Full-text audio version of this essay.

You’re walking along a forest trail. You’ve been walking for a while now, and your bladder is full. It’s been a long while since you’ve passed another person, but as a precaution, you take a few strides off the path before dropping your pants. When you’re done, you look up, and scan your surroundings. Your eye catches on something. It would be easy to miss: a small, rectangular box, disguised crudely in camo print, attached to a tree trunk with a strap. It blinks at you with its single eye.

We know that surveillance is ubiquitous in cities, but most of us rarely think about the fact that networks of surveillance extend even into forests and marshes, deserts and oceans. We tend to think of technological infrastructure as ending where the visible built environment does, which is not an unfounded assumption, especially in national parks and nature reserves which actively market themselves as places to “switch off.” When our phone ceases to pick up a signal, it corroborates the impression we’ve finally found ourselves in a place where technology has no power over us. The reality, however, is that often these places are thrumming with devices, some more visible than others, quietly monitoring soundscapes, air quality, humidity, temperature, soil acidity, the presence and absence of different species, of people.

The militaristic, colonial history surveillance belongs to is not simply waived by the fact that it operates in a forest

Monitoring — via GIS, passive or automated sampling, video and audio recording, or biomonitoring — is a vital part of ecological research, enabling researchers to track the degradation and regeneration of ecosystems, document extreme weather events, identify rapid drops or increases in animal or plant species, and determine ecosystem health. It is tempting to think that this might constitute a form of “good” surveillance. But the militaristic, colonial history this surveillance belongs to is not simply waived by the fact that it operates in a forest. Even inadvertently, many of the tools we use to do environmental science end up reproducing these systems.

The rise of surveillance technologies in conservation science corresponds with a turn towards methods that are ‘non-invasive’, allowing research to happen at a distance from its subject, and therefore resulting in minimal disturbance. One of the most effective and therefore common monitoring devices found in nature is the camera trap, also known as a trail camera. Widely used in ecology to monitor the population size, distribution and behavior of different species of animals, camera traps are cameras that are activated by a motion or infra-red sensor, trip-wires, pull-wires, pressure plates, lasers or microwave sensors. They can be left alone, for weeks or months on end, to capture images of wildlife. Inconspicuous in appearance and housed in weatherproof casing, they enable researchers to observe rare events in a “non-invasive” manner, and without the physical and mental strain of waiting and watching.

Since their origins over 100 years ago, camera traps have been adopted into widespread use, with several hundred scientific papers each year citing them as a central tool. Because camera traps rely partly on the chance of something wandering into their frame of vision, their success is contingent on them being widespread (the WWF guidelines on camera trap usage state that, as a baseline rule, one should use “as many as you can possibly get your hands on.”) Because the defining feature of a camera trap is that its trigger need not be activated by a human operator, there is nothing to stop them from watching humans.

The inadvertent capture of human subjects by camera traps is known as “human bycatch,” a name taken from the unintended capture of non-target species in fishing. In many ways, the fishing analogy is apt: Most camera traps store images on a memory card, rather than uploading them directly to a server, which means that a researcher never knows what they have captured until they “haul up the net.” The physicality of the language surrounding visual technologies — camera traps, image capture, human bycatch — is not incidental; it hints at something fundamental about the relationship between image-making and violence in contemporary society. In War and Cinema, Paul Virilio draws attention to the “deadly harmony that always establishes itself between the functions of eye and weapon.” Just as a line of sight is also a line of fire, the camouflaged box is a trap. Once one’s image is caught inside it, one’s body is also at risk.

While camera traps may not be designed to capture images of human subjects, a study carried out by researchers in the department of Geography at the University of Cambridge found that more than 90 percent of respondents using camera traps (across universities, governments, the private sector and NGOs) had inadvertently captured at least one image of a human in their most recent project. Of these, 50.7 percent reported that they’d captured images of people behaving illegally, and almost all of these (44.3 percent of respondents) had used these images for some form of management or enforcement: reporting them to the police, sharing them with conservation staff, sharing them with media, using them for research, attempting to identify those photographed, or filing them for future reference. Only 8.1 percent of projects reported deliberately trying to remove inadvertently captured images of people.

