3

Animals as Sensors

Mobile Organisms and the Problem of Milieus

YOUR DAILY WEATHER FORECAST may have been brought to you, in part, by southern elephant seals. In the Southern Ocean, seals tagged with conductivity-temperature-depth satellite relay data loggers (CTD-SRDLs) capture a detailed picture of ocean waters. Sensors travel along with seals to map their dive profiles and to gather a previously inaccessible set of oceanographic data from the ocean surface to depths down to nearly two thousand meters. Based on seals’ foraging patterns carried out in response to the Southern Ocean environment, sensor data documenting seal movements, along with temperature and salinity, can lead to further inferences about sea-ice formation, the likely movement of fronts within the Antarctic Circumpolar Current, and global circulation of ocean currents—indicators not just of weather patterns but also of longer-term shifts in climatic patterns.1

Relaying across the journeys of tagged elephant seals swimming through circumpolar seas, sensors gather data on location, depth, and temperature that circulate to Argos satellite systems (which consist of three satellites in three orbital planes),2 then filter into the institutions of environmental science to form seal dive profiles, there to complement remote-sensing and ship-gathered climatology data sets, while informing the nightly weather report. Weather forecasts and longer-term studies of environmental change come together through these multiple and distributed sensing technologies and (more-than-human) journeys.

Sensors are increasingly used in ecological study for the tracking of organisms, often with the direct outfitting of animals with sensor backpacks and radio collars, in order to understand movement and migration. Sensors typically used in tracking include data transmitted through Argos or GPS satellite systems (including environmental temperature and humidity), light levels, acceleration, location, body temperature, heart rate, orientation, altitude, and pressure.3 The latest wave of computationally enabled sensors for tracking follows from and complements multiple methods that capture movement and migration, including bird ringing or banding, bird observatories, radar and time-lapse film, and isotope and DNA analysis.4

Any number of organisms have been outfitted with computationally based tracking technologies, including honeybees with RFID tags,5 green darner dragonflies with miniature radio transmitters,6 and Arctic terns with miniature geolocators.7 Even animals that disappear or die are monitored, as with the well-known case of “Happy Feet,” the “hapless emperor penguin” who turned up on a beach in New Zealand. Fitted with an Argos satellite transmitter Sirtrack and sent on his way back to Antarctica, Happy Feet’s tracker soon ceased transmitting, which led “to the conclusion that either the satellite transmitter has detached or an unknown event has prevented Happy Feet from resurfacing.”8

The movements of organisms, from badgers to elephant seals to storks, influence understandings not only of the journeys these animals take but also the environments that they inhabit and rely upon. Tracking devices are often presented as key technologies for studying animals under threat and for informing policy and management decisions so that these organisms may be better protected.9 Disruptions to habitat, causes of mortality, loss of feeding grounds: these are environmental events that the tracked journeys of organisms may reveal. But these technologies also generate concerns about the changes in activity and performance that animals might experience as a result of wearing the devices and through the stress caused by capture in order to fit the devices.10 Sensors used for tracking raise questions about the extent to which animal movements and relations to environments shift through the machine-organism milieus that are traversed and inhabited. How are organisms, tracking devices, and milieus transindividuated, and what are the ways in which these technical objects become environmental through tracing animals’ journeys?

Drawing on fieldwork and informal interviews conducted with environmental and computer scientists, as well as a review of scientific and technical literature and symposia, this chapter examines how the movement and migration of organisms have become key sites of study facilitated through environmental sensors. It asks how understandings of environmental change have shifted through increased levels of monitoring the movements of organisms. It also attends to the ways in which sensors for environmental monitoring undertake distinct journeys and types of attachments in order to travel along with organisms. In this concrescence of machine, organism, and environment, I ask how the milieus of technical and living entities in-form and transform through the tracking process. Rather than see tracking technologies as mirroring devices that invisibly capture hitherto unknown movements and journeys, I consider how these technical objects are involved in individuating organisms and environments as entities in need of further study and protection and as concrescing computational relationships that would activate the practices necessary to achieve these objectives.

In order to investigate these individuations of sensors, organisms, and milieus, I follow three organisms on their tracked journeys. These animals include badgers with RFID collars inhabiting the well-known “ecological laboratory” of Wytham Woods in Oxford; southern elephant seals fitted with data loggers and satellite transmitters, mentioned at the beginning of this chapter, as they dive through the Southern Ocean; and white storks carrying satellite transmitters, both as they are monitored and observed by scientists and citizens in the field and as they move across the platform of an Animal Tracker app set up to engage publics in practices of tracking animals.

As the southern elephant seal that appeared at the beginning of this chapter reminds us, environmental sensor networks involve not just tracking and tracing animal activities but also new forms of computational and collaborative sens-ing, where (as discussed in chapter 1) sensing is undertaken with and through more-than-human technologies and organisms. In other words, it is not just the organisms that are sensed. Organisms are transformed into sensors that would also communicate registers of animal-based environmental sensation and inhabitation. Much like the moss discussed in the last chapter, these organisms become biosensors of sorts, and through their journeys they provide data about environmental conditions and changes.11 Through this relay of machine and organism, I then expand upon the sorts of sensing and experience that are articulated in tracked journeys. What are the nexuses, or actual worlds, of sense that occur through these mapped events?12 And how do they both expand and challenge notions of environmental participation across organisms and machines? By addressing the traversals made across technical and living milieus, I finally consider how organisms and technical objects are not only expressive of living and technical milieus but also indicative of the particular problems they encounter in these milieus.

