐analysis of 24 databases of environmental regulations, financial  incentives, overarching policies and frameworks, and voluntary approaches (Booth 2017). Overall these  databases provide comprehensive coverage of the electricity, agriculture, forestry, and fishing sectors in  OECD countries. Coverage of other sectors and regions of the world – including most BRI countries – is  less exhaustive, unfortunately. Legal Atlas is beginning to develop an interactive knowledge platform for  all environmental regulations related to infrastructure projects in every BRI country, initially focusing on  laws for linear infrastructure. Appendix 10 describes several databases that are particularly valuable for  assessing environmental policies that may have a bearing on BRI transportation infrastructure. It should  be noted that, while host country laws may constrain all the BRI investments, significant elements of the  environmental risks that are associated with BRI transport corridors are transboundary risks, including  impacts upon carbon emissions, forest cover, and biodiversity. Limiting these transboundary impacts  can fairly be considered to be global or regional public goods provision, so we should not expect host  countries to fully take these impacts into account in their own laws (Kaul, Grunberg, and Stern 1999). 32    B.  Environmental Assessments  One element repeatedly featured in many of the  national regulations and policies is the environmental  assessment, among the most commonly used planning and  management tools for addressing environmental risks for  many decades. “The ability of countries and communities  to achieve sustainable development depends in no small  measure on robust and effective EIA/SEA legislation and  implementation as a major catalyst for overcoming current  implementation gaps and achieving better environmental  outcomes.” (UN Environment 2018, p vi) In the next two  sections we will describe the use of environmental impact  assessments (EIA) and strategic environmental assessment  (SEA) for BRI transportation projects.  EIAs are typically focused on a specific development  project, such as a new highway segment or railway.  The  EIA dictates a formal administrative process that is aimed  Figure 3: EIA process flowchart. Source:  at preventing harm to the environment, often also taking  UNEP 2018.  into account related social considerations (Figure 3). EIAs  should begin in the project planning phase and thus prior to an activity taking place. The EIA identifies  potential direct (and less often indirect) environmental impacts of a project and trade‐offs between  policy goals. EIAs are aimed at informing decision‐making on how to proceed (UN Environment 2018a).  In recent years, it has become increasingly evident that EIAs have many shortcomings. Because EIAs  often occur “downstream,” well into the planning process, most decisions and commitments have  already been made and parties tend to be vested in particular outcomes (OECD 2006). Moreover, the  continual ratcheting up of EIA standards by the International Finance Corporation (IFC), World Bank, and  other multilateral development banks has had the perverse effect of promoting borrowing from other  sources with lower EIA standards, even if it means higher lending rates (Humphrey 2016).   To address these shortcomings, governments and multilateral development banks have increasingly  begun to encourage or require strategic environmental assessment (SEAs) – early‐stage analyses that  integrate environmental (and often social) considerations into policies, plans, and programs – rather  than solely relying on the single project‐based EIAs (World Bank 2012). By comparison to the project‐ focused EIA, SEAs aim to integrate environmental considerations into policies, plans, and programs (Dusi  and Xi 2009; OECD 2006). SEAs are often carried out at the regional or national scale or for an entire  sector such as transportation. They can also focus on transboundary projects in two or more countries.   Most significantly, SEAs are carried out early in the planning process when alternative transportation  routing is still feasible (Figure 4), and the full mitigation hierarchy can be followed. (See Part 2 for a  discussion of the mitigation hierarchy.) SEAs have been characterized, to some degree, as forward‐ looking, “sustainability driven” instruments, whereas EIAs are more “reactive” (Wingard et al. 2014).  33    While SEAs are increasingly being adopted and  strengthened, as compared to EIAs they are still  in a nascent stage of development. At least 40  countries have SEA systems in place including  China, Vietnam, Indonesia, Mongolia, Kenya, and  Tanzania; however, only some of these 40  countries include a formal legal requirement to  conduct SEA (UN Environment 2018a). These  assessments are designed to be flexible to allow  analysis to meet the needs of decision makers,  but as a result there is uncertainty and  inconsistency in application (UN Environment  2018a). In China, SEA implementation has been  limited by the sharing of information and data,  decision‐making process (especially the timing of  SEA planning), and legislative and political  context (T. Li et al. 2016). Gunn and Noble (2011)  note that SEA methodologies are still  Figure 4: Strategic Environmental Assessment  inadequately developed and defined and are not  process flowc4art. Source: UNEP 2018.  capable of addressing regional cumulative effects  adequately. Chaker et al. (2006) found challenges identifying appropriate leverage points in planning  and decision‐making processes for SEA input, limiting integration of SEA findings into decision making.  Despite these limitations the World Bank found sufficient evidence that SEAs can influence and inform  decisions to encourage their continued use in bank practice (World Bank 2012).    For SEAs to fulfill their promise as a format for strategic planning, they will need to be conducted within  an integrated economic planning process. The Inter‐American Development Bank’s guidance concerning  sustainable infrastructure contains two key observations related to SEA implementation: sustainability  should be considered early, “upstream” in the policy and planning stage of infrastructure development,  and environmental considerations should be incorporated into the mainstream of the planning process  so that economic development and environmental assessment are not considered within separate silos  (Serebrisky et al. 2018; UN Environment 2018b; Watkins 2014). Strategic integration of environmental  assessment with infrastructure planning has the potential to decrease the costs that are associated with  mitigation of environmental harm by addressing conflicts between environmental goals and economic  objectives at an earlier stage, when more options remain on the table. Insofar as the SEA framework  encourages early, integrated approaches to planning for sustainable development it appears to be a  promising vehicle for improving environmental outcomes associated with infrastructure investment.  C.  Environmental Policy Compliance  The adoption of comprehensive policies with appropriate guidance on implementation and  monitoring are critical first steps toward positive environmental outcomes.  However, these policies are  only as effective as their enforcement mechanisms, and the capability and political will to enforce them  (Leung et al. 2013; UN Environment 2018a). BRI countries not only differ in their environmental policy  frameworks, but there are also enormous differences in their capabilities and political will to enforce  34    environmental policies. Limited institutional capacity and generally a lack of legal requirements often  limit follow up on both EIAs and SEAs (UN Environment 2018a). Given that EIAs occur late in the  planning process, the likelihood of cancelling a project or making dramatic alterations as a result of  findings in the environmental assessment report is low. There are typically no provisions in national  legislation regarding the process for cases of non‐compliance with an environmental report or in a case  when mitigation measures prove to be ineffective (UN Environment 2018a).  D. Comparisons of National Environmental Policies and Compliance  Countries vary tremendously in the coverage and strength of their environmental and social  policies that address transportation infrastructure. Cross‐country comparisons mostly evaluate the  differences among national environmental assessment policies, in part due to their widespread use.  A  global review by UN Environment of EIAs and SEAs found that, though virtually every country studied in  the report9 has adopted EIA systems based on legislation and an increasing number of countries now  have SEA systems in place, the depth and application vary significantly among countries. The report  identified trends in EIA national legislation, including: a movement by some countries towards  decentralized oversight and implementation of EIA, though a small group of countries have moved in  the opposite direction by establishing dedicated central authorities to deal specifically with EIAs;  increased public participation requirements (but mostly limited to the scoping and review stages);  increased focus on climate change and human health and, though non‐binding, ecosystem services; and  a shift from EIAs to Environmental and Social Impact Assessments (ESIAs). Emerging trends among SEA  legislation include: the increased adoption of non‐binding procedural SEA guidelines but rarely legal  requirements; little attention to public participation; and regional rather than national level guidelines  to facilitate assessing impacts related to climate change, biodiversity, ecosystem services, and human  health (UN Environment 2018a).   In a more targeted legal analysis aimed at developing guidelines for mitigating the impact of linear  infrastructure on large migratory mammals in Central Asia, Wingard et al. (2014) compared components  of EIA and SEA legislation as it relates to linear infrastructure in eight Central Asian countries.10 Overall,  they found that, although all eight countries had EIA legislation, there were few direct references to  linear infrastructure, migratory species, or transboundary impacts within any of the legislation. EIAs in  five countries mentioned wildlife impacts, but only Kazakhstan’s legislation explicitly mentioned  migratory species and requires their consideration during the construction of linear infrastructure.  Kazakhstan, Kyrgyzstan, Tajikistan, and the Russian Federation all required the assessment of  transboundary impacts. In terms of SEAs, all Central Asia countries except Russia and China have  national legislation for both plans and program, though not all covered policies. China’s SEA provisions  are limited only to the assessment of plans and the Russian Federation assessment legislation does not  currently address SEAs. Clearly, at least along the China‐Central Asia‐Western Asia BRI Economic  Corridor, there is significant variation in national environmental legislation including some serious gaps                                                               9  The report focused on the following case study countries: Australia, Austria, Bhutan, Brazil, Cameroon,  Canada,  China, Colombia, Denmark, the European Union, Fiji, France, Germany, India, Indonesia, Kenya, Lebanon, Mexico,  Mongolia, Oman, Panama, Peru, South Africa, Tanzania, the United States, and Vanuatu (UN Environment 2018a).    10  The report focused on Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, Uzbekistan, Mongolia, China, and the  Russian Federation (Wingard et al. 2014).  35    in coverage.  While the presence of comprehensive national legislation is not the only factor responsible  for environmental performance of a country – Esty and Porter (2002) found that other important factors  include regulatory structures responsible for compliance and enforcement; availability of environmental  information; civil society presence and empowerment; administrative, scientific, and technical  infrastructure – it is clear that cross‐country differences in environmental performance are associated  with the rigor and structure of environmental regulations as well as a country’s willingness and ability to  enforce these regulations (Esty and Porter 2002).  The enormous variation in countries’ environmental policies and enforcement capabilities has real  implications for BRI transportation investments. One of the repeated concerns of environmental critics  is that BRI investors will show a preference for infrastructure projects and routes in countries with lower  standards and enforcement in order to avoid the upfront costs of carrying out a comprehensive EIA or  SEA (W. Laurance 2017; Moran et al. 2018; Shinn 2016). As China tightens its own environmental  standards, the difference between standards at home and those in many BRI countries are becoming  increasingly stark. Will China follow its longstanding commitment to rely on host‐country legislation and  norms to dictate how BRI investments address environmental impacts? Or will it require its investment  to adhere to more stringent international or Chinese standards in order to achieve its vision of a  sustainable and green Belt and Road? In the next section we will explore some of the international “best  practice” standards that could help attain a sustainable and green BRI.   II. GLOBAL ENVIRONMENTAL POLICIES   The Good Industry International Practice (GIIP) standards established by international entities –  whether international agreements, international lending safeguards, or voluntary standards from  nonprofit organizations or industry – are often more restrictive than those of individual nations,  especially countries with weak environmental policies. In this section we review a range of these  standards, from binding requirements from international lending institutions to voluntary standards that  may be advantageous to follow when considering net economic benefits or reputational enhancement.  This section also reviews the environmental standards of Chinese policy banks. Critically, while as just  noted often international practices go beyond the standards of host countries, if those practices are not  implemented in ways that actually constrain the impacts of all BRI investments, including with significant  consequences for failing to carry out improved and agreed practices, then their benefits are not realized.  Consequential monitoring and the enforcement of agreed conditions is a necessary part of all projects.  A.  International Agreements  Though there are many fewer international environmental policies than the myriad national  laws, there are still far too many that potentially bear on BRI transportation infrastructure to describe  individually here (and see Monteiro and Trachtman in review for provisions within trade agreements).  These international agreements are not likely to affect BRI transportation projects directly but through  their work with governments, international institutions, and NGOs to create appropriate relevant  policies.  Appendix 11 provides an overview of some of the critical agreements with the greatest  overlap. For more detailed information, the Green Growth Knowledge Platform meta‐analysis of  environmental databases is a good source (Booth 2017). The Ecolex (www.ecolex.org/), Environmental  Treaties and Resource Indicators (http://sedac.ciesin.columbia.edu/entri/index.jsp), and International  Environmental Agreements Database (https://iea.uoregon.edu/) also provide useful references along  these same lines.    36      B.  Lending Institution Environmental Policies  An effective means of influencing BRI transport projects has been conditions set by funding sources.    i.     Multilateral Development Banks  While the majority of BRI transportation project loans are coming from China, sometimes  multilateral development banks (MBDs) provide co‐financing for these projects. All MDBs now have  their own set of environmental and social standards, which place specific requirements on loan  recipients regarding the minimum social and environmental standards required for investments to be  allowed to progress, and BRI projects with MDB co‐financing adopt the standards of the MDB partner  (Cader et al. in preparation). Even if there is no financing from MBDs, environmental and social  standards established by MDBs – and, in particular, the IFC’s Performance Standards or the World  Bank’s new Environmental and Social Framework (ESF) – are often considered the Good Industry  International Practice (GIIP) or best practice standard for multilateral, bilateral, or commercial loans.  Appendix 12 provides an overview of these environmental and social safeguards for MDBs engaged in  BRI infrastructure loans. Still, clearly, they would have to be strongly monitored and enforced, with  consequences, to provide good incentives.  The World Bank, which was among the first MDBs to address environmental and social safeguards in the  1990s, adopted revised environmental and social safeguards in 2018 to strengthen their protections for  the environment and for people while making it easier for borrowers to comply with relevant standards.  These reforms were a response, in part, to growing demand for lending operations to be more efficient,  for due diligence to be more flexible, and for greater reliance on host country environmental and social  standards (Dollar 2018), which again are likely to be less strict. These standards parallel, to some degree,  the founding vision of the two recently established MDBs, the New Development Bank (NDB) of BRICS  countries and the Asian Infrastructure Investment Bank (AIIB). Both NDB and AIIB clearly have pledged  their commitment to sustainable development as well as efficiency gains in project processing – which  might conflict. AIIB, for example, has committed to be “lean, clean, and green,” meaning cost effective  and efficient, have zero tolerance for corruption, and a respect for sustainability and the environment.  