Existing typologies of nature-based solutions, the authors contend, have organized interventions by technical form, ecosystem type, societal challenge or spatial scale. Challenge-based classifications aligned solutions with Sustainable Development Goals priorities such as disaster risk reduction and food security; ecosystem-based typologies grounded them in ecological categories like wetlands and forests; and intervention- and scale-based schemes distinguished green infrastructure from ecosystem-based adaptation or site-level from city-wide measures. What these systems share is the treatment of governance as background context rather than a defining dimension. This omission, the study argues, reflects an implicit and flawed assumption that technically sound interventions will succeed regardless of institutional arrangements. In reality, the outcomes of nature-based solutions are at least equally dependent on governance fit—who holds authority, who mobilizes resources and how coordination is maintained over decades of maintenance and adaptation.

The conceptual core of the new framework is “implementation leadership,” defined as the actor or coalition in which the constitutive levers of governance converge at the point of institutional stabilization. The framework operationalizes governance through four analytical levers. The first, initiation, is treated as an antecedent signal, capturing who originally proposed or catalyzed the intervention. The remaining three are constitutive: decision authority, meaning the strategic power to approve or amend key elements such as site selection, design, budgeting and partnership arrangements; resource control, meaning the ability to mobilize or regulate access to land, finance, permits and technical expertise; and stewardship responsibility, meaning institutionalized long-term maintenance, monitoring and adaptive management obligations. Classification occurs at institutional stabilization—typically during or after implementation—when provisional commitments have been converted into binding contracts, mandates and budgetary routines. This methodological choice deliberately avoids classifying symbolic, temporary or transitional arrangements, and it reflects the insight that early coalition formation during planning is often an unreliable predictor of long-term governance.

To ground the typology empirically, the team conducted a qualitative, comparative analysis of 35 urban nature-based solution cases drawn from 22 countries, selected through structured searches of Scopus and Web of Science complemented by grey literature sources including the Urban Governance Atlas, ICLEI, UN-Habitat and OPPLA. Cases had to be urban, address hydro-meteorological hazards relevant to climate adaptation, and have been operational for at least three years to allow governance arrangements to stabilize. Two researchers coded the cases independently against the four levers, achieving substantial inter-coder reliability with a Cohen’s Kappa of 0.80. Analysis proceeded through a case-by-lever matrix, with archetypes identified through iterative pattern recognition of how decision authority, resource control and stewardship responsibility aligned—or failed to align—across actors. The result was five recurring governance archetypes: state-led, market-led, co-led, community-led and citizen-led.

The state-led archetype is exemplified by Switzerland’s Basel Green Roof Mandate, the Netherlands’ Room for the River programme and South Korea’s Cheonggyecheon Restoration. In each, statutory decision authority, control over core resources and stewardship responsibility are legally consolidated within public institutions, even where delivery is decentralized—regional water boards and municipalities implemented centrally planned measures in the Dutch case. Market-led solutions, by contrast, consolidate leadership within private actors: Milan’s Bosco Verticale places finance, design and long-term maintenance under a developer through condominium contracts, while Toronto’s Green Roof Bylaw represents a regulatory variant in which developers retain design and stewardship duties in response to statutory requirements. Co-led arrangements, such as Copenhagen’s Klimakvarter and Sweden’s Hammarby Sjöstad, institutionalize shared leadership, pooling municipal and utility resources while residents shape street-level design, producing asymmetrically shared authority and hybrid stewardship.

Community-led and citizen-led cases reveal subtler patterns. In Montréal’s Green Alleys, resident committees initiate proposals, co-design layouts and assume maintenance under borough guidelines with modest municipal grants of roughly CAD 7,000 to 25,000 per alley; Berlin’s Prinzessinnengarten operates on temporary land-use agreements with volunteer labour and no long-term tenure security; and Nairobi’s Kibera Public Space Project co-manages riparian land under informal tenure. Citizen-led cases such as Philadelphia’s Rain Check and Washington DC’s RiverSmart Homes show public agencies defining technical standards and approving contractors while homeowners voluntarily sign multi-year maintenance agreements on their own parcels. Critically, across all these non-state cases, formal regulatory authority remained with public actors—yet leadership still stabilized outside the state, because control over implementation resources and long-term stewardship rested with non-state actors.

Perhaps the study’s most counterintuitive finding concerns initiation. In 77 percent of cases, the intervention was externally introduced by actors other than direct local beneficiaries, yet initiation proved empirically unreliable as a classificatory signal. The Isar Restoration in Germany, born of civic mobilization in the 1980s, culminated in state-led implementation under the Isar-Plan once funding and stewardship were codified through statutory programmes. Conversely, state-initiated programmes like Rain Check stabilized as citizen-led because participation remained voluntary and stewardship household-based. Only 17 percent of cases displayed hybrid or transitional boundary conditions, where levers remained partially aligned or shifted over time through codification, consolidation or withdrawal of institutional support. The lesson, the authors conclude, is that leadership lies not in who regulates or initiates, but in who ultimately pays, implements and maintains.

The typology’s comparative power is most vivid in green infrastructure, the dominant technical orientation in the sample at 74 percent of cases, which spans all five archetypes with no single archetype accounting for more than 23 percent. Green roofs, an identical intervention form, are governed by state-led mandates in Basel, market-led compliance regimes in Toronto and voluntary household uptake in Washington DC. Similar dispersion appears within challenge-based categories: disaster risk and water management interventions range from statutory state programmes like Room for the River and Thames Estuary 2100 to co-led partnerships like Copenhagen’s Klimakvarter and citizen-led incentive models like Clean River Rewards. Shared hazard orientation, ecosystem type or spatial scale simply do not predetermine governance arrangements—and governance diversity is greatest precisely where technical replicability is highest.

The practical implications extend to policy instrument fit, equity and scaling. Instruments that succeed in one governance mode may fail in another: subsidies stimulate uptake in citizen-led programmes but have little traction in market-led models driven by compliance, while legislative mandates can overwhelm neighbourhood initiatives. Scaling pathways also prove governance-specific rather than intervention-specific. State-led solutions scale through codification into statutory frameworks and budget regimes; community-led and citizen-led forms scale through diffusion and replication across households and peer networks; and co-led configurations advance through translation, where experimental partnerships reshape broader policy via intermediaries. The typology also anticipates justice risks structurally embedded in each leadership form: market-led models tie ecological benefits to investment capacity; citizen-led incentive schemes tied to property ownership systematically exclude renters; community-led interventions struggle with insecure tenure and unstable funding; and co-led arrangements risk diffuse accountability. Meanwhile, decentralizing implementation without commensurate authority, resources or institutional backing—observed in failed pilots from Mexico City to Cape Town—produces structural misfits that voluntarism alone cannot sustain.

The authors acknowledge limitations: the study does not assess ecological effectiveness, the five archetypes necessarily abstract from institutional complexity, and classification at stabilization omits later dynamics such as elite capture or institutional drift. Future research, they suggest, should integrate leadership archetypes with ecosystem service monitoring and longitudinal designs tracing governance across the full intervention lifecycle, and test whether governance configurations consistently predispose projects to specific equity risks. Even so, the framework offers planners, funders and researchers a classification-ready lens for anticipating institutional constraints: by mapping veto points, tracing whether resources can be mobilized beyond short project cycles and testing whether stewardship is enforceable over decades, the governance-informed typology aims to help nature-based solutions scale from isolated pilots into durable, equitable urban infrastructure rather than generic green decoration.