Territory

The infrastructure of the Mississippi River was built to hold the river in place. Levees line both banks for over a thousand miles, the Old River Control Structure prevents the river from doing what rivers do, finding a shorter path to the sea. Navigation channels are dredged, spillways are opened and closed, pumping stations move water over and through barriers the river once overtopped seasonally. The system is enormous and its purpose is singular, to stabilize one of the most dynamic fluvial systems on Earth. And it worked. The river held. The certainty was structural and it was also epistemological, a belief that humanity’s technical prowess of measurement and engineering could pull the veil back on a complex, continental watershed and force it to perform predictably. Then the coast began to drown, starved of the sediment that seasonal flooding once delivered across three million acres of deltaic wetlands (USGS 1995).

Introduction

This dissertation begins with that paradox, that the infrastructure built to protect a landscape can destroy it by succeeding at what it was designed to do. Not because the engineers were incompetent but because the epistemology was myopic, because a paradigm that treats design as the production of stable solutions cannot account for what happens when the solution works perfectly and the system still fails.

What follows is not an argument against intervention. It is an argument for a different kind of knowing, one that treats design propositions as hypotheses developed through friction with dynamic systems rather than solutions derived from models that claim to contain what they study. The question that drove the research was never how do we control this? It was always, what is this trying to tell us that we don’t yet know how to hear?

I write as a licensed landscape architect whose research and practice sit at the intersection of computation, responsive technologies, and ecological infrastructures. For twenty years that work has been centered on coastal and deltaic landscapes, the territories where the failures of predictive control are most visible and most consequential. This dissertation is drawn from inside that practice. The arguments that follow are not observations made from a distance but conclusions produced through the friction of building, testing, and failing within the systems being described.

The Territory as Infrastructural Hybrid

The Infrastructural Field

The dichotomy of the natural and constructed is no longer a useful lens to focus the efforts of conservation and environmental design. The acute mosaic of fields, forests, and settlements forms a continuous infrastructural field. Interstates and railways thread through wetlands, pipelines conveying water slice through deserts, watersheds are rerouted for irrigation, and logistics platforms sprawl along spits of land in estuaries. The “natural” systems have been transformed by layers of infrastructure, now functioning as a synthetic field that is interwoven with control systems that are both physical and procedural. Foregrounding the environment as an infrastructural landscape highlights the extent to which anthropogenic needs for energy, mobility, waste disposal, and communications networks serve as the primary territorial organizers (Bélanger 2016).

Defining the Formed Condition

The territory as synthetic substrate is a formed condition, a landscape evolved by centuries of surveying, mapping, engineering, and regulation. Contemporary territories are deliberately constructed spatial organizations. The procedures of the territory, the zoning ordinances, subdivision plats, preservation policies, and highway standards, constitute an organizational space in which rules and protocols delineate the environment as decisively as physical infrastructure (Easterling 1999). The straightened channels and the Jeffersonian grid that dominates many environments are symptoms of a slowly evolving abstraction that produces spatial organization predicated on a desire for administrative predictability and ecological stasis.

A desire for abstraction runs deep as the legacy of modernism aimed to simplify complex systems into standardized, measured forms that nullify localized practices and suppress the feedbacks that allow systems to adapt to disturbance (Scott 1998). An administrative impulse drives the development of infrastructure. Systems are modeled and dimensioned according to predicted future events, the 100-year flood, projected sea level rise. An optimal configuration is calculated. Once constructed, the system is maintained in perpetuity. The future is treated as knowable, and the present is engineered accordingly.

An island or spit of land attains sovereignty when it appears on maps, receives a name, and circulates through media. Legibility, not material presence, creates territorial claim. The divergence between physical sand and cartographic inscription reveals the politics of representation, the degree to which the territory is constituted through its instruments of observation as much as through its material processes.

