This chapter gathers the six canonical frameworks developed across this body of work, Coupled Ecologies, technogeographies of sensing, generational robotics, Multiple Intelligences, wetware, and Reflexive Stewardship, into a single account of what landscape architecture becomes when it critically addresses both the indeterminacy of complex systems and the multi-species character of the territories in which it operates. The cultivant, the practice-disposition that binds these frameworks, names the ongoing negotiation between designed intention and biological agency from which adaptive epistemology is practiced. It is a disposition within all six, not a seventh framework alongside them. Adaptive epistemology is the epistemological ground from which the six frameworks take their orientation. It names the condition in which the territory participates in its own legibility, and it is constitutively plural. The chapter necessarily sharpens the distinction between adaptive management (Holling 1978; Walters 1986) and adaptive epistemology as developed here, arguing that the difference is not merely scalar or technical but ontological, and that it therefore redefines the type of practice landscape architecture is and the bodies of knowledge it produces. From this vantage, the chapter articulates this contribution to the field, outlines what is at stake for practice, ecology, and environmental justice, and names the questions this framework opens that were not previously available.
The Shape of the Argument
A synoptic view is not a closure. It arrives at a position from which the full landscape of the argument is finally visible, not as a recapitulation of settled claims, but as a position from which relations that have been in formation throughout the work become legible. Vantage points look backward, but they also look outward. The question here is not what was said, but what has come into view from this vantage that was not visible from the valley.
The argument began from a provocation drawn from practice. Landscape architecture is operating within territories that are neither natural nor artificial, neither stable nor predictable, and it is doing so with epistemological inheritances situated within modernist simplifications (Scott 1998) that are structurally inadequate to those conditions. When dealing with ever increasing complexity the dominant professional response has been technical, leading to the application of more sensors, more refined models, and more powerful computational tools. While this technical apparatus is necessary, the ever increasing application of it to the environment is not sufficient. The difficulty is not simply a deficit of data. It is a failure to rethink what it means to know, to design, and to take responsibility in systems that are inherently open-ended, nonlinear, and entangled with the autonomous agencies of organisms, materials, and machines.
In contrast, this body of work argues for a different response, not a new toolset alone, but a new epistemology. Adaptive epistemology names a way of knowing that is itself adaptive, that treats design propositions as hypotheses rather than solutions, positions monitoring and adjustment as constitutive rather than supplementary, and recognizes knowledge production as one of the primary things that design at territorial scales actually does. Working landscapes, deltas, marshes, rewilding territories, urban ecological infrastructures, are not simply objects of knowledge. They are the apparatuses through which knowledge is produced. When a sediment diversion is opened and the resulting pattern of deposition is tracked, something is learned that could not have been learned from prior simulation alone. The territory answers back. Adaptive epistemology names a practice that takes this response seriously.
This is the core claim from which the chapter proceeds. What follows is an account of how the six conceptual frameworks developed in the preceding chapters together constitute this epistemological position, why this position represents a contribution that has not previously been available to landscape architecture, what is at stake for practice, for ecology, for environmental justice, and which questions only become articulable once this framework is in place.
Six Concepts, One Framework
The six frameworks introduced in previous chapters, Coupled Ecologies, technogeographies of sensing, generational robotics, Multiple Intelligences, wetware, and Reflexive Stewardship, are not a set of discrete themes orbiting a loose topic. They form a circuit. Each is necessary and together they articulate a coherent epistemological and methodological position for landscape architecture under conditions of complexity, uncertainty, and multi-species entanglement. Adaptive epistemology is the ground from which they proceed. The condition in which design propositions function as hypotheses, and the territory’s responses are genuine epistemic events. The cultivant is the practitioner’s disposition within all six. Provisional, attentive, adjusting, oriented toward maintaining the conditions for ongoing mutual learning rather than securing any final state.
Coupled Ecologies and Synthetic Ground
Coupled Ecologies and synthetic ground provide the ontological frame within which the other frameworks operate. Territories are not blank substrates awaiting inscription. They are already active assemblages of machines in the broad sense, rivers, levees, legal codes, organisms, markets, algorithms, that act and are acted upon (Bryant 2014). Coupled ecology names the formed condition that results from centuries of these mutual actings. Landscapes that are simultaneously ecological and infrastructural, biological and computational, autonomous and regulated. To design within synthetic ground is to adjust relationships among these machines rather than impose stable form from the outside, to recognize that territories are already dynamic assemblages into which design enters as one more agent, not as sovereign author. The nature-technology distinction does not hold. The territory is already multiple, and the design task is to honor that multiplicity rather than foreclose it.