NDB, as compared to the other multilateral development banks, depends more heavily upon the host  country policies, with the aim of strengthening those policies when feasible instead of imposing bank  standards (New Development Bank 2016). This policy of “country system plus” allows for host country  regulations to be used for environmental and social safeguards except in projects or countries deemed  high risk, in which case NDB would draw on IFC Performance Standards. Some critics worry that both  AIIB and NDB efforts focus so much on improving efficiency and relying extensively on existing corporate  or host‐country standards that environmental and social safeguards may well prove to be inadequate in  practice. Critics of these policies recommend improving the definition of sustainability, increasing clarity  about how adverse impacts will be mitigated once assessed, and including a more transparent policy  about how country and client systems will be assessed and incorporated into the plan (Weiss 2017).  ii.   Chinese Policy Banks  The two largest lenders for BRI projects are both Chinese policy banks: Export‐Import Bank of  China (China Exim) and China Development Bank (CDB), which have issued $100 bln and $170 bln  respectively towards BRI projects (Gallagher and Qi 2018; Shinn 2016). Chinese policy banks are state‐ owned banks that are responsible for infrastructure lending as well as promoting foreign trade.  37    The environmental and social safeguards of these two Chinese policy banks are still nascent and less  rigorous than those of the MDBs (Appendices 12 and 13; Dollar 2018; FOE 2016; Ren et al. 2017)). CDB  was one of the first banks in China to have environmental and social policies, though. By 2005, e.g., the   CDB was requiring EIAs to be conducted in accordance with the law (Ren, Liu, and Zhang 2017). Specific  environmental regulations are not publicly available, although some of the basic environmental rules  issued by the CDB can be inferred from past lending, such as the requirement to complete EIAs by an  independent evaluator and obtain approval through the EIA process (FOE 2016). The Chinese financial  institutions investing in overseas infrastructure are also subject to environmental guidelines put in place  through the China Banking and Regulatory Commission (CBRC, now the China Banking and Insurance  Commission—CBIRC). In 2012, the CBRC issued strengthened Green Credit Guidelines (GCG) that compel  Chinese banks’ overseas lending to follow host country laws but also international norms, though this is  assessed at the institutional not the project level and implementation is still a key challenge (FOE 2017).  Again, at the project level, monitoring and consequential enforcement are likely to be needed for  impact.  Overall, environmental policies of Chinese policy banks are less mature than those of other international  financial institutions such as the World Bank and IFC.  The policy banks have not developed customized  and specific policies for environmental and social issues while the adopted environmental policies are  not strict enough; significantly, they lack environmental departments to oversee environmental issues  (Ren et al. 2017). Critics have urged the policy banks to strengthen the entire process, including the pre‐ loan review, the complaint mechanism, public consultancy with affected communities, and information  disclosure system (GEI 2018). While all banks require clients to follow host country standards, IFC and  other MDBs generally have their own standards that, when stricter than those of the host country, must  also be followed to have their expected impacts. In the case of the China Exim Bank, only when no host  country standards exist would the bank consider applying Chinese and international standards (Ren, Liu,  and Zhang 2017). CDB's environmental and social standards are also mainly based on the host country  laws and policies (Ren, Liu, and Zhang 2017). IFC and other international institutes have worked with  Chinese officials recently to build capacity in these policy banks to manage environmental and social  risks (Leung et al. 2013). The policy banks have made progress in several areas, including developing  requirements for environmental and social risk control, more stringent punishments for violating  environmental regulations,11 and measures to promote green investment (Ren, Liu, and Zhang 2017).  C.  Foreign Direct Investment Policies   Chinese state‐owned companies are so far the primary firms executing BRI projects. Besides  environmental regulations imposed by their funders, what other environmental restrictions are Chinese  firms subject to? In addition to host country laws and policies, Chinese firms – especially state‐owned  firms but also private companies – are potentially subject to regulations by their own government.   Over the last decade ‐‐ and particularly since 2015 – domestic environmental policies have been greatly  expanded and their enforcement strengthened within China. In March 2018 the government announced  institutional reform that provided expanded responsibilities to the Ministry of Ecology and Environment  (MEE) and the Ministry of Natural Resources (MNR). Appendix 14, which provides an abbreviated list of  the major Chinese national environmental policies. Two of the most significant laws, Environmental  Protection Law of the People’s Republic of China and Law of the People's Republic of China on                                                               11  For example, if an environmental impact is discovered and not rectified during monitoring, the China Exim Bank will stop  providing credit and consider the project as a non‐performing loan and bad debt. If the CDB discovers a project is violating  environmental regulations, it downgrades the project’s asset levels and limits or stops providing loans (FOE 2016; Ren et al.  2017).  38    Environmental Impact Assessment, require environmental impact assessments and plan environmental  impact assessment (otherwise known as strategic environmental assessments) for plans which affect the  environment within China. The Environmental Protection Law was strengthened in 2014 and  enforcement and monitoring has been substantially increased since then (Leung et al. 2013). However,  these laws do not apply to foreign investments outside China. Following the example of many developed  countries that have extended their national environmental standards to foreign investments, the NGO  community has been pressing for Chinese firms to be required to operate under the same regulations  for foreign projects that they must follow within China. At a minimum, some analysts feel habituation of  these firms to regulation at home will affect their work abroad. Other critics suggest that, without strict  regulations, Chinese firms are likely to “outsource” their polluting industries to other countries as  domestic environmental laws tighten (Moran et al. 2018).   Though generally Chinese domestic environmental regulations do not apply to overseas investments,  environmental regulations specific to Chinese firms investing abroad have been expanded in recent  years. The key policy that relates to Chinese overseas infrastructure construction projects, Guidelines on  Environmental Protection for Overseas Investment and Cooperation, was released in 2013 by the  Ministry of Commerce (MOFCOM) and the former Ministry of Environmental Protection (now the  Ministry of Ecology and Environment—MEE). These guidelines encourage but do not require Chinese  companies to complete environmental impact assessments (EIAs), develop environmental mitigation  measures, and work with local communities to identify potential negative environmental and social  impacts when overseas (Leung et al. 2013; Shinn 2016). What would require those steps is not clear.  In addition to environmental and social policies addressing foreign direct investment, there is a growing  body of policies that are specific to the Belt and Road Initiative. The initial vision for a Green Belt and  Road was released in 2015, followed in 2017 by a guidance document and cooperation plan.  All three  documents are ambitious and paint a picture of the BRI being able to, “…promote ecological progress in  conducting investment and trade, increase cooperation in conserving eco environment, protecting  biodiversity, and tackling climate change, and join hands to make the Silk Road an environment friendly  one.” (Chinese National Development and Reform Commission, Chinese Ministry of Foreign Affairs, and  Chinese Ministry of Commerce, 2015) All three documents are still high‐level and lack critical details on  implementation, monitoring, and enforcement (Chun 2017). These policies do depict, however, how BRI  projects with appropriate environmental mitigation mechanisms and enforcement could serve as a  model of good development practices throughout the region.  On the other hand, it is quite clear that  positive statements without enforced implementation achieve little. Again, the question is raised  concerning how compliance with related promises should be enforced.  D.  Industry and Nonprofit Standards  Firms engaged in BRI projects may voluntarily adhere to best practice environmental standards  – typically stricter than those prescribed by lenders or governments – from industry associations or  nonprofit organizations. Firms might elect to follow these guidelines because they calculate that it is in  their financial interest to invest sustainably or because they receive external pressure – from their  customers, governments, or the general public – and reputational benefit from adopting these criteria.     One useful reference concerning such voluntary standards, codes of conduct, and audit protocols  relating to sustainability in supply chains is the Standards Map database (www.standardsmap.org),  created by the International Trade Centre (Booth 2017). Again, their impact is another question in light  of the fact that many positive and well‐intentioned statements are made that do not shift outcomes.  39    Appendix 15 lists a sample of such voluntary best practice standards that are especially relevant to  transportation infrastructure projects. These can range from broad principles that firms pledge to follow  – such as the Equator Principles – to extremely detailed requirements that must be met in order to  obtain certification of sustainability, such as the CEEQUAL, Envision®, SuRe®, and Greenroads®  standards. The authors are not aware of any BRI transportation project that has received sustainable  infrastructure certification or has utilized the Natural Capital tools in their planning process, steps that in  principle could shift outcomes. The two Chinese firms that are currently members of the Equator  Principle Association, Bank of Jiangsu and Industrial Bank Co., Ltd., are both private companies.   While the adoption of best practices is still lacking for Chinese outbound infrastructure investments, the  China Chamber of Commerce for Minerals, Metals and Chemicals Importers and Exporters developed  standards that could serve as a prototype: Guidelines for Social Responsibility in Outbound Mining  Investments. These comprehensive but voluntary guidelines aim to direct Chinese mining companies  toward improving corporate social responsibility (CSR) and sustainability strategies. The original version  was produced in 2010; in 2015 a revised and more transparent version of the guidelines was released  and is currently being rolled out (CCCMC 2015; Chun 2017). Again, impacts of adoption are unclear.    Analysts generally believe that the large Chinese state‐owned enterprises (SOEs), as compared to the  smaller provincial and municipal SOEs and private firms, are more likely to have both the capacity and  inclination to follow voluntary best practices and other forms of CSR (Maurin and Yeophantong 2013; Xu  2014; Sun and Tang 2015; Wang and Hu 2017; Yuan and Landry 2018). This is due in part to the fact that  the larger SOEs have the easiest access to Chinese state financing and the most oversight from central  authorities. They are seen as representing the policies of the government. Furthermore, large SOEs are  the type of Chinese firm most likely to invest in BRI infrastructure through joint partnerships or public‐ private partnerships (PPP) rather than engineer, procurement and construction (EPC) service contracts  or build, operate and transfer (BOT) contracts. Because of the long‐term commitment, the companies  investing joint partnerships and PPPs are more likely to be responsive to some pressures for long‐term  sustainability investments (Myxter‐Iino et al. in prep). Large SOEs are also sensitive to their international  reputation, given that they do business across the globe. Because most of the large BRI projects are  carried out by large SOEs, their actions – positive or negative – do drive BRI infrastructure investing.    III. RECOMMENDATIONS: BRI CORRIDOR ENVIRONMENTAL ASSESSMENT  China’s commitment to a sustainable Belt and Road certainly is suggested by the growing body of vision  statements and guidelines outlining various conditions for green BRI investments. However, to date,  these recommendations remain high‐level and voluntary, though policy instruments and planning tools  do exist that could help achieve this vision. Whether and how they are implemented will be critical.  We recommend that policies be developed for the Belt and Road Initiative that require incorporating  environmental assessment procedures into early‐stage planning of entire BRI Economic Corridors. A BRI  Corridor Environmental Assessment (BRICEA) would focus on entire transportation corridors, thereby  taking advantage of the scale and connectivity of BRI to effectively address the cumulative direct and  indirect environmental risks from transportation infrastructure projects and following many of the  principles already developed for SEAs and regional environmental assessments. While a formal BRI  Corridor  Environmental Assessment  process does not currently exist, the United Nations Office for  Project Services (UNOPS) has developed a framework for sustainable infrastructure planning and  development which could provide many of the tools to support such a process (UNOPS 2017). Such a  40    process could follow many of the protocols already established for China’s domestic Plan for  Environmental Impact Assessments, as set forth under its EIA Law in 2003. Further, since we know that  the perfect BRICEAs will not always occur – in part because some projects are already underway – we  strongly recommend that a similar approach be applied to latter stages of required reviews of projects  over time, such as due diligence assessments.  The following attributes of a BRICEA make it distinctly well suited to reduce environmental risks from  BRI’s transportation infrastructure network:  Spatial scale Many environmental effects associated with BRI transportation projects – such as  impacts on wildlife migration, carbon emissions, and forest cover – are widespread, even  transboundary. For example, the impact of a rail or road network as a barrier to wildlife  migration can only be understood in the context of the entire migratory paths of disrupted  species and all the existing and proposed linear infrastructure within the species range.  Similarly, the assessment of indirect impacts of road and rail networks on deforestation from  land use changes is only meaningful at the scale of an entire BRI Economic Corridor. When  planning the most critical mitigation activities – avoidance actions – the scale of a BRICEA is  indispensable: Planners can identify acceptable routes that avoid sensitive areas within a  corridor and, for each of the alternative routes, prioritize their viability by their net benefits that  include economic, environmental, and social inputs (Kiesecker et al. 2010). Such an analysis  could also identify areas appropriate for restoration and compensatory offsets, which should  improve the efficiency and effectiveness of mitigation for the corridor as a whole. Each  individual BRI transportation segment would be expected to contribute to prescribed corridor‐ wide mitigation actions, either directly or indirectly (through compensatory offsets). For wide‐ ranging risks such as deforestation and biodiversity extinction, mitigation actions can only be  efficient and effective at this regional scale.   Temporal scale BRI Economic Corridors start with a series of pre‐defined endpoints and major  hubs, but have some flexibility concerning how to connect the hubs.  Integrating SEAs into early‐ stage planning for BRI Economic Corridors – before individual projects have been advanced –  can provide direction on how to place transportation infrastructures where they are likely to  cause the least environmental harm while maintaining most of the economic and connectivity  benefits. It is critical that this process be holistically integrated into the early‐stage economic  planning for the corridor, including all affected sectors. Knowing early in the process which  sensitive areas or infrastructure designs are off limits as well as which investments have positive  net benefits should result in productive investments of time, money, and political capital and  reduce risks of disruption and delay.  Regional connectivity A BRICEA will be most effective if it includes input from all stakeholders  affected within an economic corridor. Following the SEA‐type model developed for the Mekong  Basin Regional Commission, a regional commission could be established for each BRICSEA that  includes government representatives from all affected countries and a process for public  participation (see Keskinen and Kummu 2010). This commission could also collect, analyze, and  process information from each host country’s regulatory and legal systems, allowing all  stakeholders to coordinate the complex governance issues of a transboundary transportation  network. A highly visible and transparent BRICEA process should promote good relations  41    between lenders and host country governments and reduce potential local community  opposition (Leung et al. 2013).  Investor participation A BRICEA process could help lenders reduce time, money, and obstacles  dealing with serious environmental issues downstream during the EIA process. While individual  EIAs would still need to be carried out for BRI transportation projects to address local impacts,  the likelihood is that the major obstacles and need for late‐stage adjustments would have  already be addressed through the BRICEA process. With input of existing expertise from  international financial institutions and Chinese agencies such as MEE, NDRC, and policy banks,  BRICEAs should allow lenders to recognize the highest international environmental standards  while at the same time minimizing the bureaucracy, time, and cost of downstream EIAs. A well  conducted BRICEA should allow BRI transportation projects to achieve the highest GIIP  standards while at the same time minimizing the bureaucracy, time, and cost of downstream  EIAs (AIIB 2016). In other words, SEAs could provide lenders such as AIIB the opportunity to be  “lean” and “green.”     