Technology is often portrayed as a tool to measure or manage the environment: satellites map deforestation; databases record biodiversity; models predict future climate scenarios. In that view, hardware and software are infrastructural supports, external to the ecosystems they observe, props for ecological knowledge. Yet, they are also ecologies in their own right: sensors, data logs, and servers are material assemblages within flows of energy and matter. They are maintained, repaired, and adapted; they consume resources, generate waste, and leave traces. Even software grows, mutates, and proliferates across platforms, like organisms in shifting habitats.1 In turn, ecosystems, often understood as nature’s own networks, have also been engaged with as technologies, like rivers as energy and transport infrastructures, animals as instruments, air and water as genomic archives of environmental DNA, and forests as carbon flux machines. In each case, ecology is mobilized as a technological system. Technology becomes ecology when infrastructures behave like environments, with their flows, leaky metabolisms, breakdowns, and adaptations. And ecology becomes technology when organisms, landscapes, and atmospheres are instrumentalized, enrolled as sensors, processors, and machines.2
Landscapes emerge as the medium where these roles converge. More than backdrops, they serve as the dynamic sites of interactions among organisms, matter, energy, and information. Landscape ecology has long emphasized this multiscalar dimension: patches, flows, and networks interweave from the molecular to the territorial.3 Additionally, environmental data is produced locally and logged into vast networks of sensing and monitoring. Environmental knowledge is co-produced by sensing infrastructures, and as ecologies become computationally legible, the legibility manifests itself as partial and political.4 Shifting from dataset to data setting helps emphasize these local arrangements—devices, protocols, staffing, and site histories—through which local signals become data.5 A climate-monitoring network, folded into the flows it measures, is therefore not a window onto an ecosystem—it is part of it; it shares space with it. A flux tower measuring carbon exchange alters the rhythms of the forest around it; a soil sensor buried in a field participates in the microbiome it registers. When data from such points is extracted and circulated into databases and predictive models, landscapes are reconfigured as informational platforms, where ecological and technological processes are inseparable. Environments become instrumented media whose states and data are not “found,” but produced through networks.
Observation is inseparable from intervention: each act of measurement reconstitutes the environment as a model-in-use. Once physical, biological, or chemical traces circulate through sequencing pipelines, databases, and machine-learning algorithms, a stream becomes a repository of genetic signals, a beehive becomes a communication node, and information becomes one of the landscape’s ecological flows, alongside nutrients or sunlight. Nonhumans (and humans alike, including cases of citizen science) are enrolled as sensing subjects in a distributed, low-power data ecology, where air, water, and organisms “pre-process” signals before sequencing begins. Once feedback loops return this data to models, policy, and management, new landscapes are produced at the level of representation and governance. In this light, landscapes appear as technological systems because they are covered with sensors, instruments, and infrastructures, but even more fundamentally because their functioning is increasingly conceptualized and enacted as a programmable milieu. Ecosystems today are recast as distributed networks animated by recursive processes of sensing, processing feedback, and adjusting actions. To think of landscapes this way doesn’t deny their material vitality, but acknowledges that ecological and technological processes are coextensive. A wetland is at once a habitat and a data array; a coastline is both an ecosystem and a sensing network. The challenge is to read these hybrid systems as ecologies that are simultaneously material, biological, and computational. To inhabit a landscape today is therefore to inhabit a media ecology as much as a biophysical one, where what counts as environment is shaped by the infrastructures of its perception.
Sensors, platforms, and protocols register environments while also composing them into programmable milieus.6 In their becoming-environmental, computational media confront the question of place in monitoring sites. In doing so, they reveal a constitutive friction: all amassed records are no more than indexes to local knowledge,7 even as no datum can precede the infrastructures that format it.8 Monitoring infrastructures—a sensor in a riverbed, a flux tower above a forest—make this condition visible by configuring ecological processes into a system of variables and parameters. Jennifer Gabrys names these sites “ecological observatories”: forests, rivers, and experimental plots where sensors, standards, and software bind organisms, soils, and atmospheres into durable measurement media. Observatories turn ecosystems into continuous, multi-scale data infrastructures.9 They reconfigure the landscapes around them as sites where interactions unfold that can be observed, represented, and—most importantly—recomposed. With monitoring becoming operational—comprising continuous, near real-time rather than occasional or periodic measurements, often carried out by autonomous systems such as drones or terrestrial robotics—ecological observation shifts from discrete research campaigns to an ongoing service.
