Economic succession: an ecology-informed diagnostic framework for sustainable urban transformation

1 Introduction: toward a middle-range theory of sustainable urban economic transformation

Urban economics, urban science, and development economics offer influential accounts of growth and structural change. Growth theory emphasizes capital accumulation and technological change (Solow, 1956; Swan, 1956; Romer, 1986; Lucas, 1988; Banerjee and Duflo, 2005); sectoral models describe shifts among agriculture, industry, and services (Fisher, 1939; Krüger, 2008; Lutz et al., 2013; Rybski et al., 2021); urban scaling identifies regularities associated with city size (Bettencourt et al., 2007, 2010; Bettencourt, 2013; Youn et al., 2016); relatedness and economic complexity connect diversification to existing capabilities (Neffke et al., 2011; Content and Frenken, 2016; Boschma et al., 2015; Hidalgo, 2023; Li and Neffke, 2024; Muneepeerakul et al., 2013); and production-network research reveals inter-industry dependencies (Acemoglu et al., 2012; Carvalho, 2014). These literatures identify important regularities but do not provide a common account of how a particular city moves from one activity portfolio to another. Sectoral models are too coarse to identify feasible next activities, scaling relations do not specify transition sequences, relatedness can combine distinct constraints, and production networks do not explain how activities enter and persist. Conversely, place-based strategies preserve local context but are difficult to compare cumulatively without a shared causal architecture (Barca et al., 2012; Rodríguez-Pose and Wilkie, 2017; Scott and Storper, 2015).

This gap calls for a middle-range theory: one located between abstract general laws and descriptions tied to individual cases (Merton, 1968; Cartwright, 2020). We use five criteria for that designation: a bounded domain; separable causal mechanisms; observable indicators; conditional propositions with failure conditions; and a mapping from diagnosed mechanisms to interventions. Urban economic succession meets these criteria by treating city-region activity transitions as its domain, organizing six causal roles, linking them to indicators and interventions in Table 1, and stating falsifiable propositions in Section 5.

Table 1. A succession-based causal ordering for sustainable urban economic transformation.

Strategically organized institutional action is treated as cross-cutting rather than as an undifferentiated historical residual. Lobbying, vetoes, gatekeeping, and other purposive actions should be assigned to the opportunity, access, fit, or interaction mechanism they alter, while the actors' resources, political power, and institutional positions are recorded explicitly.

Important middle-range mechanisms already exist in development thought. Jacobs (1969) describes cumulative diversification through import replacement, Hirschman (1958) emphasizes backward and forward linkages, and Schumacher (1973) and Stewart (1977) stress the fit between technology and local capabilities and institutions. Evolutionary economic geography provides the closest economic antecedent by explaining branching through variety, relatedness, path dependence, and lock-in (Boschma and Frenken, 2006; Frenken and Boschma, 2007; Martin and Sunley, 2007; Boschma et al., 2015). Urban economic succession complements it by separating an economic opening, access to actors able to exploit it, and local capability fit as distinct gates, and by treating interactions, inherited feedback, and extra-local networks as modifiers of those gates.

The framework also shares Growth Diagnostics' focus on binding constraints, but shifts the unit of analysis from aggregate investment and growth to the entry, persistence, transformation, and exit of specified activities in event time (Hausmann et al., 2005). Sustainability-transitions research—including the multi-level perspective, technological innovation systems, strategic niche management, transition management, and urban-transitions scholarship—explains how niches, regimes, innovation systems, governance, and institutional change shape opportunities, access, and fit (Geels, 2002; Geels and Schot, 2007; Bergek et al., 2008; Kemp et al., 1998; Loorbach, 2010; Markard et al., 2012; Hodson and Marvin, 2010); succession supplies a complementary activity-level diagnosis within a city-region. Industrial ecology contributes analysis of material and energy flows, including industrial symbiosis, while organizational ecology models organizational founding and mortality (Chertow, 2000; Hannan and Freeman, 1977; Carroll and Hannan, 2000). Related urban-evolution research examines variation, selection, retention, and empirical trajectories in urban forms and activities (Silver et al., 2022a; Fox et al., 2022a; Silver et al., 2022b; Fox et al., 2022b). These approaches overlap with urban economic succession but do not supply its six-role causal ordering.

We define urban economic succession as ordered but non-deterministic change in the composition and organization of an urban economy, generated by opportunity availability, access to potential activities, local capability fit, interactions among activities, historical legacies and feedback, and extra-local connectivity. It encompasses entry and expansion as well as transformation, contraction, exit, and the absorption of functions into other activities. It is a theory of compositional transition, not a normative theory of desirable development. Throughout, “city” refers primarily to a functional urban area or city-region. The concept belongs to the broader view of urban economies as ecosystem-like complex adaptive systems, while remaining an economic framework whose components must be observable and testable.

This focus matters for sustainability. Decarbonization, circular production, resilience, and inclusive growth require some activities to enter and expand, incumbents to adapt, and carbon- or resource-intensive activities to transform, contract, exit, or be recombined. These processes redistribute costs and benefits within cities and across the urban–rural gradient (Le Gallo, 2026). Urban economic succession does not define the sustainable target; it diagnoses whether and how a collectively chosen transition can become locally viable and self-reinforcing.