SUMMARY AND CONCLUSION   Environmental Risks from BRI Transportation Infrastructure  We distinguish direct effects – abiotic, ecosystems, wildlife – from indirect effects induced by changes in  transport costs and land use. The former tend to be localized and unambiguously linked to a road or rail  project. The latter often have more complex connections but, still, potentially more pervasive impacts.     Direct Effects Direct environmental impacts include abiotic impacts like air and water pollution  as well as soil erosion; ecosystem impacts such as habitat destruction and fragmentation; and  wildlife impacts such as vehicle collisions, barriers to migration, and illegal wildlife trafficking.  BRI transportation investments may be especially prone to such risks because many of the BRI  Economic Corridors pass through steep terrain that is vulnerable to erosion, soil degradation  and sedimentation, and contain sensitive ecosystems with high levels of species endemism.      Indirect Effects Historically, road and rail projects can set in motion indirect and unintended  consequences. Changes in transport costs shift markets and human populations, which can open  frontiers to settlements with habitat loss, deforestation, wildlife and timber trafficking, among  other effects. The magnitudes of effects differ by context. Deforestation – a straightforward  proxy for environmental risks – has heterogenous responses to transportation investments  depending on their ecological and development settings.  BRI transportation projects that are  sited in highly developed regions with extensive prior deforestation such as southwest China,  Bangladesh, and parts of Cambodia and Kazakhstan may have less of a rise in deforestation and  it is even possible that increased economic development spurred by lowered transport costs  could result in reforestation or afforestation. BRI transport projects located in areas of medium  development – that is, areas at the margin of development frontiers that may be profitable with  further investments – can expect significant forest loss from transportation projects near  existing forests such as in Myanmar, Lao PDR, the Malay peninsula, and western Russia. Finally,  in regions with low development and little prior deforestation, such as in eastern Russia and  42    northwest Thailand, the short‐term deforestation effects of BRI road or rail projects may be  small though there remains a high degree of risk − and uncertainty − about long‐term impacts.  Roads – especially undivided highways with unrestricted access to surrounding frontiers – are  especially vulnerable to such pressures. Divided highways and railways may experience less  pressure due to their restricted access.      Environmental Mitigation  The mitigation hierarchy distinguishes four types of actions – avoid risks, reduce risks, restore  ecosystems, and offset damages – to diminish impacts while allowing economic gains from transport  investments. They are ordered in terms of desirability. In short, adjusting projects upfront to avoid  losses is crucial.     Avoidance Avoidance is by far the most important category. Selecting road or rail routes that  avoid environmentally sensitive areas is usually the most effective and least expensive way to  minimize environmental harm. Early in infrastructure planning, routes can be selected with  minimal cost to economic efficiency. Yet as planning progresses, mitigation – even minor shifts  to road or rail routes – becomes increasingly difficult. Late in the process,  there are no more  wholesale avoidance options but only remediation is available. That can be costly and often  ineffectual.  Reduction When full avoidance is not possible, mitigation should reduce environmental harms.  Engineering and design solutions are among the frequent strategies to address direct impacts of  transportation infrastructure on abiotic conditions, ecosystems, and wildlife. Reducing indirect  impacts from transport projects – such as induced land use changes – is less straightforward.  Complementary policies are the preferred strategy for modifying human decision making.   Restoration Restoration is conducted at project sites to repair either direct or indirect impacts  to biodiversity or ecosystem services. If extensive landscape transformation has occurred during  construction, then vegetative, wetland, or waterway restoration is often required.  Offsets Compensatory offsets are a final strategy, after the other actions have been exhausted.  Establishing protected areas or buying carbon or water credits can in principle counteract the  local losses of forests, biodiversity, carbon, or other environmental services due to transport  infrastructure. Because it may be difficult to fully anticipate indirect effects such as changes in  land use over the long term, and because comparing damages at project sites to gains from  offsite offsets can be difficult, we believe offsets are inherently risky and, thus, that taking this  approach should include aiming for positive environmental net impacts from BRI investments.   Environmental Policies  Policies largely determine which environmental risks are addressed and how.  For large BRI transport  investments, policies of national governments, international agreements, lending institutions,  corporations, and civil society organizations all can influence the mitigation actions taken.     National Environmental Policies In principle, host‐country policies provide at least the minimal  environmental protections for every transportation infrastructure project. China has pledged to  follow host‐country policies and norms for all Belt and Road projects, for instance. However, the  43    BRI countries differ considerably in their environmental policy frameworks and, crucially, in their  capabilities and political will to enforce even their own such policies. Policies mandating the use  of environmental impact assessments (EIA) are among the most prevalent requirement of host  country governments concerning environmental protection from infrastructure development.  However, EIA’s ability to address environmental risks is hampered because they are typically  conducted relatively late in the infrastructure planning process, when most important decisions  have already been made. Consequently, some governments have adopted early‐stage planning  such as strategic environmental assessments (SEAs). SEAs done well can address policies and  plans at a program, landscape or sector‐level, before individual projects have been advanced.  Global Environmental Policies   Best practice standards established by international entities –  agreements, lending safeguards, voluntary standards from nonprofit organizations or industry –  are often more restrictive than those of nations. An effective means of influencing BRI projects  has been conditions established by funding sources.  IFC Performance Standards or the World  Bank Environmental and Social Framework are often considered Good Industry International  Practice standards for multilateral, bilateral, or commercial loans. The environmental safeguards  of the two Chinese policy banks – the largest BRI funding sources – are less comprehensive and  rigorous, however, than those of the multilateral development banks (MDBs). Yet when MDBs  have co‐financed BRI transport projects, their more stringent standards are applied. Domestic  Chinese environmental policies, regulations, standards, and institutional reforms have been  greatly strengthened since 2015. Some regulations and standards specific to Chinese policy  banks and firms investing abroad have also been expanded in recent years.  However, much  work still needs to be done for China to expand its national environmental standards to foreign  investments, especially in the areas of monitoring and enforcement.  Recommendation: BRI Corridor Environmental Assessment  Given its scope, the BRI’s potential environmental impacts are extensive. While mitigation of these risks  is daunting, policy instruments and planning tools do exist to dramatically decrease them. We strongly  recommend that the BRI incorporate regional versions of SEA‐style approaches into the earliest‐stage  planning for entire Economic Corridors. A BRI Corridor Environmental Assessment (BRICEA) would focus  upon the entire transportation corridor, taking advantage of the scale and connectivity of the BRI to  address the cumulative direct and indirect risks from these projects. Many of the elements of a BRICEA  already exist within China’s domestic national environmental policies and could be extended to  encompass foreign investments. Further, since we know that perfect BRICEAs will not always occur – in  part because some projects are already underway – we strongly recommend that a similar approach be  applied to later stages of required reviews of projects over time, such as due diligence assessments.    The following attributes make an Environmental Assessment focused on BRI Economic Corridors  distinctly well suited to reduce risks:  Spatial scale For wide‐ranging, cumulative risks such as deforestation and extinction, mitigation  actions are most effective at a regional scale.  For avoidance, the scale of a Corridor  Environmental Assessment is indispensable, as planners can identify routes to avoid sensitive  areas and, across alternative routes, prioritize by net benefits. One can also identify areas  appropriate for restoration and offsets. Each segment would be expected to contribute to  compliance directly or indirectly (via compensatory offsets).   44    Temporal scale Integrating environmental assessments into the very earliest planning stages,  before individual projects advance, best permits guidance on how to place transport  infrastructure where it causes least environmental harm while maintaining most economic‐ connectivity benefits. It is critical that early‐stage corridor planning integrate all affected sectors.  Knowing early which sensitive areas should be off‐limits, and which investments have greatest  net benefit, should yield productive investments of time, money, and political capital and reduce  risks of disruption and delay.  Regional connectivity A BRI Corridor Environmental Assessment (BRICEA) will be most effective  with input from all stakeholders in an economic corridor. A regional commission could be  established for each BRICEA with government representatives from all affected countries, plus  broad public participation. A highly visible and transparent BRICEA process should promote  good relations between lenders and host country governments and reduce potential opposition  from local communities.  Investor participation A well conducted BRICEA should allow lenders to insist upon the highest  international environmental standards while at the same time minimizing the bureaucracy,  time, and cost of downstream EIAs. While individual EIAs would still need to be carried out to  address specific technique choices with local impacts, the likelihood is that the major obstacles  and need for late‐stage adjustments would already be addressed through such an SEA process.       45    APPENDICES  APPENDIX 1: Forest Cover Change across the Entire Belt and Road  Change in forest cover12 can be  used as a proxy for environmental risks  related to a wide range of  environmental variables of interest,  including species’ habitats, healthy  ecosystems, carbon storage, water  provision, and other ecosystem services.  Because satellite data on forest cover  are relatively easy to obtain, this is one  of the simplest and most  straightforward tools to help identify  the impact of transportation  infrastructure on ecosystems.13   Figure 5. Six economic corridors of the Belt and Road Initiative  When examining forest cover change at  Source: china‐trade‐research.hktdc.com/business‐news/article/The‐Belt‐and‐Road‐ the scale of the entire Belt and Road,  Initiative/The‐Belt‐and‐Road‐Initiative/obor/en/1/1X000000/1X0A36B7.htm    some distinct patterns emerge that  invite further exploration. The variations in forest cover and loss, for instance, highlight two important  questions: Within which of the BRI Economic Corridors (Figure 5) are forests most vulnerable to  infrastructure development? And is this change in forest cover a good proxy for vulnerability to other  environmental risks?   A cursory evaluation indicates that some corridors generate more risk than others based on the simple  fact that not all affected areas have much forest. The China‐Indochina Peninsula Economic Corridor  (CICPEC) and China‐Mongolia‐Russia Economic Corridor (CMREC) are among areas at greatest risk with  large areas facing active deforestation in the last 15 years (Figure 6). Further, across and within potential  economic corridors, we can take into account levels of prior development − e.g., high versus medium or  low – when considering impacts of transportation investment. (See Part 1, Section II.B.i.b for a  theoretical framework concerning deforestation in relation to economic development.) Within highly  developed (and deforested) regions such as southwest China, Bangladesh, and parts of Cambodia and  Kazakhstan, little natural forest remains to be cleared for agriculture. Hence, little forest impact is  expected from BRI transportation projects in these highly developed regions. It is even possible that  increased economic development spurred by lowered transport costs in such regions leads to rural                                                               12  “Forest” for continental Southeast Asia and southwest China refers to a range of different tropical forest types,  including some moist rain forests but mostly evergreen forests occupying the mountainous zones; seasonal or  deciduous forests in the monsoon dominated sub‐region; very dry forests and woodlands occupying plains,  plateaus, and other water‐limited sites; and mangrove forests along the coast (H. Stibig et al. 2007).  13  While vegetation cover is relatively easy to distinguish from barren areas in satellite imagery, primary forests can  be difficult to distinguish from tree plantations, degraded forests, and regenerating forests, thus complicating the  ability of satellite imagery to provide good estimates of changes to habitats, carbon, and other ecosystem services  (Harris et al. 2012).  46    transformations that result in reforestation or afforestation via tree plantations or natural regrowth (see  Kazcan 2016).     On the other hand,  within areas of  medium development  – that is, areas at the  margin of  development frontiers  that are close to  profitable now and will  be profitable with  further investments – significant forest loss  can be expected from  BRI transportation  projects near existing  forests. Such areas can  be found within  CICPEC (in Myanmar,  the Lao People’s  Democratic Republic,  the Malay peninsula)  Figure 6. Rail and road projects (existing, under construction, and planned) across  and CMREC (near  entire Belt and Road region in relation to forest cover, forest loss, and forest gain  Moscow).     Finally, in regions with low development and prior deforestation, such as in eastern Russia Federation  and northwest Thailand, the short‐term effects of BRI road or rail projects may be small though there  remains a high degree of risk − and uncertainty − about long‐term impacts. Should BRI road and rail  projects trigger development dynamics, with complementary investments across a decade or more  make these areas attractive for increased settlement, then such areas may experience considerable  long‐term forest loss. However, if other factors (e.g., topography, climate, and policies) dissuade such  dynamics, there could be little change.    Other environmental features – such as watershed protection which is highly valued, e.g., within big  cities, or biodiversity hotspots that are critical havens for clusters of unique and threatened species –  are not necessarily closely correlated over space with forest cover dynamics. As seen in Figure 7,  identified biodiversity hotspots may provide additional guidance at a corridor scale. At such a broad  level, the CMREC corridor, e.g., faces fewer specific hotspot risks from BRI transportation projects than  does the CICPEC corridor. Meanwhile, planned BRI road and rail projects of the China‐Central‐and‐West‐ Asia Corridor cut through biodiversity hotspots within all of Kyrzgystan, Tajikistan, and Uzbekistan.      Even such coarse patterns clearly illustrate great heterogeneities for environmental risks facing the BRI,  due to variations in both economic and ecological factors affecting the individual and cumulative risks  47    from transportation projects within different BRI Economic  Corridors. Consequently, it is imperative that integrated  planning processes at early stages of corridor development  should not consider the average or the modal BRI project  but, instead, all the corridor details relevant for assessing the  cost‐effective means for reducing environmental risk within  each corridor. This can contribute greatly, at low cost, to  avoiding societally inefficient investments in which, e.g.,  even minor economic gains (or no gains or net financial  losses from investments) could generate significant  environmental risks, i.e., a ‘lose‐lose’. Such assessments are  eminently possible, for instance along the lines illustrated for  road investments across varied landscapes in Dasgupta and  Wheeler (2016) and Danyo, Dasgupta and Wheeler (2018).  Towards application of such thinking within the BRI, in the  succeeding appendices we will explore some of these  potential risks and approaches to their potential mitigation in    greater detail for one corridor, CICPEC.  Figure 7. Rail and road projects (existing, under      construction, and planned) across entire Belt  and Road region in relation to Conservation  International’s Biodiversity Hotspots  48      APPENDIX 2: Assessing Environmental Risk from Transportation Infrastructure  Development in the China‐Indochina Peninsula Economic Corridor (CICPEC)  Many individual transportation projects or segments are  required to connect an entire BRI Economic Corridor.  