Climate Uncertainty and the Limits of Prediction

The Failure of Static Approaches

A practice that relies on environmental stability falters in an age of accelerating climate change. Global sea level rise can be measured and modeled yet remains fundamentally indeterminate in its effects and outcomes. The predictive approach fails to recognize the asymptotic relationship between modeling the future and living in an uncertain present. It abstracts historical conditions to build a narrative of a simplified and controllable present, thereby failing to recognize the agency of complex ecological reactions.

Solutions designed to resist climate change can paradoxically accelerate change in local systems. Tidal wetlands and their marshes persist when internal feedbacks, the relational underpinnings of plant productivity, sediment deposition, and hydrological inundation, keep pace with rising waters rather than retreating under accelerated rates of change (Kirwan and Megonigal 2013). In this context, the constructed levee or seawall represents a static solution that disrupts the dynamic processes enabling marsh persistence. The engineered structure, designed to protect against a predicted future, instead accelerates the degradation it was meant to prevent.

Embracing Ecological Dynamism

This critique of predictive control does not advocate for abandoning intervention but rather for reconceptualizing the relationship between design and environmental process. The challenge lies in developing approaches that work with, rather than against, the inherent dynamism of ecological systems. This requires shifting from a paradigm of control to one of cultivation and from imposing predetermined forms to nurturing adaptive capacities (Lister 2007; Raxworthy 2018).

Landscape architecture operating within this dynamism cannot rely on stable solutions maintained to function indefinitely. It calls for methods capable of learning from the systems they engage, interfaces that produce knowledge through the interaction of human, machine, and biological intelligence, and design frameworks that treat the ongoing negotiation between intention and ecological agency as the primary medium of practice. The decision about what to prototype, where to test, whose knowledge counts as evidence for revision, and who bears the cost of hypotheses that fail are not technical questions with technical answers. They determine whether a drowning island is a sediment budget problem or a question of responsibility and belonging (Allen 1999; Bélanger 2016).

My projects and research in choreographing sediment dynamics enable actors to create lands, whether topographic or bathymetric. The capability arrives before the ethics do. We learn how to reshape territory first and only afterward interrogate how that capability can be used or weaponized. That sequence is not incidental. It is the condition of practice in geopolitical projects, and it obliges designers to acknowledge that their knowledge is never structurally neutral.

The Territory as Nervous System

The territory is an evolving nervous system. Moisture probes in agricultural fields, pressure transducers in storm sewers, accelerometers across bridges, air-quality monitors on streetlights produce a continuous stream of data that redefines landscapes as arrays of measured behaviors and programmed alerts (Seibert 2021). Sensor networks track microclimates, carbon flux, animal movements, and pollutant plumes across temporal scales from seconds to decades. A rainforest wired with sensors measuring soil moisture, canopy temperature, and carbon exchange is no longer simply a patch of vegetation. It is known through its data and is tunable within that abstraction.

Bratton (2025) argues that computation is a planetary phenomenon, not merely a human industrial product, that it was discovered as much as it was invented. The territory as computational substrate is not a metaphor. It is the condition within which environmental design now operates. The question is not whether to engage this condition but how, and toward what ends.

The consequences of inattention are not abstract. The choice of where sensors are placed, what phenomena they measure, how frequently they report, and who has access to the data determines which communities receive protection and which face exposure. Centralized sensor networks often serve institutional aims while rendering the data less accessible to the communities most affected by its implications. Digital twins tighten the feedback loop between measurement and autonomous action in ways that can further mediate how environmental phenomena are experienced by inhabitants (Ye et al. 2023). When life is represented through dashboards, the experience of wading through waterlogged streets or watching a shoreline erode over generations risks being understood solely through abstraction rather than lived encounter (Greenfield 2013).

The sensing infrastructures that constitute the contemporary territory are not neutral instruments applied to a pre-existing condition. They are expressions of values that actively shape what the territory becomes. What is monitored becomes legible, what is legible becomes governable, and what is not monitored falls outside the frame of institutional attention entirely (Gabrys 2016). A marsh without a tide gauge is not merely unobserved. It is administratively invisible, excluded from the models that allocate protection and the budgets that fund it. The nervous system of the territory does not simply report on a landscape. It determines which landscapes are allowed to matter.