Sensing Movement and Migration

Animals are on the move, and have always been so, but their movements and migrations are emerging as different and more detailed events vis-à-vis data gathered through tracking studies. Why do organisms undertake these journeys? And how do these journeys shift understandings of environments not as fixed territories but as fluctuating zones of inhabitation, sporadic meeting points, feeding, and mating grounds—essential stopovers on some far-flung journey? Studies of animal movement and migration have been undertaken for some time, with tracking and nomadism as much indicators of different practices for studying (and even living with) animal movements as scientific approaches to tracking and tracing organisms through computational sensors and data. The comings and goings of insects, birds, mammals, and fish have indicated seasonal change, habitat disruption, or even impending disaster. Movement, it seems, provides an indication of animal behavior and routine.13

Indeed, the emerging field of “movement ecology” is seen to address some of the “unanswered questions in ecology,” including reasons for fluctuations in animal populations, where animals are and when they are located in distinct areas, as well as when, where, and why they die.14 If animals are on the move, so too is ecology adopting more mobile methods in order to understand these questions. As historian of science Etienne Benson discusses in his study of wildlife tracking, Wired Wilderness, Denali National Park in Alaska is one such site that at one point went from a practice of reluctant animal tracking to one of perpetual observation, where “new kinds of radio collar” were being deployed that would make it possible “to track wildlife 24 hours per day, 7 days per week.”15 In some cases, arguments are made not just for mapping and tracking animals in relation to distinct research questions but also for “life tracking” animals in order to understand movement, behavior, ontogeny, dispersal, and mortality across lifetimes.16 In such a scenario, animals would be tagged at birth and spend their entire lifecycle wearing a tracking device that monitors and records their activities.

In a parallel way, tracking animals is a practice that provides more information on the environmental selections and constraints that animals encounter. Movement and migration data gathered from tracking devices can be related to environmental conditions, for instance, as one environmental data platform, MoveBank, makes readily available.17 In the context of an experimental ecology that translocates tagged birds away from their migration routes and then observes their eventual (correcting) movements, it is possible to draw inferences about the way that birds pick up on the movements in winds, the possible cues made available through shifts in atmospheric conditions, and the flight paths taken, as well as wing beats, heart rate, and energy expended along a chosen path. As ornithologist and advocate of animal tracking Martin Wikelski has indicated in relation to this example, “We can use individual animals as sentinels for the atmosphere if we understand how they use the atmosphere.”18 Through flight path, heart rate, and acceleration data (among other variables), the journeys of tagged animals can provide indirect data on the churnings of air. These data can be further corroborated by comparing environmental data to flight conditions, creating a record of the particular milieus that animals might be navigating through and inhabiting.

Every Animal with a Cell Phone

The technologies used to track animals typically consist of devices that communicate through radio and satellite telemetry, including 3G and 4G mobile phone signals, GPS, Argos satellite systems, RFID tags, and data loggers.19 While most tracking devices are attached to the external bodies of organisms as collars, backpacks, and epoxied antennae in the case of insects and smaller organisms, some tracking devices are subcutaneously injected (not unlike the RFID tagging of many household pets) and are especially focused on logging heartbeat and temperature. Monitoring technologies are becoming increasingly miniaturized, and the expectation is that a greater number of animals may be monitored through more sensory variables over longer periods of time.20

Presenting research at the Symposium on Animal Movement and the Environment 2014, Wikelski recounted a project focused on studying families of geese in Siberia in order to understand their social interaction. Describing the process of tagging the geese, he noted, “So you catch them, you put tags on them and this is really how the world should look like: every animal has a little cell phone, they talk with us.”21 Wikelski indicates, however, “There are still some problems because these things should be much smaller, and they should all be on necklaces.” Despite these considerations, all in all, the tracking of geese was “working well,”22 and was providing new data about their social interactions, as well as the ways in which rising temperatures lead to migration activities. This is seen to be a way, ultimately, to “get closer to the decision-making process of these geese.”23

Within the proliferating range and type of tracking devices in use, it is possible to combine sensor data with a wider range of data. In addition, Wikelski has noted that there is a further need “to have very miniaturized cameras on these animals,” similar to those used on the CritterCam project (and as discussed in chapter 2) in order “to ground truth other sensor data.”24 As Greg Marshall of the CritterCam project has noted, initially the idea to attach cameras and monitoring devices to animals seemed improbable, and the consensus was that animals would not tolerate carrying the devices. However, as he suggests, “most animals seem to care little about the unusual electronic remora appended to their backs. This unexpected finding has increasingly emboldened researchers to consider use of animal-borne imaging tools to study difficult-to-observe animal behavior and ecology.”25 As a result of this and the miniaturization of monitoring technologies, there are now

With such extensive plans for instrumentation, across sensor modalities, animals move from being cell-phone equipped to becoming multidimensional sensing nodes that communicate their own bodily conditions, interactions with neighboring animals, movement, and environmental conditions. In this sense, Marshall notes in relation to marine organisms, “animals themselves can now serve as remote ocean observation platforms carrying instruments to characterize habitat over temporal and spatial scales relevant to their basic biology and life histories.”27 Through this ongoing data collection, animals become sensor nodes and platforms, from which some of the “unanswered questions in ecology” are meant to be addressed, while also creating an expansive and even global animal-sensor network that functions as the “pulse of the living planet.”

Pulse of the Planet

The notion that animal-sensor networks are providing the “pulse of the living planet” comes from one of the most notable and sizeable initiatives to undertake animal tracking on a global scale: the satellite-based project International Cooperation for Animal Research Using Space, or ICARUS. A global small-animal tracking system led by Wikelski, ICARUS seeks to set up a “remote sensing platform for scientists world-wide that track[s] small organisms globally, enabling observations and experiments over large spatial scales.”28 While the Argos satellite system has primarily been used for tracking and sensing larger animals, as it requires larger transmitters for communication,29 ICARUS is able to work with smaller tags that communicate locations across shorter distances. The anticipation with this technology is that ICARUS will allow insights into the “dispersal and migration” of smaller organisms, which will “provide a seeing-eye dog for humankind”—in other words, it will enable the use of “the evolved senses of animals for remote sensing.”30

While ICARUS is set to be launched as an antenna added to the International Space Station (ISS) in 2015, it is also seeking to partner with Russian, Chinese, and European Space Agency (ESA) satellite launches. The ISS antenna is proposed to communicate with small tags weighing less than five grams (and which Wikelski projects will further reduce in size) and which consist of a logger tag with GPS.31 There are many perceived possible uses of ICARUS that “will enable researchers to answer some ‘grand challenges in environmental sciences,’” including understanding the spread of diseases carried by animals, protecting sites key to migration pathways, and establishing relationships between biological diversity and ecosystems, as well as establishing a disaster network based on animal sensing.32

These multiple uses of ICARUS for research are based on animals serving as continual generators and collectors of data, which can further be collected in the MoveBank database and compared to corresponding environmental data. All together, these animal-generated and sensor-based data are meant to provide an ongoing picture of the pulse of the planet. The movement of organisms, the fluctuations of animal populations, the responses to deforestation or other land-use based events and changes, may all register as flickering patterns of information, the cadence of a living planetary body: here is another iteration of “program earth,” articulated through the digital monitoring of the movements of innumerable earthly organisms.

