Border Fences and Their Impacts on Large Carnivores, Large Herbivores and Biodiversity: An International Wildlife Law Perspective

Fences, walls and other barriers are proliferating along international borders on a global scale. These border fences not only affect people, but can also have unintended but important consequences for wildlife, inter alia by curtailing migrations and other movements, by fragmenting populations and by causing direct mortality, for instance through entanglement. Large carnivores and large herbivores are especially vulnerable to these impacts. This article analyses the various impacts of border fences on wildlife around the world from a law and policy perspective, focusing on international wildlife law in particular. Relevant provisions from a range of global and regional legal instruments are identified and analysed, with special attention for the Bonn Convention on Migratory Species and the European Union Habitats Directive.


INTRODUCTION
The last few decades have witnessed a proliferation of security and other fences along jurisdictional boundaries across the globe, culminating in the recent, hasty construction of border fences by several European countries to stem refugee flows. 1 Besides their intended consequences for people, it is becoming increasingly clear that border fences have unintended consequences for wildlife as well. Such fences can inter alia curtail animals' mobility, fragment populations and cause direct mortality. Large carnivores and large herbivores are especially vulnerable to these impacts.
This article analyses the various impacts of border fences on the natural environment from an international wildlife law and policy perspective. First, it highlights and illustrates the surprisingly severe and pervasive impacts of border fences on wildlife around the world. This exercise reveals that, from a wildlife law and policy point of view, border fencing is all but a marginal issue, despite the scant attention paid to this topic in the scholarly literature so far. Subsequently, the article identifies and discusses a range of relevant international legal instruments, paying attention both to legally binding obligations and their interpretation, and to non-binding guidance. Separate, detailed analyses are devoted to two particularly significant regimes, namely the legal framework of the Bonn Convention on Migratory Species (CMS) 2 and its subsidiary instruments, and the European Union (EU) Habitats Directive. 3 The article's final section contains concluding observations and recommendations.

BORDER FENCING AS A WILDLIFE LAW AND POLICY ISSUE
The construction of barriers along borders is a longstanding practice, from the Great Wall of China and Hadrian's Wall to the present day. Besides the highprofile fences along the United States (US)/Mexico, North/South Korea and Israel/West Bank borders, fences currently separate Malaysia and Thailand, India and Pakistan, Iran and Iraq, China and Mongolia, and Botswana and Zimbabwe, to provide a small sample. Whereas the global amount of border fences briefly stagnated in the years following the fall of the Berlin Wall, the terrorist attacks of 11 September 2001 heralded a stark increase of border fence construction projects that continues to date. 4 Most recently, hundreds of kilometres of fences were hastily erected along external and internal EU borders to stem refugee flows. A recent estimate puts the total length of border fences in Eurasia alone (not counting the Middle East) in the order of a staggering 30,000 km. 5 *Corresponding author. Email: a.trouwborst@tilburguniversity.edu 1 The term 'fence' is used in this article in a broad sense, as encompassing fences in a narrow sense as well as walls and other artificial barriers.
International law does not preclude States from constructing border fences on their own territories, save when said construction would under the circumstances violate particular international obligations of the State involved, 6 for instance under human rights law, migration law or indeed international wildlife law. The latter field is aimed at the conservation of wild flora and fauna, the ecosystems they compose and biodiversity at large. The relevance of this legal discipline to border fences might at first sight seem to be marginal at best. Yet, a recent increase in attention for the impacts of border fences on the natural world in the conservation biology literature warrants a serious examination of the role reserved for international wildlife law in this regard.
Usually, border fences are built to keep people out, such as armed forces, terrorists, drug smugglers, economic migrants and refugees. Sometimes, they are built to keep people in as well, as with the Iron Curtain. A fence may, furthermore, be intended to mark a border or to reinforce a territorial claim. Only in exceptional cases have border fences been intended to impede the movement of (wild or domesticated) non-human creatures. Reportedly, in 2008 the authorities of a Chinese district of the Inner Mongolia region erected a 100 km fence along the border with Mongolia in order to stop wolves (Canis lupus) from crossing over into China and devouring local livestock. 7 Likewise, Botswana's primary stated purpose for erecting a 500 km fence along the Zimbabwean border in 2003 was to keep out cattle infected with foot and mouth disease, 8 repeating prior, smaller-scale veterinary cordon fencing projects along the Namibian border. At any rate, in almost all cases, the impacts of border fences on biodiversity are unintended by-products. These impacts, however unintended, can be significant. Border fences can block or hamper animals' movements and can also injure or kill animals attempting to cross. The various types of impacts are concisely discussed below.