Recognizing Non-Human Actors

The formed condition of the territory is not solely the product of human intention and action. Non-human actors, from soil microorganisms to hydrological processes to plant communities, actively shape territorial development (Seibert 2021). Recognizing these material agencies requires moving beyond anthropocentric frameworks that position humans as the sole agents of environmental change.

This recognition has methodological implications. Design approaches must account for the ways in which materials and processes resist, exceed, or transform human intentions. The behavior of sediment in a tidal marsh, the growth patterns of vegetation, the movement of water through a watershed, these are not passive responses to human intervention but active processes with their own logics and tendencies. When a sediment diversion on the Mississippi Delta delivers material to a receiving basin, the sediment does not simply accumulate where engineers projected it would. It sorts itself by grain size according to flow velocity. It colonizes in patterns shaped by wave exposure and salinity gradients. It builds landforms whose geometry reflects the interaction of hydraulic forces and biological activity rather than the specifications of the diversion’s design. The resulting landscape is co-authored by the infrastructure and the material, and the material’s contribution cannot be fully anticipated.

Designing within these material processes requires a form of ecological intelligence that recognizes the relational nature of material behavior. The performance of a living shoreline depends not on the properties of its materials alone but on their interactions with tidal patterns, wave energy, sediment supply, and biological processes. Mical (2012) describes soft infrastructures that function more like ecological processes than engineered machines. They adapt and evolve in response to changing conditions rather than maintaining a fixed configuration. Maintenance, in this frame, is not the servicing of a finished form but an ongoing negotiation between designed intention and biological agency, what this dissertation calls the cultivant, a concept developed fully in Chapter 11. The territories examined in this work are always in process, always being tended, and the question of who tends, to what standard, and in whose interest is inseparable from the act of design.

The territory is a formed condition. Not a natural landscape with infrastructure laid on top but a product of centuries of surveying, engineering, regulation, and computation that has made the ecological and the administrative, the biological and the digital, inseparable. The levee and the marsh it starves are parts of the same formation. The sensor network and the community it renders visible or invisible are parts of the same formation. The zoning ordinance and the settlement pattern it produces are parts of the same formation. The territory was not formed once and left to persist. It is being formed continuously, by the infrastructure embedded within it, by the policies governing it, by the organisms inhabiting it, and by the data systems now reading it. This chapter describes that condition and the failures that emerge when design treats a living formation as a problem to be solved.

The formed condition of the territory cannot be addressed through solutions in the conventional sense. Solutions assume bounded problems with optimal configurations, yet predictive solutions have a greater chance of displacing or obscuring underlying conflicts than ameliorating them (Holmes 2020; Morozov 2013). What the condition demands instead are interfaces. Designed connections through which human intention, ecological process, and computational infrastructure meet and influence one another across time. An interface in this sense is not a screen or a control panel but a site of exchange, a point where different agents, biological, technical, institutional, encounter one another and where the terms of their interaction can be negotiated and revised (Allen 1999; Robinson and Davis 2018).

The distinction between territory and landscape is not merely conceptual. It carries different orientations to the communities, nonhuman species, and ecological processes that a site sustains. Territory is an abstracted space composed of objects and processes for the purposes of state administration. Borders, census tracts, tax parcels. Landscape is the living medium, the evolving array of materials and relationships composed and choreographed to produce life-catalyzing forms. Designing in the landscape rather than the territory means accepting the obligation to work with the site’s own logic rather than imposing the state’s management framework on it.

Synthetic Ground and Coupled Ecologies

Interfaces accumulate and over decades, technical objects layer with computational outputs and regulatory protocols until what emerges is a synthetic ground that is not a passive substrate, but a ground that acts. This is a socio-cultural and techno-ecological regime made material, shaped as decisively by extraction corridors and fiber optic cables as by the topographies that environmental phenomena produce (Shannon and Smets 2010).