More Data = More Engagement

The impetus to collect more data as a way to achieve greater insights is one that runs through environmental sciences and is a key way to address environmental change. As discussed in chapter 1, the proliferation of more detailed and more real-time sensor data is also meant to provide fundamentally new insights into ecological processes. These data-based insights are further meant to bolster conservations projects, so that more sound and effective decisions can be made. Monitoring, in this sense, is a practice that is undertaken ultimately to protect organisms.33 The “great migrations” are declining,34 the extinction of species is occurring at an unprecedented rate, and organisms that are vital to food chains are collapsing.

One project, the Tagging of Pacific Predators, or TOPP, deployed 4,306 tags across 23 species to study events such as the possible collapse of bluefin tuna.35 In a video outlining the aims of the project, the point is made that these tagging initiatives will not only “unlock the mysteries of the deep” but will also provide new information that can be communicated to publics. As Warner Chabot notes, “The more that we can give the public the facts they will respond. If you inform and inspire the public you will empower the public to respond and, frankly, act in its goodwill and future.”36 More detailed information is a resource that is meant to motivate conservation efforts, inform policy, and stir publics into action.

The drive to collect more data on organisms as they move around and undertake migrations has in this way influenced numerous citizen-sensing projects. While animal spotting and bird ringing are long-standing practices within citizen science, many more projects are emerging that make use of digital devices to track, record, monitor, and observe moving and migrating organisms. Indeed, in one example of GPS-based citizen science and environmental monitoring, I can recall a campaign made by the Wildfowl and Wetlands Trust (WWT) in 2002 for citizens to “adopt a goose,” which would be fitted with a GPS tracker to relay data about a particular goose’s journey to the citizen-adopter. As a sequel, perhaps, to the transmitted heartbeats of the Russian dog Laika launched in Sputnik 2, the WWT conservation project monitored light-bellied brent geese and their migrations from Canada to Ireland. The public could adopt geese with such auspicious names as Major Ruttledge and Arnthor, and in exchange receive up-to-the minute information on position, speed, and heading, delivered via email or to mobile phone.37

Many citizen-sensing engagements then focus on both gathering data as well as interacting with the greater stores of data now available on animal movements. Gathering information in some cases is the trigger for particular conservation-based actions, including BirdReturns, a project led by the California Nature Conservancy that incorporates citizen-science data from the eBird project out of the Cornell Lab of Ornithology, which maps the migratory patterns of birds as they move through the Central Valley of California. Participants in the project can further submit observations through a web platform or app, BirdLog North America,38 and all together this citizen-supplied information is mapped onto critical wetland zones. The Nature Conservancy then rents the space from farmers via a “pop-up habitat” scheme, where the farmers allow their land to remain flooded during critical times when birds are migrating through the valley.39

The Migratory Connectivity Group notes that multiple citizen-science and citizen-sensing projects actually tend to concentrate on migratory organisms, including birds, invertebrates and fish.40 Projects in this area working through distinctly digital methods and devices further include the “Tag a Tiny Program,” which enrolls the help of recreational fishermen to catch, measure, tag, and release juvenile Atlantic bluefin tuna in order to study their “annual migration paths and habitat use.”41 Many more citizen-sensing projects focus on animals as they move and migrate, from tagged sharks that send tweet alerts when they approach popular beach areas in Australia, to records of migrating eels in the River Thames, as well as projects such as Roadkill Garneau, which asks citizens to record and locate roadkill sightings using EpiCollect, an online platform and mobile app, so that critical sites where movement has gone wrong can be recorded.42

There are even videos instructing lay audiences on how to build their own DIY GPS tracking kit by hacking a GPS device to build a bespoke radio collar.43 And creative-practice projects have created distinct ways of engaging with animal movement by, for instance, being able to engage in a text-message exchange with fish in the Hudson River in New York City. “Amphibious Architecture,” a project by Natalie Jeremijenko, David Benjamin, and Soo-In Yang, uses motion sensors, LEDs, and a text-message system to trigger alerts to passersby who may tune into the movements of fish. Sample text messages note, “Underwater, it is now loud. To find out more, text ‘HeyHerring’ or ‘AhoyAnchovie’ or ‘GreatEast.’” This proof-of-concept project enters into those communicative exchanges with animals that tracking and sensing technologies are meant to enable, albeit with a slightly different approach to the messages that might be shared.44 In a different way, in her “Extreme Green Guerilla” project, Michiko Nitta has proposed that we might harness the movements and migrations of animals as an alternative and even “green” communication system, where our messages might be more efficiently and ecologically carried by animals crossing oceans and continents.45

A common thread across these scientific and creative-practice projects is that communicative exchange unfolds not through speech, but rather through perceptive engagements built up through environmental inhabitations. The prevailing sense with tracking projects seems to be that this is a mode of communication that may be readily accessible to us, where by observing organisms it may be possible to deduce their environmental requirements. Watching, spotting, and reporting journeys; tagging and contributing to scientific monitoring; and amassing collections of migratory data—within and through the interstices of movement ecology projects—multiple projects are contributing to building up more detailed accounts of animals’ movement and migration.46 And in this watching and encountering of organisms, humans, more-than-humans, and organisms are moving through intersecting milieus, forming new nexuses of sense.