Of course, regular (non-border) fences and other linear infrastructure such as highways, railroads, pipelines, cables and canals, can also hinder or harm wildlife. 9 For instance, the world's longest fence is not a border fence: the 5,614-km-long Dingo Fence built across Australia in the nineteenth century to protect domestic sheep from predation by dingoes (Canis dingo). 10 Besides, many fences around the globe have been erected expressly for wildlife conservation purposes, for instance to keep poachers outside and animals safely inside of protected areas. 11 Nevertheless, the present article focuses exclusively on border fences, for various reasons. For highways, railroads and similar infrastructural projects, impacts on biodiversity are often considered as part of the planning process, and mitigated through measures like animal crossing structures or wildlife-friendly fence design. For border fences, things tend to be different. Whereas they, too, may cut through wildlife habitat over huge distances, border fences are meant to be impenetrablefor people, to be sure, but as a consequence also for many animals, especially large-bodied onesand wildlife overpasses are generally incompatible with this purpose. In addition, the construction of border fences tends to be motivated by security concerns that are considered paramount over most other considerations. This means that any potential impacts on wildlife may not be contemplated in decision making, or simply be taken for granted. Thus, national environmental legislation which might impede or delay a border fence's construction may be inapplicable, overruled or just ignored, to the effect that no environmental impact assessment (EIA) or strategic environmental assessment (SEA) is made and protected species legislation is not applied. A striking example is offered by US federal legislation adopted in 2005, which sidelines all environmental laws, such as the Endangered Species Act (ESA), which might interfere with the construction of the Mexican border fence: 'The Secretary of Homeland Security shall have the authority to waive, and shall waive, all laws such Secretary, in such Secretary's sole discretion, determines necessary to ensure expeditious construction of the barriers and roads under this section.' 12 Added impetus for writing this article was provided by the recent flurry of border fence construction in Europe in response to the influx of refugees from Syria, Iraq and Afghanistan, and other migrants. Early 2016, scientists raised the alarm concerning the consequences of these refugee fences for biodiversity in the journal Nature, and subsequently conducted a more comprehensive review of the available knowledge regarding the implications of border fences for wildlife conservation across Eurasia. 13 Naturally, the particular features of each border fence determine its permeability and overall consequences for different species of wildlife. These features vary widely. Whereas man-made border walls of ice remain the stuff of fiction for the time being, 14 it seems that most other eligible materials have actually been put to use to solidify boundaries in one way or another. Existing border fences include (combinations of) chain link, barbed or razor wire fences, electrified fences, steel fences, concrete walls, sand walls (e.g., Morocco/ Western Sahara), mud walls (e.g., Pakistan/Afghanistan), trenches and even underground metal walls (e.g., Egypt/Gaza Strip). Common accessories include roads, floodlights, human guards, dogs and landmines. Besides its design, the biodiversity impacts of a border fence depend on other variables as well, which obviously include the distribution and characteristics of the species present.
The potential impacts of border fences on wildlife are various. 15 Foremostly, border fences impede mobility. Fences can either be literal barriers, when animals are unable to cross or go around them, or functional barriers, when they (due to physical obstacles and/or a deterrent effect of their appearance) reduce the frequency of animal movements, even if some individuals may occasionally get across. 16 Whereas these effects are easily imagined for large mammals like bears or deer, both barrier effects may also affect smaller creatures like reptiles, insects and even birds. 17 This way, border fences fragment habitats, split populations, cause genetic isolation and alter behaviours that may be important to the longterm survival of the populations or species involved. 18 Examples of such behaviours include seasonal migrations, nomadic movements in search of food or water, and the dispersal of adolescent animals from their birth grounds in search of territories and partners of their own. For instance, when migratory or nomadic journeys are cut short by a border fence, the animals involved may weaken or die through starvation or dehydration. Border fences may thus cause gradual population reductions or even quite sudden mass mortality. 19 Climate change has made the importance of mobility for wildlife all the greater, as the distributions of many species, including sedentary ones, are shifting in response to changing climatic conditions. 20 Another way in which border fences may indirectly contribute to increased wildlife mortality is through the ready supply of wire they may offer, which can be turned into snares by poachers. 21 Poachers may also take advantage of border fences when pursuing animals, by chasing and trapping fast-moving animals against the fence. 22 Border fences can even give natural predators an unfair advantage, as animals like African wild dogs (Lycaon pictus), wolves and coyotes (Canis latrans) can also learn to improve their hunting success by chasing prey into fences. 23 Furthermore, and significantly, border fences may injure or kill animals directly, when they get entangled in barbed or razor wire, get electrocuted or step on landmines.
All in all, border fences can cause declines and even local disappearance of species. Impacts appear to be most severe for large animals, both large carnivores with their low-density occurrences, huge home ranges and long-distance dispersaland large herbivores, especially those inclined to traveling far and wide. 24 This, in turn, can have further effects rippling through ecosystems, especially given the influential roles of large carnivores and large herbivores in the greater scheme of things. 25 Border fences may also influence plant life by affecting processes like seed dispersal and pollination. Indeed, the ancient Great Wall of China appears to have caused notable genetic differences between plant species on either side. 26  3 clearly suggests that the sum of border fences' impacts on wildlife across the globe is overwhelmingly negative, it should be noted that border fences may occasionally provide unintended benefits for wildlife as well. For instance, the no-entry security zones along fences can provide quiet havens for wildlife, and border fences can prevent animals from roaming into countries with higher levels of poaching or prevent poachers from roaming into countries with higher levels of wildlife. 27 Botswana's veterinary border fences have had adverse consequences for many wild animal populations, including giraffe (Giraffa camelopardalis), African elephant (Loxodonta africana), steppe zebra (Equus quagga) and many species of antelope, both by ensnaring animals and by cutting them off from vital resources. 28 Fences erected in 1996 at the Namibian border, along the entire West Caprivi Strip (190 km), pose a clear example, as they closed off crucial migratory routes. 29 During a brief inventory in 1997, the following animals were encountered dead along the fence itself, whereby it should be realized that these merely constitute the tip of the iceberg of the fence's overall toll: five giraffes, one elephant, two elands (Taurotragus oryx), one roan antelope (Hippotragus equinus), one sable antelope (Hippotragus niger), five kudus (Tragelaphus strepsiceros) and several smaller antelope. 