2 What is ecological succession?

Ecological succession is the process by which a biological community changes through time, especially after disturbance or the creation of new habitat. Primary succession begins on newly exposed surfaces with little soil or biological inheritance. Secondary succession begins where a shock or disturbance leaves soils, organisms, seeds, microbes, or physical structures in place; these legacies usually speed recovery and make later development depend on what came before. Although succession is often illustrated with plants, it includes the joint reassembly of plants, animals, microorganisms, soils, and the networked flows of energy and materials among them (Walker and del Moral, 2003; Prach and Walker, 2011; Chang and Turner, 2019; Dini-Andreote et al., 2015).

Succession is not simply a sequence in which one group of species replaces another. It explains why that turnover occurs. Three broad conditions are required: space or resources must become available; potential colonists must be present in the surrounding landscape and able to reach the site; and those arrivals must be able to establish and persist under local conditions (Pickett et al., 1987, 2009). Their success is then shaped by interactions with other organisms. Some early colonists improve difficult conditions by stabilizing soil, adding nutrients, or creating shade and shelter. Others slow later arrivals by occupying space or consuming scarce resources. Organisms also change the environment they inhabit, so the community already present helps determine what can arrive next. Climate, repeated disturbance, and connections to nearby habitats continue to influence the pathway (Connell and Slatyer, 1977; Walker and Chapin, 1987; Fukami, 2015; Poorter et al., 2024).

Several patterns recur without forming a universal sequence. Early colonists often spread quickly, whereas later-dominant organisms tend to persist under stronger competition or lower resource availability. Biomass, soil development, physical structure, and sometimes internal nutrient cycling increase, while diversity and food-web complexity may rise, level off, or decline. These tendencies depend on tradeoffs and local conditions rather than a fixed timetable (Bazzaz, 1979; Huston and Smith, 1987; Odum, 1969; Pregitzer and Euskirchen, 2004; Purschke et al., 2013; Poorter et al., 2021, 2023).

Modern theory accordingly rejects a single progression toward a stable “climax.” Outcomes depend on disturbance, surviving legacies, arrival order, landscape context, and chance; biomass, soils, composition, and food-web structure may recover at different rates, and trajectories may converge, diverge, stall, reverse, or be reset (Clements, 1916; Gleason, 1926; Fukami, 2015; Walker et al., 2010; Wardle et al., 2004; Chang and Turner, 2019). Succession is therefore best understood as a middle-range theory: it offers a common causal framework for explaining particular pathways rather than one universal path. Its central questions concern opportunities for establishment, access by potential occupants, local performance, interactions, feedback, history, and outside connections (Pickett et al., 1987, 2009; Poorter et al., 2023, 2024; van Breugel et al., 2024).

For the purposes of this paper, succession theory provides five practical lessons:

• A visible sequence is a pattern to be explained, not an explanation by itself.
• Broad causes can be separated into observable mechanisms, such as lack of access, poor local conditions, competition, or repeated disturbance.
• Predictions must be conditional on history, context, and available participants.
• Departures from an expected path can reveal which mechanism is missing or binding.
• Diagnosis can guide intervention: change local conditions, restore connections, introduce missing participants, manage harmful interactions, or allow more time.

This conditional account of compositional change—rather than a simple analogy of orderly stages—is the part of succession theory that we translate to sustainable urban economic transformation.

3 The shared foundation: complex adaptive systems

3.1 Complex adaptive systems

The comparison between ecological succession and urban economic development is useful only if it rests on more than metaphor. Complex adaptive systems provide a basis for disciplined comparison. They are composed of heterogeneous, partly autonomous elements whose local interactions generate aggregate patterns, and those aggregate patterns feed back to change the behavior, performance, and composition of the elements themselves. Cross-domain theory is possible at this level of causal organization: a model transfers not because two systems use the same vocabulary or look alike, but because comparable relations and mechanisms organize their change (Simon, 1962; Holland, 1992).

Ecosystems are canonical complex adaptive systems. Local interactions and selection generate system-level structure, while nonlinearity creates historical dependence, self-organization, and multiple possible trajectories (Levin, 1998). Economies have an analogous architecture of heterogeneous firms, workers, consumers, investors, and institutions that adapt under incomplete information, interact through networks, and continually alter the environment to which they respond (Arthur, 1999, 2021; Foster, 2005). Urban and regional research extends this reasoning spatially by treating cities as open, bottom-up systems whose trajectories emerge from interacting agents and institutions rather than from a central controller (Batty, 2008; Martin and Sunley, 2007).

This shared classification makes a conditional transfer of candidate mechanisms worth testing; it does not imply that cities are biological ecosystems. Economic actors anticipate, strategize, create formal rules, and deliberately redesign their environments. A mechanism from ecology therefore becomes economically informative only when its variables can be redefined in observable economic terms, it remains plausible under purposeful human action, and it produces predictions that can fail. Work on coupled social–ecological systems shows how feedback, strategic interaction, spatial heterogeneity, and processes operating at different time scales can be represented within a common framework while retaining domain-specific assumptions (Levin et al., 2013). Ecology supplies a mature causal architecture; economics determines how that architecture must be translated, measured, and bounded.

3.2 Parallel patterns of compositional change

The most visible parallel is patterned turnover in system composition. A disturbed or abandoned patch of land may move from sparse pioneer vegetation toward increasingly woody vegetation. An economy may move from agriculture through industrialization toward a larger service share. Figure 1 places these familiar patterns side by side.