Technogeographies of Sensing
Technogeographies of sensing specify the apparatus without which adaptive epistemology remains a formal gesture. If design propositions are hypotheses, they must be testable, and testability at territorial scales requires distributed sensing infrastructures, networks of sensors, remote-sensing platforms, citizen monitoring, and data aggregation systems. Yet the argument of this body of work has insisted that sensing is not simply instrumental. No single sensing framework captures what a territory is, because a territory is not a single kind of thing. When two sensors disagree, the gap between readings carries information about a process that neither sensor was designed to capture alone. What the territory has been working out in that gap is what a more homogeneous sensing apparatus would never have made available. Choosing what to sense is the first epistemological act. Attending to divergences between what different instruments reveal is the second.
The spatial distribution of sensing is also a political act. What is measured shapes what is known, what is known shapes what is protected, and the geography of sensors is a geography of care and neglect (Gabrys 2016). A sensing grid that carefully tracks salinity and turbidity in commercial fishery zones while leaving adjacent subsistence fisheries unmeasured is not a neutral technical decision. It encodes interests into the territory’s nervous system. Technogeographies of sensing therefore push adaptive epistemology toward questions of epistemological plurality, asking who decides what gets measured, whose environments become legible, and what the territory holds that current instrumentation has not yet been designed to receive.
Generational Robotics
Generational robotics addresses the temporal mismatch between technological obsolescence and ecological succession. Landscapes develop across decades and generations. Robotic and computational platforms turn over on product cycles of years. Generational robotics names the design problem of ensuring continuity of function, accumulated behavioral knowledge, and relational learning across platform generations, so that the knowledge produced in the relationship between a robot and its territory is not lost when hardware is replaced. The robot’s becoming and the territory’s becoming are aspects of a single relational process (Simondon 1958). The behavioral histories particular to specific territories that a generational robot develops over years constitute a form of knowledge that only staying produces, the accumulated record of a conversation across time that no initial specification could have anticipated and no single human career can span.
Multiple Intelligences
Multiple Intelligences translates the distributed cognitive character of adaptive landscape systems into a design principle. The designer’s tacit judgment, the algorithm’s pattern recognition, the biological community’s evolutionary responsiveness, the robot’s accumulated behavioral history are all structurally in play simultaneously in these systems. None is reducible to the others, and none is sufficient alone. The coupling itself is cognitive (Bach 2009). The distributed knowledge production described in the Multiple Intelligences framework above, biological, material, and computational agents conducting inquiry simultaneously, is how managed ecosystems actually operate. A practice adequate to this reality must develop methods for reading and responding to what non-human agents learn.
Wetware
Wetware names the operative layer of coupled assemblages, where biological processes whose metabolisms and behaviors are entangled with digital and robotic systems constitute a medium that is neither purely organic nor purely computational. The design problem in wetware is the quality of the coupling itself. Tight enough for legibility, flexible enough for surprise. Organisms whose metabolic responses constitute evidence that no prior model generated are not exceptions to the design framework. They are the framework’s primary yield, what the territory decided, given the conditions the design provided. The productive zone is genuine mutual tension between the designed and the living, and the designer’s task is to maintain the conditions under which that tension remains generative rather than collapsing into either inert control or unreadable chaos.
Reflexive Stewardship
Reflexive Stewardship is the practical disposition that follows from taking the other five frameworks seriously. The practitioner is a participant, not an external observer, and her knowledge is shaped by her vantage point within the assemblage she tends. She sees clearly from where she stands, and that position makes certain things visible while structurally obscuring others. Reflexive Stewardship is the active cultivation of the knowledge that vantage point makes invisible. The deliberate, sustained effort to bring community knowledge and biological responsiveness into the practice as constitutive inputs rather than supplementary consultation. The goals of a project are themselves hypotheses, subject to revision as the assemblage reveals what the designer’s position could not have shown her from the start.
Reflexive Stewardship is not a technique but an ethos, a commitment to ongoing accountability, to monitoring not only the landscape but one’s own assumptions, and to treating management as a form of inquiry. Robbins’ political ecology (2004) foregrounds landscapes as political formations in which actors hold unequal power to influence management and unequal vulnerability to outcomes. The reflexively attentive practitioner asks whose knowledge is being produced, whose environments are being made legible, and who controls resulting data, not as an additional layer of ethics applied after the design is done, but as a constitutive dimension of how she builds her own understanding of what she is working with and what she cannot yet see.
Divergence as Knowledge
Adaptive epistemology is constituted to receive divergent accounts and to treat their divergence as information rather than error. When the satellite tide gauge and the in-situ conductivity sensor produce a gap between their readings, when the waterman’s knowledge of the water and the salinity gauge’s reading do not coincide, when one instrument’s record contradicts another’s, the gap is not a calibration problem to be resolved. It is information about a process that neither account was designed to capture alone. Knowledge lives in the divergence between accounts rather than in either account separately. Every territory holds at least two territories that do not fully coincide. The framework developed here is built to sustain that plurality rather than reduce it to a single authoritative reading. The circuit of six frameworks remains productive precisely because each attends to the territory from a different angle, and the disagreements among those angles are themselves part of what the territory has to teach.