Animals as Sensors: Badgers, Elephant Seals, and White Storks

Data sets that are more complete and comprehensive are meant to fill in the blank spaces on our maps of animal movement so that we might “build a global picture of the creatures with which we share this world.”47 With animals serving as sensors and sensor networks, sensor data is meant to function not only as descriptive data but also as material that allows us to infer events from what animals might be sensing and responding to in environments. Animals-as-sensors become subject-superjects in a particular way within tracking projects, where their journeys are meant to communicate the experiences of their environmental encounters. The becoming environmental of computation here occurs through the journeys and tracking that unfold as sensors travel with organisms, as well as through the ways in which organisms become computational both as carriers of sensors and through the ways in which their sensory ecologies are meant to provide data and information on environmental conditions. Organisms are thus made to be computational twice over, as they sense and are sensed. I now turn to consider three specific journeys or movements of animals that attend to the ways in which animals-as-sensors concresce as indicators of specific engagements with milieus.

Badgers Socializing in Wytham Woods

WildSensing, an interdisciplinary collaboration between computer scientists and ecologists based at the University of Cambridge and Oxford University that took place between 2007 and 2010, involved a study of badger activity in Wytham Woods near Oxford—a highly instrumented test site known for its ongoing ecological experiments from at least the days of Charles Elton, an ecologist well known for his studies of population ecology and animal invasions in the early to mid-twentieth century.48 Wytham Woods is a 390-hectare landscape that is “one of the most researched areas of woodland in the world,” with numerous monitoring projects underway at any given time.49 But many of these projects are often set up in relation to distinct research questions and concerns and do not join up data sets collected from the site. At the same time, because ecological study and experimentation have taken place over several decades at Wytham Woods, there are extensive data sets and histories of animal observation. With badgers, for instance, data collection extends over the past twenty years, although it may have a larger granularity due to manual observation methods; and badgers have been trapped and released in Wytham Woods for the past thirty years (which has been the usual way of studying animal movement).

The WildSensing project was initiated to establish whether to and to what extent badgers transmitted tuberculosis, for instance, to livestock. Data from these observations were meant to aid in policy and management of badgers at agricultural edges.50 To undertake this research, the project focused on the social networks of badgers, since as it turns out they have distinct modes of interaction and cooperation. In total, eighty badgers were tagged and caught once every six months over the duration of the project. Animals were tagged with RFID radio collars, which would be released when badgers where thinning. As a result of using RFID for detection, badgers could be sensed underground as well as above ground, but only within the sensor area and not across the entire forest.51

In the first iteration of the WildSensing project, badgers were tagged with RFID radio collars that communicated with fixed sensor detection and storage nodes located within a zone of the forest. From these points, field researchers could conduct mobile data collection (which could, theoretically, also be carried out by mobile robots). Within one year, the project collected over twenty-five million records, and so the gathering and transmission of data presented issues for how to structure these networks.52 Due to the quantities of data collected and transmitted, much of the project focused on ways of duty-cycling data more efficiently in order to save power, which is an ongoing issue within sensor networks.

In the second iteration of the project, an increasing emphasis was placed on working with off-the-shelf sensor equipment. Rather than having fixed sensor nodes in the network, the project instead used the badgers as the mobile sensor network across which data circulated to fixed collection nodes triggered by presence detectors with a fifty-meter radius. The data from these nodes were then either stored on SD cards or transmitted via 3G mobile phone networks several times per day to servers. On the one hand, this approach focused on how sensors learn and adapt to animal behavior. Working with RFID sensors and machine learning in the form of an adaptive algorithm, this approach focused on having sensors operate in response to and at key moments of animal activity. On the other hand, as sensors and animals were paired in this form of environmental monitoring, sending new software over wireless networks to the animal collars also became a way to reprogram sensors without having to catch the animals or adjust the sensor hardware or infrastructure so that the network could be adapted to animal activities.53

Emerging within this approach is the use of sensors not just to describe and capture environmental events but also to develop a dynamic evolution of sensors in response to animal behavior such that computation and the distribution of sensation are ontogenetic. While critiques of early tracking devices suggested that they were “‘mere descriptions of movement and activity,’”54 and hence at times considered to be relatively static renderings of environmental processes, increasingly sensor systems are regarded as generating more integrated, adaptive, and actuated approaches to environmental monitoring.

As the WildSensing mobile network developed, it became a system for relating information from animal to animal via radio collars and then on to collection nodes. Animals became sensors and operators in the network, at once collecting data about their activities and location, while also becoming part of the extended computational infrastructure. The network patterns were ad hoc, based on the badger activity, and were not entirely preestablished configurations. The social behavior of the badgers, as well as the microclimate and other environmental conditions at Wytham Woods, contributed to the intersections of technical and living milieus. The sensors and computational network necessary to capture phenomena had to emerge along with ecological events and animal activity, where, for instance, practices of relaying data across organisms and storing sensor data in nodes, then capturing the data through mobile collection, developed as a more effective configuration for sensing the badger activity.

Machine learning here extended not just to parsing environmental data but also to learning animal behavior and reprogramming sensing and collection methods accordingly. In this sense, sensors became organismal and environmental. While this was not a completely open process, as sensors are configured to detect certain variables and not others, it was also not a process of complete automation, where sensors might be preprogrammed to detect phenomena according to fixed configurations. If we were to follow Simondon in this regard, how might this contingent approach to sensing shift both technical object and technical milieu in relation to the individuations that occur through encounters with living entities? Rather than approach sensors as “prosthetic” devices, moreover, might we find it more accurate to consider the ways in which these sensor technologies reorganize, in-form, and transform along with the organisms they would track?