The resemblance is informative but deliberately limited. The three-sector sequence conceals large differences within manufacturing and services, and cities may bypass manufacturing, reindustrialize, undergo premature deindustrialization, or develop unusual combinations of primary, industrial, and service activities (Lutz et al., 2013; Rybski et al., 2021). Ecological succession is likewise ordered without following a universal timetable or endpoint. In both domains, the broad sequence is therefore a pattern to be explained, not a rule that identifies which specific occupant or activity comes next.

3.3 From composition to interaction networks

The comparison extends beyond changing shares to the organization of flows among components. Ecological food webs and economic input–output systems can both be represented as weighted directed networks, allowing link density (the share of possible links that are present), two-way connections, dependency, and cycling to be measured in comparable terms (Iskrzyński et al., 2022). Classical succession theory proposed that developing ecosystems tend to retain nutrients more effectively, route a larger share of material through internal and decomposition-based pathways, and support more biomass per unit of energy throughput (Odum, 1969). Later work qualified any simple monotonic account: nutrient retention, for example, may be greatest while biomass is accumulating rather than at a final steady state (Vitousek and Reiners, 1975). The transferable proposition is therefore not that every mature system becomes progressively “closed,” but that development reorganizes flow structure along recurrent, context-dependent trajectories.

Economic production networks provide a direct arena for testing this proposition. Higher-income economies tend to contain more active intersectoral links, and ecological measures of cycling and reciprocity have already been applied to input–output networks (Gloria et al., 2024; Braun et al., 2021; Iskrzyński et al., 2022). At the urban or city-region scale, development may alter network density, the share of intermediate demand met locally, import leakage, and the relationship between output and direct or embodied energy and material inputs. This intersects with urban-metabolism research, which treats cities as open systems organized by exchanges of energy and materials and calls for integrated analysis of economic activity, resource throughput, environmental burden, and well-being (Ulgiati and Zucaro, 2019; Zucaro and Agostinho, 2025). These dimensions need not move together, and their trajectory need not match ecological succession. That is what gives them diagnostic value: two cities with similar sectoral composition may differ because one has formed locally embedded production linkages while the other remains dependent on external inputs or contains activities that generate few local spillovers. Figure 2 presents this comparison as a hypothesis to be tested rather than an assumed equivalence.

The figures identify a common object of explanation—ordered change in a portfolio and in the relationships among its members—but they do not identify the cause of any particular transition. A restoration manager facing failed recovery must distinguish among missing seeds or other dispersing organisms, poor soils, competition, consumption by herbivores, repeated disturbance, and landscape isolation. A development practitioner facing a stagnant urban economy must distinguish among absent demand, inaccessible firms or workers, weak capabilities, missing suppliers, finance, institutions, and inherited industrial structure. In both cases, similar outcomes can arise from different constraints and therefore require different interventions. The shared complex-system properties justify asking whether succession's causal architecture can organize that diagnosis; the next section translates it into economic terms.

4 Urban economic succession for sustainable transformation

4.1 From economic regularities to a causal ordering

Urban economic succession organizes the literatures reviewed above as parts of a common transition problem. Sustainable transformation requires new activities to emerge, incumbents to adapt, local linkages to form, and environmentally damaging activities to transform, be absorbed, contract, or exit. Sectoral transition, scaling, complexity, relatedness, and production-network research each illuminate part of this process, but none supplies a common causal ordering for why a specified activity enters and becomes established in one city but not another (Fisher, 1939; Balland et al., 2020; Rybski et al., 2021; Bettencourt et al., 2007; Youn et al., 2016; Neffke et al., 2011; Hidalgo, 2023; Li and Neffke, 2024; Acemoglu et al., 2012; Carvalho, 2014; Muneepeerakul et al., 2013; Hong et al., 2020; Arbabi and Punzo, 2023).

Recent work provides more directional building blocks: diversification can be modeled through prerequisite capabilities, labor-skill networks, pioneering vs. path-following entry, place-specific relatedness, and intercity connectivity (O'Clery et al., 2021; Shutters and Waters, 2020; Corradini and Vanino, 2022; Barbero et al., 2024; Liang et al., 2024). Succession contributes not another universal trajectory but a causal ordering for identifying and explaining recurring pathways (van Breugel et al., 2024).

Rather than posit either a single sequence or unlimited local uniqueness, the framework anticipates that comparable cities may exhibit a limited set of recurrent transition types. Tracking structural change in 938 U.S. urban economies from 2001 to 2017, Shutters et al. (2021) identified six response archetypes around the Great Recession, ranging from little change or return to a prior state to continuous change and rapid shifts followed by stabilization. This typology does not establish universal stages; it shows that responses to a common shock or disturbance can form a small set of recognizable pathways. Urban economic succession seeks to explain which causal constraints select among them and when they recur.