While  each of these segments of rail or road may cross a relatively  short distance, as a whole the full economic corridor can  traverse many different ecosystems and political boundaries.   At what scale should environmental risks be assessed – at  the level of the entire corridor or the individual projects?  There is not one answer to this question, as it will depend on  the types of risks being evaluated; invariably assessments  need to be made at multiple scales to truly capture all risks.  Carrying out an assessment at the project level through an  environmental impact assessment makes sense when  considering the impacts of construction on localized harms  such as particulate pollution or changes in soil stabilization.  However, the effect of infrastructure operations on many of  the most pervasive and long‐lasting factors – such as habitat  fragmentation and land use change – can only be properly  evaluated at the broader landscape or entire‐corridor scale,  Figure 8. Existing and BRI road and rail  ideally through an early‐stage strategic environmental  segments along the eastern, central, and  assessment process. Over the course of a series of  western transportation routes connecting  Appendices we take a closer look at one of the BRI Economic  Kunming, China and Bangkok (highlighted in  yellow).   Corridors, the China‐Indochina Peninsula Economic Corridor  (CICPEC), to illustrate a few of the potential environmental  risks from alternative rail and road routes at the scale of the entire corridor. Specifically, we will analyze  recently constructed and planned BRI high‐speed rail and highway projects connecting Kunming, China  to Singapore, via Bangkok, Thailand.    The vision of improving the transportation network between Kunming and Singapore is not new to BRI  and CICPEC. Efforts to connect southwest China and Southeast Asia through a rail network go back to  the early 1900s during the colonial period. Three potential routes have been envisioned:  A western  route passing through Myanmar; a central route running through the Lao People’s Democratic Republic  and Thailand; and an eastern route that crosses through Vietnam and Cambodia (Figure 8). All three  routes are currently passable at least by road for passengers but none of the three routes has a  complete a high‐speed rail line capable of efficient freight transportation.    In the following appendices, we focus primarily on risks to forests and biodiversity conservation. This is  one group of risks among many, including environmental and social risks related to water provision and  quality, air and noise pollution, GHG emissions, fire vulnerability, earthquake vulnerability, resettlement,  and indigenous rights. We focus on biodiversity and conservation as distinctive and vulnerable natural  assets of CICPEC. It has by far the highest levels of biodiversity of any of the BRI Economic Corridors.  In  addition, habitat and biodiversity risks are sometimes quite spatially specific, offering the opportunity to  avoid or at least greatly reduce damages through choices across routes or adjustments in chosen routes.  49    To assess the conservation risks from rail and road improvements or new construction along these three  alternative routes, we will explore five potential indicators across the entire CICPEC corridor: forest  cover, intact frontier landscapes, biodiversity hotspots, umbrella species, and illegal wildlife trafficking.  50    APPENDIX 3: Environmental Indicator: Ecosystem Effects as Measured by Forest Cover  Change  Change in forest cover14 can be used as  an indicator of environmental risk that  relates to a wide range of environmental  variables of interest, including species’  habitats, healthy ecosystems, carbon  storage, water provision, and other  ecosystem services. Because satellite data  on forest cover are relatively easy to  obtain, this is one of the simplest and  most straightforward tools to help  identify the impact of transportation  infrastructure on ecosystems.15   In recent years, most of the change in  forest cover within the CICPEC corridor  has occurred in Cambodia, the Lao  People’s Democratic Republic, and  Myanmar, especially in mountainous  areas (H. Stibig et al. 2007; H. J. Stibig et  al. 2014; Zeng et al. 2018; Figure 9).  Interestingly, many of the most affected  areas have been along the boundaries  between countries, such as between  Yunnan Province, China and Myanmar;  Lao PDR, Cambodia and Vietnam; and  Figure 9. Rail and road projects within CICPEC and forest in  between Cambodia and Thailand. Most  relation to forest cover, forest loss, and forest gain  afforestation and reforestation has  occurred in Vietnam and northern  Thailand. The primary driver of change in forest cover is the conversion to cash crop plantations,  followed by conversion to timber plantations and logging (H. Stibig et al. 2007; H. J. Stibig et al. 2014).  What role have roads played in forest conversion?  Stibig and colleagues (2007) found that road  construction was responsible for some forest change, particularly in the region from China to Thailand  through the northwest of Lao PDR; from eastern Thailand through Lao PDR to the coast of central  Vietnam; and in Cambodia. These authors did not identify any forest change that resulted from rail                                                               14  “Forest” for continental Southeast Asia and southwest China refers to a range of different tropical forest types,  including some moist rain forests but mostly evergreen forests occupying the mountainous zones; seasonal or  deciduous forests in the monsoon dominated sub‐region; very dry forests and woodlands occupying plains,  plateaus, and other water‐limited sites; and mangrove forests along the coast (H. Stibig et al. 2007).  15  While vegetation cover is relatively easy to distinguish from barren areas in satellite imagery, primary forests can  be difficult to distinguish from tree plantations, degraded forests, and regenerating forests, thus complicating the  ability of satellite imagery to provide good estimates of changes to habitats, carbon, and other ecosystem services  (Harris et al. 2012).  51    construction (H. Stibig et al. 2007). This may be because there have been many fewer railways than  roads constructed in the last two decades in continental Southeast Asia; because roads were more likely  to be built into previously inaccessible frontier landscapes; or because railways provide less access to  open forested areas because passengers and freight are limited to station access.   Looking forward, would forest  cover be equally affected by  BRI transportation projects  along the eastern, central, and  western routes? Many of the  recent and proposed BRI  projects along these three  corridors are sited in areas of  high prior deforestation or on  top of footprints of existing  smaller roads or rail (Figure 9),  which lessens the degree to  which they could further  fragment the landscape,  degrade existing forests, or  provide new access. All three  of these routes, though, pass  through some forested  habitats that may be  vulnerable to habitat  destruction as a result of a new  or upgraded rail or road. As  can be seen in Figure 9, BRI rail  lines could create extensive  Figure 10. Rail and road projects within CICPEC in relation to forest cover,  risk of deforestation in  forest loss, forest gain, and protected areas. Inset of proposed BRI Burma  northern Myanmar (Western  rail Nam Tok ‐ Thanbyuzayat project   route) and Lao PDR (Central  route). One segment that merits particular flagging for careful and early investigation is the proposed  BRI Burma rail Nam Tok ‐ Thanbyuzayat project in the Thailand section of the western route crosses one  of the most forested and biologically important areas of Thailand (Figure 10 inset), cutting through a  national park and skirting alongside several other protected areas. By contrast, the Thailand segment of  the central route passes through areas with extensive prior deforestation (Figure 9) and poses no threat  of further forest cover loss.     52    APPENDIX 4: Environmental Indicator: Ecosystem Effects as Measured by Intact Frontier  Landscapes in CICPEC  Intact forest landscapes (IFLs) can be  among the most critical areas to  conserve because they frequently  serve as refuges for native  biodiversity and produce extensive  ecosystem services. Because of  extensive logging and land use  change over many decades  throughout continental Southeast  Asia, few IFLs remain within CICPEC  (Figure 11). It is not surprising that  BRI rail or road lines mostly avoid  these scarce IFLs because the  eastern, central, and western routes  are largely built along the footprint  of existing, though less developed,  transportation corridors.16 Because  the tropical region of Southeast Asia  is among the most diverse and  productive regions of the world for  biodiversity and carbon productivity  (Li et al. 2016), it is especially critical  to safeguard the integrity of the few  remaining IFLs in the region. Of the  BRI rail and road projects along the  CICPEC corridor, two are of concern:   The upgrading and widening of a  highway project and potential HSR  Figure 11. Rail and road projects within CICPEC in relation to intact  projects continue to expand near the  frontier landscape and protected areas.   UNESCO World Heritage site Dong  Phayayen‐Khao Yai Forest Complex  (a rare IFL near Bangkok) and the proposed Burma rail Nam Tok‐ Thanbyuzayat project that bisects  several Thai national parks near the Myanmar border.                                                                   16  By definition, an IFL is roadless.  53    APPENDIX 5: Environmental Indicator: Wildlife Effects as Measured by Biodiversity  Hotspots and Umbrella Species  While forest cover change can be a good  indicator of wildlife habitat, animals are not  equally distributed across these areas.  Other, complementary indicators are  needed to identify high‐value biodiversity  habitats. IFLs capture high‐value areas that  serve as critical refuges for many animal  species that require extensive undisturbed  habitat, but there are other species that  might be threatened precisely because  their habitat has been reduced in size and  are not found in IFLs (many of which are  mountainous). Here we introduce two  alternative indicators that can help identify  some of the most high‐value biodiversity  areas that should be prioritized when  mitigating risks created by transportation  corridors. The first indicator, “biodiversity  endemism hotspots,” delineates areas with  high levels of faunal species richness,  species endemism, and species  endangerment.  Li and colleagues (2016)  developed such a metric for continental  Figure 12. BRI roads and rail projects in CICPEC as they  Southeast Asia utilizing data from the IUCN  intersect with hotspots9 of biodiversity endemism for  Red List of Species of Concern for  threatened species (after Li et al. 2017)  mammals, birds, and amphibians,  augmented with existing, publicly available remote sensing data (Figure 12). Overlapping distributions of  species of concern, Li and colleagues prioritized areas for conservation. For the CICPEC corridor, the  highest value, most sensitive areas for birds, mammals, and amphibians mostly occur in mountainous  areas of Southeast Asia due to their high endemism and species richness but can also be found in lower  elevations, especially in northern regions of the Lao People’s Democratic Republic, Vietnam, and  Thailand. The planned BRI rail projects along the central route in Lao PDR clearly would cut directly  through this biodiversity endemism hotspot while the other two routes largely skirt these hotspots.  The distribution of umbrella species can serve as an alternative indicator for identifying high‐value areas  for wildlife. “Umbrella species” are wildlife species that have large habitat needs or other requirements  and whose conservation results in many other species being conserved at the ecosystem or landscape  level. For the CICPEC corridor, large cats such as tigers (Panthera tigris) and clouded leopards (Neofelis  nebulosi ) can serve this role nicely (Figure 13). For many decades, tigers have been threatened due to  the expansion of human development including an extensive road system throughout most of the tigers’  range that created access for poachers and fragmented tiger habitat (Mcmillan 2018; WWF 2016). Due  to overhunting and loss of habitat, fewer than 3,500 tigers lived in the wild, occupying less than 7% of  54    their historical range despite  extensive efforts to set up  protected areas to conserve these  large cats. Tigers are listed as  endangered by the IUCN Red List  (Mcmillan 2018; Walston et al.  2010). The smaller clouded leopard  is also threatened with extinction,  but less so than tigers. The clouded  leopards are listed as vulnerable  because the species has lost an  estimated 30% of its adult  population in the last two decades  due to poaching, habitat loss,  fragmentation, and other human  pressures (Mcmillan 2018). The only  places where BRI projects intersect  with the tiger habitats are the  Burma rail Nam Tok‐Thanbyuzayat  near the Thailand‐Myanmar border  and the Bangkok‐Kuala Lampur HSR  project in Malaysia. The former  project bisects the tiger range,  potentially separating populations  while the latter skirts the side of the  tiger range, thus potentially further  Figure 13. Tiger and clouded leopard ranges and BRI road and rail  restricting tiger habitat. Clouded  infrastructure in CICPEC. Inset show potential habitat corridor that  leopards have a larger range and  could connect two disjunct habitat ranges.  more populations that are more  resilient to human development; their habitats are at risk from BRI transportation projects along all  three corridors. Of greatest concern are the Kunming‐Vientiane railway that would bisect extensive  clouded leopard habitat in Laos and the Kunming‐Mandalay corridor bisecting the cat’s range in  Myanmar.  Which of these two indicators – biodiversity endemism hotspots or umbrella species – is a better tool  for identifying sensitive and high‐value areas when assessing transportation infrastructure risks? The  umbrella species ranges are relatively simple to use because they require the range of only single or  several species. They have the added value of generating public awareness and potentially greater  public support because of the high‐profile nature of these species. However, individual species will  invariably have ranges that are specific to their habitat requirements and threats. The absence of tigers  from Lao PDR, a hub of biodiversity endemism, shows the limitation of using only umbrella species. As a  composite indicator, biodiversity hotspots are more likely to capture areas of high value to a wider  community of species at risk.  Ideally, both indicators would be used, utilizing their comparative  advantages.  55    APPENDIX 6: Illegal logging and wildlife trade  Improving transportation  corridors can contribute  significantly to Illegal logging  and wildlife trade, especially  along the CICPEC corridor. The  Southeast Asia region is both a  source for illegal wildlife and  timber and a hub for  international trafficking  (Broussard 2017; Chouvy 2013;  Felbab‐Brown 2013a). The  threat posed by transportation  infrastructure development is  twofold:  First, new and  improved roads in Southeast  Asia have been opening up new  frontiers for illegal logging and  wildlife poaching (Clements et  al. 2014; Felbab‐Brown 2013b,  2013a). Additionally, these  improved transportation  networks facilitate the  movement of timber and  wildlife (BenYishay et al. 2016;  Felbab‐Brown 2013b, 2013a).  Given that much of the illegal  timber and wildlife trade are  destined for East Asia,  facilitating the movement of  Figure 14. Pangolin trafficking routes and BRI infrastructure in CICPEC.  goods along BRI transportation  Inset shows proximity of proposed BRI Sihanoukville Port Connector  Road to the Bokor National Park, a source of pangolins (after Chouvy  corridors in CICPEC will  2013).  undoubtedly also facilitate  wildlife smuggling. Consider the  particular case of the pangolin,  a species highly endangered primarily because of extensive poaching for the international illegal wildlife  markets. As can be seen in Figure 14, the proposed BRI Sihanoukville Port Road and Rail extensions in  Cambodia are located very close to one of the prime source areas for illegal pangolin poaching, the  nearby Bokor National Park and surrounding forests (Chouvy 2013; Felbab‐Brown 2013a). An upgraded  port link would not only facilitate smuggling access to the current pangolin trafficking routes, it could  potentially open a new maritime hub for smuggling traffic.    56    APPENDIX 7: Environmental Indicator: Indirect Effects as Measured by Forest Cover  Change  Southworth et al. (2011) find evidence for a scarcity‐based regional forest transition in continental  Southeast Asia by comparing Cambodia – the country with the highest proportion of intact frontier  landscape in the region and also the fastest rate of deforestation – with Thailand as well as Nepal and  India – where deforestation appears to have slowed as forest becomes scarcer. Cropper et al. (2001)  also find suggestive evidence that forest transition effects may be present in Thailand where they find  that the roads built farther away from intact forest  fringes have little effect on forest clearing (consistent  with Amazonian evidence). As can be seen in Figure  15, reforestation in northern Thailand and the  contrasting deforestation across the border in the Lao  People’s Democratic Republic provides further  support that the forest transition is underway in  Thailand but Lao PDR and Cambodia are still in the  early stages of deforestation. Meyfroidt and Lambin  (2009) give an account of forest transition in Vietnam,  which is characterized by approximately 40% leakage  to nearby Lao PDR and Cambodia where illegal forest  harvesting is more common. This regional experience  of weaker protection within some countries,  particularly those with weaker states, such as  Cambodia, underscores the role of trade in displacing  deforestation. It could be expected that improved  transportation would increase the salience of this  mechanism as the region becomes more connected.  