In the Danish context, recent studies deploy in situ monitoring with modular sensor networks, at times enrolling living organisms as sensing nodes.10 Across aquatic and terrestrial settings—marine coasts, freshwater streams and lakes, wetlands, forests, agricultural land, and rewilded sites—researchers combine environmental DNA (metabarcoding, qPCR, CRISPR), camera traps, and networked sensors to capture ecological signals. Citizen-collected eDNA, for instance, has detected coastal marine fish at national scale, while camera trapping has recorded wild vertebrates in wetlands.11
This work moves beyond treating organisms as mere objects of study to enlist them in the architecture of measurement itself. Under the rubric of distributed sensing, the environment is configured as a dispersed apparatus: aerosols, water bodies, and citizen observers become collection surfaces and relays, folding social participation and airborne particles into the same networked ecology.12 Plants and animals, like bees, are fitted into modular arrays—carrying temperature, vibration, or chemical signals—whose telemetry and storage pipelines mirror other infrastructural systems.13 Human–environment interfaces couple ubiquitous sensing to perception, casting monitoring as an immersive—at times empathic—mode of encounter rather than a purely extractive one.14 Taken together, these projects conceptualize landscapes as distributed information systems in which high-throughput sequencing, real-time platforms, and machine learning interleave with field protocols. Organisms and ecosystems appear not simply as subjects of observation but as co-constitutive components of the networks that register, transmit, and act upon ecological change.
Experimental observatories, and by extension, the landscapes they refer to, function as media-infrastructures through three concrete modes: observation of the long-term infrastructures that format metabolism into durable signals; simulation, where life is rendered as algorithms that produce emergent environments; and manipulation in environments programmed to perform futures in situ. Together, these modes enact the landscape as an ecology and a recursive informational model characterized by distributed sensing, embedded actuators, and feedback loops that both reveal and shape environments and environmental processes.
ICOS: Long-Term Observation
Since 1996, the beech forest at Sorø, located in the Sjælland region west of Copenhagen, has hosted one of the world’s longest continuously operating carbon-flux towers. Run by DTU Sustain as part of the European Integrated Carbon Observation System (ICOS), the site records more than a hundred ecosystem and atmospheric parameters every thirty minutes. Its instruments register carbon dioxide, methane, and water vapor exchanges between forest and atmosphere, alongside meteorological conditions and soil processes. Each half-hourly data point accumulates into a series that now spans almost three decades, one of only three flux records of such duration worldwide.15
The forest canopy and the tower’s sensors form a coupled system: the flux tower needs the forest’s respiration and photosynthesis to yield data, while the forest is continuously reframed through the apparatus’s metrics. Photosynthetic uptake becomes a seasonal curve; drought stress appears as a deviation in the carbon balance; an extreme weather event is inscribed as a spike. What emerges is a long-term negotiation of what counts as “normal” forest function. The forest, through the tower, becomes an algorithmic object, its ecological processes formatted into standardized variables and data streams.
At the same time, the Sorø observatory is more than a local experiment. As part of ICOS, it contributes to a pan-European greenhouse gas research infrastructure that standardizes methods, calibrates instruments, and makes data openly accessible. The data harvested several times daily from Sorø is uploaded to ICOS servers, where it joins parallel series from Finnish peatlands, French dairy farms, and Alpine meadows. These standardized data flows form the basis for climate modeling, international negotiations, and carbon accounting regimes. The beech forest thus participates in a continental sensor network, its metabolism rendered commensurable with other ecosystems through protocols of observation and exchange.
The political significance of ICOS lies in this coupling of temporal depth with infrastructural governance. Besides what it can reveal about forest acclimation to climate change, the value of the Sorø series lies in its continuity and comparability, which make it authoritative. Long-term, standardized data is a prerequisite for both scientific generalization and political action. The forest is therefore enrolled as witness in debates about national carbon budgets and global climate policy, its cycles and shifting rhythms mobilized as evidence in arenas far from Sjælland. In this way, the Sorø beech forest is an example of infrastructural globalism, focusing on lasting standards, calibrated instruments, and shared archives. ICOS Denmark introduces infrastructures of duration and standardization within the environment, which itself becomes a recording device equipped with sensors and protocols. Its metabolism is then formatted into stable, transportable data—durable series that move into models and policy.
ALMaSS: Simulating the Countryside
Developed since the late 1990s, the Animal, Landscape, and Man Simulation System (ALMaSS) project has been one of the most ambitious efforts to simulate ecological processes across agricultural landscapes. Originating at Aarhus University, ALMaSS combines detailed geospatial maps of Danish farmland with behavioral algorithms for species ranging from skylarks and beetles to hares and voles. Its core idea is simple yet generative: by encoding organisms and environments as digital agents governed by rules, it becomes possible to observe emergent ecological dynamics under varying conditions of farming, pesticide use, and climate.