30 In Asia, border fences have proliferated right across the continent. These fences are a particular concern in Central Asia, which has been dubbed the 'Serengeti of the North', 31 as it is still home to a range of large migratory and nomadic mammal species. By splitting populations, impeding migrations and killing animals attempting to cross, border fences pose an actual or potential threat to many of these, including the Mongolian gazelle (Procapra gutturosa), saiga antelope (Saiga tatarica), Asiatic wild ass (Equus hemionus, also known as khulan), Bactrian camel (Camelus ferus), argali sheep (Ovis ammon) and snow leopard (Panthera uncia). 32 Illustrations include statistics showing the stark impact of the Kazakhstan/Uzbekistan border fence on a transboundary saiga population; 33 GPS data showing that fences along the Mongolian/Chinese border have effectively separated the remaining and dramatically declining herds of Asiatic wild ass into distinct subpopulations on either side of the border; 34 and photos of Mongolian gazelles hopelessly caught up in barbed wire during attempts to negotiate a Mongolian/Russian border fence. 35 The following example concerns a more surprising way in which border fences can diminish animals' chances of survival. An increase in fatal attacks on people by Asiatic black bears (Ursus thibetanus) and leopards (Panthera pardus) in the Kashmir region has been linked to the border fence constructed by India along its contested border with Pakistan, which has disrupted the predators' wandering patterns and diminished their access to natural prey, ultimately causing them to enter villages and target humans. 36 Needless to say, this situation has increasedand quite understandably sothe likelihood of bears and leopards being killed by local people, whether to prevent further attacks or in retaliation. As for a positive effect of Researchers attribute this to a combined effect of the border fence and more effective legal protection in Israel. 37 The mixed effects of some border fences are particularly well illustrated with reference to yet another gazelle from the region. Conservationists have been struggling to determine whether the various border security fences constructed along Israel's borders are a good or a bad thing for the endangered Israeli gazelle (Gazella gazella). To be sure, the barriers impede the animals' cross-border mobility, degrade its habitat and fragment its populations further. 38 Nonetheless, these adverse impacts are overshadowed by the safety the barriers offer the gazelles from their gravest threat, namely death at the hands of Palestinian hunters. 39 As most surviving gazelles remain on Israeli territory, the species' advocates would actually welcome the closing of remaining gaps in the West Bank separation barrier. 40 In Europe, the security fences that have stood along the western borders of the Russian Federation and Belarus for decades have apparently left their marks on populations of wolf, brown bear (Ursus arctos), Eurasian lynx (Lynx lynx) and European bison (Bison bonasus) in the region, primarily through fragmentation. 41 In Central Europe, since the Iron Curtain was taken down a quarter of a century ago, wolves and other large carnivores have found their way across the old fence lines again. However, the Iron Curtain also serves as a reminder that the effects of border fences can linger for generations, not only for humans but for animals as well. As a recent GPS-tracking study of red deer (Cervus elaphus) living on both sides of the border between Germany and the Czech Republic demonstrates, a stunning 25 years after the complete removal of the electrified border fences, the deer still do not cross the boundary. 42 This is particularly fascinating when realizing that none of the deer alive today have ever seen the fence, as red deer do not tend to live beyond 15 years. Mother deer apparently still teach their fawns that the old fence line is a no-go area. 43 Meanwhile, the brand new razor wire fences erected by Slovenia along the Croatian border create fresh challenges. Significantly, they bisect the transboundary Dinaric-Balkan populations of wolf, lynx and brown bear. 44 Besides, dead deer have been found entangled and badly damaged in the same fences' coiled wires, amidst evidence of prolonged struggles, the rather gruesome images of which have caused a public outcry. 45 Ample footage, to end with the pre-eminent American example, is also available of all manner of creaturesincluding cougars (Puma concolor), mule deer (Odocoileus hemionus), roadrunners (Geococcyx californianus), snakes, lizards and frogswhich have been stopped in their tracks by the US/Mexico fence. 46 This border fence is believed to prejudice myriad other species besides, such as the fragile jaguar (Panthera onca) population in the region. 47

PREVENTION, MITIGATION AND THE TRANSBOUNDARY PARADIGM
To summarize, border fences can have an array of impacts on wild fauna and flora, at individual, demographic and genetic levels, and ultimately on ecosystems and overall biodiversity. This state of affairs places the border fence topic squarely within the field of wildlife law and policy. For instance, according to the strategic 'Aichi Targets' adopted by the parties to the Convention on Biological Diversity (CBD), 48 by 2020 the 'degradation and fragmentation' of natural habitats must have been 'significantly reduced', and the conservation status of threatened species improved. 49 Moreover, the urgent need to address the adverse impacts of international border fences on wildlife conservation, as reviewed above, is a textbook example where transboundary, intergovernmental cooperation is of the essence. Hitherto, however, the wildlife dimension of border fences has hardly received any attention in international law scholarship, 50 in contrast with the implications of such fences in terms of humanitarian law, human rights, migration law and the law of territory, which have been more extensively addressed in the scholarly literature, 51 and even by the International Court of Justice. 52 As indicated above, the present article is intended as a contribution to filling this gap. Incidentally, whereas the focus of this article is on wildlife conservationconcerned with the fate of populations, species and ecosystems rather than individual animalsit should be noted that border fences clearly raise distinct animal welfare issues as well.
It is important to note that, depending on the circumstances, some of the adverse impacts of border fences can in principle be prevented, eliminated or mitigated. A striking example is the incidental, temporary removal of some sections of border fence in Central Asia to enable seasonal migratory movements of Mongolian gazelle and saiga antelope herds. 53 Another instance concerns the aforementioned Botswana/Namibia Caprivi veterinary fence, as a 30-km stretch of it was eventually removed in order to restore the most critical portion of the affected wildlife corridor. 54 Besides such temporary or partial fence removal, potential mitigation measures include animal-friendly fence design; 55 careful planning of fence routing in the landscape; the compensation of inaccessible resources by providing, for instance, for artificial waterholes; the creation of wildlife crossing structures where compatible with security requirements; the adjustment of species conservation and management plans to reflect the population isolation caused by fences; and the translocation of individuals as a form of assisted dispersal to counter genetic fragmentation. 56 Generally speaking, conducting an EIA when border fence construction is contemplated would evidently be conducive to the prevention and mitigation of unwanted impacts on wildlife. By way of a final illustration, court rulings against the Israeli Ministry of Defence concerning the wildlife impacts of the West Bank barrier's construction have recently led to the implementation of mitigation measures such as special underpasses for small animals. 57 Evidently, a potentially crucial role is reserved for law in the present context, as it can require and/or facilitate any such preventive and mitigation measures.