4.2 The causal components

In ecological succession, compositional change requires an opportunity for establishment, access to a pool of potential colonists, and differential performance of those colonists under local conditions (Pickett et al., 1987, 2009). Interactions, feedbacks, disturbance history, and landscape context then shape the realized pathway (Fukami, 2015; Poorter et al., 2024). The economic translation preserves this ordering. These labels identify causal roles rather than permanent categories: infrastructure or finance, for example, may create an opening in one transition and determine local establishment in another. To prevent post-hoc reassignment, an empirical study should specify the focal transition, entry threshold, time window, and hypothesized causal mechanism before observing outcomes; assign each variable to a role through that mechanism; and hold the assignment fixed in the primary analysis. Pre-registration is preferable, and any alternative assignments should be reported as sensitivity analyses rather than substituted after results are known. The framework tracks activities and the capabilities that support them rather than treating industry codes as fixed units, because capabilities and occupations can move across industry boundaries as an economy develops (Duranton and Puga, 2005).

The three proximate gates are opportunity availability, activity availability, and local capability fit. Opportunity is a market, policy, technological, or incumbent-exit opening—including procurement, emissions standards, brownfield redevelopment, or demand for repair and reuse—rather than the mere presence of related capabilities. Activity availability asks whether the needed firms, founders, investors, technologies, or workers can be assembled locally or accessed through mobility and intercity networks (Liang et al., 2024). Capability fit asks whether an arriving activity can persist under local skills, suppliers, infrastructure, institutions, finance, land, regulation, and, for low-carbon or circular activities, electricity, grid, logistics, certification, and material conditions. Relatedness is an important proxy for this multidimensional fit (Neffke et al., 2011; Li and Neffke, 2024), but appearance without persistence does not demonstrate establishment.

Three groups of conditions modify those gates. Interactions include facilitation through shared labor, suppliers, knowledge, infrastructure, and demand, as well as competition for land, labor, capital, water, energy, attention, and political support. Priority effects and historical feedback arise because early specializations leave labor pools, assets, institutions, reputations, and constituencies that alter later opportunity and fit. These inherited conditions must be distinguished from ongoing environment modification by current participants. Reflexive strategic action by incumbent firms, coalitions, or public actors is not subsumed under historical feedback: it should be coded according to the opportunity, access, fit, or interaction mechanism that it directly alters, while actor identity, resources, political power, and institutional position remain explicit parts of the explanation. The portfolio at time t therefore changes the filters at t+1, consistent with ecological priority effects and asymmetric capability accumulation (Fukami, 2015; O'Clery et al., 2021). Landscape and network context captures the trade, commuting, migration, ownership, knowledge, supply-chain, and multilevel-policy connections through which an open city-region can relax, intensify, postpone, or externalize local constraints.

4.3 A schematic statement

Urban economic succession includes entry and expansion as well as contraction, exit, and the absorption of functions into other activities. Equation (1) formalizes only the entry side of that broader turnover process. Let h_*ict* denote the hazard that activity i enters functional urban area or city-region c, or expands there beyond a stated threshold, during period t, conditional on not having already done so. A succession-based specification is

hict=g(Oict,Aict,Fict;Iict,Hict,Nict),(1)

where O is opportunity availability, A is access to the potential activity pool, and F is local capability fit. These are the three proximate gates to entry. I represents interactions with incumbent activities, H represents historical sequence, inherited legacy, and nonstrategic feedback, and N represents extra-local network context. The subscripts recognize that historical exposure and network position may differ across activities within the same city. Where a component is genuinely common to every activity in a city-period, it may instead be represented by a city-level term such as H_*ct* or N_*ct*. These modifiers alter the strength and meaning of the proximate gates.

The function is not a universal production equation, an estimable specification by itself, or a formal proof of sequence. It is organizing notation that makes the proposed components explicit, directs measurement toward separable mechanisms, and supports comparison across cities. The substantive theory lies in the conditional and temporal relationships stated in Section 5. Table 1 summarizes the translation into candidate indicators and interventions.

An empirical implementation would define the activity–city–periods at risk and use event-history analysis: a Cox proportional-hazards model for continuous event times or, more commonly with annual regional data, a discrete-time logit or complementary-log–log model (Cox, 1972). O, A, and F enter as pre-specified time-varying covariates, and interactions such as O test whether the effects of the gates are conditional rather than merely additive. I_*ict*, H_*ict*, and N_*ict* can enter as measured modifiers through interaction terms; city-wide components may be entered separately as H_*ct* or N_*ct*. City and activity heterogeneity can be represented by fixed effects, multilevel random effects, or shared frailties, and non-proportional effects can be modeled with time interactions. The variables and interaction structure must follow the mechanism specified for the focal transition.

The contraction side can be operationalized in parallel rather than deferred entirely, but contraction should not be treated as the preferred sustainability outcome. For most stable or growing city-regions, the relevant objective is a managed transition away from carbon-intensive production. Once an activity is established above a stated threshold, it can enter a second risk set with at least four distinguishable outcomes: within-activity transformation, in which the incumbent adopts lower-carbon technologies or production processes while retaining much of its economic function; substitution or functional absorption, in which expanding lower-carbon activities absorb workers, capabilities, assets, or functions released by the incumbent; managed contraction or exit, in which an activity is deliberately phased down because transformation is infeasible or its function is no longer socially desirable; and unmanaged collapse, in which activity, employment, and capabilities disappear before viable alternatives are established. Sustainability transitions may require both the creation of alternatives and the destabilization of carbon-intensive regimes, but their success also depends on the procedural and distributive treatment of workers and communities whose livelihoods depend on those regimes (Kivimaa and Kern, 2016; Newell and Mulvaney, 2013).