What does this imply for the proposed BRI  transportation projects?  Three planned projects have  been highlighted in earlier appendices for their high  conservation value: The Nam Tok‐Thanbyuzayat rail  along the western route in Thailand for its extensive  forest cover, IFLs, and big cat habitat; the Lao PDR  section (Boten‐Vientian) of the central  route for its  endemism biodiversity hotspot and extensive forest  cover; and the Cambodian Sihanoukville Port road and  rail extension in the western route for its IFL and role  Figure 15. Rail and road projects within CICPEC  in  in supporting endangered pangolin habitat. While all  relation to cropland, forest cover, forest loss, and  of these deserve careful attention and carefully  forest gain  planned mitigation strategies, the increased pressure  for land use change and deforestation created by BRI transportation network improvements, increased  trade, and urbanization would likely result in more deforestation in the Cambodian and Lao PDR sites.   The Thai and Cambodian sites both have existing protected areas in place, but much greater  enforcement is needed in Bokor National Park to effectively safeguard this area.  57    APPENDIX 8: Mitigation Strategies: Protected Areas  Protected areas can provide an important  tool for mitigating environmental risks from  BRI rail and road projects. Establishment or  increased enforcement of a protected area  near a transportation corridor can help  reduce the potential impact from land use  change from human settlement, illegal timber  extraction, and wildlife poaching. Figure 16  shows that, while there are many protected  areas along existing roads and rail in CICPEC  as well as those under construction or  planned, they are strikingly absent from the  Lao People’s Democratic Republic.17 This  clearly flags the need to consider this as part  of a mitigation strategy for the planned rail  line that will pass through areas of high  endemism for threatened biodiversity.    Establishing a protected area does not  provide automatic safeguard against human  settlement and destructive activities such as  wildlife poaching and illegal logging. A road or  rail bisecting a protected area will create an  edge effect that can penetrate deep into the  Figure 16. Rail and road projects within CICPEC in relation to  protected areas (IUCN Categories I‐IV), forest cover, forest loss,  zone. Moreover, the protected status is only  and forest gain  as strong as its enforcement.  Consider the  case of Snuol Wildlife Reserve in Cambodia.  Landsat imagery reveals that in less than a  decade after a road was constructed passing  directly through the park, extensive clearing  occurred within the park along the road in part  due to poor enforcement (Clements et al. 2014;  Figure 17).   Two of the planned BRI projects – a segment of  the Thai  Nam Tok‐ Thanbyuzayat rail (Figure 10  in Appendix 3) and the Cambodian Sihanoukville  Port road (Figure 11 in Appendix 6) – both pass  Figure 17. Land cover change of Snuol Wildlife Reserve,  Cambodia. Landsat images were obtained for three time  through existing national parks. The degree to  points: when the road was (1) absent (1990), (2) recently  which these parks can provide sufficient  completed (2001), and (3) had existed for some time  (2009). From Clements, et al. 2014.                                                               17  While there are a few areas protected within Lao PDR, none is within the IUCN Categories I‐IV with its higher  levels of protection.  58    deterrence will depend largely on enforcement capacity.      APPENDIX 9: Mitigation Strategies: Using Umbrella Species for Mitigation Planning  Appendix 5 described how umbrella species ranges could be used to help identify high‐value areas for  wildlife, using the example of tiger and clouded leopard ranges. This same information can be used to  help guide the development of mitigation strategies using a mitigation hierarchy framework (Figure 1).  Given the critically endangered status of tigers, their habitats should be considered no‐go or  “avoidance” areas, with no BRI road or rail projects  permitted within a several kilometer buffer of their  range. Clouded leopards, with substantially larger  populations, a lower endangerment status, and more  resilience to human development may allow for  more flexible mitigation strategies. When possible,  BRI rail and road projects should also avoid the  clouded leopard ranges and buffers, but if not,  strategies should be put in place to reduce impacts.  Possible reduction options include designating or  strengthening protected areas that safeguard habitat  affected by the range and creating tunnels or  overpasses for passage around transportation  corridors. Because these reductions in the threat  may not completely eliminate the risk from  transportation infrastructure on the affected  clouded leopard populations, compensatory offset  action can also be considered. One potential offset  Figure 18. Potential habitat corridor that could  might be the creation and protection of habitat  connect two disjunct habitat ranges for clouded  corridors that link disjunct habitat patches within the  leopards.  CICPEC region but are distant from the busy  transportation corridors (Figure 18).       59    APPENDIX 10: Databases of National Environmental Policies  Databases of national environmental regulations, financial incentives, overarching policies and frameworks, and  voluntary approaches  Database   Website  Data Description  Faolex   faolex.fao.org  National laws, regulations and policies relating to food, agriculture,  and renewable natural resources  Ecolex  www.ecolex.org  National environmental legislation, court decisions, and literature  Climate Change  web.law.columbia.edu/climat Information on country‐specific climate change policies, laws, plans,  Laws of the  e‐change/resources/climate‐ and initiatives by sector (climate, energy, air pollution, forest and land  World  change‐laws‐world  use, environmental impact assessments, and adaptation and  resiliency)  New Climate  climatepolicydatabase.org  Information on GHG mitigation policies by sector (electricity and heat,  Policy Database  industry, buildings, transport, and agriculture and forestry)  Legal Atlas  www.legal‐atlas.net/  Dynamic database of national environmental legislation that can be  compared across topics and countries      60    APPENDIX 11: International Agreements      Agreement  Description  Date Signed  Convention on  Treaty established by UNESCO for the classification, conservation, and sustainable use  1971*  wetlands (Ramsar  of wetlands. The convention identifies wetlands of international importance  Convention)   (especially those important for waterfowl) and places restrictions on development at  those sites.  Convention  Treaty to protect natural and cultural heritage through the designation of World  1972*  concerning the  Heritage Sites (WHS). WHS are overseen by UNESCO. Signatory countries must “take  protection of the  the appropriate legal, scientific, technical, administrative and financial measures  world cultural and  necessary for the identification, protection, conservation, presentation and  natural heritage   rehabilitation of this heritage.”  Convention on  International agreement to ensure that international trade of plant and animal  1973*  international trade  species does not threaten their survival. Species covered by CITES are listed in three  in endangered  appendices, with each appendix requiring a different level of protection. Trade of  species of wild  species in Appendix I is prohibited completely, while trade of Appendix II and III  fauna and flora  species is closely regulated but allowed within certain limits and with proper  (CITES)   permitting.  European  AGR defines the E‐road network of routes of strategic importance for international  1975  agreement on  traffic flows within Europe. The agreement sets the standards, including consideration  main international  of the direct and indirect effects of roads and traffic on people, fauna and flora; soils,  traffic arteries  sub‐soils, water, air, microclimate; landscape, physical property and cultural heritage.  (AGR)  EIAs are mentioned as measures to address adverse impacts.   Basel convention  Aims of the Basel convention are threefold, 1) to reduce the generation of hazardous  1989*  on the control of  waste and promote sound management of any hazardous waste that is generated, 2)  transboundary  to restrict transboundary movement of hazardous waste, unless moving the waste  movements of  across borders is for environmentally sound management, and 3) to develop a  hazardous wastes  regulatory system for the transboundary movement of hazardous waste.  and their disposal  (Basel Convention)  The convention of  The treaty was created by the United Nations Economic Commission for Europe. It  1991*  environmental  obligates Parties to conduct Environmental Impact Assessments at an early stage of  impact assessment  planning for certain activities that are expected to cause environmental harms. It also  in a transboundary  requires that States must notify each other when activities are expected to result in  context (Espoo  transboundary environmental harms.  Convention)  United Nations  The objective of the treaty is to stabilize the greenhouse gas concentration of the  1992 (Kyoto  framework  atmosphere to prevent anthropogenic interference with the climate system by  Protocol)*  convention on  limiting country‐level emissions of these gasses. UNFCCC Conferences of the Parties    climate change  are held annually to address current issues of climate change and have resulted in the  2015 ( Paris  (UNFCCC)  Kyoto Protocol and the Paris Climate Agreement.  Agreement)*  Convention on  The treaty has the goals of conserving biodiversity, sustainably using the components  1992*  biological diversity  of biodiversity, and equitably sharing the benefits provided by genetic resources of  (CBD)   biological diversity. Signatories must develop national strategies detailing how the  country will achieve these goals. The 2011‐2020 strategic plan for biodiversity  (created by the CBD) includes the Aichi biodiversity targets, which are measurable  targets for 2020 that will help achieve the convention’s goals.  United Nations  The convention aims to combat desertification and mitigate the effects of drought  1994*  convention to  through sustainable land management. The convention encourages cooperation  combat  between developed and developing countries to enhance knowledge sharing and  desertification  technology transfer for sustainable land management. The treaty calls for national  and regional action programs to implement the convention; these programs call for  specific measures that can be taken to combat desertification.   UNECE convention  The convention establishes rights of the public (individuals and their associations) to  1998  on access to  receive environmental information that is held by public authorities; to participate in  information, public  environmental decision‐making by commenting on and receiving feedback on, for  61    participation in  example, proposals for projects, plans, and programs affecting the environment; and  decision‐making  to challenge public decisions that have been made without respecting the two  and access to  aforementioned rights or environmental law in general.  justice in  environmental  matters (Aarhus  Convention)  * China is a signatory to agreement.      62    APPENDIX 12: Multilateral Development Bank Environmental Policies  MDB  Key Aspects of Relevant Environmental Policy  Year  Introduced  International  The IFC Performance Standards on Environmental and Social Sustainability (PS) have gained  Revised  Finance  recognition as the global best practice standard for assessing and mitigating negative  2012  Corporation  environmental and social outcomes related to large infrastructure projects. The standards  (IFC)  adhere strictly to the mitigation hierarchy, placing high importance on avoidance of impact  Performance  if possible. Especially relevant for the environment risks related to transportation  Standards   infrastructure are Resource Efficiency and Pollution Prevention (PS3) and Biodiversity  Conservation and Sustainable Management of Living Natural Resources (PS6). Each    performance standard has an accompanying guidance note that provides more technical  details about how borrowers should adhere to the PS.  World Bank  In addition to protecting the poor and the environment and ensuring sustainable  Revised  Environmental  development, WB’s Environmental and Social Framework (ESF) addresses, among other  2018  and Social  things, transparency, non‐discrimination, social inclusion, public participation, and  Framework  accountability. The Environmental and Social Standards mirror the IFC’s Performance  Standards very closely.  Asian  ADB's Safeguard Policy Statement (SPS) governing the environmental and social safeguards  Revised  Development  of ADB's operations are a cornerstone of its support to inclusive economic growth and  2009  Bank (ADB)  environmental sustainability in Asia and the Pacific.  The objectives of the SPS are to avoid,  or when avoidance is not possible, to minimize and mitigate adverse project impacts on the     environment and affected people, and to help borrowers strengthen their safeguard  systems and develop the capacity to manage environmental and social risks.  African  AfDB’s Integrated Safeguards System consists of four interrelated components: Integrated  2013  Development  Safeguards Policy Statement (PS), Operational Safeguards (OS), Environmental and Social  Bank (AfDB)  Assessment Procedures. The PS describes common objectives of the Bank’s safeguards and  lays out policy principles. The OS are a set of five safeguard requirements that Bank clients    are expected to meet when addressing social and environmental impacts and risks. The  Impact Assessment Guidance Notes provide technical guidance to the Bank’s borrowers or  clients on standards of sector issues, such as roads or fisheries, or on methodological  approaches clients or borrowers are expected to adopt to meet OS standards. OS3,  Biodiversity, Renewable Resources, and Ecosystem Services, is especially relevant to  addressing environmental risks from BRI transport infrastructure.  European Bank  EBRD’s Environmental and Social Policy (ESP) puts safeguards in place to prevent or  Revised  for  minimize any adverse environmental or social impacts, to improve the project’s efficiency,  2014  Reconstruction   and maximize benefits for the wider community and future generations. ESP outlines how  and  the EBRD will address the environmental and social impacts of its projects by defining the  Development  respective roles and responsibilities of both the Bank and its clients in designing,  (EBRD)   implementing and operating projects; setting a strategic goal to promote projects with high  environmental and social benefits; and mainstreaming environmental and social  sustainability considerations into all its activities.  Asian  The recently established AIIB released the first draft of its Environmental and Social  2016  Infrastructure  Framework in February of 2016, which includes an Environmental and Social Exclusion List— Investment  a list of project types or activities that the bank refuses to finance on environmental or  Bank (AIIB)  social grounds. In many ways, the AIIB Environmental and Social Framework aligns with  similar standards released by other banks, but it also relies heavily on its partners’ standards  (Weiss 2017).   New  The NDB’s Environmental and Social Framework (ESF) includes an environmental and social  2016  Development  policy as well as environmental and social standards (ESS). ESS1, the Environmental and  Bank (NDB)  Social Assessment, is particularly relevant.         63    APPENDIX 13: Chinese Policy Bank Policies  China Policy Banks  Key Aspects of Relevant Environmental Policy  Year  Introduced  Export‐Import Bank  The 2007 Guidelines for Environmental and Social Impact Assessment of China Export  2007; 2015  of China (China  and Import Bank’s Loan Projects requires environmental impact assessments,  Exim Bank)  monitoring, and review of project impacts for all projects before a project gains  approval. When deemed necessary, environmental and social responsibilities may be  included in the loan contract. The Exim Bank also has the right to monitor the client’s  implementation of the mitigation activities (FOE 2016; Leung et al. 2013).   China Development  CDB has transparent sustainable development objectives – including an objective on  2004  Bank (CDB)  environmental protection for climate, ecology, clean energy, and low‐carbon living –  but specific environmental policies and their content are not available to the public  (FOE 2016). In 2006, CDB pledged to abide by the United Nations Global Compact 10  principles in human rights, labour standards, environment, and anti‐corruption. CBD  produced a series of non‐binding frameworks to promote environmentally‐friendly  businesses, including an annual Work Plan for Loans to Reduce Pollution and  Emissions, Guidelines on Environmental Protection Project Development Review, and  Guidelines on Special Loans for Energy Conservation and Emission Reduction (Friends  of the Earth (FOE) 2016; Ren et al. 2017).         64    APPENDIX 14: Chinese Policies  Domain  Specific  Key Aspects of Relevant Environmental Policy  Year  Policies  Introduced  General  Environmental  The original 1989 law provided the impetus for environmental impact  1989/  Environmental  Protection  assessments for construction projects within China. The EIA process was  amended  Laws and  Law of the  administrative, not statutorily. In 2014, China updated the law to include,  2014  Guidelines  People’s  among other provisions, stricter penalties and greater opportunities for  Republic of  public environmental litigation.  China  Law of the  The EIA Law sets forth requirements for several types of strategic  2003  People's  environmental assessments (SEAs) within China.  