In the main, many international legal instrumentsin their binding provisions and/or in subsequent decisions adopted by their partiesemphasize the need to prevent and mitigate the fragmentation of wildlife populations and habitats and to ensure the maintenance or restoration of adequate connectivity, for instance through well-connected protected area networks extending across international frontiers. 58 Correspondingly, States have increasingly embraced the notion of shared responsibility for the conservation of transboundary natural areas and species, 59 including through joint site designation under treaties like the Ramsar Wetlands Convention 60 and the UNESCO World Heritage Convention; 61 through the designation of Transfrontier Conservation Areas (TFCAs) within the context of the Southern African Development   63 Likewise, transboundary population-level management has been gaining acceptance, including in the context of international legal instruments, 64 as an overarching conservation paradigm for wildlife (sub)populations shared by several countriessuch as large carnivores in Europe 65 and elephants in southern Africa 66so as to adjust the planning and implementation of conservation and management measures to the contours of transboundary animal (sub) populations rather than national jurisdictions. The recent worldwide increase in border fencing has put a huge spanner in the works of many transboundary conservation efforts. As noted by Linnell et al., it is 'somewhat ironic' that during the last 15 years or so, while transboundary conservation paradigms were taking centre stage in wildlife law and policy, the 'global trend has been for an unprecedented increase in the unintended prevention of wildlife from moving across borders'. 67

A GLOBAL INVENTORY OF INTERNATIONAL WILDLIFE LAW OBLIGATIONS VIS-A-VIS BORDER FENCES
At any rate, numerous obligations under environmental agreements are of apparent relevance to the effects of border fences on wildlife conservation. The same is true of the international customary law obligation of all States to 'ensure that activities within their jurisdiction and control respect the environment of other States'. 68 Below, a global (but not necessarily exhaustive) list is provided of binding provisions in global and regional legal instruments which, depending on the circumstances, may be infringed through the construction of border fences: Each Party shall take measures in its national territory with the objective to improve and ensure continuity and connectivity of natural and semi-natural habitats in the Carpathians, thus allowing dispersal and migration of wild species populations particularly of large carnivores, and genetic exchange between such populations. 84 In a case where the natural habitat of the endangered species is located on both sides of the state border between the Parties, such concerned Parties shall cooperate on ensuring the conservation and, as may be necessary, recovery of those species and their natural habitats. 85 Whether constructing or retaining a particular fence is at odds with any of the international obligations laid down in the listed provisions or elsewhere depends on a range of variables. These include whether the country or countries involved are bound by relevant legal instruments in the first place; the extent to which (potentially) affected species or sites are covered by these instruments; the phrasing of the specific provisions involved; and the particular features of the fence in question and its (expected) impacts. 86 It should be borne in mind that the provisions in question ought to be interpreted in light of the treaties' overall objectives and relevant subsequent decisions by the parties. 87 Thus, to illustrate, in a case involving border fencing, the interpretation of the listed provisions of the Alpine Biodiversity Protocol is informed by the subsequently recorded resolve of contracting parties to 'preserve and restore wildlife as wildlife to the extent possible by assuring their free movement in space and time'; and to 'preserve and connect wildlife habitats and ensure the permeability of the landscape'. 88 Likewise, the indicated provisions of the Bern Convention must be understood and applied in view of relevant guidance recorded in the Recommendations adopted over the years by the Convention's Standing Committee. One of these commits parties to the following course of action: Taking measures to restore or to compensate for the loss of ecological corridors caused by the building of new roads and other constructions that prevent animals from migrating or interchanging. In these cases, the responsible authority has to safeguard such crossing routes, for example, by building special tunnels for otters and badgers, by building so-called cerviducts for deer, . . . or by any other appropriate means. 89 Special attention should also be paid to the existence and applicability of exception clauses in legal instruments, and to any reservations that may have been submitted by the State(s) involved. Regarding the former, the Kiev SEA Protocol provides that plans and programmes 'whose sole purpose is to serve national defence or civil emergencies' are 'not subject to this Protocol'. 90 The 1968 African Nature Conservation Convention offers another example by stating that the 'provisions of this Convention shall not affect the responsibilities of Contracting States concerning i) the paramount interest of the State, ii) "force majeure", iii) defence of human life'. 91 Evidently, the question may also arise whether any circumstances precluding wrongfulness might apply under general international lawfor instance, whether the erection of a particular border fence in breach of a treaty obligation can be said According to Article III.4 of the CMS, parties that are range States of an Appendix I species 'shall endeavour' inter alia to 'prevent, remove, compensate for or minimize, as appropriate, the adverse effects of activities or obstacles that seriously impede or prevent the migration of the species'. 98 Whilst of clear relevance to the border fence issue, the use of the word 'endeavour' appears to afford a margin of discretion to the party concerned, making this an obligation of effort rather than result. 99 The erection of a border fence affecting Appendix I wildlife would therefore not necessarily in all circumstances constitute a violation of Article III.4.