The Circuit
The circuit that these frameworks form is not linear. Adaptive epistemology as the epistemological ground is only operational with technogeographies of sensing to make propositions testable. Technogeographies of sensing gain their critical force from the recognition of Multiple Intelligences operating simultaneously in the territory. Multiple Intelligences are materially grounded in wetware, where biological autonomy is entangled with computational mediation. Wetware is expressed within Coupled Ecologies, the synthetic ground from which it is inseparable. And Coupled Ecologies are navigated through Reflexive Stewardship, which feeds back into the epistemological orientation from which design propositions are formed. Generational robotics names the temporal dimension that all five require. Without strategies for continuity across platform generations, the accumulated knowledge that adaptive epistemology produces cannot persist. Remove any one framework and the circuit loses coherence. Together, they supply something landscape architecture has lacked. A unified epistemological position adequate to the conditions of the Anthropocene.
The Mississippi River Basin Model, one of this body of work’s recurring touchstones, makes this circuit visible in negative. Cast in concrete at 1:2,000 horizontal and 1:100 vertical scales, the MRBM functioned as adaptive management avant la lettre. Engineers proposed interventions, pushed water through the model, and observed effects before acting in the river. It was also an epistemological trap. By stabilizing the river as a hydraulic system rendered in concrete channels, the model hardened its own assumptions. The territory continued to do what the territory does. Move sediment, sustain microbial ecologies, carry socio-political histories at human scales. None of that activity was legible to the apparatus. The technogeography of the model determined what could be known. The biological was excluded from the apparatus, synthetic ground was reduced to hydraulics, and the reflexivity the model might have enabled was displaced by the confidence that comes with legibility. The river was made knowable in a single register and continued to act across all the others. De Monchaux (2025) traces this enclosure as a persistent trajectory in the history of simulation, models becoming more closed to view as they become more powerful. The MRBM is the exemplary case, and the framework developed here defines itself against that shadow.
The contemporary computational apparatus is the MRBM’s successor, not its correction. More powerful, more distributed, more capable of processing the territory’s signals, and structurally susceptible to the same closure if deployed within the same epistemological commitment. The framework developed here defines itself against that inheritance.
The Decisive Distinction: Adaptive Management and Adaptive Epistemology
The distinction drawn in Chapter 2 between adaptive management and adaptive epistemology can now be stated with greater precision, because the six frameworks and the practice that generated them provide its empirical content.
The framework developed here is indebted to adaptive management, the tradition of experimental environmental governance led by C. S. Holling, Carl Walters, and their collaborators from the 1970s onward (Holling 1978; Walters 1986). Adaptive management’s central insight, that management actions can be treated as experiments generating information about system behavior, is foundational to this body of work. Panarchy theory (Gunderson and Holling 2002), resilience thinking (Folke et al. 2010), and civic science (Lee 1993) are explicit influences. Yet the position developed here is not simply an application of adaptive management to landscape architecture. The difference between adaptive management and adaptive epistemology is not just one of scale or disciplinary translation. It is a difference in ontological scope and epistemological ambition.
Classically formulated, adaptive management remains within the horizon of scientific management. Managers formulate hypotheses about system behavior, implement interventions that test those hypotheses, monitor outcomes, and revise subsequent actions. Management becomes more systematic, more rigorous, more responsive. But goals remain largely exterior to the learning process, defined in advance as ecological outcomes or resource yields, and the locus of intelligence remains the human management team. Uncertainty is acknowledged, learning is institutionalized, yet the basic relation between knowing subject and known object persists. The ecosystem is managed and humans are those who know.
Adaptive epistemology, as developed here, goes further in three directions.
First, it extends the experimental subject to include design propositions themselves, the spatial, material, and infrastructural configurations that landscape architects create, not just management protocols. The morphology of a restored marsh, the geometry of a sediment diversion, the arrangement of a monitoring network are all articulated as hypotheses about how biological, material, and hydrological processes will respond under given conditions. Design becomes a mode of knowledge production in its own right. Built form is not only outcome but epistemic instrument.
Second, adaptive epistemology extends the category of intelligence beyond the management team to encompass biological, material, and machine intelligences. Marsh plants distributing root architecture in response to pore-water chemistry, machine-learning models detecting unexpected correlations between wind direction and sediment capture, sediments sorting by grain size through self-organization. Each is a form of inquiry. An epistemology that counts only human learning as knowledge production misses the majority of information being generated in managed landscapes. The framework developed here insists that this distributed knowledge production is not metaphorical. It describes how managed ecosystems actually operate. Non-human agents conduct inquiry, and a practice adequate to this reality must develop methods for reading and responding to what they learn.
Third, adaptive epistemology turns back on the epistemological status of design itself. Adaptive management asks how managers can learn more effectively from the consequences of their actions. Adaptive epistemology asks what kind of knowing landscape design at territorial scales constitutes, and how that knowing reshapes relationships among designers, territories, and non-human coinhabitants. This second question opens onto philosophy of science, posthumanist theory, and critical social science in ways that the managerial orientation of adaptive management does not require. It is a question about the nature of design knowledge, not only about improving management performance.