Under this law, plan  Republic of  environmental Impact assessments (PEIA) are legally required for major  China on  economic development activities, integrated plans (such as land use,  Environmental  regional development, and watershed development).  These PEIAs is used  Impact  to integrate environmental considerations into all phases of the  Assessment  preparation of spatial and land use plans. A second type of strategic  (EIA Law)  environmental assessment are required for sectoral plans (for example,  industry, agriculture, husbandry, forestry, energy, water conservancy,  transportation, and natural resources development) within China (Dusi and  Xi 2009; Wu et al. 2010).  Guidelines on  These voluntary guidelines represent a key policy released in 2013 by  2013  Environmental  MOFCOM and the former MEP (now MEP). These guidelines recommend  Protection for  that companies complete EIAs, develop environmental mitigation  Overseas  measures, and work with local communities to identify potential negative  Investment  environmental and social impacts (Leung et al. 2013).    and  Cooperation  Guidelines for  The China Chamber of Commerce of Metals, Minerals and Chemicals  2014  Social  Importers and Exporters (CCCMC) developed these guidelines to assist  Responsibility  Chinese companies improve their environmental and social performance in  in Outbound  overseas mining investments. Though these guidelines are not strictly  Mining  focused on transportation infrastructure, they represent the first detailed  Investments  voluntary standards for foreign infrastructure investment by Chinese  companies.  Guidelines of  The China International Contractors Association (CHINCA) developed  2017  Sustainable  guidelines to assist Chinese international contractors in building  Infrastructure  infrastructure that meets leading sustainability standards. The guidelines  for Chinese  cover five major phases of infrastructure development: funding, planning  International  and design, building, operation and maintenance, and closure.  Contractors   BRI‐Specific  Vision and  This vision document outlines China’s vision and mission of the BRI,  2015  Guidelines  Actions on  highlighting sustainable development, cooperation between nations, and  Jointly  the importance of free trade. Issues related to the environment are  Building Silk  mentioned only briefly in document, signaling priorities for cooperation in  Road  natural resource investments and industries; cooperation in ecological  Economic Belt  conservation, biodiversity protection, and strategies for climate change;  and 21st  and organizing public interest in ecological protection for the benefit of the  Century  public.  Maritime Silk  Road  Guidance on  The high‐level and non‐binding document outlines a broad range of  2017  Promoting  significant environmental issues that will be addressed under BRI. The  Green Belt  document directs Chinese enterprises engaged in BRI projects to  and Road  “voluntarily obey local environmental protection laws, regulations,  standards, codes, honor environmental and social responsibilities and  release annual environmental reports.” The documents promote a range of  tasks including green supply chains; partnerships with NGOs and research  65    organizations; environmental protection platform cooperation; capacity  building; green technology transfers; sharing of environmental protection  information and big data; and enhancing green guidance for corporate  behavior.   Vision for  This vision document acknowledges that oceans as an ecosystem  2017  Maritime  contributes valuable natural assets. The document provides specific, non‐ Cooperation  binding recommendations for the protection of the marine environment.  under the Belt  The green development section (4.1) highlights ensuring the health of the  and Road  ocean for present and future generations; cooperation among nations to  Initiative  undertake conservation and preservation of marine ecosystem services,  ecosystems, and species; establishing efforts to monitor, evaluate,  preserve, and restore marine and coastal systems; jointly tackling marine  pollution issues; demonstrating low‐carbon development in maritime  sectors; supporting small‐island nations in adapting to climate change; and  strengthening cooperation for international blue carbon programs.   Belt and Road  This cooperation plan was written to fit in with the previous Vision and  2017  Ecological and  Actions on Jointly Building Silk Road Economic Belt and 21st Century  Environmental  Maritime Silk Road, Guidance on Promoting Green Belt and Road, and  Cooperation  China’s 13th Five‐Year Plan. The Environmental cooperation plan promotes  Plan  activities such as cooperation between countries for environmental  protection, applying and establishing green financial instruments,  increasing NGO and think‐tank involvement in environmental planning and  partnerships, strengthening green corporate behavior, and enhancing  green supply chains.        66    APPENDIX 15: Industry and NGO Environmental Standards  Standard  Description  Equator Principles  The Equator Principles (EPs) is a risk management framework for private financial institutions to  determine, assess, and manage environmental and social risk in projects. The EP framework is  modelled closely on the IFC’s performance standards. Currently 94 global financial institutions  from 37 countries have adopted the framework for managing environmental and social risk.   SuRe® Standard  The Standard for Sustainable and Resilient Infrastructure (SuRe®), developed by Global  Infrastructure Basel, is a global voluntary standard which integrates key criteria of sustainability  and resilience into infrastructure development and upgrade, through 14 themes covering 61  criteria across governance, social and environmental factors.   CEEQUAL assessment  CEEQUAL is an evidence‐based sustainability assessment, rating, and awards scheme for civil  rating  engineering, infrastructure, landscaping, and public realm projects.  Envision® rating  The Envision® system provides a set of guidelines to aid and quantify the sustainability of an  system  infrastructure project during the planning and preliminary design phases.  Greenroads® rating  The Greenroads Rating System is measures and manage sustainability on transportation projects  system  using environmental, social, and economic performance measures. Projects are evaluated by an  independent, expert, third‐party review.  Natural Capital  Framework created by the Natural Capital Coalition to provide guidance on how to incorporate  Protocol  human interaction with nature and natural capital into actionable information for business  managers to inform decisions.      67    DATA LINKS/CITATIONS FOR MAPS  All maps were created by Dr. Andrew Jacobson, Duke University, unless otherwise noted.  Cities: Reed, T. and S. Trubetskoy (2018) The Belt and Road Initiative and the Value of Urban Land. World  Bank mimeo.Ports: Reed, T. and S. Trubetskoy (2018) The Belt and Road Initiative and the Value of  Urban Land. World Bank mimeo.  Roads & rails: Reed, T. and S. Trubetskoy (2018) The Belt and Road Initiative and the Value of Urban Land.  World Bank mimeo.Administrative boundaries: GADM v2.8 ‐ https://gadm.org/index.html .    Forest change: Global Forest Watch 2015 v1.3 ‐ https://www.globalforestwatch.org/ .   Hansen, Matthew C., et al. "High-resolution global maps of 21st-century forest cover change." science 342.6160 (2013): 850-853.  Cropland 2015: ESA CCI land cover V2.0 ‐ https://www.esa‐landcover‐cci.org/?q=node/175  Cite – ESA Climate Change Initiative – Land Cover Project 2017   Copyright notice: © ESA Climate Change Initiative ‐ Land Cover project 2017 Should you write any  scientific publication on the results of research activities that use one or several CCI‐LC products as  input, you shall acknowledge the ESA CCI Land Cover project in the text of the publication and provide  the project with an electronic copy of the publication (contact@esalandcover‐cci.org).  Protected areas: UNEP & IUCN. 2018. The World Database on Protected Areas (WDPA). Cambridge: UNEP‐ WCMC. https://protectedplanet.net/  Pangolin hub & trafficking routes: Chouvy, P‐A. 2013. An Atlas of Trafficking in Southeast Asia. London:  I.B. Tauris.  Intact Forest Landscape 2013: Potapov, Peter, et al. "The last frontiers of wilderness: Tracking loss of intact forest landscapes from 2000 to 2013." Science Advances 3.1 (2017): e1600821.  http://www.intactforests.org/    Centers of Endemism: Li BV, Hughes AC, Jenkins CN, Ocampo-Peñuela N, Pimm SL (2016) Remotely Sensed Data Informs Red List Evaluations and Conservation Priorities in Southeast Asia. PLoS ONE 11(8): e0160566. https://doi.org/10.1371/journal.pone.0160566   Species distributions (clouded leopard & tiger): Goodrich, J., Lynam, A., Miquelle, D., Wibisono, H., Kawanishi, K., Pattanavibool, A., Htun, S., Tempa, T., Karki, J., Jhala, Y. & Karanth, U. 2015. Panthera tigris. The IUCN Red List of Threatened Species 2015: e.T15955A50659951.   http://dx.doi.org/10.2305/IUCN.UK.2015-2.RLTS.T15955A50659951.en. Downloaded on 10 September 2018.  Grassman, L., Lynam, A., Mohamad, S., Duckworth, J.W., Bora, J., Wilcox, D., Ghimirey, Y., Reza, A. & Rahman, H. 2016. Neofelis nebulosa. The IUCN Red List of Threatened Species 2016: e.T14519A97215090.   http://dx.doi.org/10.2305/IUCN.UK.2016-1.RLTS.T14519A97215090.en. Downloaded on 10 September 2018.     68    LITERATURE CITED  AfDB. 2015. Payment for Enviromental Services: A Promising Tool For Natural Resources Management In  Africa. Abidjan, Cote d’Ivoire.  Ahmed, S. E. et al. 2014. “Road Networks Predict Human Influence on Amazonian Bird Communities.”  Proceedings of the Royal Society B: Biological Sciences 281(1795): 20141742–20141742.  http://rspb.royalsocietypublishing.org/cgi/doi/10.1098/rspb.2014.1742.  AIIB. 2016. AIIB: Environmental and Social Framework.  Albers, H. J. 2010. “Spatial Modeling of Extraction and Enforcement in Developing Country Protected  Areas.” Resource and Energy Economics 32(2): 165–79.  Alix‐Garcia, Jennifer, Katharine RE Sims, Patricia Yanez‐Pagans, and Elizabeth N Shapiro. 2013. “Only One  Tree from Each Seed? Environmental Effectiveness and Poverty Alleviation in Programs of  Payments Fro Ecosystem Services.” 7(4): 56.  Alpízar, Francisco, Anna Nordén, Alexander Pfaff, and Juan Robalino. 2017a. “Spillovers from Targeting  of Incentives: Exploring Responses to Being Excluded.” Journal of Economic Psychology 59: 87–98.  ———. 2017b. “Unintended Effects of Targeting an Environmental Rebate.” Environmental and  Resource Economics 67(1): 181–202.  Ambani, Mildred M. 2017. “GIS Assessment of Environmental Footprints of Standard Gauge Railway  (SGR) on Nairobi National Park, Kenya.” University of Nairobi, School of Engineering.  Andam, Kwaw S et al. 2008. “Measuring the Effectiveness of Protected Area Networks in Reducing  Deforestation.” Proceedings of the National Academy of Sciences of the United States of America  105(42): 16089–94.  Andersen, Lykke E. 1996. “Causes of Deforestation in the Brazilian Amazon.” Journal of Environment &  Development 5(3): 309–28.  Anderson, Christa M., Christopher B. Field, and Katharine J. Mach. 2017. “Forest Offsets Partner Climate‐ Change Mitigation with Conservation.” Frontiers in Ecology and the Environment 15(7): 359–65.  Andrews, Kimberly M., Tom A. Langen, and Richard P.J.H. Struijk. 2015. “Reptiles: Overlooked but Often  at Risk from Roads.” In Handbook of Road Ecology, eds. Rodney van der Ree, Clara Grilo, and Daniel  J. Smith. West Sussex, U.K.: John Wiley & Sons, Incorporated.  Angelo, Gino J D et al. 2006. “Evaluation of Wildlife Warning Reflectors for Altering White‐Tailed Deer  Behavior Along Roadways.” Wildlife Society … 34(4): 1175–83.  http://onlinelibrary.wiley.com/doi/10.2193/0091‐ 7648(2006)34[1175:EOWWRF]2.0.CO;2/abstract%5Cn%3CGo to ISI%3E://WOS:000243173600036.  Angelsen, Arild. 2007. World Bank Policy Research Working Paper Forest Cover Change in Space and  Time : Combining the von Thünen and Forest Transition Theories.  Armsworth, P. R., G. C. Daily, P. Kareiva, and J. N. Sanchirico. 2006. “Land Market Feedbacks Can  Undermine Biodiversity Conservation.” Proceedings of the National Academy of Sciences 103(14):  5403–8.  Asher, Sam, Teevrat Garg, and Paul Novosad. 2017. The Ecological Footprint of Transportation  69    Infrastructure.  Asher, Sam, and Paul Novosad. 2015. “Market Access and Structural Transformation: Evidence from  Rural Roads in India.”  Babińska‐Werka, Joanna, Dagny Krauze‐Gryz, Michał Wasilewski, and Karolina Jasińska. 2015.  “Effectiveness of an Acoustic Wildlife Warning Device Using Natural Calls to Reduce the Risk of  Train Collisions with Animals.” Transportation Research Part D: Transport and Environment 38: 6– 14.  Barber, Christopher P., Mark A. Cochrane, Carlos M. Souza, and William F. Laurance. 2014. “Roads,  Deforestation, and the Mitigating Effect of Protected Areas in the Amazon.” Biological  Conservation 177: 203–9. http://dx.doi.org/10.1016/j.biocon.2014.07.004.  Baylis, Kathy, Don Fullerton, and Payal Shah. 2016. What Drives Forest Leakage?  Bechtel, Peter, and Sean Nazerali. 2016. A National Biodiversity Offset System: A Road Map for  Mozambique. https://www.profor.info/sites/profor.info/files/MozambiqueOffsetRoadmap_1.pdf.  Benítez‐López, Ana, Rob Alkemade, and Pita A. Verweij. 2010. “The Impacts of Roads and Other  Infrastructure on Mammal and Bird Populations: A Meta‐Analysis.” Biological Conservation 143(6):  1307–16.  Bennett, Genevieve, Melissa Gallant, Contributing Author, and Kerry Ten Kate. 2017. “State of  Biodiversity Mitigation 2017: Markets and Compensation for Global Infrastructure Development.” :  92. http://www.forest‐trends.org/dir/em_newsletter%0Ahttp://www.forest‐ trends.org/dir/em_newsletter%0Ahttp://www.forest‐trends.org/dir/em_newsletter.  Bennett, Genevieve, and Franziska Ruef. 2016. Alliances for Green Infrastructure: State of Watershed  Investment 2016. http://www.forest‐trends.org/documents/files/doc_5463.pdf.  Bennett, M T et al. 2014. “China’s Conversion of Cropland to Forest Program for Household Delivery of  Ecosystem Services: How Important Is a Local Implementation Regime to Survival Rate Outcomes?”  Forests 5(9): 2345–76.  BenYishay, Ariel, Bradley Parks, Daniel Runfola, and Rachel Trichler. 2016. “Forest Cover Impacts of  Chinese Development Projects in Ecologically Sensitive Areas.” AidData Working Paper (December  2016).  Bhattarai, K, D Conway, and M Yousef. 2009. “Determinants of Deforestation in Nepal’s Central  Development Region.” Journal of Environmental Management 91(2): 471–88.  Blackman, Allen. 2015. “Strict versus Mixed‐Use Protected Areas: Guatemala’s Maya Biosphere  Reserve.” Ecological Economics 112: 14–24.  Blackwell, Bradley F., Travis L. DeVault, and Thomas W. Seamens. 2015. “Understanding and Mitigating  the Negative Effects of Road Lighting on Ecosystems.” In Handbook of Road Ecology, eds. Andrew  van der Ree, Daniel J. Smith, and Clara Grilo. West Sussex, U.K.: John Wiley & Sons, Incorporated,  143–50.  Booth, Sarah. 2017. Analysis of Existing Environmental Policy Databases.  Bremer, Leah L. et al. 2016. “One Size Does Not Fit All: Natural Infrastructure Investments within the  Latin American Water Funds Partnership.” Ecosystem Services 17: 217–36.  70    http://dx.doi.org/10.1016/j.ecoser.2015.12.006.  Broussard, Giovanni. 2017. “Building an Effective Criminal Justice Response to Wildlife Trafficking:  Experiences from the ASEAN Region.” Review of European, Comparative and International  Environmental Law 26(2): 118–27.  Brown, Phillip H., and Christopher L. Lant. 1999. “The Effect of Wetland Mitigation Banking on the  Achievement of No‐Net‐ Loss.” Environmental Management 23(3): 333–45.  Bruschi, Daniele, Davide Astiaso Garcia, Franco Gugliermetti, and Fabrizio Cumo. 2015. “Characterizing  the Fragmentation Level of Italian’s National Parks Due to Transportation Infrastructures.”  Transportation Research Part D: Transport and Environment 36: 18–28.  http://dx.doi.org/10.1016/j.trd.2015.02.006.  Bull, Joseph W. et al. 2013. “Biodiversity Offsets in Theory and Practice.” Oryx 47(3): 369–80.  Bulte, Erwin H., Leslie Lipper, Randy Stringer, and David Zilberman. 2008. “Payments for Ecosystem  Services and Poverty Reduction: Concepts, Issues, and Empirical Perspectives.” Environment and  Development Economics 13(3): 245–54.  Business and Biodiversity Offsets Programme (BBOP). 2009. Business, Biodiversity Offsets and BBOP: An  Overview ,. Washington, DC.  ———. 2012. Biodiversity Offset Design Handbook‐ Updated. Washington, DC.  Cader, Masud, Alexander Cantor, Shujie Shao, and Mingqing Liu. Economics of the BRI: Co‐Investment  and Creating Markets along the Belt and Road. Washington, DC.  Caro, Tim, Andrew Dobson, Andrew J. Marshall, and Carlos A. Peres. 2014. “Compromise Solutions  between Conservation and Road Building in the Tropics.” Current Biology 24(16): R722–25.  http://dx.doi.org/10.1016/j.cub.2014.07.007.  Carter, Mason C., R.C. Kellison, and R. Scott Wallinger. 2015. Forestry in the U.S. South. Baton Rouge:  Louisiana State University Press.  CCCMC. 2015. Guidelines for Social Responsibility in Outbound Mining Investments. https://www.emm‐ network.org/wp‐content/uploads/2015/03/CSR‐Guidelines‐2nd‐revision.pdf.  Chai, Jian, Quan Ying Lu, Shou Yang Wang, and Kin Keung Lai. 2016. “Analysis of Road Transportation  Energy Consumption Demand in China.” Transportation Research Part D: Transport and  Environment 48: 112–24. http://dx.doi.org/10.1016/j.trd.2016.08.009.  Chaker, A., K. El‐Fadl, L. Chamas, and B. Hatjian. 2006. “A Review of Strategic Environmental Assessment  in 12 Selected Countries.” Environmental Impact Assessment Review 26(1): 15–56.  Chen, Zhenhua, Junbo Xue, Adam Z. Rose, and Kingsley E. Haynes. 2016. “The Impact of High‐Speed Rail  Investment on Economic and Environmental Change in China: A Dynamic CGE Analysis.”  Transportation Research Part A: Policy and Practice 92: 232–45.  http://dx.doi.org/10.1016/j.tra.2016.08.006.  Chomitz, Kenneth M, and David A Gray. 1996. “Roads, Land Use, and Deforestation: A Spatial Model  Applied to Belize.” The World Bank Economic Review 10(3): 487–512.  Chomitz, Kenneth M, and Timothy S Thomas. 2003. “Determinants of Land Use in Amazônia : A Fine‐ 71    Scale Spatial Analysis.” American Journal of Agricultural Economics 85(4): 1016–28.  Chouvy, Pierre‐Arnaud, ed. 2013. An Atlas of Trafficking in Southeast Asia: The Illegal Trade in Arms,  Drugs, People, Counterfeit Goods and Natural Resources in Mainland Southeast Asia. New York: I.B.  