A sediment pipeline feeding a marsh creation site offers one way into this problem. The pipeline itself is infrastructure and the deposited soils are situated in ecology. But the pipeline belongs to the same operational web as the material it delivers and the landscape that receives it. The soils support vegetation and that same vegetation traps more sediment. The question of where infrastructure ends and ecology begins assumes a boundary that no longer corresponds to system behavior. Territories operate as coupled ecologies, assemblages of mechanical, digital, and biological components bound through feedbacks that resist categorical separation. The coupling names the condition in which each component’s behavior is constituted through its relationships with the others, and in which the system’s intelligence exceeds what any component produces alone.

Tidal wetlands illustrate this condition, though the vividness should not be mistaken for simplicity. Whether a marsh persists or drowns depends on couplings the marsh cannot control. The internal processes of plant productivity, organic matter accumulation, sediment trapping matter, as do decisions made in offices and legislatures miles away, including levee placement, sediment diversion schedules, and dredging policy (Kirwan and Megonigal 2013; Temmerman and Kirwan 2015). Engineering choices alter plant physiology. Policy instruments entangle with sediment dynamics. Clear attribution of cause becomes impossible because the system’s behavior emerges from interactions that no single variable controls.

From an ontological standpoint, and the ontology matters here, these territories are assemblages of machines, where the definition of machine reads broadly. Rivers act in the sense that they constrain and enable other entities. Similarly ports act, legal codes act, wetlands act, and markets act. All can be understood as entities shaping each other’s possibilities without reserving agency for humans or devices alone (Bryant 2014). The patterns observable at any moment emerge from intersecting machine logics whose interactions exceed any single designer’s intentions.

Bach (2009) extends this insight from ontology to epistemology. If mind is software running on hardware, a pattern of information processing that is substrate-independent, then systems biological, computational, and ecological can produce outputs that exceed the specifications of their designers. A marsh that reorganizes its sediment dynamics in response to altered hydrology is not merely reacting. It is computing a response of its own, working out what the altered conditions require in ways that exceed what any model was built to project. The territory is not just an assemblage of machines. It is an assemblage of machines capable of generating knowledge that the design did not put there.

Working within synthetic ground means adjusting relationships among machines rather than imposing stable form. The question confronting territorial practice is not whether to engage these coupled systems, as that decision was made centuries ago and is inscribed in every levee, drainage district, and straightened channel. The question is how intentionally to engage, toward what ends, and with what capacity for knowledge production.

The profession has begun to answer that question, though incompletely. Living infrastructures, constructed wetlands, beneficial sediment placement, engineered shorelines, have become standard practice in coastal management, an acknowledgment that biological processes can perform work that engineered systems cannot sustain alone. But these approaches still operate largely within a management paradigm, biology enrolled to serve predetermined outcomes. What this dissertation calls wetware, developed in Chapter 09, goes further. It embeds computation into the biological relationship, coupling organisms with sensing and actuation so that the living system is not merely performing but producing knowledge through its own responses. Whether biology serves as a tool or participates as a partner in the design process depends on this difference.

Cultural Landscapes at Risk: Islands, Shoals, and Thresholds

The stakes of coupled ecologies are not abstract to humanity. They are places with names and people with histories.

In estuarine regions like the Chesapeake Bay, subsidence, sea-level rise, and modified sediment logistics have steadily eroded the terrestrially low islands over the past 150 years. These islands are homes to communities of watermen, their families, churches, and cemeteries, which have steadily disappeared into the bay. Some islands persist in diminished form, ringed by bulkheads and revetments that slow but cannot stop the inevitable retreat (Cronin et al. 2005). Every year the perimeter of the islands shrinks and the question of what comes is urgent.

Ethnographic accounts from Tangier Island depict what the process of an eroding home feels like from inside a community experiencing it. Houses flooded repeatedly, raised on new foundations, and then flooded again. Churchyards where graves are undermined by wave action and the buried being exposed by the sea. The daily rhythms of crabbing and oystering adjust, decade by decade, to shifting channels and beds of eelgrass (Swift 2018). Decisions about protection, accommodation, or relocation are entangled with deep attachments, sometimes religious in character, always constrained by the economic realities that limit choice.