The distinction is consequential. Adaptive management makes resource management more experimentally robust. Adaptive epistemology reconceives design practice as a form of knowledge production irreducibly entangled with the multi-species, computational, and political assemblages in which it operates. The former improves management. The latter makes a different claim about what design is. A landscape architecture that merely adopts adaptive management is a more rigorously experimental practice. A landscape architecture that takes up adaptive epistemology becomes a different kind of practice altogether, one that understands itself as producing not only functional, beautiful, or ecologically sound landscapes, but knowledge of how territories behave under human engagement, and of what it means to engage them responsibly.
Practitioners will recognize this distinction in the difference between two approaches to coastal restoration. An adaptive management approach establishes monitoring protocols, sets performance thresholds, and commits to adjusting management if targets are not met. An adaptive epistemological approach does this and also conceives the project’s spatial form as a set of propositions, including what the particular geometry of restored marsh edge proposes about tidal energy, what the phasing of planting proposes about succession, and what the configuration of sensors proposes about which processes are treated as significant. Adaptive managers learn from outcomes and adaptive epistemologists design in ways that make projects inherently legible to learning. This is a different relationship between design intent and design form.
The Contribution
A Methodology for Posthumanism
The theoretical traditions that run through this body of work, Haraway’s Staying with the Trouble (2016), Bennett’s Vibrant Matter (2010), Latour’s actor-network thinking (1987, 2004), Bryant’s ontology of machines (2014), have become central to the environmental humanities and design theory. What they have largely lacked is a methodology for practice, a framework that tells designers not only how to think about multispecies entanglements, vibrant matter, or actor-networks, but what to do.
Haraway’s call to “make kin in the Chthulucene,” to take seriously entangled human and non-human lives in a damaged world, is philosophically and politically compelling. Bennett’s insistence on thing-power challenges the instrumentalization of matter. Yet both leave largely open the question of how these orientations translate into site analysis, material specification, monitoring regimes, and governance arrangements. These theoretical tools were, in many ways, waiting for a method.
The six-part framework developed here, Coupled Ecologies, technogeographies of sensing, generational robotics, Multiple Intelligences, wetware, and Reflexive Stewardship, creates a way forward. Coupled Ecologies translates Latour’s actor-network thinking into an ontological understanding of territories as assemblages in which biology, computation, and infrastructure are constitutively entangled, each with its own agency and none simply inert. Technogeographies of sensing translate this recognition into practical questions about who and what is measured, whose agencies become legible to governance, and whose remain invisible. Multiple Intelligences translates Haraway’s commitment to partial, situated knowledge and Bennett’s recognition of nonhuman agency into a design protocol that treats human judgment, machine cognition, and biological response as co-producers of landscape knowledge. Wetware names the biological medium through which knowledge is produced, the living systems whose metabolic responses constitute what the territory decided given the conditions the design provided, and that only the ongoing relationship yields. Generational robotics addresses the temporal mismatch between technological obsolescence and ecological succession, designing continuity of function across platform generations so that landscape knowledge accumulates beyond any single human career. Reflexive Stewardship translates Haraway’s ethos of staying with the trouble into a practical disposition, remaining with complex systems rather than seeking neat resolution, and treating management as a continuous negotiation rather than a one-off intervention. The cultivant, the practice-disposition that binds these six, names the ongoing relationship between designed intention and biological agency in which maintenance is the primary design act, not a framework itself, but the posture from which all six are practiced.
This is this body of work’s most original move, not the extension of posthumanist theory itself, that work is already substantial, but its translation into a coherent methodology for landscape architecture. The contribution runs in two directions. It brings posthumanist theory into design practice, giving practitioners conceptual tools drawn from feminist science studies, new materialism, and multispecies ethnography, and demonstrating what those tools enable on the ground. And it brings design practice into posthumanist theory, insisting that the delta, the rewilding zone, the instrumented watershed are not objects of reflection but sites where posthumanist commitments are already being negotiated among designers, managers, organisms, algorithms, and materials.
Haraway reminds us that it matters what thoughts think thoughts, what descriptions describe descriptions (Haraway 2016). This framework adds that it matters what designs make possible. The epistemological orientation from which designers approach a territory shapes not only the forms that emerge, but the future conditions of knowing that those forms enable or foreclose. A landscape stabilized under a predict-and-control paradigm is not only a different spatial product than one designed within adaptive epistemology. It is a different epistemic infrastructure. It closes down forms of learning that adaptive design would leave open. Epistemological commitments are thus materialized in levees, planting plans, monitoring networks, and management protocols, and persist in those artifacts long after the theories that produced them have shifted.