By contrast, the obligation in Article III.5, concerning the 'taking' of Appendix I animals, does not afford parties any discretion whatsoever: 100 Parties that are Range States of a migratory species listed in Appendix I shall prohibit the taking of animals belonging to such species. Exceptions may be made to this prohibition only if: (a) the taking is for scientific purposes; (b) the taking is for the purpose of enhancing the propagation or survival of the affected species; (c) the taking is to accommodate the needs of traditional subsistence users of such species; or (d) extraordinary circumstances so require; provided that such exceptions are precise as to content and limited in space and time. Such taking should not operate to the disadvantage of the species. 101 Parties must report any such exceptions as soon as possible to the CMS Secretariat. 102 'Taking' is defined as 'taking, hunting, fishing, capturing, harassing, deliberate killing, or attempting to engage in any such conduct'. 103 As others have rightly observed, this definition has a 'very wide scope indeed' 104 and can have 'farreaching implications for parties'. 105 Notably, whereas 'killing' must be intentional to be covered by Article III.5, the 'taking', 'capturing' and 'harassing' of Appendix I wildlife are within the scope of the obligation in Article III.5 even when they are unintentional. 106 This is where border fences enter the picture. It is instructive in this regard to consider an example given by Simon Lyster, concerning the incidental capture of Atlantic ridley turtles (Lepidochelys kempii)an Appendix I speciesin shrimp fishing nets, which constitutes a significant threat to the species: Since the entanglement of turtles in the trawls clearly constitutes 'capturing' or 'harassing', even if the killing of turtles is deemed not to be 'deliberate', it is probably fair to conclude that Article III(5) imposes a legal duty on parties that are Range States of the Atlantic ridley to prohibit the use of shrimp trawls in areas where the turtle occurs unless the trawls are fitted with 'Turtle Excluder Devices'. 107 By analogy, if it can be reasonably foreseen that the construction of a particular border fence may lead to Appendix I wildlife becoming entangled or otherwise 'taken' by the fence, or if an existing border fence is taking such toll, then the construction or maintenance of such fence would seem to be incompatible with the obligations of the contracting party involved under Article III.5 of the CMS. Moreover, even if a fence does not lead to 'capture' or physical injury of any kind, it would still be at odds with Article III.5 if its effects on Appendix I animals constitute 'harassing'. The US ESA provides an interesting parallel, as it also includes 'harass' in its definition of 'take', and defines the former as an act or omission that creates the likelihood of affecting wildlife by 'annoying it to such an extent as to significantly disrupt normal behavioral patterns which include, but are not limited to, breeding, feeding or sheltering'. 108 This is understood to include the annoying effects of persistent noise, light or motion, but not so much the physical modification of habitat, which latter is covered by different terms. 109 Regarding the CMS, in the absence of concrete guidance it is hard to say to what degree a border fence may result in the 'harassing' of Appendix I wildlife, although a case can clearly be made that the term covers the annoying effects of fence attributes like floodlights and patrolling vehicles, and perhaps also the stress caused to an animal when it finds the fence blocking its intended movement in search of water, food or a mate.
As regards the exceptions that Article III.5 allows from the required prohibition to take Appendix I wildlife, quoted above, reasons (a), (b) and (c) are unlikely to arise in connection with a border fenceexcept perhaps reason (b) in a rare scenario where the fence expressly serves to protect an Appendix I species, for instance, by keeping foreign poachers out. It may, however, obviously be possible for a contracting party to argue under (d) that the erection of a particular border fence is required by 'extraordinary circumstances'. Parties would seem to have quite a degree of discretion in this regard. 110 The requirement that an exception from the prohibition to 'take' Appendix I animals must be 'precise as to content', 111 is evidently more problematic, as it will be virtually impossible to predict with any degree of precision what toll a border fence will be takingfor instance, what number of animals it will affect, and how. Next, in cases where a border fence can as such be considered to violate the prohibition of Article III.5, the condition that exceptions must be 'limited in time' entails that the party involved cannot justify the building of a permanent fence. Lastly, the fence in question should generally 'not operate to the disadvantage of the species', although it is unclear what this implies precisely. 112 Various resolutions adopted by the Conference of the Parties (COP) of the CMS are of relevance to the border fence issue. For instance, the COP has requested contracting parties to conduct an EIA or SEA for potentially harmful projects and plans, including assessment of any 'effects involving impediments to migration' and any 'transboundary effects on migratory species'. 113 The COP has also urged parties to cooperate over transboundary areas, 'ensuring that barriers to migration are to the greatest possible extent eliminated or mitigated'. 114 In the context of climate change, another resolution calls on parties to 'strengthen the physical and ecological connectivity between sites, permitting dispersal and colonization when species distributions shift'. 115 Furthermore, a resolution on the impact of power lines on migratory birds provides for interesting parallels with the impacts of border fences on migratory wildlife. 116 Of special relevance to border fences is Resolution 11.24 of 2014, establishing the Central Asian Mammals Initiative (CAMI). 117 The CAMI is a so-called Special Species Initiative (SSI), a cooperative and flexible species conservation arrangement, involving both governmental and nongovernmental stakeholders, providing coordination between and building on pre-existing CMS instruments for Central Asian species. 118 It covers 15 Central Asian species 119 -13 large herbivores and two large carnivores, not all of which are CMS-listed, occurring across 14 countriesmany of which are vulnerable to the threats posed by border fences. The resolution recognizes that 'fences can have a particularly detrimental impact on the conservation status of migratory mammals', in particular through 'direct mortality and fragmentation of habitats' and by 'disrupting essential movement from one place to another'. 