In ALMaSS, every element of the model is constituted through available knowledge, derived either from published literature or from field-based ecological data.16 A vole is a mammal implemented as a collection of routines: metabolism, reproduction, dispersal, and death. A field is a polygon coded with crop types, rotation schedules, and spraying regimes. When these digital entities interact, a landscape emerges in silico, as a synthetic ecology, wrong in its reduction of complexity, yet valuable for exploring potential outcomes.17 The vole population may collapse if pesticide regimes intensify; skylark nesting success may fluctuate with crop rotations; beetle survival may depend on field margins.
This computational countryside has been used to explore biodiversity conservation and pesticide regulation under scenarios that cannot be trialed ethically or practically in the field. In this way, the model becomes an experimental ground for environmental stewardship. It does not produce truth in the sense of correspondence, but plausibility in the sense of futures rehearsed and compared. Its generativity rests on standards and model “gateways” that let species routines, farm practices, and weather link up—a soft infrastructure for testing futures, where validation practices counter the opacity of complex simulations.
By translating ecology into code, ALMaSS enacts landscapes as assemblages of algorithms, recasting life as a computational process. Yet it also has an aesthetic dimension: its visualizations make ecological abstraction legible to researchers and regulators alike. By staging the countryside as a recursive space of computation, knowledge, and imagination, ALMaSS exemplifies how simulation operates not only as a scientific tool, but also within a media ecology that includes material infrastructures, a wide range of practices, and configurations of political power.
AnaEE: Programming Environments
If ALMaSS enacts ecology in silico, the Danish node of the European Research Infrastructure Consortium Analysis and Experimentation on Ecosystems (AnaEE) turns environments themselves into laboratories. AnaEE Denmark operates nine experimental platforms across forests, fields, heathlands, lakes, and streams. Unlike observatories that passively monitor, AnaEE’s sites manipulate ecological conditions to stage climate scenarios in situ.18
On the Risø peninsula in Roskilde Fjord, near a decommissioned nuclear research facility originally founded by Niels Bohr, Risø Fields spanned an eleven-hectare arable ecosystem, where crops would grow under a movable Free-Air CO₂ Enrichment (FACE) system that releases carbon dioxide into the air, raising concentrations to future climate levels. Warming arrays controlled soil and canopy temperatures, while irrigation and nutrient inputs were systematically varied. The fields acted as a programmable environment, a platform where possible futures were implemented in real time. Now, on the same site, an Ecotron—a controlled-environment ecosystem analyzer comprising replicated, enclosed mesocosms that enable precise control of conditions and continuous measurement of ecosystem processes—is set to open.19 This infrastructural shift reconfigures the landscape for enclosed, high-resolution experiments; an autographic environmental simulation.20 At Vestskoven, a newly planted forest, similar manipulations test how afforestation projects will sequester carbon or alter biodiversity: warming cables, CO₂ rings, and automated flux towers integrate the forest into a techno-ecological assemblage.21 AnaEE programs experimental conditions in situ: controllers regulate CO₂ concentrations, heaters adjust temperature, and irrigation systems manipulate soil water balance.
These infrastructures demonstrate that to monitor is always also to intervene, collapsing the gap between model and milieu. FACE rings and warming cables are not external devices but components of the ecosystem they act upon. They configure the field as an experimental system, a landscape that is at once habitat, apparatus, and model. In this sense, AnaEE exemplifies what it means for landscapes to become media infrastructures, sites where ecology is engineered to perform futures in advance.
Like ALMaSS, AnaEE depends on the coupling of hardware and software. Sensors log data continuously, databases aggregate them into series, and models extrapolate from them to broader scales. Similar to ICOS, information streams flow from the manipulated plot to national and continental networks, where they feed into policy scenarios and adaptation strategies. The site is therefore both local and distributed, experiment and infrastructure.
AnaEE rehearses climate change. Its interventions are approximations, simplifications that are useful in their performativity. While they do not predict the future with certainty, they generate plausible futures to guide decision-making. Politically, they transform climate change from abstraction into experiment, offering evidence that can be mobilized for stewardship, management, and adaptation. It therefore embodies a mode of experimental manipulation. While ICOS Denmark stabilizes fluctuations and gradients into durable datasets, and ALMaSS reconstructs landscapes in code, AnaEE programs landscapes directly, embedding them within infrastructures that remake environmental conditions. Such approaches underscore that ecology today is always enacted while observed, a recursive practice of modeling that composes environments as much as it describes them.