Tauris & Co. Ltd.  Chun, Zhang. 2017. “China Needs Urgent Oversight of Investments.” Chinadialogue.  https://www.chinadialogue.net/article/show/single/en/10258‐China‐needs‐urgent‐oversight‐of‐ investments.  Clements, Gopalasamy Reuben et al. 2014. “Where and How Are Roads Endangering Mammals in  Southeast Asia’s Forests?” PLoS ONE 9(12): 1–26.  Cochrane, Mark A., and William F. Laurance. 2002. “Fire as a Large‐Scale Edge Effect in Amazonian  Forests.” Journal of Tropical Ecology 18(3): 311–25.  Coffin, Alisa W. 2007. “From Roadkill to Road Ecology: A Review of the Ecological Effects of Roads.”  Journal of Transport Geography 15(5): 396–406.  Correa, Juan et al. 2018. Evaluating International Climate Finance at the Local Level: Impacts of the  Olhos D’Agua Project, an Amazon Fund/REDD Initiative.  Cropper, Maureen, and Charles Griffiths. 1994. “The Interaction of Population Growth and  Environmental Quality.” The American Economic Review 84(2).  Cropper, Maureen, Charles Griffiths, and Muthukumara Mani. 1999. “Roads, Population Pressures, and  Deforestation in Thailand, 1976‐1989.” Land Economics 75(1): 58.  Cropper, Maureen, Jyotsna Puri, and Charles Griffiths. 2001. “Predicting the Location of Deforestation:  The Role of Roads and Protected Areas in North Thailand.” Land Economics 77(2): 172.  D’Alfonso, Tiziana, Changmin Jiang, and Valentina Bracaglia. 2016. “Air Transport and High‐Speed Rail  Competition: Environmental Implications and Mitigation Strategies.” Transportation Research Part  A: Policy and Practice 92(635867): 261–76. http://dx.doi.org/10.1016/j.tra.2016.06.009.  D’Angelo, Gino, and Andrew van der Ree. 2015. “Use of Reflectors and Auditory Deterrents to Prevent  Wildlife‐Vehicle Collisions.” In Handbook of Road Ecology, eds. Andrew van der Ree, Daniel J.  Smith, and Clara Grilo. West Sussex, U.K.: John Wiley & Sons, Incorporated, 213–18.  Dalgaard, Carl‐Johan, Nicolai Kaarsen, Ola Olsson, and Pablo Selaya. 2018. “Roman Roads to Prosperity:  Persistence and Non‐Persistence of Public Goods Provision.” University of Copenhagen, mimeo: 1– 50. http://www.econ.ku.dk/pabloselaya/papers/RomanRoads_v9‐6_Feb2018.pdf.  Danyo, Stephen, Susmita Dasgupta, and David Wheeler. 2018. Potential Forest Loss and Biodiversity  Risks from Road Improvement in Lao PDR.  Dasgupta, Susmita, and David Wheeler. 2016. Minimizing Ecological Damage from Road Improvement in  Tropical Forests.  Defries, Ruth S., Thomas Rudel, Maria Uriarte, and Matthew Hansen. 2010. “Deforestation Driven by  Urban Population Growth and Agricultural Trade in the Twenty‐First Century.” Nature Geoscience  3(3): 178–81.  Deininger, Klaus, and Bart Minten. 2002. “Determinants of Deforestation and the Economics of  72    Protection: An Application to Mexico.” American Journal of Agricultural Economics 84(4): 943–60.  Deng, Xiangzheng et al. 2011. “Pressure Cookers or Pressure Valves: Do Roads Lead to Deforestation in  China?” Journal of Environmental Economics and Management 61(1): 79–94.  Dobson, Andrew et al. 2006. “Habitat Loss, Trophic Collapse, and the Decline of Ecosystem Services.”  Ecology 87(8): 1915–24. https://esajournals.onlinelibrary.wiley.com/doi/abs/10.1890/0012‐ 9658%282006%2987%5B1915%3AHLTCAT%5D2.0.CO%3B2.  Dobson, Andrew P. et al. 2010. “Road Will Ruin Serengeti.” Nature 467(7313): 272.  Dollar, David. 2018. 13 Brookings Institution Is China’s Development Finance a Challenge to the  International Order? Washington, DC.  Donaldson, Dave. 2010. NBER Working Paper Series Railroads of the Raj: Estimating the Impact of  Transportation Infrastructure.  Dong, Suocheng et al. 2018. “An Evaluation of the Economic, Social, and Ecological Risks of China‐ Mongolia‐Russia High‐Speed Railway Construction and Policy Suggestions.” Journal of Geographical  Sciences 28(7): 900–918.  Dorsey, Benjamin, Mattias Olsson, and Lisa J. Rew. 2015. “Ecological Effects of Railways on Wildlife.” In  Handbook of Road Ecology, eds. Rodney van der Ree, Daniel J. Smith, and Clara Grilo. West Sussex,  U.K.: John Wiley & Sons, Incorporated, 219–27.  https://ebookcentral.proquest.com/lib/duke/detail.action?docID=1895495.  Dusi, Jiri, and Jian Xi. 2009. Strategic Environmental Assessment and Southeast Asia: A Progress Review  and Comparison of Country Systems and Cases. Washington, DC.  Ekstrom, Jon, Leon Bennun, and Robin Mitchell. 2015. A Cross‐Sector Guide for Implementing the  Mitigation Hierarchy. Cambridge: The Biodiversity Consultancy Ltd.  Engel, Stefanie, Stefano Pagiola, and Sven Wunder. 2008. “Designing Payments for Environmental  Services in Theory and Practice: An Overview of the Issues.” Ecological Economics 65(4): 663–74.  Esty, Daniel C.E., and Michael Porter. 2002. “Regulation and Performance: A Leading Indicator of Future  Competitiveness?” In The Global Competitiveness Report 2001–2002, eds. Michael Porter et al.  New York: Oxford University Press, 78–101.  Federal Transit Administration. 2009. FTA Rail Modernization Study: Report to Congress.  Felbab‐Brown, Vanda. 2013a. “The Illegal Trade in Wildlfie in Southeast Asia and Its Links to East Asian  Markets.” In An Atlas of Trafficking in Southeast Asia, ed. Pierre‐Arnaud Chouvy. New York: I.B.  Tauris & Co. Ltd., 137–54.  ———. 2013b. “The Jagged Edge: Illegal Logging in Southeast Asia.” In An Atlas of Trafficking in  Southeast Asia, New York: I.B. Tauris & Co. Ltd., 113–36.  Ferraro, Paul J. 2008. “Asymmetric Information and Contract Design for Payments for Environmental  Services.” Ecological Economics 65(4): 810–21.  ———. 2009. “Regional Review of Payments for Watershed Services: Sub‐Saharan Africa.” Journal of  Sustainable Forestry 28(3–5): 525–50.  ———. 2013. “More Strictly Protected Areas Are Not Necessarily More Protective: Evidence from  73    Bolivia, Costa Rica, Indonesia, and Thailand.” Environmental Research Letters 8(2).  Ferraro, Paul J., and R. David Simpson. 2002. “The Cost‐Effectiveness of Conservation Payments.” Land  Economics 78(3): 339–53.  Ferretti‐Gallon, Kalifi, and Jonah Busch. 2014. Center for Global Development Working Paper What  Drives Deforestation and What Stops It? A Meta‐Analysis of Spatially Explicit Econometric Studies.  Forman, R. T T. 1998. “Road Ecology: A Solution for the Giant Embracing Us.” Landscape Ecology 13(4).  Forman, Richard T.T., D Sperling, JA Bissonette, and AP Clevenger. 2003. Road Ecology : Science and  Solutions. Washington, DC: Island Press.  Foster, Andrew D, and Mark R Rosenzweig. 2003. “Economic Growth and the Rise of Forests.” The  Quarterly Journal of Economics 118(2): 601–37.  Freitas, Simone R., Todd J. Hawbaker, and Jean Paul Metzger. 2010. “Effects of Roads, Topography, and  Land Use on Forest Cover Dynamics in the Brazilian Atlantic Forest.” Forest Ecology and  Management 259(3): 410–17.  Friends of the Earth (FOE). 2016. Emerging Sustainability Frameworks: China Development Bank and  China Export‐Import Bank. Washington, DC. https://1bps6437gg8c169i0y1drtgz‐wpengine.netdna‐ ssl.com/wp‐content/uploads/wpallimport/files/archive/emerging‐sustainability‐frameworks‐CDB‐ ChinaExim.pdf.  ———. 2017. Investing in a Green Belt and Road? Assessing the Implementation of China’s Green Credit  Guidelines Abroad. Washington, DC. https://1bps6437gg8c169i0y1drtgz‐wpengine.netdna‐ ssl.com/wp‐content/uploads/2017/12/ChinaGreenBelt_Dec‐2017.pdf.  Fyumagwa, R. et al. 2013. “Roads as a Threat to the Serengeti Ecosystem.” Conservation Biology 27(5):  1122–25.  Gallagher, Kelly Sims, and Qi Qi. 2018. Policies Governing China’s Overseas Development Finance  Implications for Climate Change. Boston, MA.  https://sites.tufts.edu/cierp/files/2018/03/CPL_ChinaOverseasDev.pdf.  Geisler, Charles, and Ragendra de Sousa. 2001. “From Refuge to Refugee: The African Case.” Public  Administration and Development 21(2): 159–170.  Geist, Helmut J, and Eric F Lambin. 2001. What Drives Tropical Deforestation?  ———. 2002. “Proximate Causes and Underlying Driving Forces of Tropical Deforestation.” Bioscience  52(2).  Givoni, Moshe. 2006. “Development and Impact of the Modern High‐Speed Train: A Review.” Transport  Reviews 26(5): 593–611.  Global Environmental Institution (GEI). 2018. “How Policy Banks Can Go for Green: Case Study on Green  Finance.” http://www.geichina.org/en/how‐policy‐banks‐can‐go‐for‐green‐case‐study‐on‐green‐ finance/ (October 3, 2018).  Goosem, Miriam. 2015. “Vulnerability and Climatic Conditions: Particular Challenges for Road Planning,  Construction and Maintenance.” In Handbook of Road Ecology, eds. Rodney van der Ree, Daniel J.  Smith, and Clara Grilo. West Sussex, U.K.: John Wiley & Sons, Incorporated, 397–406.  74    Gray, Thomas N.E. et al. 2018. “The Wildlife Snaring Crisis: An Insidious and Pervasive Threat to  Biodiversity in Southeast Asia.” Biodiversity and Conservation 27(4): 1031–37.  Gunn, Jill, and Bram F. Noble. 2011. “Conceptual and Methodological Challenges to Integrating SEA and  Cumulative Effects Assessment.” Environmental Impact Assessment Review 31(2): 154–60.  http://dx.doi.org/10.1016/j.eiar.2009.12.003.  Gunson, Kari E., Giorgos Mountrakis, and Lindi J. Quackenbush. 2011. “Spatial Wildlife‐Vehicle Collision  Models: A Review of Current Work and Its Application to Transportation Mitigation Projects.”  Journal of Environmental Management 92(4): 1074–82.  http://dx.doi.org/10.1016/j.jenvman.2010.11.027.  Haddad, Nick M et al. 2015. “Habitat Fragmentation and Its Lasting Impact on Earth’s Ecosystems.”  Science Advances 1(2): 1–9.  Hamer, Andrew J., Thomas E.S. Langton, and David Lesbarreres. 2015. “Making a Safe Leap Forward:  Mitigating Road Impacts on Amphibeans.” In Handbook of Road Ecology, eds. Rodney van der Ree,  Clara Grilo, and Daniel J. Smith. West Sussex, U.K.: John Wiley & Sons, Incorporated.  Harangozo, Gabor, and Cecilia Szigeti. 2017. “Corporate Carbon Footprint Analysis in Practice – with a  Special Focus on Validity and Reliability Issues.” Journal of Cleaner Production 167: 1177–83.  http://dx.doi.org/10.1016/j.jclepro.2017.07.237.  Hargrave, Jorge, and Krisztina Kis‐Katos. 2013. “Economic Causes of Deforestation in the Brazilian  Amazon: A Panel Data Analysis for the 2000s.” Environmental and Resource Economics 54(4): 471– 94.  Harris, Nancy L et al. 2012. “Baseline Map of Carbon Emissions from Deforestation in Tropical Regions.”  Science 336(June): 1573–76.  Haya, Barbara Kresch. 2010. “Carbon Offsetting: An Efficient Way to Reduce Emissions or to Avoid  Reducing Emissions?” University of California, Berkeley. http://escholarship.org/uc/item/7jk7v95t.  Helsingen, Hanna (WWF‐Myanmar), Ben (University College London) Milligan, Michele (WWF Global  Science) Dailey, and Nirmal (WWF‐US) Bhagabati. 2018. Greening China’s Belt & Road Initiative in  Myanmar.  Herrera, Diego. 2015. “Spillover Conservation around Protected Areas in the Brazilian Amazon: Frontier  Dynamics and Variation in Nearby Deforestation ‘Blockage’ (Reverse Leakage).” Duke University.  Hillman, Jonathan. 2018. ““ China ’ s Belt and Road Initiative : Five Years Later ” A Testimony By :” : 1– 12.  Holmgren, Peter, Edward Juma Masakha, and Håkan Sjöholm. 1994. “Not All African Land Is Being  Degraded: A Recent Survey of Trees on Farms in Kenya Reveals Rapidly Increasing Forest  Resources.” Ambio 23(7): 390–95.  Honey‐Rosés, Jordi, Kathy Baylis, and M. Isabel Ramírez. 2011. “A Spatially Explicit Estimate of Avoided  Forest Loss.” Conservation Biology 25(5): 1032–43.  Hoyos, Laura E, Marcelo R Cabido, and Ana M Cingolani. 2018. “A Multivariate Approach to Study  Drivers of Land‐Cover Changes through Remote Sensing in the Dry Chaco of Argentina.”  International journal of Geo‐Information 7(170): 1–15.  75    Humphrey, Chris. 2016. “Time for a New Approach to Environmental and Social Protection at  Multilateral Development Banks.” (April).  Ibisch, Pierre L. et al. 2016. “A Global Map of Roadless Areas and Their Conservation Status.” Science  354(6318): 1423–27.  ICMM, and IUCN. 2013. Independent Report on Biodiversity Offsets.  https://www.icmm.com/website/publications/pdfs/biodiversity/biodiversity‐offsets%0A.  IFC. 2012. “Guidance Note 6 Biodiversity Conservation and Sustainable Management of Living Natural  Resources.” (2006): 1–70.  http://www.ifc.org/wps/wcm/connect/a359a380498007e9a1b7f3336b93d75f/Updated_GN6‐ 2012.pdf?MOD=AJPERES.  IUCN. 2016. IUCN Policy on Biodiversity Offsets.  http://cmsdata.iucn.org/downloads/iucn_biodiversity_offsets_policy_jan_29_2016.pdf.  Iuell, Bjørn et al. 2003. COST 341‐Habitat Fragmentation Due to Transportation Infrastructure: The  European Review. https://handbookwildlifetraffic.info/handbook‐wildlife‐traffic/.  Jackson, Nathan D., and Lenore Fahrig. 2011. “Relative Effects of Road Mortality and Decreased  Connectivity on Population Genetic Diversity.” Biological Conservation 144(12): 3143–48.  http://dx.doi.org/10.1016/j.biocon.2011.09.010.  Janic, Milan, and Tomislav Jovanović. 2012. “Estimating Some Social and Environmental Effects from  Rail/Road Substitution in the Trans‐European Transport Corridors.” PROMET ‐  Traffic&Transportation 24(4): 335–41.  http://www.fpz.unizg.hr/traffic/index.php/PROMTT/article/view/438.  Jayachandran, Seema et al. 2017. “Cash for Carbon: A Randomized Trial of Payments for Ecosystem  Services to Reduce Deforestation.” Science 357(6348): 267–73.  Johnson, Michael, Rufus Edwards, and Omar Masera. 2010. “Improved Stove Programs Need Robust  Methods to Estimate Carbon Offsets.” Climatic Change 102(3): 641–49.  Joppa, Lucas N., and Alexander Pfaff. 2009. “High and Far: Biases in the Location of Protected Areas.”  PLoS ONE 4(12): 1–6.  ———. 2011. “Global Protected Area Impacts.” Proceedings of the Royal Society B: Biological Sciences  278(1712): 1633–38.  Joppa, Lucas, and Alexander Pfaff. 2010. “Reassessing the Forest Impacts of Protection: The Challenge of  Nonrandom Location and a Corrective Method.” Annals of the New York Academy of Sciences  1185: 135–49.  Joshi, Prathibha, and Kris Beck. 2017. “Environmental Kuznets Curve for Deforestation: Evidence Using  GMM Estimation for OECD and Non‐OECD Regions.” iForest 10(1): 196–203.  Kaczan, David James. 2016. “Can Roads Contribute to Forest Transitions?” 2016.  https://davidkaczan.com/job‐market‐materials/.  Kaczan, David, Alexander Pfaff, Luz Rodriguez, and Elizabeth Shapiro‐Garza. 2017. “Increasing the Impact  of Collective Incentives in Payments for Ecosystem Services.” Journal of Environmental Economics  and Management.  76    Kaul, I., I. Grunberg, and M. Stern. 1999. “Defining Global Public Goods.” In Global Public Goods:  International Cooperation in the 21st Century, Oxford University Press.  Kelly, Peter, and Xuexi Huo. 2013. “Land Retirement and Nonfarm Labor Market Participation: An  Analysis of China’s Sloping Land Conversion Program.” World Development 48: 156–69.  Keskinen, Marko, and Matti Kummu. 2010. ResearchGate Impact Assessment in the Mekong: Review of  Strategic Environmental Assessment (SEA) & Cumulative Impact Assessment (CIA).  http://www.researchgate.net/publication/266415643.  Kiesecker, Joseph M., Holly Copeland, Amy Pocewicz, and Bruce McKenney. 2010. “Development by  Design: Blending Landscapelevel Planning with the Mitigation Hierarchy.” Frontiers in Ecology and  the Environment 8(5): 261–66.  Koop, Gary, and Lise Tole. 1999. “Is There an Environmental Kuznets Curve for Deforestation?” Journal  of Development Economics 58(1): 231–44.  Laurance, Susan G.W., Philip C. Stouffer, and William F. Laurance. 2004. “Effects of Road Clearings on  Movement Patterns of Understory Rainforest Birds in Central Amazonia.” Conservation Biology  18(4): 1099–1109.  Laurance, William. 2017. “The Dark Legacy of China’s Drive for Global Resources.” YalEnvironment360,  Yale School of Forestry & Environmental Studies.  Laurance, William F. et al. 2014. “A Global Strategy for Road Building.” Nature 513: 229–33.  ———. 2015. “Bad Roads, Good Roads.” In Handbook of Road Ecology, eds. Andrew van der Ree, Daniel  J. Smith, and Clara Grilo. West Sussex, U.K.: John Wiley & Sons, Incorporated, 10–15.  ———. 2015. “Reducing the Global Environmental Impacts of Rapid Infrastructure Expansion.” Current  Biology 25(7): R259–62. http://dx.doi.org/10.1016/j.cub.2015.02.050.  Laurance, William F., and Irene Burgués Arrea. 2017. “Roads to Riches or Ruin?” Science 358(6362): 442– 44.  Leblois, Antoine, Olivier Damette, and Julien Wolfersberger. 2017. “What Has Driven Deforestation in  Developing Countries Since the 2000s? Evidence from New Remote‐Sensing Data.” World  Development 92: 82–102.  Ledec, George, Sally Johnson Campos, and Diana Reay Reay. 2016. Biodiversity Offsets: A User Guide.  Washington, DC.  Leung, Denise, Yingzhen Zhao, Tao Hu, and Athena Ballesteros. 2013. World Resources Institute Issue  Brief Environmental and Social Policies in Overseas Investments: Progress and Challenges for China.  papers2://publication/livfe/id/144506.  Li, Binbin V. et al. 2016. “Remotely Sensed Data Informs Red List Evaluations and Conservation Priorities  in Southeast Asia.” PLoS ONE 11(8): 1–18.  Li, M., J. Wu, and X. Deng. 2013. “Identifying Drivers of Land Use Change in China: A Spatial Multinomial  Logit Model Analysis.” Land Economics 89(4): 632–54.  Li, Man et al. 2014. “Impacts of Road Expansion on Deforestation and Biological Carbon Loss in the  Democratic Republic of Congo.” Environmental and Resource Economics 60(3): 433–69.  77    Li, Tianwei et al. 2016. “Strategic Environmental Assessment Performance Factors and Their Interaction:  An Empirical Study in China.” Environmental Impact Assessment Review 59: 55–60.  Li, Xin, Yingling Fan, and Lan Wu. 2017. “CO2emissions and Expansion of Railway, Road, Airline and in‐ Land Waterway Networks over the 1985–2013 Period in China: A Time Series Analysis.”  Transportation Research Part D: Transport and Environment 57(September): 130–40.  http://dx.doi.org/10.1016/j.trd.2017.09.008.  Litvaitis, John A., and Jeffrey P. Tash. 2008. “An Approach toward Understanding Wildlife‐Vehicle  Collisions.” Environmental Management 42(4): 688–97.  Liu, Hong, and Guanie Lim. 2018. “The Political Economy of a Rising China in Southeast Asia: Malaysia’s  Response to the Belt and Road Initiative.” Journal of Contemorary China: 1–16.  Liu, Shiliang et al. 2011. “Effects of Road Network on Vegetation Pattern in Xishuangbanna, Yunnan  Province, Southwest China.” Transportation Research Part D: Transport and Environment 16(8):  591–94. http://dx.doi.org/10.1016/j.trd.2011.08.004.  