120 Through Resolution 11.24, the COP adopted (i) a CAMI Programme of Work, (ii) a guidance document on linear infrastructure and (iii) a Single Species Action Plan for the argali sheep. All three expressly address border fences. Regarding the first, specifically with regard to Mongolian and goitred gazelle, Asiatic wild ass, Przewalski's horse and Bactrian camel in the 'Gobi Desert -Eastern Steppes Ecosystem', the Programme of Work accords high priority to addressing the impacts of (border) fences through the following actions: (a) Map existing fences across the landscape in a spatial (GIS) database, including important meta-data, (b) mitigate impact of existing fences through removal or modification to wildlife-friendly designs, (c) strengthen EIA requirements so that fences that are required or proposed are assessed as to their necessity and if so, ensure that they are wildlife friendly and appropriate to all species affected. (i) Assess the legal framework which exists (joint ownership of railroads, border security policies). (ii) Create working group to assess best practice standards or take the lead in defining new ones, and (d) explore issues/options related to increasing border fence permeability. 121 More detailed and technical guidance to implement these recommendations, and also to address impacts on other species, has been elaborated in the Guidelines for Addressing the Impact of Linear Infrastructure on Large Migratory Mammals in Central Asia, the second document adopted through Resolution 11.24. 122 The third is the International Single Species Action Plan for the Conservation of the Argali. 123 The threats posed to argali by border fencesmortality through entanglement, curbed mobility and increased poaching, as well as population fragmentation and genetic isolationfeatured amongst the reasons for listing the argali sheep under the CMS in 2011, and are consequently also addressed in the Single Species Action Plan adopted in 2014. 124 Besides the CAMI, there is another SSI of potential relevance, which focuses on Sahelo-Saharan Megafauna and covers six of the aforementioned North-African ungulates. 125 These overarching SSIs should be distinguished from the array of subsidiary instruments which have been adopted under Article IV of the Convention. Presently, these comprise seven treaties and 19 nonbinding Memoranda of Understanding (MoU), with each instrument tailored to the conservation needs of a particular (sub)species or species group. Several of these instruments are, or may become, relevant to the border fence issue. These include the Gorilla Agreement, which has a strong emphasis on connectivity conservation; 126  For reasons of space, the relevance of only one of these is concisely illustrated here, namely the Saiga Antelope MoU. The MoU itself expresses the concern that 'obstacles preventing natural dispersion and migration' adversely affect the saiga's conservation status. 131 Since the MoU's adoption, the signatories have devoted express attention to the impact of border fences on saiga antelope, resulting inter alia in the development of detailed technical guidance to abate the adverse consequences of the Kazakhstan/Uzbekistan border fence. 132 Furthermore, the recently adopted International Work Programme for 2016-2020 urges the saiga range States to '[r]emove barriers impeding the movement of saiga antelopes or, if not possible, alter fences . . . and other linear infrastructure to allow saigas to pass freely', under express reference to the aforementioned CMS Guidelines on linear infrastructure in Central Asia. 133 The Work Programme also urges the performance of EIA and SEA prior to the implementation of 'all infrastructure and other development likely to impact saigas'; the designation of 'trans-frontier protected areas where appropriate, in light of saiga distributions'; and the establishment of a 'network of ecological corridors in order to protect saiga populations during migration'. 134

BORDER FENCES AND THE EU HABITATS DIRECTIVE
In a book on the Habitats Directive published in 2015, one of the authors asserted that: 'In the European Union, most national borders are merely virtual. There are no man-made barriers within the EU, like walls or fences. And very often, the same habitat and the same species of fauna and flora can be found on both sides of the border.' 135 Within a very short time frame, the situation has changed. Nowadays, for instance, wolves, lynx, bears and deer intent on crossing the border between EU Member States Slovenia and Croatia are likely to run into a very real man-made barrier. Indeed, the main reason for singling out EU wildlife law for a closer look in this article, besides the elevated enforceability and other special features of the legislation itself, 136 is the special history of the initial disappearance of border barriers within Europe, followed by the unexpected recent surge of border fence creation along the external and internal borders of the Union. Over the past years, the EU has established an elaborate framework to ensure an integrated management of its external borderswith a view to managing migration effectively and ensuring a high level of securitywhile safeguarding the free movement of persons within the Union. 137 A Regulation introduced in 2013 enabled the reintroduction by Member States of border controls at internal borders in exceptional circumstances, particularly when the Member State in question faces a serious threat to public policy or internal security. 138 The Regulation provides for various safeguards to ensure that this is only done as a last resort, and on a temporary basis. 139 In 2015, an extraordinary influx of refugees and other migrants led several Member States to reintroduce or reinforce border controls, including the rapid erection of hundreds of kilometres of border security fences on both the external and internal borders of the EUwithout prior EIAs, let alone SEAs, regarding their design or placement. 140 Curiously, fences of whatever length or nature do not occur in the lists of projects for which an EIA is, or may be, required under the EU EIA Directiveeven if these lists contain categories of linear infrastructure with potentially similar effects on wildlife, such as roads and railways. 141 Under the EU SEA Directive, however, an SEA may be required, in particular when a 'plan' or 'programme' to construct one or more border fences 'has been determined to require an assessment pursuant to Article 6' of the Habitats Directive (see below). 142 Leaving aside the question to what degrees the construction of border fences complies with the aforementioned EU legal framework on borders, immigration and free movement, the analysis below exclusively addresses their compatibility with the Habitats Directive. The most relevant provisions from the Habitats Directive are (i) the duties in Article 6 to avoid and remedy harmful impacts on protected nature within 'special areas of conservation' (SAC) designated as part of the Natura 2000 network; and (ii) the rules in Article 12 concerning the killing and capture of strictly protected species. Both provisions are examined below.