The surveillance of human activity is in fact a dual function of camera traps. Human bycatch images are used by government bodies, researchers and NGOs to monitor human activity in sites of ecological interest, and prosecute activities deemed illegal, unecological, or simply undesirable. Of course, judgements about what are desirable or even ecological ways to behave in “wildlife areas” are incredibly subjective, incredibly political, and tend to reflect the values of the ruling order. The very idea of “national parks” is a modernist concept grounded in nationalist ideology and settler colonialism (the expansion of highly curated protected forest areas under Nazi Germany is one such example; the expansion of national parks on indigenous land in North America and Australia is another), and the designation of wildlife areas in the Global South by European and North American NGOs often constitutes a violent form of neo-colonialism. It follows, then, that the policing of who enters these spaces, and what they do there, is by no means less violent nor less political than the policing of behavior in urban areas.

The camouflaged box is a trap. Once one’s image is caught inside it, one’s body is also at risk

The University of Cambridge study coins the term “surveillance conservation” to refer to conservation practices whose primary or secondary function is the shaping of “disciplined conservation actors.” This can take explicit, neo-colonial and violent forms, such as in the use of military tactics against poaching (a phenomenon that is, itself, a direct result of colonial rule). It can also take subtler forms, as in the way national parks and nature reserves might promote a certain idea of what the ideal visitor looks like, and how that person should behave. The woodland near where I live in London, for example, has historical importance as a site for queer cruising, a shelter for rough sleepers, a venue for illegal raves. Like many urban green areas, it offers refuge not only to wildlife, but to those marginalized by enforced social norms. What does it mean for images of people using these areas in different ways to be captured and held, alongside images of small-to-medium-sized mammals, inside a camo-print box?


Wildlife areas, like public space more widely, are becoming increasingly militarized; and conservation surveillance, like surveillance more widely, is expanding. What is surprising is that the camera trap has mostly escaped scrutiny, despite its obvious resemblance to CCTV and other technologies of surveillance that have come to be widely mistrusted. This speaks to the trust we place in projects that affiliate themselves with the ecological sciences, a trust stemming from the very real urgency of the climate crisis and biodiversity loss. But the ease with which devices such as the camera trap are able to straddle ecology and security hints at an important, but overlooked history: the environmental sciences have evolved in a startlingly close relationship with the military-industrial complex.

The way that environmental science is done, and the assumptions and aims that underpin it, have changed over time. As Jennifer Gabrys writes in Program Earth, today’s iteration figures the earth as made of data, and therefore as “an object of management and programmability.” (Ecosystems can be programmed to function “correctly,” and people can be programmed to behave in ways deemed ecologically appropriate.) The earth-as-data conception of ecological science has its origins — at least in part — in the Cold War, when large amounts of military money were invested into environmental data gathering. The pinnacle of this military-funded global data drive was the International Geophysical Year, an international project running from July 1957 to December 1958 in which data from 67 countries was collected. Data from the IGY was held in one of three World Data Centers, and a clear plan for the emerging “world order” can be read in their designations: the United States hosted World Data Centre “A,” the Soviet Union hosted World Data Centre “B,” and World Data Centre “C” was subdivided between Australia, Japan and various countries in Western Europe.

One major impetus for the IGY was the growing threat of nuclear attacks: understanding “normal” geophysical conditions was essential to detect areas where radiation testing by enemy states had led to detectable abnormalities of climate or in the chemical composition of soil, oceans and atmosphere. Another impetus was the serious consideration of environmental warfare as a military tactic — in 1974, for example, the public became aware of a Pentagon scheme to seed clouds in Vietnam and Cambodia, triggering rainfall and landslides that would disrupt the transportation of supplies to guerrilla fighters. In order for these attacks to be effective, complex understandings of ecosystem science needed to be developed.

It is certainly true that the environmental data collected over the past 70 years has been hugely important for our understanding of earth systems and our sense of our place within them (it was during the IGY, for example, that some of the first data was gathered about the accumulation of carbon dioxide in the atmosphere). But the price we have paid for this information — the deep entanglement of environmental and military science — is hard to know, and probably hard to understate.