Refraction, the third core contribution alongside adaptive epistemology and the cultivant, is the method that made this framework legible. The same projects appeared in Chapter 5 as practice, in Chapter 6 as modeling genealogy, in Chapter 7 as sensing politics, in Chapter 9 as adaptive management, in Chapter 10 as temporal infrastructure, and in Chapter 11 as co-authorship. Each retelling, each refraction, revealed properties that were always present in the work but traveling invisibly within the instrumental contexts that commissioned it. The framework did not emerge from theory applied to practice. It emerged from practice systematically retold from vantage points its original framings never demanded.
The case studies woven through this body of work, the Mississippi River Basin Model as a caution about the epistemological limits of stabilizing representations, the Chesapeake Bay islands as sites of ongoing negotiation between human infrastructure and non-human dynamics, the various computational monitoring and management projects drawn from the author’s own practice, are not solely illustrations. They indicate that this framework is already emergent in the world, that practitioners are reaching toward it under the pressure of conditions the inherited paradigm cannot handle, and that explicit articulation can sharpen and extend what practice is attempting. It is theory from practice returning to practice, the model of knowledge production that adaptive epistemology recommends.
This body of work also pushes back against a tendency in posthumanist design theory to remain at the level of ontology and neglect the institutional, material, and political conditions within which design actually operates. To say that territories are assemblages of human and non-human agencies is necessary but insufficient for a practitioner navigating liability regimes, permitting processes, client expectations, financing, and professional cultures calibrated to another epistemological paradigm. The work engages these conditions directly, showing the misfit between adaptive approaches and conventional practice, and sketching concrete mechanisms such as process-based liability, performance-based contracting, and service-oriented business models through which the transition might be navigated. This is posthumanist theory with its boots on.
Stakes: Practice, Ecology, and Justice
For Practice
For landscape architecture, the stakes are immediate and material. A discipline that continues to design for conditions that no longer exist, sizing infrastructure from historical records under non-stationary climates, setting restoration targets against reference ecosystems already made unattainable by warming trends, treating monitoring as a dispensable add-on, will produce landscapes that fail. They will fail not only aesthetically or culturally but materially. Levees overtopped, marshes drowned, restoration investments unmade by shifting baselines. The stationarity assumption that Milly and colleagues declared dead in 2008 remains embedded in design standards and professional norms (Milly et al. 2008). Adaptive epistemology offers a different trajectory.
That trajectory does not promise success in the sense of control. Complex systems do not afford guarantees. What it offers instead is a framework for productive failure. When a sediment diversion finds a path different from what the model projected, the question becomes. What did the landscape reveal? What hypothesis was tested and what was learned? The project is not simply a failure that should never have been built. It is an experiment that has generated knowledge. The investment is converted into understanding that can inform subsequent propositions.
Enabling this conversion requires institutional adjustment across four domains that landscape architecture has barely begun to address. First, liability. The professional negligence standard assumes that “reasonably prudent” practice can be defined against fixed benchmarks and that outcomes can be attributed to specific design decisions. Adaptive epistemology destabilizes both assumptions. When landscapes emerge from interactions among biological, material, computational, and human agents over extended timescales, isolating the contribution of any single decision to any particular outcome may be impossible. The alternative is a process-based standard of care, evaluating whether practitioners employed defensible methods for characterizing uncertainty, designing monitoring, and implementing adaptive responses, rather than an outcome-based standard that demands predictive certainty no one possesses (Williams and Brown 2014). Medical practice has moved in this direction and landscape architecture can follow.
The urgent real-world context for these frameworks is already unfolding. Professor and Urban Planner Jesse Keenan documents how North America is entering what he terms “a great domestic climate migration (climigration) that may very well reshape everything from our physical landscape to our electoral politics” (Keenan 2025, 3). Crucially, predictive models cannot reliably forecast where and when this will occur. As Keenan (2025, 17) observes, policymakers might never be able to reliably forecast climigration but must prepare for the possibility that some places will hollow out while others explode with the weight of climate-driven population growth. This is the condition adaptive epistemology attempts to address, designing under irreducible uncertainty by creating conditions for continuous learning rather than predicting a stable outcome.
Second, financing. Capital budgets fund construction and operating budgets fund maintenance, but neither accommodates the sustained monitoring, periodic adjustment, and iterative redesign that adaptive management requires. Performance-based contracts that tie payments to demonstrated outcomes over defined periods, endowment models that fund long-term stewardship through capital reserves, and blended finance structures that use public funds to de-risk private investment in ecosystem restoration offer pathways, none fully mature, all requiring experimentation across diverse project types and institutional contexts.
Third, business models. The conventional project-delivery model cannot accommodate engagements that extend into indefinite futures of monitoring and adjustment. Service-based and stewardship models where firms provide sustained engagement rather than discrete deliverables, compensated through retainers or performance-based payments, represent the structural alternative. Some firms already operate partially within this model through campus management and post-construction monitoring services.