The decision to leave is rarely dictated by physical conditions alone and social networks, place attachment, and signals from policymakers matter. When a government buyout program is implemented some leave who might otherwise stay, and when none exists, some stay who might otherwise leave. The threshold of inhabitability is as much social as it is geomorphic (Richter 2015). The drowning of the Chesapeake Bay is simultaneously a matter of sediment budgets and sea-level rise entangled with narratives that frame stories of loss, identity, responsibility, and belonging.

These islands sit within long histories of colonization, extraction, and racialized geographies that cartography has inscribed into imperial projects of exploration and resource capture (Smith and Hole 1624). In more recent scholarship these littoral zones are framed as sites of Black and Indigenous life that unsettle colonial, land-based imaginaries (King 2019). What is lost when an island drowns is never only the land itself.

The Chesapeake Bay islands frame an ethical demand that the research program takes seriously as a design question, not only as a political observation. Prototyping the Bay, a design research studio I have taught at the University of Virginia since 2018, engages the Pocomoke Sound (Cantrell 2025, course syllabus), a marginalized environment of sea-level rise, land subsidence, and shifting ecological communities. Its first course objective asks students to develop coherent design values that speak to their convictions regarding the cultural, technical, and philosophical basis of their design research. This is not a soft opening exercise. It is the insistence that adaptive design methods are not ethically neutral, that the choices embedded in every design proposition carry consequences for the communities and organisms the site sustains.

When design propositions function as hypotheses, the ethical dimension is not separable from the technical content. The decision about what to prototype, where to test, and who bears the cost of revision determines whether adaptation serves the communities most at risk or repeats historic patterns of dispossession under the banner of progress. Monitoring that reveals disproportionate burdens demands trajectory adjustment, not a note in a report but a redesign. Strategies addressing land loss in the Chesapeake and elsewhere must hold this demand as a design constraint, not an afterthought.

Political Ecologies of the Interface

Sensors, models, and automated controls have become central tools and methods for environmental governance. The interface where data is collected, processed, and acted upon has emerged as a technological object where power concentrates.

The decisions of what to monitor, how to model, and who has access to the levers of control determines who receives protection and who faces exposure. Understanding the political ecology provides tools for parsing these dynamics and confronting the question of who wins, who loses, and by what mechanisms (Robbins 2012).

Green infrastructure is promoted as a technical solution for flooding and water quality, yet its siting, design, and branding often channel investment into some neighborhoods while leaving others underserved. Gentrification haunts these projects as amenities raise property values and displace the communities most at risk from a changing climate. The metrics of success measure acres treated and gallons detained while rarely counting who was displaced.

The promise of the smart city adds another layer by treating the city as an abstraction of data that privileges efficiency and investment over lived experience and democratic contestation (Greenfield 2013). When life is represented through dashboards, residents become data points defined as users, consumers, and risks rather than as political subjects capable of challenging the terms under which they are governed.

The same frameworks operate in flood and coastal management where model outputs become the primary basis for risk maps, insurance rates, or investment priorities and the assumptions embedded in those models gain exponential influence while remaining opaque to the most affected (Collier, Mizes, and von Schnitzler 2016).

A politically attuned approach treats interfaces as negotiable constructs. Where the sensors go, which variables are measured, how results are visualized, and who has a voice when protocols are revised become explicit design topics. The aim is not to abandon technological tools but to embed them in processes that foreground equity, accountability, and the possibility of contestation, keeping multiple voices and multiple species inside the system as it adapts.

Adaptive Epistemology

The condition described across this chapter, territories formed by centuries of engineering and computation, ecological systems that resist prediction, sensing infrastructures that constitute as much as they report, communities whose survival depends on design decisions made in their name or without their knowledge, points toward a fundamentally different relationship between design and knowledge. Design propositions cannot function as static solutions delivered to stable sites. They are better understood as hypotheses, developed through friction with the systems they engage, revised through monitoring, and held provisionally against the certainty that the landscape will do things no model anticipated.