Most tangibly, it can be seen in the fact that most planetary environmental data is still held by U.S. federal agencies with close ties to the military. Today, for example, the largest global provider of weather and climate data is the United States NCEI (National Centres for Environmental Information; formerly the National Climatic Data Centre). The NCEI is a service of the National Oceanic Administrative Association, which calls itself, proudly, “America’s environmental intelligence agency” working in service “to protect life and property” (the NOAA is, in turn, part of the department of commerce). The NOAA holds an archive of data collected by the U.S. Navy, U.S. Air Force, the Federal Aviation Administration, and international meteorological services: meteorological stations in the U.S. are given a WBAN number, which stands for Weather Bureau Army Navy. These sources feed into the NCEI (National Centres for Environmental Information), a subdivision of the NOAA, which is the largest global provider of weather and climate data. The large-scale environmental data collection that we rely on for tracking environmental challenges that are increasingly global in scale — such as climate change, biodiversity loss, and the toxicity left in the wake of military activity — therefore comes, at least in part, directly from the U.S. military, which is more culpable than any other single institution for these problems.

There is, or should be, an obvious conflict of interest here. Instead, the greenwashing of the military has been so successful that alignments between military and environmental projects are becoming normalized. One example of this is the phenomenon of “Military to Wildlife Conversions” (often called M2W), by which previous military testing sites are handed over to environmental bodies and designated wildlife areas. As many critics of military-environment interactions have pointed out, this is a convenient way for the military to avoid cleaning up after itself: the Rocky Mountain Arsenal site, for example, now described as a “National Wildlife Refuge,” functioned as a test site for chemical and biological weapons including VX nerve gas, mustard gas, chlorine gas and rice blast spores. Today, it is managed by the Fish and Wildlife Service, which is still largely dependent on the Department of Defence for funding. Data about lasting contamination and its effects on the ecosystem, therefore, is scarce.

Part of the reason that “human bycatch” has not received more criticism is perhaps because it is unlikely to impact many white people

In many cases, M2W sites also function as a way of allowing military and colonial occupation to continue. The U.S. Navy used the Island of Vieques, Puerto Rico, as a testing ground for over 60 years. Following military occupation, large areas were handed over to the Fish and Wildlife Service, which in many ways imposes tighter restrictions on land use than the military. Some have speculated that this is a way of keeping the region under U.S. federal control, so that the Navy can return there with greater ease. Similarly, Bikini Atoll was rebranded a “pristine wilderness” after inhabitants were dispossessed so that the island could be used for nuclear testing. According to Rachel Woodward, such cases can be seen as examples of “military creationism”: the myth that we have the military to thank for the “preservation” of natural landscapes.

The tendency to associate ecological science with social justice isn’t naïve: in theory, ecological science — with its understandings of mutuality, cooperation, and its celebration of diversity – should be the antithesis of military values. In the 1960s and ’70s, a popular movement began to take shape in opposition to the massive military undertakings of the Cold War. This movement is widely credited to the publication of Rachel Carson’s Silent Spring, which drew attention to the extractivist logics underpinning industrial agriculture, and demonstrated that environmental data collection could be used as a force for good. Her investigation of the use of synthetic pesticides, many of which had been developed through military funding, was backed up by four years of research, made possible by a book deal, for which she drew on her own connections with government scientists, and was able to identify a small camp of experts willing to talk to her, confidentially, about the detrimental effects of the widespread use of DDT. The book received fierce backlash.

If it was difficult for Carson to gain access to the scientific data she needed as a white, wealthy (albeit female) scientist with a university degree, then it was even more so for those without these privileges. More often than not, in cases of toxicity and contamination, data is embodied. Even Carson herself was diagnosed with breast cancer after spending large periods of time in DDT-contaminated areas. In the 1970s, residents of a housing development known as “Love Canal” reported high rates of cancer and children being born with birth defects. It was later revealed that Hooker Chemical Company had sold the site to the council after using it as a dump for toxic waste. In the absence of environmental data, residents set up the Love Canal Housing Association and developed its own health survey, accounting for deaths, birth defects, and other health complications (its findings were known as “housewife data,” due to the women driving the initiative). Only through this was the community able to gain support for an investigation and eventually secure reparations.

Of course, science does not have to take place in universities, and it doesn’t necessarily need an enormous budget. An emerging movement called “citizen science” — sometimes given the more inclusive name of “participatory science” — considers ways in which non-professionals can collect data using cost-efficient devices. But democratizing access to a technology isn’t a guarantee that this data will be used for socially just purposes. When a device is developed as part of a militarized vision, this context leaves traces in the device’s hardware and software. And our own internalization of a culture of policing and surveillance can influence their functions.