These outcomes have different regional consequences. Within-activity transformation or well-sequenced substitution can preserve more employment, productive capability, supplier relationships, and fiscal capacity than net contraction, although neither eliminates adjustment costs. Managed contraction can reduce disruption relative to abrupt collapse, but it still entails a net loss unless workers, assets, and capabilities are redeployed. A contraction or transition hazard should therefore measure not only market decline and regulatory phase-out pressure, but also the timing and scale of alternative entry, residual demand, sunk-asset commitments, capability transferability, worker access to alternative employment, institutional buffering, incumbent resistance, and network dependencies. Competing-risk or multistate models can then distinguish transformation, absorption, managed exit, and collapse rather than treating every decline below an activity threshold as equivalent.

A different case arises in persistently shrinking cities. Where long-term demographic and economic decline leaves an inherited urban footprint, infrastructure network, or service capacity that cannot be maintained, planned contraction or right-sizing may be an appropriate citywide adaptation. Planning research has proposed right-sizing strategies that coordinate vacant-land reuse, green infrastructure, neighborhood planning, and the adjustment of infrastructure and services to a smaller population (Schilling and Logan, 2008; Haase et al., 2014). This differs from sectoral decarbonization in an otherwise stable or growing city-region. Even here, contraction is not an end in itself: the objective is to align infrastructure, land use, services, and economic activity with a smaller population while protecting welfare, reducing abandonment costs, and preserving capabilities that can support subsequent adaptation.

This ordering turns the ecological comparison into a falsifiable urban framework. It predicts that broad regularities such as relatedness and scale should operate conditionally through distinct mechanisms; that cities with similar portfolios may diverge because different gates or network conditions bind; and that effective policy depends on diagnosing the constraint before selecting an intervention. The following section derives testable propositions from those claims.

5 Testable propositions

A useful theoretical framework must generate claims that can fail. The first four propositions concern activity entry and persistence; the fifth concerns the system-level reorganization of flows that may result from repeated entry and interaction. The preceding subsection already provides a conceptual multistate account of within-activity transformation, functional absorption, managed contraction or exit, and unmanaged collapse. Developing estimable competing-risk specifications and comparative evidence for those outcomes remains an empirical research priority. Each proposition below is stated together with the observation that would discriminate between the framework and simpler accounts already available.

5.1 Joint availability

Industry entry should depend jointly on the existence of an opportunity and on access to a relevant activity pool, including access through extra-local networks (Liang et al., 2024). A positive interaction between opportunity and access would demonstrate complementarity, but it would not by itself establish that both operate as necessary gates. The stronger gate interpretation predicts that the marginal effect of opportunity should approach zero when access is below a pre-registered low threshold, and that the marginal effect of access should approach zero when opportunity is similarly low. Entry should become materially more likely only when both gates are sufficiently open. If opportunity and access instead have stable additive effects across the full range of the other variable, the evidence would support separate additive causes rather than joint necessity.

To make these empirical implications concrete, consider annual entry of a specified green activity into U.S. metropolitan areas, as modeled in Equation (2). Define entry as employment in the focal activity exceeding a pre-registered minimum level and location quotient in two consecutive years. The event is dated in the first qualifying year; the second-year requirement is used retrospectively to distinguish persistent entry from a 1-year fluctuation. All predictors are measured before that first qualifying year. A pre-specified sensitivity analysis may instead date the event in the second qualifying year. Measure opportunity (O) before entry with activity-specific local demand or policy openings, such as public procurement awards, relevant construction permits, or binding regulatory changes. Measure access (A) as the prior stock of focal firms and occupations in migration-, commuting-, ownership-, or collaboration-connected regions. Measure fit (F) by comparing the activity's national occupation vector with the city's pre-entry industry-by-occupation portfolio, optionally weighting occupational overlap with O*NET skill similarity. Incumbent suppliers and competitors provide I; lagged specialization and entry sequence provide H; and intercity connectivity provides N.

With annual observations, estimate a complementary log–log entry model,

log[-log(1-pict)]=αi+γc+τt+βOOict-1+βAAict-1+βFFict-1+βOA(O×A)ict-1+δ′Zict-1,(2)

where α_*i*, γ_*c*, and τ_*t* absorb activity, city, and year heterogeneity; Z contains pre-specified modifiers and controls such as population, current portfolio diversity, and state-level conditions; and pict denotes the discrete-time probability of entry among activity-city pairs still at risk. Opportunity and access should be placed on substantively interpretable scales with pre-registered low and high values. Evidence of complementarity requires β_*OA*>0, substantively meaningful conditional effects when both gates are high, and improved held-out prediction relative to an additive model. The stronger joint-gate interpretation additionally requires a low predicted entry hazard when either gate is below its pre-registered threshold and a marginal effect of each gate that approaches zero when the other is low but becomes positive when the other is high. The proposition is not supported if the interaction is null or unstable, if additive effects predict equally well, or if these conditional patterns disappear after the planned controls.

5.2 Conditional relatedness

Relatedness should raise the probability of entry and persistence, with an effect that varies by type of transition. It should matter most for activities extending an established trajectory, and least for pioneering activities, which should depend instead on external connectivity, policy support, or new institutional combinations (Corradini and Vanino, 2022). Relatedness is the best-established regularity in the field, so reproducing it would tell us nothing. The framework's distinct claim is that relatedness proxies one component of the ordering, local capability fit, and should therefore weaken precisely where other components carry the weight. A relatedness coefficient of similar magnitude across rooted and pioneering entries would show that relatedness is not decomposable in this way, and the proposed decomposition is not supported.