Fourth, professional culture. A discipline that rewards design authorship through photogenic completion images and penalizes epistemological humility must learn to value demonstrated learning over the performance of certainty. Awards programs that assess adaptive capacity alongside design quality, pedagogies that develop skills in monitoring and systems dynamics alongside form-making (Felson and Pickett 2005), and honest long-term case studies of how projects evolved after completion are the cultural infrastructure this transition requires.
The framework also offers a sharper answer to the question of what landscape architecture contributes that other disciplines cannot. Landscape architects work at the intersection of design and ecology, of spatial form and temporal process, of human intention and non-human agency. The adaptive epistemological position treats this intersection not as a compromise but as a productive tension from which a distinctive knowledge emerges, simultaneously spatial and temporal, aesthetic and ecological, human-centered and multi-species. This is not knowledge that ecology, engineering, or the environmental humanities can produce alone. It requires the integrative practice of landscape architecture. The framework is a claim that this practice can be made epistemologically rigorous.
For Ecology
For ecology, the stakes concern the fate of multi-species territories under Anthropocene conditions. Ecological science has long emphasized the non-equilibrial, path-dependent, threshold-sensitive character of ecosystems (Pickett and White 1985; Cook 1999). Yet the implications of this shift have not fully propagated into design practice. Restoration ecology still often operates under the reference-condition paradigm, seeking to return systems to pre-disturbance baselines that climate change increasingly renders fictive and unreachable (Hobbs, Higgs, and Harris 2009; Higgs et al. 2014). The framework developed here takes ecology seriously not by importing it unchanged but by forging a design epistemology commensurate with ecological insight.
The concept of the neo-wild is crucial. Neo-wild landscapes, computationally mediated, algorithmically managed, infrastructure-supported yet ecologically autonomous in their outcomes, constitute a new category of environmental condition that landscape architecture must learn to create, manage, and evaluate (Cantrell, Ellis, and Martin 2017). The neo-wild refuses the binary between management and wildness, infrastructure and autonomy, human agency and non-human freedom. It does so not as a thought experiment but as a set of designed and built systems that are simultaneously maintained and wild, intentional and emergent. Neo-wildness is what Haraway’s sympoiesis looks like when materialized at territorial scale, making-with rather than making-by-oneself (Haraway 2016).
Central to this ecological reframing is the cultivant, the ongoing relationship between designed intention and biological agency in which maintenance is the primary design act rather than a secondary service to a completed designed object. The cultivant, as developed in Chapter 11, extends Raxworthy’s viridic toward a relational conception of design authorship, not the designer making a landscape but the designer entering into an ongoing negotiation with biological agency whose outcomes cannot be fully anticipated. For ecology, this means that the designer’s obligation does not end at construction. It persists across the landscape’s operational life, monitoring how biological agency responds to designed conditions, adjusting those conditions as species assemblages develop, revising design intentions as the co-authorship reveals what the biological systems need and what the designed infrastructure cannot provide. The Bayou Bienvenue restoration in Louisiana’s Lower Ninth Ward exemplifies the stakes. A landscape where biological autonomy, cypress regeneration, marsh accretion, microbial community assembly, met designer assumptions about planting, hydrology, and salinity management, and where the gap between what was specified and what the wetware actually did constituted the most productive knowledge the project generated. The cultivant names a practice adequate to this gap, not the elimination of surprise but the design of conditions in which surprise becomes legible and actionable.
This opens a substantial research program within ecology. How do species assemblages unfold in neo-wild environments compared to unmanaged or conventionally managed ones? How do vegetation communities develop under open-objective management regimes rather than reference-condition targets? How do migratory species respond to territories whose hydrological regimes are modulated by algorithmic management instead of historical variability? What ecological signatures mark the “third wave” of cybernetics in which uncertainty becomes generative rather than problematic (Zhang 2025)? These are empirical questions that neither ecology nor landscape architecture can answer alone and they are the questions this framework makes visible.
For Environmental Justice
A territory is always a plurality. Its soils hold multiple temporalities simultaneously, geological and hydrological and biological. Its knowledge is produced by multiple parties simultaneously, the designer’s instruments, the manager’s protocols, the resident’s embodied memory, the non-human community’s metabolic history. No single sensing apparatus enrolls all of these. When communities whose knowledge is not enrolled in the sensing apparatus are structurally excluded from the apparatus’s field of legibility, the apparatus does not simply miss their knowledge. It produces outcomes that cannot account for their conditions, their vulnerabilities, or their claims. This is the epistemological mechanism through which unjust outcomes are produced, and it is why the political dimension of technogeographies of sensing follows directly from the epistemological claim, not the other way around.