Locatelli, Bruno, and Raffaele Vignola. 2009. “Managing Watershed Services of Tropical Forests and  Plantations: Can Meta‐Analyses Help?” Forest Ecology and Management 258(9): 1864–70.  Mahmoud, Mahmoud I. et al. 2017. “Alternative Routes for a Proposed Nigerian Superhighway to Limit  Damage to Rare Ecosystems and Wildlife.” Tropical Conservation Science 10.  Maliszewska, Maryla, and Dominique van der Mensbrugghe. In preparation. The Belt and Road Initiative:  Macro and Sectoral Impacts. Washington, DC.  Maurin, Cristelle, and Pichamon Yeophantong. 2013. “Going Global Responsibly? China’s Strategies  Towards ‘Sustainable’ Overseas Investments.” Pacific Affairs 86(2): 281 – 303.  McKenney, Bruce, and Jessica Wilkinson. 2015. Achieving Conservation and Development: 10 Principles  for Applying the Mitigation Hierarchy. https://www.nature.org/ourinitiatives/urgentissues/land‐ conservation/smart‐development/applying‐the‐mitigation‐hierarchy.pdf.  Mcmillan, Madeleine. 2018. “Threats to Big Cats in Southeast Asia.” Duke University.  Mena, Carlos F., Richard E. Bilsborrow, and Michael E. McClain. 2006. “Socioeconomic Drivers of  Deforestation in the Northern Ecuadorian Amazon.” Environmental Management 37(6): 802–15.  Meyfroidt, Patrick, and Eric F Lambin. 2009. “Forest Transition in Vietnam and Displacement of  Deforestation Abroad.” 106(38).  Meyfroidt, Patrick, Thomas K Rudel, and Eric F Lambin. 2010. “Forest Transitions , Trade , and the Global  Displacement of Land Use.”  Miteva, Daniela A., Brian C. Murray, and Subhrendu K. Pattanayak. 2015. “Do Protected Areas Reduce  Blue Carbon Emissions? A Quasi‐Experimental Evaluation of Mangroves in Indonesia.” Ecological  Economics 119: 127–35.  Monteiro, J. A., and J. P. Trachtman. In Press. “Environment‐Related Provisions in Preferential Trade  Agreements.” In Handbook of Deep Trade Agreements, eds. A. Mattoo, Nadia Rocha, and Michele  Ruta.  Moran, Daniel, Ali Hasanbeigi, Cecilia Springer, and Global E Intelligence. 2018. The Carbon Loophole in  78    Climate Policy: Quantifying the Embodied Carbon in Traded Products.  Morelli, Federico et al. 2014. “Can Roads, Railways and Related Structures Have Positive Effects on  Birds? ‐ A Review.” Transportation Research Part D: Transport and Environment 30: 21–31.  http://dx.doi.org/10.1016/j.trd.2014.05.006.  Myczko, Łukasz et al. 2017. “Effects of Local Roads and Car Traffic on the Occurrence Pattern and  Foraging Behaviour of Bats.” Transportation Research Part D: Transport and Environment  56(August): 222–28.  National Rail. 2009. “Comparing Environmental Impact of Conventional and High Speed Rail.” Network  Rail, UK, August, Available at: http://www. ….  http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:Comparing+environmental+imp act+of+conventional+and+high+speed+rail#0.  Nelson, Andrew, and Kenneth M Chomitz. 2011. “Effectiveness of Strict vs . Multiple Use Protected  Areas in Reducing Tropical Forest Fires: A Global Analysis Using Matching Methods.” PLoS ONE  6(8).  Nelson, G. C., and D. Hellerstein. 1997. “Do Roads Cause Deforestation? Using Satellite Images in  Econometric Analysis of Land Use.” American Journal of Agricultural Economics 79(1): 80–88.  New Development Bank. 2016. Environment and Social Framework V4. Shanghai, China.  https://www.ndb.int/wp‐content/uploads/2017/02/ndb‐environment‐social‐framework‐ 20160330.pdf.  Nolte, C., A. Agrawal, K. M. Silvius, and B. S. Soares‐Filho. 2013. “Governance Regime and Location  Influence Avoided Deforestation Success of Protected Areas in the Brazilian Amazon.” Proceedings  of the National Academy of Sciences 110(13): 4956–61.  OECD. 2006. Assessment Applying Strategic Environmental Assessment: Good Practice Guidance for  Development Cooperation. http://www.oecd‐ilibrary.org/development/applying‐strategic‐ environmental‐assessment_9789264026582‐en.  Pagiola, Stefano, Agustin Arcenas, and Gunars Platais. 2005. “Can Payments for Environmental Services  Help Reduce Poverty? An Exploration of the Issues and the Evidence to Date from Latin America.”  World Development 33(2 SPEC. ISS.): 237–53.  Panlasigui, Stephanie et al. 2017. Impacts of Certification, Uncertified Concessions and Protected Areas  on Forest Loss in Cameroon, 2000 to 2013.  Parris, Kirsten. 2015. “Ecological Impacts of Road Noise and Options for Mitigation.” In Handbook of  Road Ecology, eds. Andrew van der Ree, Daniel J. Smith, and Clara Grilo. West Sussex, U.K.: John  Wiley & Sons, Incorporated, 151–58.  Paton, Dale G., Simone Ciuti, Michael Quinn, and Mark S. Boyce. 2017. “Hunting Exacerbates the  Response to Human Disturbance in Large Herbivores While Migrating through a Road Network.”  Ecosphere 8(6).  Pfaff, Alexander et al. 2007. “Road Investments, Spatial Spillovers, and Deforestation in the Brazilian  Amazon.” Journal of Regional Science 47(1): 109–23.  ———. 2009. “Park Location Affects Forest Protection : Land Characteristics Cause Differences in Park  79    Impacts across Costa Rica Park Location Affects Forest Protection : Land Characteristics Cause  Differences in Park Impacts across Costa Rica ∗.” The B. E. Journal of Economic Analysis & Policy  9(2).  ———. 2014. “Governance, Location and Avoided Deforestation from Protected Areas: Greater  Restrictions Can Have Lower Impact, Due to Differences in Location.” World Development 55: 7–20.  ———. 2018. “Roads & SDGs, Tradeoffs and Synergies: Learning from Brazil’s Amazon in Distinguishing  Frontiers.” Economics 12: 1–25.  Pfaff, Alexander, and Juan Robalino. 2012. “Protecting Forests, Biodiversity, and the Climate: Predicting  Policy Impact to Improve Policy Choice.” Oxford Review of Economic Policy 28(1): 164–79.  ———. 2017. “Spillovers from Conservation Programs.” Annual Review of Resource Economics 9(1):  299–315.  Pfaff, Alexander, Juan Robalino, Diego Herrera, and Catalina Sandoval. 2015. “Protected Areas’ Impacts  on Brazilian Amazon Deforestation: Examining Conservation ‐ Development Interactions to Inform  Planning.” PLoS ONE 10(7): 1–17.  Pfaff, Alexander, Juan Robalino, Catalina Sandoval, and Diego Herrera. 2015. “Protected Area Types,  Strategies and Impacts in Brazil ’s Amazon: Public Protected Area Strategies Do Not Yield a  Consistent Ranking of Protected Area Types by Impact.” Philisophical Transactions of the Royal  Society B (370).  Pfaff, Alexander S P. 1999. “What Drives Deforestation in the Brazilian Amazon?” 43: 26–43.  Pfaff, Alexander, Francisco Santiago‐Ávila, and Lucas Joppa. 2017. “Evolving Protected‐Area Impacts in  Mexico: Political Shifts as Suggested by Impact Evaluations.” Forests 8(1): 1–14.  Pfaff, Alexander, and Robert Walker. 2010. “Regional Interdependence and Forest ‘Transitions’:  Substitute Deforestation Limits the Relevance of Local Reversals.” Land Use Policy 27(2): 119–29.  Pilgrim, J. D., and Jon Ekstrom. 2014. Technical Conditions for Positive Outcomes from Biodiversity  Offsets.” An Input Paper for the IUCN Technical Study Group on BiodiversityOffsets.: IUCN. Gland,  Switzerland.  Popp, J. N., and S. P. Boyle. 2017. “Railway Ecology: Underrepresented in Science?” Basic and Applied  Ecology 19: 84–93. http://dx.doi.org/10.1016/j.baae.2016.11.006.  Porras, I, M Greig‐Gran, N Neves, and International Institute for Environment and Development. 2008.  Natural Resources Issues All That Glitters: A Review of Payments for Watershed Services in  Developing Countries. http://www.iied.org/pubs/pdfs/13542IIED.pdf.  Potapov, Peter et al. 2017. “The Last Frontiers of Wilderness: Tracking Loss of Intact Forest Landscapes  from 2000 to 2013.” Science Advances 3(1).  Rainey, Hugo J. et al. 2015. “A Review of Corporate Goals of No Net Loss and Net Positive Impact on  Biodiversity.” Oryx 49(2): 232–38.  van der Ree, Rodney, Jeffrey W. Gagnoy, and Daniel J. Smith. 2015. “Fencing: A Valuable Tool for  Reducing Wildlife‐Vehicle Collisions and Funnelling Fauna to Crossing Structures.” In Handbook of  Road Ecology, eds. Andrew van der Ree, Clara Grilo, and Daniel J. Smith. West Sussex, U.K.: John  Wiley & Sons, Incorporated.  80    Van Der Ree, Rodney, Daniel J. Smith, and Clara Grilo. 2015. Handbook of Road Ecology Handbook of  Road Ecology.  Reed, Tristan, and Alexandr Trubetskoy. 2018. The Belt and Road Initiative and the Value of Urban Land.  Washington, DC.  Ren, Peng, Chang Liu, and Liwen Zhang. 2017. “China’s Involvement in Coal‐Fired Power Projects along  the Belt and Road. Global Environmental Institute.” (May).  Ren, Peng, Liwen Zhang, Rogn Zhu, and Jingwei Zhang. 2017. How China’s Policy Banks Can Support  Sustainable Foreign Investment.  Renaud, Fabrice G., Karen \ Sudmeier‐Rieux, and Marisol Estrella, eds. 2013. The Role of Ecosystems in  Disaster Risk Reduction. Tokyo: United Nations University Press.  http://dx.doi.org/10.1016/j.ijdrr.2015.08.001.  Rico, Jimena et al. 2018. Deforestation Impacts of Conservation & Development in the Peruvian Amazon:  Evaluating Land‐Use Restrictions in Combinations with Rights to Extraction *.  Robalino, Juan A. 2007. “Land Conservation Policies and Income Distribution: Who Bears the Burden of  Our Environmental Efforts?” Environment and Development Economics 12: 521–33.  Robalino, Juan, and Alexander Pfaff. 2013. “Ecopayments and Deforestation in Costa Rica: A Nationwide  Analysis of PSA’s Initial Years.” Land Economics 89(July 2015): 432–48.  Robalino, Juan, Alexander Pfaff, and Laura Villalobos. 2015. “Deforestation Spillovers from Costa Rican  Protected Areas.” Documentos de trabajo U. Costa Rica.  ———. 2017. “Heterogeneous Local Spillovers from Protected Areas in Costa Rica.” Journal of the  Association of Environmental and Resource Economists 4(3): 795–820.  Ruan, Xiaofeng, Feng Qiu, and Miles Dyck. 2016. “The Effects of Environmental and Socioeconomic  Factors on Land‐Use Changes: A Study of Alberta, Canada.” Environmental Monitoring and  Assessment 188(8).  Rudel, Thomas K., Laura Schneider, and Maria Uriarte. 2010. “Forest Transitions: An Introduction.” Land  Use Policy 27(2): 95–97.  Sanchez‐Azofeifa, G. Arturo, Alexander Pfaff, Juan Andres Robalino, and Judson P. Boomhower. 2007.  “Costa Rica’s Payment for Environmental Services Program: Intention, Implementation, and  Impact.” Conservation Biology 21(5): 1165–73.  Sausmikat, Nora et al. 2017. Silk Road Bottom‐up: Regional Perspectives on the ‘Belt and Road Initiative.’  Cologne, Germany.  Sawyer, Hall et al. 2013. “A Framework for Understanding Semi‐Permeable Barrier Effects on Migratory  Ungulates.” Journal of Applied Ecology 50(1): 68–78.  Seiler, Andreas, and J‐O Helldin. 2006. “Mortality in Wildlife Due to Transportation.” In The Ecology of  Transportation: Managing Mobility for the Environment, eds. J. Davenport and J.L. Davenport.  Netherlands: Springer. http://link.springer.com/10.1007/1‐4020‐4504‐2.  Selva, Nuria, Adam Switalski, Stefan Kreft, and Pierre L. Ibisch. 2015. “Why Keep Areas Roadfree? The  Importance of Roadless Areas.” In Handbook of Road Ecology, eds. Andrew van der Ree, Daniel J.  81    Smith, and Clara Grilo. West Sussex, U.K.: John Wiley & Sons, Incorporated, 16–26.  www.wiley.com%5Cgo%5Cvanderree%5Croadecology.  Serebrisky, Thomas et al. 2018. Technical Note IDBG Framework for Planning, Preparing, and Financing  Sustainable Infrastructure Projects: IDB Sustainable Infrastructure Platform.  Shapiro‐Garza, Elizabeth. 2013. “Contesting the Market‐Based Nature of Mexico’s National Payments for  Ecosystem Services Programs: Four Sites of Articulation and Hybridization.” Geoforum 46: 5–15.  http://dx.doi.org/10.1016/j.geoforum.2012.11.018.  Shinn, David H. 2016. “The Environmental Impact of China’s Investment in Africa.” Cornell International  Law Journal 49: 25–67.  Sidle, R. C., M. Ghestem, and A. Stokes. 2014. “Epic Landslide Erosion from Mountain Roads in Yunnan,  China‐Challenges for Sustainable Development.” Natural Hazards and Earth System Sciences  14(11): 3093–3104.  Sills, Erin O et al. 2015. “Estimating the Impacts of Local Policy Innovation: The Synthetic Control Method  Applied to Tropical Deforestation.” PLoS ONE 10(7): 1–15.  Sims, Katharine R E. 2010. “Conservation and Development: Evidence from Thai Protected Areas.”  Journal of Environmental Economics and Management 60(2): 94–114.  Sims, Katharine R E, and Jennifer M. Alix‐Garcia. 2015. “Parks versus PES: Evaluating Direct and  Incentive‐Based Land Conservation in Mexico.” Journal of Environmental Economics and  Management.  Smith, Daniel J., Rodney Van Der Ree, and Carme Rosell. 2015. “Wildlife Crossing Structures: An Effective  Strategy to Restore or Maintain Wildlife Connectivity Across Roads.” In Handbook of Road Ecology,.  Smith, M., D. de Groot, and G. Bergkamp. 2006. Pay. Establishing Payments for Watershed Services.  Gland, Switzerland. https://www.iucn.org/sites/dev/files/import/downloads/pay.pdf.  Soanes, Kylie, and Andrew van der Ree. 2015. “Reducing Road Impacts on Tree Dwelling Animals.” In  Handbook of Road Ecology, eds. Andrew van der Ree, Daniel J. Smith, and Clara Grilo. John Wiley &  Sons, Incorporated, 334–40.  Song, X. P., C. Huang, and J. R. Townshend. 2014. “An Integrated Framework for Evaluating the Effects of  Deforestation on Ecosystem Services.” IOP Conference Series: Earth and Environmental Science  17(1).  Southworth, Jane, Harini Nagendra, and Lin Cassidy. 2011. “Forest Transition Pathways in Asia–studies  from Nepal, India, Thailand, and Cambodia.” Journal of Land Use Sciences 4: 1–15.  de Soyres, François et al. 2018. How Much Will the Belt and Road Initiative Reduce Shipment Times and  Trade Costs? Washington, DC.  Stibig, H. J. et al. 2014. “Change in Tropical Forest Cover of Southeast Asia from 1990 to 2010.”  Biogeosciences 11(2): 247–58.  Stibig, H, F Stolle, R Dennis, and C Feldkötter. 2007. JRC Scientific and Technical Reports, Forest Cover  Change in Southeast Asia–the Regional Pattern.  http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:Forest+Cover+Change+in+South east+Asia+‐+The+Regional+Pattern+‐#0.  82    Symes, William S., Madhu Rao, Michael B. Mascia, and L. Roman Carrasco. 2015. “Why Do We Lose  Protected Areas? Factors Influencing Protected Area Downgrading, Downsizing and Degazettement  in the Tropics and Subtropics.” Global Change Biology 22(2).  The World Bank. 2017. “Environmental and Social Framework. Second Draft.” : 141.  The World Bank Group. 2015. A National Biodiversity Offset Scheme: A Road Map for Liberia’s Mining  Sector. Washington, DC.  TNC. 2016. Capacity Building for Mongolian Ministry of Environment and Green Development (MEGDT)  in Relation to Biodiversity and Conservation in the Southern Gobi Desert: Final Report.  UN Environment. 2018a. Assessing Environmental Impacts ‐ A Global Review of Legislation. Nairobi,  Kenya. https://www.unep‐ wcmc.org/system/dataset_file_fields/files/000/000/494/original/Environmental_Impacts_Legislati on.pdf?1518007077%0Ahttp://web.unep.org/publications/.  ———. 2018b. 14 Mainstreaming Biodiversity in the Infrastructure Sector : Fostering System‐Level  Approaches.  UNOPS. 2017. Evidence‐Based Infrastructure: NISMOD‐International.  Vessali, Kaveh V. 1996. “Land Use Impacts of Rapid Transit: A Review of the Empirical Literature.”  Berkeley Planning Journal 11(1).  Viana, Vm, and Mc Cenamo. 2008. “Railroads in the Amazon: A Key Strategy for Reducing  Deforestation.” Carbon & Climate L. Rev … 1: 290–97. http://heinonlinebackup.com/hol‐cgi‐ bin/get_pdf.cgi?handle=hein.journals/cclr2§ion=52.  Walston, Joe et al. 2010. “Bringing the Tiger Back from the Brink‐the Six Percent Solution.” PLoS Biology  8(9): 6–9.  Wang, Peishen, Ning Yang, and Juan D Quintero. 2012. China : The Environmental Challenge of Railway  Development.  Watkins, Graham. 2014. No. IDB‐TN Inter‐American Development Bank TECHNICAL NOTE Approaches to  the Assessment and Implementation of Sustainable Infrastructure Projects in Latin American and  the Caribbean.  Watson, James E.M. et al. 2018. “The Exceptional Value of Intact Forest Ecosystems.” Nature Ecology  and Evolution 2(4).  Weiss, Martin A. 2017. Congressional Research Service Asian Infrastructure Investment Bank (AIIB).  https://fas.org/sgp/crs/row/R44754.pdf.  Weller, Cameron. 2015. “Construction of Roads and Wildlife Mitigation Measures: Pitfalls and  Opportunities.” In Handbook of Road Ecology, eds. Andrew van der Ree, Daniel J. Smith, and Clara  Grilo. West Sussex, U.K.: John Wiley & Sons, Incorporated, 60–64.  Westin, Jonas, and Per Kågeson. 2012. “Can High Speed Rail Offset Its Embedded Emissions?”  Transportation Research Part D: Transport and Environment 17(1): 1–7.  http://dx.doi.org/10.1016/j.trd.2011.09.006.  Wingard, J. et al. 2014. Guidelines for Addressing the Impact of Linear Infrastructure on Large Migratory  83    Mammals in Central Asia. Convention on the Conservation of Migratory Species of Wild Animals  (CMS) Technical Report.  World Bank. 2012. “Strategic Environmental Assessment in the World Bank: Learning from Recent  Experiences and Challenges.” : 103.  http://documents.worldbank.org/curated/en/729811468331017746/pdf/728950ESW0whit0200E NV0SEA0pub0final.pdf.  World Bank Technical Assistance Program. 2009. Strengthening China’s Environmental Protection  Administrative System: Analysis and Recommendations. Washington, DC.  Wu, Jing et al. 2010. “Strategic Environmental Assessment Implementation in China ‐ Five‐Year Review  and Prospects.” Environmental Impact Assessment Review 31(1): 77–84.  Wunder, Sven. 2007. “The Efficiency of Payments for Environmental Services in Tropical Conservation.”  Conservation Biology 21(1): 48–58.  WWF. 2016. The Road Ahead: Protecting Tigers from Asia’s Infrastructure Development Boom. Gland,  Switzerland.  Xu, Jintao, Ran Tao, Zhigang Xu, and Michael T Bennett. 2010. “China’s Sloping Land Conversion  Program: Does Expansion Equal Success?” Land economics 86(2): 219–44.  Xu, Zhigang, M T Bennett, Ran Tao, and Jintao Xu. 2004. “China’s Sloping Land Conversion Programme  Four Years on: Current Situation and Pending Issues.” International Forestry Review 6(3–4): 317– 26.  Yu, He et al. 2017. “Effects of the Qinghai‐Tibet Railway on the Landscape Genetics of the Endangered  Przewalski’s Gazelle (Procapra Przewalskii).” Scientific Reports 7(1): 1–14.  Zeng, Zhenzhong et al. 2018. “Highland Cropland Expansion and Forest Loss in Southeast Asia in the  Twenty‐First Century.” Nature Geoscience 11(8): 556–62. http://dx.doi.org/10.1038/s41561‐018‐ 0166‐9.  Zhou, Lihuan et al. 2018. Moving the Green Belt and Road Initiative: From Words to Actions.    84