5.3 Endogeneity

The order in which industries and institutions enter should influence later development even after the current portfolio is controlled, and successful entrants should change the conditions governing subsequent entry. These are the two directions of a single claim, and together they separate succession from a static suitability model, which holds that a city's present portfolio determines what can enter next. Cities matched on current composition but differing in the sequence by which they acquired it should diverge in subsequent entry probabilities and capability accumulation. Entry events should be followed by measurable shifts in occupational composition, supplier presence, and the relatedness of remaining opportunities, with larger shifts for entrants that buy and hire locally than for externally supplied operations (Hirschman, 1958). Should matched cities fail to diverge, the current portfolio summarizes history completely and history adds no explanatory power beyond current composition.

5.4 Portfolio interaction

Entry probability should depend on the configuration of the existing portfolio, and not on bilateral similarity between the entrant and each incumbent taken separately. Relatedness density averages pairwise proximities and so treats a portfolio as a collection of independent industries, whereas a missing complement or a contested input is a property of the set. The discriminating cases are entries with high relatedness density that nonetheless fail, and the proposition holds if those failures are systematically accounted for by input–output position, labor market competition, or land competition. Should pairwise relatedness already predict which of them fail, the interaction structure adds nothing that existing measures do not carry.

5.5 Network-flow trajectories

Repeated entry and interaction should reorganize production and resource flows in recurrent, stage- and context-dependent ways. Conditional on city size, sectoral composition, and openness, changes in the activity portfolio should be associated with changes in input–output link density, reciprocity and cycling, the share of intermediate inputs sourced within the city-region, import leakage, and output per unit of direct and embodied energy or material input (Braun et al., 2021; Gloria et al., 2024; Iskrzyński et al., 2022). The claim is not that all cities become more internally closed, nor that more cycling is always more efficient. It is that these dimensions should trace a limited family of trajectories and that departures from the trajectory expected for a city's composition and network position should be informative.

The micro-to-macro link is conditional rather than automatic. When an entrant constitutes a newly represented activity category, representing it separately adds a row and column of purchases and sales to the local transactions matrix. Entry within an existing input–output category does not change the matrix dimensions, but it may change transaction values, technical coefficients, sourcing patterns, or the distribution of purchases and sales within that category. In either case, entry produces a system-level change only when entrants are economically material, persist, and form local backward or forward linkages. Small entrants, externally supplied branch operations, or enclave activities may change composition without appreciably changing aggregate local flows. The proposition therefore predicts network reorganization after clusters of persistent, locally embedded entry events, not after every entry. Its sign is also conditional: new local substitution may raise link density or local intermediate use, whereas integration into extra-local supply chains may reduce them.

For example, diversification without the formation of backward and forward linkages should be followed by weaker local capability accumulation and greater external dependence than diversification that becomes locally embedded (Hirschman, 1958). Conversely, unusually high internal cycling could indicate productive local integration, but it could also reflect inefficient lock-in or weak access to external capabilities. Interpretation must therefore depend on the combination of network structure, resource intensity, productivity, and extra-local connectivity rather than on any one measure alone.

The discriminating test is longitudinal. Researchers should estimate network-flow trajectories for comparable city-regions and ask whether departures predict subsequent activity entry, productivity, resilience, or resource intensity after controlling for scale and sectoral composition. If network measures show no temporal regularity beyond those controls, or if deviations have no relationship to subsequent outcomes, the network-flow extension of urban succession is unsupported.

6 Discussion: from theory to sustainable urban transformation

The causal ordering also has a practical use. Like Growth Diagnostics, it asks which locally specific constraint binds before recommending intervention (Hausmann et al., 2005). Restoration managers use succession theory to diagnose why recovery has stalled and to decide which constraint to remove; sustainable urban transformation needs an equivalent capacity focused on activity portfolios and sequences. Placing a city on a single ladder from agriculture to industry to services does not do this, nor does treating a relatedness map as an automated list of industries to subsidize. Instead, a succession-based diagnostic would proceed through a series of questions.

• What is the target? State what the economy is meant to deliver: employment resilience, export capacity, decarbonization, wage growth, broader participation, and so on. Naming a fashionable industry is not a target. Ecological restoration likewise begins from a reference condition rather than a demand for maturity in the abstract.
• What history and agency produced the current portfolio? Past specialization, prior shocks, land use, infrastructure, institutions, and policy choices reveal which assets are inherited and which are lock-ins. The diagnosis should distinguish passive legacy from active resistance by actors who can suppress openings, restrict access to land or finance, or use regulatory vetoes. The current portfolio should not be interpreted independently of its sequence or its power relations.
• Is there an economic opportunity? Separate a missing opening from a missing capability. A city may hold the right skills and still lack demand, procurement, market access, or a technological or regulatory opening. Relatedness analysis speaks primarily to capability fit, so it may point to a skills gap even when the real constraint is that nobody is buying.
• Can relevant activities reach the city? The reachable pool includes incumbent firms, potential founders, investors, technologies, skilled migrants, university spin-offs, and external partners. Connectivity may be as important as local stock, but a rooted firm and a remotely managed branch plant differ in their effect on endogeneity.
• Can the activity perform locally? Relatedness, occupational overlap, suppliers, institutions, finance, suitable sites, infrastructure, and permitting should be treated as performance filters. Low performance may reflect one critical bottleneck rather than a globally deficient system.
• What interactions will the transition create? A target activity may complement incumbent industries, compete for scarce labor or land, depend on a vulnerable supplier, or displace an existing economic function. Development policy should assess portfolio effects, not only direct job counts.
• What sequence of interventions is required? Some interventions should precede others. Training without employers may cause out-migration; firm recruitment without suppliers may produce an enclave; research investment without commercialization institutions may fail to generate local industry. Succession theory directs attention to ordering.