Tangier Island in the Chesapeake Bay makes the argument concrete. As subsidence and shoreline erosion accelerate, the island’s political legibility depends partly on who is measuring what, and with which instruments. Remote sensing can document shoreline retreat and tidal gauges can quantify sea-level trends. But the lived history of a community watching its landscape dissolve over generations, the local knowledge of what the island was, how it has changed, what it has cost, is not automatically registered in the computational systems modeling future conditions or informing interventions. This is not simply a missing data layer. It reveals how technogeographies structurally produce certain knowledges as legible and leave others outside the apparatus. The pattern extends beyond the Chesapeake. The Isle de Jean Charles and the communities of Plaquemines Parish in Louisiana’s coastal zone face managed retreat from landscapes whose sediment dynamics, subsidence rates, and cultural geographies have been systematically underrepresented in the sensing infrastructures that govern federal investment and intervention. The communities most at risk are the communities whose knowledge is least enrolled in the instruments that allocate protection. An adaptive epistemology that ignores this is simply a more efficient version of modernism (Scott 1998).
Reflexive Stewardship, as the ethos that closes the circuit, must therefore be epistemologically conscious as well as ecologically literate. The practitioner’s vantage point shapes what she can see from where she stands. The deliberate cultivation of what that vantage makes invisible requires bringing in community knowledge not as supplementary input but as constitutive, because what communities know about their territories carries information the sensing apparatus cannot generate on its own. Designers who deploy sensing infrastructures without asking whose knowledge is being produced, whose environments are being made legible, and who controls resulting data are not practicing Reflexive Stewardship. They are reproducing the epistemological signature of state simplification under new technical conditions.
Against solutionism, the tendency to recast political problems as technical puzzles for computational resolution (Morozov 2013), the framework developed here insists that adaptive epistemology is irreducibly political. It does not only ask how we can learn more effectively from landscapes, but also who is authorized to learn, whose knowledge counts, and in whose interest that knowledge is deployed. These questions cannot be answered by infrastructure or algorithms alone. They require deliberative, power-conscious, participatory processes of the sort articulated in political ecology and environmental justice scholarship and operationalized in collaborative planning frameworks (Innes and Booher 1999). A fully adaptive epistemology is thus also a democratic epistemology, not only because legitimacy demands it, but because the kinds of knowledge complex territories require cannot be produced without diverse perspectives, local knowledges, and sustained, shared attention.
Trajectories
A framework that is itself adaptive does not conclude with closure. It opens trajectories. The final gesture of this body of work is therefore toward the directions of inquiry that the framework makes available, questions that could not have been clearly formulated before the conceptual work was done.
Multi-Species Governance
First, multi-species governance. If territories are assemblages of biological, material, and computational agents, and if practice takes seriously the agencies and intelligences of non-human participants, then governance, who has standing in decisions, whose interests are represented, becomes more complex than conventional environmental frameworks can accommodate. Human stakeholders are represented, unevenly, through democratic and regulatory processes. But how are the interests of marsh plant communities, migratory birds, microbial assemblages, or future human generations represented? Haraway’s call to make kin points toward this question without specifying institutional form (Haraway 2016). Designing governance frameworks adequate to multi-species assemblages is a research program this body of work can open. It will require collaboration between landscape architects, political theorists, ecologists, legal scholars, and speculative designers willing to prototype governance as carefully as they prototype form.
The Epistemology of Machine Learning in Landscape Management
Second, the epistemological status of machine learning in landscape management. As ML models trained on sensor data become central to managing complex territories, the question of what kind of knowledge they produce, how it relates to human, biological, and material knowledges, becomes pressing. The framework for distributed cognition developed in Chapter 11 is a beginning, not a complete epistemology of machine learning in environmental management. If landscape architecture is a form of knowledge production, then algorithmic recommendation systems are part of the design of knowing. They prefigure which futures can be imagined and which interventions can be justified. When algorithms recommend actions, under what conditions are they producing knowledge? With what limits, and for whom? These questions are central to accountability and to the politics of algorithmic governance in environmental systems, an emerging area this framework is positioned to inform.
Temporal Politics and Institutional Design
Third, temporal politics and institutional design. Adaptive epistemology commits to long timescales, multi-decadal landscape trajectories, successional arcs, sediment budgets, intergenerational justice. Political institutions, funding cycles, careers, and attention spans operate on much shorter clocks. The structural mismatch between the timescales of adaptive ecological management and the timescales of institutions is one of the deepest obstacles to implementing this framework. How can adaptive commitments survive leadership changes, budget crises, regulatory shifts, and institutional amnesia? Research is needed into governance structures, financing mechanisms, and organizational forms that can hold adaptive commitments across temporal gaps no single practitioner can span. This is political science and organizational theory, but it is also landscape architecture, insofar as the spatial and material design of infrastructures, how visibly they encode monitoring, how they invite or resist engagement, bears on institutional durability.
Keenan documents that adaptation will involve irreversible loss. “The relocation and climigration of people, firms, species, and ecologies will no doubt accelerate the micro-collapse of local cultures and communities. Cultural heritage and practices, foods and recipes, and places of spiritual resonance will be lost” (Keenan 2025, 22). Reflexive Stewardship requires humility about this, an acknowledgment that design cannot prevent all loss, that some of what we love will be transformed beyond recognition, and that our responsibility is to center the voices of those bearing the greatest costs.