Evidence Ecologies
Experimental observatories and, by extension, the landscapes they refer to, act less as transparent representations and more as media in their own right. They generate surplus knowledge by enabling experiments, simulations, and projections that exceed immediate empirical circumstances. A vole in ALMaSS is a medium for testing pesticide policy. A tree in Sorø is a medium for inscribing decades of respiration and drought into standardized data. Sensors generate streams, models convert streams into knowledge, and archives stabilize memory for circulation. The surplus—plausible futures and governable variables—arises from the recursive triad of observation (ICOS), simulation (ALMaSS), and manipulation (AnaEE). Observations supply the parameters and baselines for calibrating simulations, while manipulations generate complementary data for both. These three interconnected modes each produce knowledge through different means: by maintaining long-term, standardized data series; by coding agents and environments to observe emergent dynamics; and by staging interventions in real-world settings. Each mode generates productive frictions: in the maintenance of archives, in the adjustments for homogeneity and instrument drift; against computational opacity and in the demands for validation; along experimental standards, logistical constraints, and in the challenge of ensuring comparability across locations.
Interoperability and shared protocols enable data to travel across different contexts, transforming localized measurements into global evidence. ICOS’s strength lies in comparability across sites; AnaEE coordinates experiments across nations and ecosystems; ALMaSS is legible to regulators through conventions that facilitate scrutiny. These infrastructures serve as tools of governance just as much as they are of science. Standardization is thus not a neutral background but an epistemic practice in itself. It determines what counts as valid observation, which variables are measured, how uncertainties are handled. Standardization also raises questions about what forms of knowledge become authoritative, and which remain marginal. These questions underscore that instruments of science are also infrastructures of governance, shaping how environmental futures are known, debated, and managed.
Acting as such interfaces, infrastructures like ICOS, ALMaSS, and AnaEE translate place into policy. One of the key consequences of these processes is the reconfiguration of scale. If the miner’s canary, a common instance where life itself becomes an instrument of proxy sensing, operated at the scale of a single mine, warning a handful of workers, ICOS inscribes the metabolism of a forest into global climate models, linking local respiration to planetary carbon budgets; ALMaSS collapses the Danish countryside into agents and pixels, allowing policy-makers to play out scenarios across regions; and AnaEE stages climate change futures on ten-hectare fields, extrapolating to national strategies. This scaling up requires decisions about representation, interpolation, and generalization. It transforms local ecologies into global evidence, often at the cost of erasing specificity. Yet it also enables forms of stewardship that would be impossible otherwise, like the coordination of carbon policies, the anticipation of climate impacts, and the design of conservation measures. Technological ecologies thus mediate between the intimacy of local environments and the abstraction of planetary futures.
Stewardship itself shifts in this context. Caring for an environment isn’t just about preserving habitats or species, but also about maintaining infrastructures, databases, and protocols. The technicians calibrating sensors at Sorø, the gardeners hosting FACE rings at Risø, and the programmers debugging ALMaSS code are all engaged in forms of ecological care. Stewardship becomes infrastructural: a matter of sustaining the conditions under which knowledge can be produced and futures imagined. They encourage us to adopt more local methods and treat environmental data as local—focusing on the data setting, not just the data set. How were AnaEE plots wired, staffed, and repaired? What is Sorø’s site history? They ask us to make place part of the data presentation, avoiding frictionless universals to reattach data to their specific settings and stakes, using data to build relationships and treat nonhuman participants as co-interpreters.
Infrastructural Stewardship
Taken together, long-term observation, simulation, and manipulation organize—through their descriptions—the conditions under which the environment is known and governed. They also co-produce the very ecologies they measure, installing the variables, thresholds, and time series through which landscapes become legible—and actionable.
Their frictions—opacity, standardization, maintenance—should be seen as political levers rather than mere technical nuisances, as they can prevent technocratic anaesthetics from displacing politics.22 Making archives durable, complex models interpretable, and interventions accountable is inseparable from deciding whose worlds are rendered visible, comparable, or disposable.
From this follows a practical program. If evidence infrastructures shape environmental futures, then environmental activism must be infrastructural: funding maintenance; mandating openness; demanding repairability; insisting that standards be publicly governed. Observatories and databases should be treated as commons, not as black boxes. Open, accountable, repairable ecologies of evidence that are wired to the climates they must help change must be built and cared for. These are not technical niceties; they are conditions for livable futures.
Paolo Patelli and Jussi Parikka, “Model Ecologies—On Disassembling and Reassembling Landscapes,” in Media matters in landscape architecture, ed. Karen M’Closkey and Keith VanDerSys (San Francisco: Applied Research and Design Publishing (AR+D), 2025).