For sustainability policy, this diagnostic ordering guards against treating a socially desirable technology as locally feasible merely because its environmental benefits are clear. A heat-pump, building-retrofit, battery-recycling, or circular-manufacturing strategy can fail for different reasons: weak demand, inaccessible firms and workers, missing grid or logistics capacity, competition for land and skills, poor connections to incumbent suppliers, or an intervention sequence that creates training before jobs or facilities before markets. The same technology may therefore require different policy portfolios in different city-regions.

This diagnostic interpretation aligns with recent economic complexity research. Li and Neffke (2024) argue that relatedness is most useful when it identifies anomalies: activities that appear well matched to a local economy and remain unexpectedly weak. Such anomalies point to a missing mechanism. The question becomes not “Which related industry should we pick?” but “Why has an apparently feasible activity not established?” That is a distinctly succession-based question. Hidalgo (2023) policy framework similarly distinguishes what, when, where, and who dimensions of structural transformation. The succession framework adds a complementary how: the sequence of openings, arrivals, establishment filters, interactions, and feedbacks through which a transformation becomes self-sustaining.

Consider a city seeking to build a circular construction economy. A coarse sector model classifies the relevant activities, and a relatedness model identifies proximity to construction, demolition, logistics, materials processing, architecture, and engineering. A succession diagnosis asks further questions. Is there effective demand for low-carbon construction and reused materials? Can processing firms, workers, technologies, and finance reach the city? Are material-grading standards, suitable sites, logistics, certification, and procurement rules in place? Will architects, contractors, public procurers, and recovery firms form local linkages? Which missing function binds, and what must be established first? A missing market standard requires a different intervention from a missing recovery facility, inaccessible finance, or a shortage of skilled workers. The output is therefore a conditional intervention sequence specific to that city.

This example also shows what the ecological lens contributes beyond vocabulary. Suppose two cities have the same construction skills, demolition stream, final procurement mandate, and measured relatedness to material-reuse activities. In one, procurement is imposed before recovered-material grading, deconstruction capacity, and exchange infrastructure exist; demand is met by imports, local entrants fail, and incumbents organize against the mandate. In the other, those colonization and establishment conditions are built first, early reuse firms form local linkages, and procurement then expands a viable activity pool. A static inventory of the final conditions treats the cities as equivalent. Succession predicts different outcomes from the order of arrival, facilitation by early entrants, and feedback into the next event. If sequence and feedback add no explanatory power once contemporaneous capabilities are controlled, this claimed contribution of the ecological transfer fails.

7 Research priorities

The causal ordering identifies several research priorities that, taken together, would allow findings from separate cities to support a coherent middle-range theory of urban economic succession and sustainable transformation.

7.1 Measurement and identification

Longitudinal research should identify sequences of industry, occupation, technology, and institution entry rather than relying only on cross-sectional portfolios. Parallel event-history models should estimate contraction and exit, distinguishing genuine capability loss from the absorption of functions into other industries. Sequence analysis, event-history models, and change-point detection can distinguish recurrent pathways from retrospective narratives. Alternative components of the causal hierarchy should also be measured separately. Relatedness is a valuable composite indicator, but it may reflect shared knowledge, supply-chain relationships, labor flows, demand, institutions, or infrastructure. Distinguishing these channels is essential for intervention. Measurement must also account for professional functions that move across industry boundaries as an economy matures: legal services begin as general practice, differentiate into specialisms, and then move inside client firms as in-house counsel. American Community Survey industry-by-occupation data could test whether lawyers employed outside the legal services sector increase with city size and economic complexity, as would be consistent with such absorption and with the scaling of professional differentiation more generally (Bettencourt et al., 2014). Industry-by-occupation data can distinguish a capability that has been lost from one that has been absorbed, and this distinction is required before relatedness measured on industry shares can be trusted for mature economies.

7.2 Transition mechanisms and network context

Studies should compare rooted and pioneering transitions. The mechanisms reinforcing an established path may differ from those enabling departure from it (Corradini and Vanino, 2022). This distinction is central to policies for green transition, technological upgrading, and escape from regional lock-in. Because urban sustainability policies distribute costs and benefits unevenly within city-regions and across urban systems, studies should also examine who gains, who bears transition costs, and how network position conditions access to new opportunities (Le Gallo, 2026). Locally rooted entry and externally managed branch operations should likewise be tracked separately, and their downstream effects on local capability compared (Hirschman, 1958). Network context should also be incorporated explicitly. Intercity mobility, scientific collaboration, ownership, trade, and digital connectivity may substitute for or complement local capabilities (Liang et al., 2024).