Elephant Seals Diving in the Southern Ocean

If the badgers of Wytham Woods presented a quite local and land-based sensor study, then the elephant seals of the Southern Ocean offer up a much different milieu in the form of underwater spaces, relatively obstreperous temperaments in comparison to badgers accustomed to recurrent catch-and-release, and sensor systems that communicate via satellite rather than more proximate RFID nodes. Led through the Natural Environment Research Council Sea Mammal Research Unit at the University of St. Andrews, the elephant seal study set out to research how these animals respond to environmental variability and how this variability might in turn influence population fluctuations.55

As a study of telemetry and marine mammals, the project was situated within a larger study of over one hundred mammals and marine species. The elephant seals were tagged with CTD-SRDLs in the first iteration of the study. Data gathered included depth, conduction, temperature, pressure, and acceleration, as well as stereo sound captured at 500khz. The tagged seals not only revealed profiles of their diving habits, where everything from buoyancy to fat reserves can be assessed based on diving details, but they also captured data on the temperature of the Southern Ocean at depths not typically monitored.56 Translations could then made from elephant seal data to the Argos satellite system to St. Andrews to the Met Office to the BBC weather forecast.57 Elephant seal data have also been integrated into ocean observing systems, so that the complex conditions of oceans become more thoroughly monitored through the underwater activities of tagged marine animals.58

During a 2010 Mammal Society conference focused on techniques for sensing and tracking animals, one of the presenters discussing this elephant seal project considered the possibility for moving from Argos to a GPS/GSM mobile phone system to relay data. Such a system effectively would involve “stick[ing] a mobile phone on seals as a point of connection,” and could become a method for communicating with publics.59 Here, by bringing animal tracking and communication into the realm of mobile phone networks a more immediate contact with the animals would appear to unfold—similar to the geese with cell phones discussed earlier in this chapter. Such a strategy of communicating with publics was in fact implemented when Argos was first used to monitor animals, where emails and updates of tracked animals were regularly sent to schoolchildren and publics.60 Yet the imaginary of moving from a satellite connection to mobile phone exchanges of data seems to bring the immediacy of animal communication even closer, particularly when animals are sensors of environments.

This is the common thread that arises in public presentations of animal tracking projects, where the ability to communicate with animals is frequently referred to as one of the benefits of these sensor systems. Indeed, the ICARUS promotional material notes, “With the help of new technology, animals will be able to communicate with us, revealing changes, dangers and connections that will help us to have a better understanding of the fabric of life on earth.”61 The ambition is that animals will “talk” to us, and in so doing that they will communicate the distinct sensory inhabitations that they experience. Animals equipped with sensors become, in turn, active sensors able to perform heightened modes of communication. Through this talking, the hope is that we will finally be guided toward making better decisions for preserving the planet.

Tagged elephant seals communicate not only the specific data points of temperature and downward acceleration when they dive but also provide an indication of the multiple milieus that they cross, from the technical milieus of sensor devices, to the lived milieus of the seal, and the transformative or associated milieus across and through which new becomings concretize. This is a becoming environmental of computation and a becoming computational of organisms. Perhaps it is also the becoming organismal of machines. These points will be discussed in the section below “The Problem of Milieus,” but I here turn to consider the final animal-sensor journey in the form of white storks that are tracked with satellite transmitters and which also feature on the Animal Tracker app, which is oriented toward engaging citizens in monitoring animals.

White Storks Navigating Aerial Ecologies

The final sensor journey that I discuss in this chapter involves the tracking of white stork migrations across two main flyways through Germany, Greece, Turkey, Tunisia, and South Africa. Based at the Max Planck Institute for Ornithology in collaboration with researchers across multiple countries in the flyways, this study of white storks attempts to gather more complete lifetime tracking data across populations.62 The storks are fitted with solar-powered GPS transmitters that capture location and body-acceleration data every five minutes. Updates on the locations of the white storks are sent by SMS messages, but the majority of data are downloaded from relatively nearby VHF radio connections, which need to be no more than three hundred meters away from the storks. This means that scientists need to follow—and even “chase”—the white storks from their breeding areas and along their flyways in order to download data.63

Part of the interest in studying white storks is in relation to their importance as “sentinels” for environmental events. For example, storks can be found in areas where there are outbreaks of desert locusts, as they will congregate in these areas for feeding. In this way, the storks have been described as “advanced” remote sensors for providing insights into environmental events. Storks are studied for the patterns and energy expenditure of their migrations, how they interact, and where they stop to rest. Body-acceleration data that are gathered from the storks’ journeys allow researchers to estimate energy expenditure and behavior, which can further indicate environmental events.64 In addition, attention is given to where and why animals are dying, as this could influence land-use decisions about which habitats to conserve.

As many storks die in remote places in Africa, there is a need to retrieve transmitters and the high-resolution data they contain. In this way, another sort of citizen sensing emerges, although in this case participants are referred to as “collaborators,” who help to find and return the transmitters from the dead birds. From Malawi to Sudan, people return the birds and transmitters, often accumulating stories of how and where the storks may have died, including becoming entangled in debris from rubbish dumps.65

It is these practices of attending to where storks are, and retrieving them when they die, that Wikelski has suggested is the “future” of citizen science, when people are specifically engaged with observing animal movement. Wikelski has further suggested that animals may be better protected if people know where they are, since they are less likely to be hunted if they are watched over. To this end, the Max Planck Institute for Ornithology has developed a wildlife tracking app, Animal Tracker, which is a tool for citizen engagement that provides relatively real-time data about animal locations and allows us to “observe animals . . . virtually in our cell phones every day.”66

Some of the first tagged and tracked animals to feature on the Animal Tracker app are the eighty white storks that are under study.67 Users of the app can see the migratory routes of the white storks and put them on a “watch list” so that notifications are sent “when they do something”—and news of the storks can also be shared on social media while records of the data are kept in the MoveBank archive.68 As I click through the Animal Tracker app, I find a white stork located just outside of Nuremburg, Germany, in the small town of Forcheim. Named “Aldo,” this white stork last registered activity the previous evening, and is tagged with a DER AT881 (eobs 3946) sensor. I learn through searching outside the app that the digital telemetry company e-obs makes “high-end digital tags for the study of animal behavior,” with a focus on “lightweight GPS tags.”69 This is the device that allows the white stork’s movements to be transmitted and eventually displayed through the interface of the app.

Aldo was born in 2014 in Vorra, Oberfranken, Bavaria, and was one of three chicks in the nest. His siblings are Amos and Resl. I am able to click on Aldo’s two-week and one-year movement data, and I see from the two-week record of his movements that he has hovered around the town of Hochstadt for a while, then quickly moved over to Forcheim, dipping down to Erlangen, and then back up to Forcheim, where he currently rests. If I look at the record of his one-year movement data, I can see that he has not traveled far from the place of his birth. I can also mark Aldo as a favorite stork so that I can receive updates. Several weeks later, I see that Aldo and many other storks are on the move from Germany to the south of France, on to Spain and Morocco. While Aldo is near the Spanish Pyrenees I can zoom out even further to see stork movement down the second flyway, across Greece and Turkey in to Sudan and down to Tanzania.