The aforementioned razor wire fences installed along large tracts of the Slovenia/Croatia border, pursuant to a decision by the Slovenian government in November 2015 to fence off the entire 670-km border between the two countries, constitute an apposite scenario to bear in mind during this examination. As documented by Linnell et al., some 350 km of this partly constructed and partly projected border fence cuts across 'some of the best preserved natural areas of the region', where it runs between a number of contiguous SACs on both sides of the border that are part of the Natura 2000 network and have been designated inter alia for wolf, lynx and brown bear. 143 Wolves and bears in this border area are part of interconnected populations which extend southeast into Greece and cover nine countries. 144 The fence threatens to cut the northwesternmost animals off from these Dinaric-Pindos wolf and bear populations. Once separated from the core population, biologists fear that Slovenian wolves would face rapid inbreeding, 'making viability of such a population fragment questionable'. 145 For lynx, the picture is bleaker still. Lynx in the region are part of the already vulnerable Dinaric population, shared between Slovenia, Croatia, and Bosnia and Herzegovina. 146 The Slovenian border fence 'may just be the last push for the population to spiral down the extinction vortex'. 147 Besides these impacts at the population level, animals may be individually affected. Large carnivores have large home ranges, and studies of GPS-tracked bears, wolves and lynx in Slovenia and Croatia have documented many cross-border journeysprior to the arrival of the border fence, that is. 148 For instance, of the 10 or 11 wolf packs currently existing in Slovenia, five have (or had) dual nationalities, so to speak, with home ranges straddling the border with Croatia. 149 As regards direct mortality, the entanglement of wolves, bears or lynx in the fence's razor wire coils is clearly conceivable, just like the unfortunate deer mentioned above. At the time of writing, however, large carnivore casualties had not yet been recorded 150 (which as such is not surprising, given that they occur at much lower densities than deer).
Article 6.2 of the Habitats Directive requires Member States to take appropriate steps to avoid, in the SACs, the deterioration of species' habitats as well as 'disturbance of the species for which the areas have been designated, in so far as such disturbance could be significant in relation to the objectives of this Directive'. This obligation has repeatedly been interpreted by the Court of Justice of the EU (CJEU) as a requirement to 'do what it takes': what ultimately counts is the result. 151 Besides, concrete plans or projects that are potentially harmful to the protected nature within SACs are subject to the stringent authorization scheme set out in Article 6.3-4 of the Directive: 3) Any plan or project not directly connected with or necessary to the management of the site but likely to have a significant effect thereon, either individually or in combination with other plans or projects, shall be subject to appropriate assessment of its implications for the site in view of the site's conservation objectives. In light of the conclusions of the assessment of the implications for the site and subject to the provisions of paragraph 4, the competent authorities shall agree to the plan or project only after having ascertained that it will not adversely affect the integrity of the site concerned and, if appropriate, after having obtained the opinion of the general public.
4) If, in spite of a negative assessment of the implications for the site and in the absence of alternative solutions, a plan or project must nevertheless be carried out for imperative reasons of overriding public interest, including those of a social or economic nature, the Member State shall take all compensatory measures necessary to ensure that the overall coherence of Natura 2000 is protected. It shall inform the Commission of the compensatory measures adopted. Where the site hosts a priority habitat type and/or a priority species, the only considerations which may be raised are those relating to human health or public safety, to beneficial consequences of primary importance for the environment or, further to an opinion from the Commission, to other imperative reasons of overriding public interest.
According to the CJEU, the criterion from Article 6.3 implies that plans or projects may in principle be authorized only 'where no reasonable scientific doubt remains as to the absence' of adverse impacts. 152 The prior assessment to determine this needs to contain 'complete, precise and definitive findings and conclusions capable of removing all reasonable doubt as to the effects of the work proposed'. 153 If there is no such certainty that significant adverse impacts will not occur, only Article 6.4 can offer a way out for the authorities seeking to go ahead with a project anyway. Notably, both Articles 6.2 and 6.3 can require addressing developments taking place outside of an SAC if those developments threaten the species within the SAC. 154 It should also be noted that both wolf and brown bear are marked as priority species in Annex II of the Directive.
Article 12.1, which as a matter of principle applies both within and beyond SACs, commits Member States to taking 'the requisite measures to establish a system of strict protection' for the animal species listed in Appendix IV of the Directiveincluding the three aforementioned large carnivores. 155 This requires the establishment and enforcement of prohibitions on, inter alia, 'all forms of deliberate capture or killing' and 'deliberate disturbance' of individual animals. Importantly, the term 'deliberate' in the sense of this provision covers not only actual intent, but also the acceptance of reasonably foreseeable results. 156 For instance, in the words of the CJEU, for capture or killing to be 'deliberate'and thus covered by the prohibition of Article 12.1it is sufficient that the author of the act concerned 'accepted the possibility of such capture or killing'. 157 Furthermore, the Court has emphasized that Member States must not merely prohibit the acts enumerated in Article 12, but also take all measures necessary to ensure that the prohibitions in question are not violated in practice. 158 163 These operations affected an ecological corridor connecting two bear subpopulations. A report is cited establishing that the noise and vibrations caused by the mining operations 'will prevent access for the brown bear to that corridor, or make it much more difficult'. 164 The Court qualifies these impacts as 'disturbances' of the Natura 2000 site 'which are significant having regard to the conservation of the brown bear', thus amounting to a violation of Article 6.2 of the Habitats Directive. 165 Comparable conclusions are reached in respect of the capercaillie, as the mining operations are deemed 'capable of producing a barrier effect likely to contribute to the fragmentation of the habitat of the capercaillie and the isolation of certain sub-populations of that species', 166 one of which is located outside the SAC. As the Court observes: there is a risk that operations currently being carried out, in conjunction with projects the implementation of which is imminent, form a continuous east-west barrier for the capercaillie, capable of leading to the isolation of population pockets of that species, and, over time, to the disappearance of pockets located to the south of that barrier. 