This dissertation develops that reorientation through six frameworks distilled from twenty years of practice. Multiple Intelligences describes the co-production of knowledge by human, machine, and biological agents. Technogeographies of Sensing addresses the political and spatial implications of what gets measured and by whom. Wetware names the coupling of biological and computational systems as a knowledge-producing medium. Generational Robotics extends design learning across timescales exceeding human institutional memory. Coupled Ecologies identifies the territorial condition in which biology, computation, and infrastructure are bound through feedback into a single operative system. Reflexive Stewardship is the ethical commitment to remaining answerable to what the landscape is becoming and to the question of who benefits and who bears the cost. Chapter 02 develops adaptive epistemology, the epistemological commitment that organizes all six, with the individual frameworks developed in the chapters that follow. Running through all six is a practitioner’s disposition this dissertation calls the cultivant, developed fully in Chapter 11, the ongoing negotiation between designed intention and biological agency in which maintenance is the primary design act.

This reorientation should not be confused with adaptive management. Adaptive management, as developed in conservation biology from Holling’s resilience work forward (Holling 1973, 1978), proposes iterative cycles of action, monitoring, and adjustment, but its epistemological assumptions remain within the predictive paradigm, the goal is still to reduce uncertainty and improve model accuracy over time. Adaptive epistemology, as this dissertation develops it, makes a different claim. That design practice is itself a mode of knowledge production, that the design proposition generates categories of knowledge that cannot be produced in advance through modeling alone. The distinction is not between managing adaptively and managing statically. It is between treating knowledge as something that precedes action and treating it as something produced through action. Chapter 02 develops this distinction and its implications.

The contemporary response to the failures described across this chapter has been overwhelmingly technical. More sensors, more refined models, more powerful computational tools applied to the same territorial systems. This response is necessary but not sufficient, because the difficulty is not a deficit of data. It is the epistemological frame within which the data is collected, interpreted, and acted upon. Computation deployed within a predictive paradigm does not correct the paradigm’s limitations. It compounds them. Models that run at higher resolution produce predictions that are more precise and equally wrong when the system crosses a threshold no historical record anticipated. Sensing infrastructure that monitors more variables more frequently generates institutional confidence that can override the embodied knowledge of communities who have inhabited the territory across generations. Algorithms that optimize toward predetermined objectives narrow the range of futures the territory can explore, collapsing the adjacent possible into the single trajectory the model specified. The infrastructure tradition is not failing for lack of technology. It is failing because the epistemological commitment that organizes the technology has not been examined. This dissertation examines it.

The political context of territorial design cannot be separated from the design research it enables. The NEOM consultation (2022–25), developed with Adam Mekies through Sherwood Design Engineers, makes this explicit. NEOM is a project of an authoritarian state, funded by sovereign wealth derived from the extraction and sale of fossil fuels, displacing Indigenous Huwaitat communities from their ancestral territories to make way for development that replicates in built form the political absolutism of its client. These facts are not footnotes to the design research. The political economy of the project shaped what was possible to investigate. The scale of resources, the speed of decision-making, the absence of democratic contestation all enabled design propositions that would have been difficult or impossible to develop under different institutional conditions. The knowledge produced in that context carries the imprint of how it was produced.

The design propositions developed in the consultation make intellectual contributions to the discipline, contributions developed fully in Chapters 05 and 09. The doctoral research productively holds both. The contribution and the political conditions that produced it, the adaptive epistemology and the authoritarian epistemology that funded the resources to develop it. Reflexive stewardship is not the elimination of this tension. It is the commitment to remaining honest about it and to asking, throughout the work, how findings generated under one set of political conditions might be translated into contexts governed by different values.

If this is the epistemological problem, prediction and control failing at the scale of territorial dynamic systems, how did this kind of practice actually develop? What communities, collaborations, and institutional conditions make an alternative form of knowing possible? And what does it look like to produce knowledge through design rather than to apply knowledge through it?