If I happen to meet Aldo in the field, I can also add my own observations of his activities. The “add observation” section of the app asks:

I mark Aldo as a favorite stork, and begin to click through the map looking for other birds: they are everywhere, in small towns and mountain valleys, some even appear to be waiting at cafes, pinpointed on top of restaurant signs and in town centers. I learn about Wanderer, the stork born in a birch tree, and Isolde, who was born in southwest Germany but is now located in Parc Natural Régional de Camargue in the south of France.

Commenting on the possibilities of tagging and so sensing with and through animals, Wikelski notes,

Indeed, in this project of tracking animals and thereby “decoding the intelligence of animal behavior,” even further applications have been proposed for using animals as sensors. If white storks are sentinels for certain types of population fluctuations of organisms they feed on, for instance, other animals could be understood for the clues they provide about possible disasters that may be imminent. Wikelski has thus tested the anecdote of whether animals are able to predict disaster by tagging goats near Mount Etna and testing their movement patterns in relation to volcanic eruptions As the goats are sensitive to the eruptions, they demonstrate “strange and erratic behavior” that can be captured through tagging. The data gathered from mapping the goats’ behavior in real time could be used to trigger an alert for an imminent volcanic eruption up to 4 to 6 hours before the event occurs.

A patent is pending on this “Disaster Alert Mediation Using Nature,” which is an invention for “a method forecasting an environmental event” that involves collecting behavioral or physiological data from a population of animals, comparing it to a baseline data set, and establishing an alert for moments when a threshold is crossed in the comparison between baseline and current data. The patent includes a software program that is able to execute the steps necessary to analyze data and trigger an alert.72 Wikelski sees this alert system as “really useable” as a “technical device,” and is seeking investment in order to create “a global animal observation system” especially useful for “areas where people don’t have much money.”73

From sentinel white storks to citizen-sensing apps to global animal observation systems, animals are increasingly made into sensor nodes and networks that would inform us about critical environmental conditions and their responses. Yet what are the implications of these burgeoning animal-sensor networks? And what sorts of animal-human-milieu interactions might unfold through the more pervasive project of tagging numerous organisms? I take up these questions for the remainder of this chapter, specifically attending to the traversals made across organisms, sensing, data, and milieus.

The Problem of Milieus

The ways in which animals are becoming both sensor nodes and parts of extended sensor networks raise questions about how these tagged and tracked individuals traverse and inhabit milieus. In this discussion of milieus, both technical and living, I am drawing on the work of Simondon and Canguilhem, who in varying but shared ways were interested to account for the ways in which individuals (per Simondon) and organisms (per Canguilhem) are formed and in-formed by encountering “problems” in their milieus. As Canguilhem has suggested in his analysis of milieus, how organisms encounter the problem of their milieu is how they become. Yet these problems are different for different organisms.74 As Simondon similarly articulates, the problem of the milieu is a condition for inventive responses, which is also a condition for individuation.75 As milieus are sites of inventive encounters and responses to problems, moreover, it is not possible to limit the relations and capabilities that individuals might draw on and express in addressing the problems of their milieus.76

This approach to organisms/individuals and milieus has several points of resonance for thinking about the implications of tracking animals and using their movement patterns as extended sensor networks. Humans in the form of scientists and citizen scientists have largely formed the problem of milieus as one of gathering more data in order to address environmental change. In this sense, understanding how to respond to the problem of our shifting milieus has become a project of ensuring there are no “blank spots” on our maps of environmental change. This problem-logic is influenced by the notion that when data sets are the most complete we will assumedly have the most advanced ability to manage environments. In turn, the problem of our milieus has also become one of monitoring all manner of environmental phenomena, including tracking organisms for the clues they provide about the worlds that they inhabit and how their worlds may be changing.

There are a curious series of translations that take place across animal-sensed milieus, tagged organisms, and generated data, since we could ask whether organisms are having to inhabit our encounters with our problem-milieus by living with tags and tracking devices, potentially for their entire lifetimes. Yet how do these intersections of encounters with milieus transform animals as they encounter their milieus and the problems of their milieus: Does the situation of wearing tags and tracking devices change the ways in which organisms encounter their milieus, while also in-forming their problems? It has been recognized in scientific literature on movement ecology that tagging can and does change the activities of organisms.77 Questions have also arisen as to whether it is always instructive to tag organisms that are under threat, as the process of capturing, tagging, releasing, and monitoring may contribute to the stress of animals.78

But tagging and tracking are not just issues of intervention in order to gain a more accurate picture of organismal activity. There are also points of consideration about how monitoring devices and practices in-form the milieus and perceptive exchanges of organisms with those milieus, since this is also the very thing that would be mobilized, whether for conservation and policy or for disaster networks. Canguilhem has critically noted that a danger with some forms of science, such as physics, is that they can be based upon a universal milieu that speaks neither to the perceptive experiences of organisms nor humans. If science is in the world, however, as Canguilhem suggests, it must admit to a diversity of milieus.

Perception is the way in which organisms go about encountering and fashioning their milieus. Sensing is then a key practice for working through problems of milieus.79 As Canguilhem writes, “In fact, as a proper milieu for comportment and life, the milieu of man’s sensory and technical values does not in itself have more reality than the milieus proper to the woodlouse or the gray mouse.”80 No milieu or experience of a milieu is more real than any other, unless we adhere to the universal milieu of science, which establishes a version of the real that disqualifies all others.81 Following Whitehead, to account for the experience of the woodlouse and the scientist, we would have to make room for the “pluralistic realism” of environments and inhabitations.82 Yet this is not a description of an absolute relativism, but rather of accounting for the a/effects that different inhabitations within distinct milieus express.