Jussi Parikka, “Medianatures,” Zeitschrift für Medien- Und Kulturforschung 9, no. 1 (2018): 103–6.
Monica G. Turner and Robert H. Gardner, Landscape Ecology in Theory and Practice: Pattern and Process, 2nd ed. (New York: Springer, 2015).
Michael Hamilton et al., “New Approaches in Embedded Networked Sensing for Terrestrial Ecological Observatories,” Environmental Engineering Science 24, no. 2 (2007): 192–204.
Yanni Alexander Loukissas, All Data Are Local: Thinking Critically in a Data-Driven Society (Cambridge, MA: The MIT Press, 2019).
Jennifer Gabrys, Program Earth: Environmental Sensing Technology and the Making of a Computational Planet (Minneapolis, MN: University of Minnesota Press, 2016).
Dietmar Offenhuber, Autographic Design: The Matter of Data in a Self-Inscribing World (Cambridge, MA: The MIT Press, 2024), 11–12. Loukissas, All Data Are Local, 161.
Paul N. Edwards, A Vast Machine: Computer Models, Climate Data, and the Politics of Global Warming (Cambridge, MA: The MIT Press, 2013).
Jennifer Gabrys, “Ecological Observatories: Fluctuating Sites and Sensing Subjects,” in Field_Notes – From Landscape to Laboratory, ed. Laura Beloff et al. (Helsinki: Finnish Bioart Society, 2013), 178–87.
Dirk Zeuss et al., “Nature 4.0: A Networked Sensor System for Integrated Biodiversity Monitoring,” Global Change Biology 30, no. 1 (2024).
Sune Agersnap et al. “A National Scale ‘BioBlitz’ Using Citizen Science and eDNA Metabarcoding for Monitoring Coastal Marine Fish,” Frontiers in Marine Science 9 (March 2022). Rubæk Holm et al., “Holistic Monitoring of Freshwater and Terrestrial Vertebrates by Camera Trapping and Environmental DNA,” preprint, bioRxiv, November 24, 2022.
Christina Lynggaard et al., “Airborne Environmental DNA Captures Terrestrial Vertebrate Diversity in Nature,” Molecular Ecology Resources 24, no. 1 (2024).
Maximilian Drentschew et al., “Automated Environmental Monitoring with Remote Biological Sensors for Large Areas,” IFAC Proceedings Volumes 43, no. 23 (2010): 89–94.
Patricia Flanagan and Raune Frankjaer, “Rewilding Wearables: Sympoeitic Interfaces for Empathic Experience of Other-than-Human Entities,” in Proceedings of the Twelfth International Conference on Tangible, Embedded, and Embodied Interaction (New York: Association for Computing Machinery, 2018), 611–16. Brian Mayton et al., “The Networked Sensory Landscape: Capturing and Experiencing Ecological Change Across Scales,” Presence: Teleoperators and Virtual Environments 26, no. 2 (2017): 182–209.
Kim Pilegaard and Andreas Ibrom, “Net Carbon Ecosystem Exchange during 24 Years in the Sorø Beech Forest – Relations to Phenology and Climate,” Tellus B: Chemical and Physical Meteorology 72, no. 1 (2020).
“Short Introduction to ALMaSS,” ALMaSS, accessed August 22, 2025, ➝.
Wrong as in George E.P. Box’s “All models are wrong.” George E. P. Box, “Science and Statistics,” Journal of the American Statistical Association, 71 (1976): 791–99. C. J. Topping, “The Animal Landscape and Man Simulation System (ALMaSS): A History, Design, and Philosophy,” Research Ideas and Outcomes 8 (2022): 6. C. J. Topping, et al., “ALMaSS, an Agent-Based Model for Animals in Temperate European Landscapes,” Ecological Modelling 167, no. 1 ( 2003): 65–82.
Hans De Boeck et al., “AnaEE: A European Infrastructure for Future-Oriented Experimental Ecosystem Research,” EGU General Assembly 2020, May 4–8, 2020, ➝.
Jacques Roy et al., “Ecotrons: Powerful and Versatile Ecosystem Analysers for Ecology, Agronomy and Environmental Science,” Global Change Biology 27, no. 7 (2021): 1387–407.
Offenhuber, Autographic Design, 155.
“AnaEE Denmark.” University of Copenhagen, ➝.
Nicholas Mirzoeff, “Visualizing the Anthropocene,” Public Culture 26, no. 2 (2014): 213–32.