Let’s return to the example of the camera trap. Alongside their use in university research settings and NGO-led conservation projects, camera traps have long been in consumer use as a gadget for amateur wildlife photography, a hunting tool, and even a tool for tracking paranormal activity. A quick search for camera traps on Amazon yields a range of similar-looking boxy cameras ranging from $25–$200. In the comments sections, people commonly report purchasing the camera trap to catch footage of wildlife in their garden, but finding it incidentally useful for home security, or visa versa. The product names are made up of various arrangements of keywords including “hiking,” “hunting,” “game,” “surveillance,” “wildlife,” “home security,” “monitoring,” “observation.” Some of these words seem to belong to one of two distinct applications (e.g. “wildlife” or “home security”), but others, such as “monitoring” and “observation,” point to a shared vocabulary between military and ecological applications. The camouflage-print casing that houses most of these cameras, too, is indicative of a long history of overlapping tactics. Unlike the drone, however, which carries its association with military enterprise even into its use as a consumer good, camera traps have managed to maintain a degree of mundanity that renders this element invisible.

What happens to all the photographs of people captured by consumer camera traps? Do they sit inside hard drives and USB sticks, or do they circulate online, becoming memes, evidence, or powerpoint slides? In December last year, a non-profit organization called Conservation International launched the world’s largest public database of camera trap images, crowdsourced from professional and non-professional users. The database, called Wildlife Insights, aims to address the fact that despite the increasingly widespread use of camera traps, “photos and data are not effectively shared or analyzed, leaving valuable insights just out of our grasp.” The project is backed by Google, and uses Google’s Artificial Intelligence technology to sift through, organize and analyze the data, which consists of both the images themselves, and the valuable metadata (location, time) attached to them. The website encourages its users to upload photos in bulk: it has a feature for filtering photos by “all taxonomies including human categories,” and classes to describe different types of humans “(park ranger, tourist, etc).”

Wildlife Insights demonstrates several ways in which conservation surveillance, even when crowdsourced by regular people, can be said to bring together the violence of government surveillance and the violence of conservation. Google’s artificial intelligence technology — particularly its facial recognition technology, which would presumably be central in detecting and filtering out photos of human subjects — is notoriously racist (in 2015, a Black software developer reported that Google’s Photos app had misidentified himself and his Black friend as “gorillas”). If Wildlife Insights only recognizes white people as people, it follows that Black, brown and Indigenous people will be disproportionately affected by the problem of “human bycatch,” just as Black, brown and Indigenous people are disproportionately targeted by surveillance in wider society. As a tool for policing, surveillance is entangled with white supremacy, beneficial for some at the expense of others. Part of the reason that the phenomenon of human bycatch has not received more criticism is perhaps because it is unlikely to have an impact on the lives of white people.

Conservation surveillance fulfills many of the same aims as surveillance more broadly. The sanctuary offered by the presence of trees and the absence of police recedes

Wildlife Insights promises to harness “the power of big data” to “craft smart conservation policies.” It identifies the eight key stakeholders that the database will ostensibly serve: land managers, government, companies, scientists, indigenous communities, citizen scientists, non-profits, and the general public. In doing so, it makes the dangerous claim that all of the listed stakeholders have the same, nebulous shared aim of “conservation,” ignoring the ways in which this term has been mobilized to enforce white, Western ideas about what is worth conserving and who is best placed to do it. It is essential, then, to think about what kind of narratives such a database will generate, and what kind of policies it will work to support. When “conservation” is figured as a neutral aim, just like when land is figured as a public good, it enforces a settler colonial notion of the untouchable “wilderness” that must be protected for the benefit of all. Such generalizations are a form of violence, erasing centuries of human habitation while enforcing a very specific conception of nature and the human relationship to it.

Conservation surveillance essentially fulfills many of the same aims as surveillance more broadly. It encourages the homogenization of behavior and reinforces the existing social order by outlining the “proper” use of the areas it surveils. As the dual use of camera traps becomes increasingly normalized, it may become more common to see small, one-eyed boxes hidden in the trees in our local parks, woodlands and wildlife areas. It may also be more common for us to see regular CCTV cameras or home security systems dressed in camo-print and labelled as “ecological monitoring devices.” Meanwhile, the sanctuary offered by the presence of trees and the absence of police recedes.