Indeed, proposals to use animals as sensor networks on one level seems to take on the approach of diversifying the sensing-milieu exchanges that occur across individuals. The encounters of organisms with their milieus provide another empirical basis for understanding environments and make room for the experiences of other organisms. And yet, in attending to the diversity of exchanges within milieus, a consistent if universal mode of capture is employed in the form of sensing and tagging devices. Here, one might ask whether it is perception (rather than a milieu) that has been transformed into a universal reality, whereby sensing devices, the variables they would measure, and the unfolding of sensing processes are made generalizable across organisms as an exchange of information. These generalized modes of information-based perception, furthermore, might be described as distinctly cybernetic operations, where sensing of milieus produces information that is the basis for actuating and producing further effects in milieus. Rather than the physics of a universal milieu, sensors might have given us the cybernetics of generalizable perception and experience.

Working across Canguilhem and Simondon, one could then ask: How do milieus and perception shift, both for organisms and devices, when sensing is primarily undertaken and filtered through tracking and tagging technologies? Working laterally with Simondon’s discussion of the associated milieu, we could say that technical objects concretize technical milieus in a way that could be compared to Canguilhem’s articulation of how organisms at once encounter and concretize their milieus. The difference, following Simondon, between technical and living milieus would be the way in which living milieus can be self-reproducing, whereas technical milieus are self-reproducing only in distinct circumstances where they operate as natural objects, and even then they imply the contribution and intermediation of the living entities that made them—in other words, humans.

In traversing these different milieus, we could say that it is the living milieus of tracked organisms that begin to resemble the operations of the technical milieus of technical objects, since animal sensing becomes equated with computational sensors. By virtue of being equipped with sensors, animals’ perceptive encounters with their milieus are transformed into informational exchanges through computational sensor networks. A response to an environmental event is a sensor-actuator exchange of information. An adaptation to an environmental event is a calculative decision, arrived at through an analysis of energy expenditure and environmental cues. Organisms’ perceptual engagements with their milieus become informational not simply in the way in which they are in-formed but also as digital operations generative of computational data. Such an approach in part fits with the more recent notion that all of “nature” is composed of information and so is inherently computable.83 But it also coincides with the longer histories of cybernetics where informational exchanges have been put to work to explain everything from ecosystems to population collapse.

Sensing of environments is then generally understood with tagging and tracking studies to fit within an informational logic of sensing stimuli, transferring signals, and actuating responses. Yet in what ways might this informational approach to perception preconstitute the possible modalities and relations of individuals as they interact with their worlds? A flight path chosen becomes a matter of a response to wind direction and speed and an organism’s internal calculation about energy to be expended to reach a particular destination. Rather than this being a question of what is captured and what is not—a usual way of attempting to make room for all that is in “excess” of scientific endeavor—one might suggest this is a way of making particular worlds and milieus in which the problems of organisms are articulated and acted upon. Environments and environmental change become informational problems. These are the informational-environmental-organismal processes, in other words, whereby we are working through the problem of our milieus, which are increasingly sites of environmental concern, as well as presupposing the perceptual-milieus of other organisms. We might then ask how such an approach to working out the problems of our milieus might also in-form our possible becomings in relation to how to “protect” organisms and their milieus. The becoming environmental of computation and the becoming computational of environments are processes that concretize these extended political and ecological effects.

Machine and Organism

While we could discuss the animal-sensor networks that come together in movement ecology as hybridities or infoldings of sense, as discussed in chapter 2 (and throughout this study), I am interested to maintain a focus on the environmental operations of perception (rather than attend to different conjugations of subjects and objects, nature and culture). At the same time, it is useful here to turn to a particular discussion that Canguilhem raised in relation to machines and organisms that provides insights into the ways in which perceptive capacities may be understood, potentially through machinic, and later cybernetic, forces.

Organisms have circulated through computational and cybernetic imaginaries for some time now, from dolphins studied for sonar sensing and later taken up as a topic of interest by Gregory Bateson, to Nicholas Negroponte’s gerbil-based interests as displayed in the “Software” exhibition, and many more besides.84 Automata studies have looped, continuously it seems, across organismal and technological modalities of sensing: linking, comparing, and fusing these to arrive at a more perfect union.

In his chapter “Machine and Organism,” Canguilhem works through “the mechanical theory of the organism” to consider how philosophers and scientists alike often “have taken the machine to be a given,” not only as though it is the concretization of scientific theory but also as though it provides an originary template for explaining the functions of organisms. But he sets out to demonstrate how “biological organization” is anterior to machines, so that life cannot simply be described through reference or analogy to machines. Across Descartes to Taylor there unfolds a certain mechanistic analysis of organisms that accounts for some outputs and not others. From Canguilhem’s perspective, there is a need to “inscribe the mechanical within the organic.”85 He writes:

I read this assertion less as insisting on an essential organic foundation against and with which mechanical operations unfold and more as claiming that life is not reducible to any one version of mechanical rationalization (such as automatism, about which Simondon had much to say) because this approach would delimit what is otherwise an opening into experience and potential. The “function” of organisms cannot be definitively narrowed down to a singular process, since this would be to further obviate the potential for inventive encounters with the problem of milieus.

Canguilhem’s grappling with machine and organism extends to Simondon’s consideration of technical objects and speaks in particular to his critique of automatism (and Wiener’s version of cybernetics) as reducing technical engagements to limited functions without the potential for becoming or invention.87 Simondon was critical of both the conflating of animals and machines as basic units of responsiveness within Wiener’s cybernetic theory (which seems a distinct continuation of the Cartesian legacy of automata),88 as well as the ways in which—as automata—organisms could be “capable only of adaptive behavior.”89

With sensors and tracking devices for studying animal movement, what emerges in part is just this cyberneticization of environments and organisms, where sensing becomes a way to home in on responsiveness, where movement and migration are assumed to be largely indicative of a series of adjustments and programs of responses that organisms make in relation to environments, and where sensing is an exchange of information. In this way, computation becomes the unquestioned originary machine that would in-form how we understand organismal and environmental processes. The “behavior” of organisms then becomes a series of information-based calculations and adaptations.90 For both Canguilhem and Simondon, however, living is an inventive perceptual response to milieus. The problem of milieus gives rise to becomings, and not mere adaptations. Technical objects and technical milieus, moreover, can be understood as particularly human-oriented ways of working through problems of milieus in ways that might be inventive, but which are inevitably expressions of value: of which problems matter, and how they are to be addressed.91 The potentialities of organisms (and machines) are thus individuated in these shared but differently articulated problems, milieus, and relations.