The Pedagogy of Adaptive Epistemology
Fourth, pedagogy. If this framework is to matter beyond theory, designers must be educated to practice within it. Studio pedagogy has historically cultivated form-making, representation, and articulacy in projecting clear design intentions. It rewards decisive visions and compositional mastery. Adaptive epistemology requires overlapping but different capacities. Tolerance for ambiguity, skills in experimental design, fluency with environmental monitoring and data, facility with multi-stakeholder processes, and the humility to revise judgments in light of what landscapes reveal. These are not well served by studio formats organized around the terminal critique, which tends to evaluate the proposal as artifact rather than the design process as inquiry.
Developing a pedagogy adequate to adaptive epistemology is an open problem. Felson and Pickett’s “designed experiments” (2005) sketch one approach, design research more broadly offers templates for treating projects as instruments of inquiry. The Prototyping the Bay studio represents a first systematic attempt to translate the adaptive epistemology framework into foundation-level design pedagogy, requiring students to frame proposals as structured experiments, articulate what questions the design is asking, and identify monitoring conditions that would constitute evidence requiring design revision. Translating these into a curriculum that cultivates the full range of capacities adaptive practice requires, spatial, ecological, computational, political, and cultural, is a substantial project, but an urgent one. The designers who will be practicing through the critical decades of the twenty-first century are being trained now and the epistemologies they internalize will shape how they respond to unprecedented conditions.
Representation and the Provisional
Fifth, representation. Landscape architecture is a representational discipline. It draws, models, maps, and visualizes possible worlds. Its representational conventions were largely developed for predict-and-control paradigms, master plans, planting plans, and construction documents that presuppose future states that can be specified in advance and built accordingly. They are representations of outcomes, not of inquiries. Adaptive epistemology needs representations that are themselves provisional. Depictions of ranges rather than certainties, visualizations that foreground monitoring and feedback, formats that can evolve alongside landscapes rather than becoming archival at the moment of construction.
Corner’s agency of mapping (1999) and the projective mapping practices that followed are partial responses, representations that generate new understandings rather than simply recording existing conditions. Yet projective mapping has not fully absorbed the feedback loops, monitoring data, and algorithmic management layers that adaptive epistemology demands. The interfaces described throughout this body of work, dashboards, data visualizations, multi-scale monitoring displays, are nascent representational forms moving in this direction. Developing them into a coherent, shareable practice, and a set of conventions legible across design, ecology, engineering, management, and community contexts, is a design research task of its own.
Coda: Landscapes of Becoming
This body of work began in territory, in marshes, deltas, watersheds, and the infrastructures threaded through them. It ends there as well. Not in abstraction, but in the specific, irreducible condition of landscapes that are always in the process of becoming something they have not yet been.
Technogeographies of sensing stretched across these territories are not, at their best, instruments of surveillance or optimization alone. They are instruments of attention, ways of sustaining care for environments whose transformations unfold across timescales longer than any project, career, or life. The designer who builds monitoring into a project makes a commitment not that they know what will happen, but that whatever happens will be worth knowing. This is a different relationship to the future than predict-and-control regimes offer. It is less certain, more honest about the limits of expertise, and more responsive to the conditions in which twenty-first-century landscape architecture is compelled to work.
The neo-wilds that emerge from these entangled systems, the marshes wired with sensors and fed by calibrated diversions, populated by organisms whose behaviors exceed any prior specification, are landscapes of emergence, surprise, and ongoing becoming. They are never finished, never fully ours. They belong to the assemblages of agents, biological, computational, material, and human, whose interactions produce them. Yet they are also places of care, sites of sustained, adaptive, justice-conscious attention to multi-species territories that human action has shaped and that human practice must continue to tend.
The argument this body of work advances is that care and knowledge, at these scales, are not separable. To know a territory adaptively is to design within it as inquiry, to sense it with awareness of the politics of sensing, to enable its biological autonomy while maintaining the infrastructures that make that autonomy possible, to steward it reflexively across decades and generations. This is already a form of care. And care, practiced at territorial scales with the discipline and rigor that adaptive epistemology demands, is itself a form of knowing. This is the framework. Landscapes are already practicing it, unevenly and imperfectly, in the way that all genuine learning is imperfect and all the more valuable for that.
The question that opened this body of work, how landscape architecture should practice within territories whose futures cannot be known, has not been resolved. It has been transformed. What began as a crisis of method has become a claim about what design is. A form of engaged inquiry distributed across human and non-human intelligences, proceeding through experimental propositions that landscapes revise and extend, and oriented toward forms of care that sustain multi-species territories across the timescales of their becoming. The work continues. The landscapes are already waiting.
Not a closure but an opening. The framework does not resolve the tensions it has named between control and autonomy, between human intention and biological agency, between justice and scale. It positions landscape architecture to inhabit those tensions productively. The questions it leaves open are the ones the field now has to work with.