It is of course not the case that all environmental data collection projects should be dismissed. The power of ecological data, as a force for social and environmental justice, makes it all the more important to interrogate its entanglement in the military-industrial complex; also, to outline ways in which the tools for turning data into evidence can be handed back to communities, especially those who bear the brunt of extractive capitalism and environmental degradation. While participatory science is a move in the right direction, the extent to which scientific processes and devices are entangled in the military industrial complex, dependent on its allocations of funding and informed by its logics, means that such an approach can end up simply perpetuating the expansion of militarized natures.

One way in which this can be rectified is by projects that are genuinely community-oriented at every stage of the process, from collection, to storage, to analysis of the data. One example is in the use of monitoring devices by Indigenous groups in the Peruvian and Ecuadorian Amazon to take action against the environmental crimes enacted by corporations. In Loreto, Peru, the Kichwa, Tikuna, Yagua and Maijuna peoples have a monitoring system comprising drones and satellite maps to defend the Napo and Amazon basins from illegal loggers. In Ecuador, along the Aguarico River, Siekopai and Cofàn communities have taken action against mining companies with evidence gathered from drones and camera traps.

Another way is by ensuring that the hardware and software of the data-gathering devices themselves are developed with this kind of locally-grounded usage in mind. In Newfoundland, a laboratory called CLEAR (Civic Laboratory for Environmental Action Research) develops cheap, easy to use devices for the monitoring of plastic pollution in waterways. At its helm is Max Liboiron, a fierce critic of traditional scientific methodologies on the grounds that they end up reproducing the status quo, even if well-intentioned. Another example is MyNatureWatch, a project which provides step by step instructions for building cheap, open source DIY camera traps. By creating new devices which are widely modifiable and adapted to the purposes of specific communities, such projects resist the drift towards generalization and mega-narratives, and actively interrogate the inscribed values and functions of devices used to do science.

Paired with community ownership of data at a local level (as opposed to large, centralized, corporate-backed databases), the DIY and open-source approach to environmental sensing could broaden the remit of environmental research, breaking its dependence on military or corporate funding, and allowing research to take place beyond the traditional boundaries of Western ecological science. Such an approach would give us new stories about the world we live in. Having data collected, managed and owned by communities rather than corporations could lead to important changes in the way power is distributed in society, giving localities the evidence they need to advocate for greater decision-making power at a local level. The sharing of data between communities could strengthen the role of the local in global politics and facilitate connections between geographically disparate groups.

But it could also lead to slower, less tangible changes. Environmental sensing devices have the potential to tap into our instinct to care, our sense of curiosity, and our awareness of our own entanglement with other bodies and earth systems. They can be playful and exploratory, opening up new ways of thinking about ways in which humans, technologies and non-humans can co-exist. The word “sensing” has its origins in the Latin sentire, to feel: sensors help us to feel out the world around us, and to build meaning in it. Today, more so than any of the other senses, vision has become militarized and monetized. Platforms compete for “eyeballs” and institutions compete for data. Underlying this is a belief that looking is something that one body does to another, and that with total vision comes total control. This is not the only way to think about vision, though, nor is it the only way to think about technologies of vision. Like the other senses, vision is inherently relational; it comprises moments of encounter between multiple subjects, and therefore multiple subjectivities. It is time we started building technologies that stemmed from and supported this conception of vision.

In other words, what we should aim for is not necessarily a forest where we can have a wild pee, peacefully, in the knowledge that our wilderness will not be disturbed by technological blinking eyes. As long as these eyes are entangled in murky regimes striving towards omnipresent vision, we have ample reason to distrust them. But it is possible to build a world in which we don’t have to distrust our technological devices; in which we walk through the forest, and hear the whirring and clicking and silent heartbeat of various monitoring devices, and feel comfortable in the knowledge that these devices are part of the ecosystem of the forest itself, operated by — and working in service of — the people, animals and plants who live there. When this is the case, their watching, their listening and their sensing should be no more unnerving to us than the watching, listening and sensing of the trees, the birds and the river.