Sharing Perceptual Worlds, Environment as Participation

Where does this discussion bring us in considering the implications of computational sensing practices for tracking migrating and moving organisms? I have not sought to articulate a position for or against these practices. Instead, I have attempted to find a way to address how the activities of animals are parsed, what this means for our understandings of their perceptual worlds and milieus, and how we might return to consider that which our informational-based ambitions and methods tune us into. If these data gathered through tracking techniques are meant to influence citizen sensing, policy, and conservation, then how might we also make room for attending to the very particular worlds and inhabitations that are accounted for, and the milieus, inventions, and becomings that might remain off the computational map, no matter how many more data points we add? The point this raises in relation to monitoring the movements of organisms is that tracking technologies are expressive of computational and cybernetic logics that parse organismal “behavior” as a particular response to milieus. But this articulation of relations is not definitive. Instead, it makes particular worlds within which we understand the goings-on of animals. It is another versioning of a programmable and programmed earth.

Part of this project might then involve attending to the nexuses of sense that are formed through distinct subject-superjects, as Whitehead puts it in his “philosophy of organism,” since it is at these points where “actual worlds” form that are specific to perceiving and experiencing subjects.92 This is also why, as I articulated in the introduction to Program Earth, we might consider the multiple earths that concresce through the project of programming the planet, since not only is the planet “wired up” in numerous ways but also there are many organisms that differently express their experiences and attendant worlds as part of their inhabitations.

In this study that works across Simondon and Whitehead, among others, it becomes ever more evident that what we take for a subject or individual is not a pregiven matter. Instead, individual entities are articulated through these processes of individuating and concrescing. Why is it important to return, continually, to this point? Because it is through addressing the entities, experiences, and worlds that are engaged in processes of becoming that we might attend to the ways in which feelings traverse worlds, organisms are affected, and individuals feel themselves to be more-than-one. As Combes writes in relation to Simondon:

We might say that sensor technologies—across uses that involve tracking and much more—assume that the perceptive problematic is settled, and the task that remains is to document the facts of movement in order to account for behavior. In this respect, we might say that a certain cybernetic mechanization of organisms and their milieus has settled into practice, whereby informationally based modes of sensing eliminate both perceptive problematics and the possibility for inventive experience. As Simondon reminds us, the perceptive problematic is also a matter of participation, where perceptual encounters with milieus are articulations of a feeling for more-than-one, and a feeling for worlds.

Simondon has then suggested that the ongoing resolution of problems in a milieu is the basis by which an organism continues to individuate itself and be individuated. Perception, in this case, is not a matter of an organism decoding an external form for calculative gain but of articulating relations within their milieus.94 All of the entities involved in individuating, perceiving, encountering, resolving, relating, and worlding can shift in these processes, which concretize through participation in milieus. As Simondon writes:

From this perspective of encountering individuals through processes of individuation, Simondon further notes, “It now becomes feasible to think of both the internal and external relationship as one of participation, without having to adduce new substances by way of explanation.”96 Participation is a way of working through the problems of individual entities, milieus, and relations. These are conditions and entities that form through the very processes and experiences of participation. Participation is in-formative and inventive, rather than a register of programmed behavior or responsiveness.

For this reason, organisms might also disrupt the projects and studies in which they find themselves, from destroying cameras to removing tags to disappearing from the radar.97 As Stengers has noted, “The construction of an experimental device” in no way “ensures that the being we wish to mobilize will agree to show up.”98 Rather than seeing these as instances of bad data or outliers in a research study, we might instead tune into these events as instances where participation in actual and perceptual worlds occurs in different registers, often in relation to different problematics, where one milieu or perceptual problematic is no more real than another. The worlds and milieus to which we attend are specific expressions of problems and commitments. The fact that these are more-than-one makes for what Stengers terms “cosmopolitics,” which admits the coexistence of different and even contrasting practices and worlds.99

In order to move beyond the delimitations of machines and organisms, technical and living milieu, we might finally consider how, following Simondon, it might be possible to adopt an approach to the technicity of sensors and tracking devices in order to consider the sorts of relations that they put into play and how these relations are not the only possible modalities for being and becoming.100 By addressing the technicity of sensors, we might find it is possible to articulate different perceptual problematics as well as different potentialities of sense, while considering how these potentialities are bound up with making milieus and actual worlds.

Even more than studying the computational object of sensors, by account-ing for the technicity of sensors we might consider the participatory relations of sense, as well as the becoming environmental of sensor devices across particular modalities and with specific organisms and milieus. While tracking devices provide one approach to encountering the perceptual world of organisms through measuring variables as indicators of behavior, there also are ways in which we might consider how, following Anna Tsing, “our observations of non-humans present continual challenges to our cultural agendas that require new inflections and transpositions of our cultural ‘sense.’”101

Organisms are tuned to particular problems in their milieus. They are affected in differing ways according to their interests as well as what the milieu proposes—and they have effects on their milieus. This is another way of coming around to the discussion of affect and being affected, a topic traversed by Spinoza and Simondon, as well as Latour, Stengers, and even writers on environmental topics, where the capacity to be affected might further in-form our conservation and environmental practices. Affect has to do with participation, and it may further spark an ethos in relation to environments. Encounters are critical to this process, since as Didier Debaise has written, “The relation between the ability ‘to be affected’ (passive potentiality) and ‘to affect’ (active potentiality) is complex, as the living can neither be explained by its environment nor by its components. Everything happens in the encounter.”102 We might, from this perspective, approach animal tracking as well as citizen sensing as practices that unfold as necessarily inventive encounters. At the same time, at these nexuses of sense and in this becoming actual of worlds, we might consider how our information-based environmental problems affect and in-form the problems and milieus of other organisms, both in their lived actuality and in the ways in which we generate problems to be acted upon, of and for these organisms.