167 In the words of Verschuuren, 'the conclusion cannot be other than that destroying a corridor that leads to the deterioration of a site is prohibited by Article 6(2)'. 168 For border fences, the preceding analysis demonstrates that the maintenance of an existing fence or the construction of a new one may, depending on the circumstances, constitute a violation of Article 6 and/or Article 12 of the Directive. Regarding Article 6.2, a border fence may produce deterioration or disturbance in parallel to the Spanish bear and capercaillie case, amounting to a violation of the Directive. Regarding Article 6.3-4, it is evident that a border fence can qualify as a 'plan or project'. 169 Whether the construction of a border fence infringes on Article 6.3-4, however, depends on a range of factors. These may include whether the potential wildlife consequences of the fence were contemplated in advance of its construction; what those consequences are; whether any underpasses or other wildlife corridors in the fence qualify as mitigation or compensation; 170 whether the purpose of the fence can be said to be 'public safety', or another 'imperative reason of overriding public interest'; whether alternative solutions have been proven absent; whether 'all compensatory measures necessary' have been taken; 171 and whether soliciting an 'opinion' from the Commission was required and, if so, whether it was sought. 172 Needless to say, a crucial factor would be whether the procedure of Article 6.3-4 was followed at all. If only for the latter reason, it would seem that Article 6 of the Habitats Directive has been violated on more than one occasion through the hasty erection of border fences potentially affecting SACs in various EU Member States. Regarding Article 12, whether a particular border fence is at odds with this provision would again depend on the circumstances. Pertinent factors include whether it is reasonably foreseeable that the fence may entangle Annex IV animals, resulting in 'capture or killing'; whether any other anticipated impacts may be qualified as 'disturbance' in the sense of Article 12; whether, accordingly, in light of the available information the decision to construct the fence can be considered as 'accepting the possibility' of such capture, killing and/or disturbance; and, if so, whether the three cumulative conditions of Article 16 have been met, so as to justify a derogation.
It should be stressed that the Habitats Directive, unlike some of the legal instruments discussed above and unlike the EU EIA and SEA Directives, 173 does not contain any safeguard clause concerning emergency measures. Likewise, regarding general defences that have been invoked by Member States in the past to justify non-compliance with EU obligationssuch as force majeure or other equivalents of the circumstances precluding wrongfulness that apply under general international lawthe CJEU has so far been extremely reluctant to let Member States off the hook. 174 Thus, it seems that in order to comply with their obligations under EU law, Member States must follow and abide by the procedure of Article 6.3-4 of the Habitats Directive for all projects, no matter how urgent, that are potentially harmful to wildlife within Natura 2000 sitesincluding border fences. Similar considerations apply to the generic strict protection of Annex IV wildlife, where Article 16 is the sole basis for derogations from the prohibitions of Article 12. Interestingly, in December 2015 the Croatian government sent a letter to the European Commission, alleging that the erection by Slovenia of razor wire fences along the Croatian border represents a 'clear breach of the Habitats Directive', and requesting the Commission to 'discharge its duties as the "Guardian of the Treaties" and take swift action to ensure effective implementation of EU nature conservation legislation'. 175 The above analysis appears to support the position taken by the Croatian authorities.
Before concluding, it is appropriate to highlight another implication of border fences for the implementation of the Habitats Directive concerning large carnivores. This concerns the scale at which Member States are expected to achieve and maintain a favourable conservation status (FCS) for species covered by the Directive, and may be illustrated with reference to the somewhat enigmatic condition from Article 16 that a derogation from strict protection may only be allowed if it is 'not detrimental to the maintenance of the populations of the species concerned at a favourable conservation status in their natural range'. 176 In the absence of unambiguous case law from the CJEU, the appropriate scale for assessing whether this condition is fulfilled remains the subject of debate. 177 According to a quite broadly supported interpretation, however, for wide-ranging species like large carnivores whose populations extend across international borders, the appropriate benchmark is the transboundary populationon condition that a transboundary population-level management plan or similar safeguard is in place. 178 If, however, a border fence effectively cuts off one part of a population from the other, then to all intents and purposes the former transboundary population ceases to exist. For each Member State involved, the FCS benchmark would then revert to being the population segment on its own side of the fence, resulting in more onerous and potentially unworkable requirementsespecially bearing in mind that some countries may be too small to harbour a viable population of wolves, bears or lynx by themselves. 179

CONCLUSIONS AND RECOMMENDATIONS
If the analysis above demonstrates one thing, it is that the issues raised by border fences deserve serious attention from policy and decision makers, conservationists, researchers and other stakeholders in the field of (inter)national wildlife law and policy. These fences, as Linnell et al. observe, have the potential to 'undo decades of conservation and international collaboration efforts', and their proliferation entails a need to 'realign our conservation paradigms with the political reality on the ground'. 180 This realignment and the evident need to keep the adverse impacts of border fences on wildlife to a minimum, seem to imply that the role reserved for international cooperation and international law is becoming greater rather than smaller.
As outlined above, many already existing obligations in international legal instruments have a bearing on border fences as they relate to wildlife, even if they do not specifically address the fences in so many words. At the same time, there is apparent room for improvement regarding the contribution of said instruments to preventing and ameliorating the harmful impacts of border fences on wildlife around the world. Apart from raising awareness concerning the existence and potential of these instruments and obligations, much could be gained through the further clarification and interpretation of their implications for border fences, including through the elaboration of authoritative guidance expressly addressing the issue. The development and adoption of a CBD COP Decision, CMS COP Resolution and/or Bern Convention Standing Committee Recommendation expressly addressing border fences would appear to constitute particularly promising avenues to take this forward. Similar considerations apply to the adoption of guidance within the various other legal frameworks discussed above. Such decisions could usefully build on the CMS guidance already developed for Central Asia and the aforementioned mitigation measures proposed in the scientific literature.