7.3 Outcomes, intervention, and comparative learning

Development outcomes should be multidimensional. Forest recovery research shows that structure, composition, biomass, and function recover at different rates (Poorter et al., 2021). Urban economies likewise may progress in productivity while lagging in inclusion, resilience, wage quality, entrepreneurship, decarbonization, or resource efficiency. Sustainable-city outcomes should therefore be evaluated jointly across environmental, social, and economic dimensions rather than inferred from compositional change alone (Griffith, 2021; Zucaro and Agostinho, 2025). There is no single scalar measure of economic maturity. Policy evaluation should instead test whether interventions selected through Table 1 and the diagnostic workflow actually relax the diagnosed constraint and improve the targeted outcomes. Comparative research should ultimately identify the level at which generalization is useful: not a universal sequence for all cities, but a library of bounded models linking recurrent configurations to mechanisms, outcomes, and policy levers.

A further measurement priority is to reconstruct longitudinal input–output and resource-flow networks for cities or functional urban regions. This would connect economic succession to urban-metabolism research by testing whether the network-flow measures introduced above follow recurrent trajectories as industrial composition changes, and whether deviations predict resource use, waste, emissions, resilience, or dependence on external systems (Ulgiati and Zucaro, 2019; Braun et al., 2021; Iskrzyński et al., 2022; Gloria et al., 2024). The measures should remain separate rather than being collapsed into a single index of economic maturity: stronger local linkage or cycling need not imply greater productivity, resilience, or sustainability. The objective is to establish boundary conditions and determine whether departures diagnose weak linkage formation, external dependence, inefficient lock-in, or other constraints.

This priority faces a considerable data constraint: official, longitudinal input–output tables rarely exist at a functional urban-region scale. Researchers may regionalize national tables with non-survey approaches such as FLQ, update or balance partial matrices with GRAS, or embed local sectors in multi-regional input–output systems, but the assumptions and uncertainty of reconstructed links must be reported (Flegg and Webber, 2000; Temurshoev et al., 2013; Timmer et al., 2015). Where those data are too weak, pragmatic proxies include localized buyer–supplier records, inter-industry labor flows, industry-by-occupation matrices from the U.S. Census Bureau's American Community Survey (ACS), O*NET-weighted skill networks, patent or collaboration networks, and occupation-space transitions. Such proxies test linkage formation and capability movement without claiming to recover a complete city-region transaction table.

8 Limits and boundary conditions

Urban economic succession is not a claim that cities are biological ecosystems. Several differences are fundamental. First, cities are radically open. They draw labor, capital, knowledge, energy, materials, and demand from distant places, so a local absence need not be a functional absence. Second, human agency is strategic. Firms, governments, universities, workers, and investors anticipate outcomes and deliberately alter constraints. Ecological organisms can modify environments, but they do not formulate industrial policy. Third, political power and distribution matter. A development pathway may be feasible and profitable while remaining socially undesirable. Succession theory does not settle whose objectives should define the target state. Fourth, economic innovation can occur over short time scales. New technologies and organizational forms can alter the activity pool far more quickly than evolution alters a regional species pool. Fifth, there is no universal urban climax. Mature forests are themselves dynamic, and urban economies even more so. A service-dominated metropolis is not the necessary or normatively superior endpoint for every city. Sixth, greater diversity, local linkage density, or internal cycling is not automatically sustainable. A more interconnected economy may remain carbon intensive, shift environmental burdens outside the city-region, or deepen inequality. Sustainability must therefore be evaluated with explicit environmental and distributive criteria rather than inferred from network complexity or economic maturity (Ulgiati and Zucaro, 2019; Zucaro and Agostinho, 2025).

Strategic resistance must therefore be distinguished from inherited historical feedback, but this distinction should not be framed as active economic agency vs. passive ecological modification. Organisms can alter their environments through behavior, niche construction, and ecosystem engineering, changing habitat structure, resource availability, disturbance regimes, and the conditions governing the establishment and persistence of other organisms (Jones et al., 1994; Odling-Smee et al., 2003). At an abstract causal level, these processes are not wholly dissimilar from political-economic resistance: in both cases, incumbent system participants can modify the environment in ways that facilitate some potential successors and inhibit others. The analogy has an important boundary, however. Firms, coalitions, and public actors can anticipate future and distributive consequences, coordinate around explicit objectives, and deploy formal institutions—including law, regulation, property rights, standards, procurement, finance, and administrative vetoes—to determine who may enter and on what terms. Organized political resistance is therefore a distinctive form of environment modification, but environment-shaping activity itself is not unique to human economies.

The relevant distinction is consequently threefold. First, inherited legacies such as a sunk refinery, established zoning, or a specialized labor pool may persist without current coordinated action. Second, ongoing environment-modifying feedback is generated by participants in both ecological and economic systems. Third, strategically organized institutional action occurs when incumbent firms, coalitions, or public actors lobby against clean procurement, restrict access to land or infrastructure, control finance or standards, or delay permits in order to suppress an opening or raise an entrant's costs. These are observable actions and veto points, not merely “history.” In the framework, such actions should be coded according to the gate they alter—opportunity, access, or local capability fit—while actor identity, resources, political power, and institutional position remain explicit parts of the explanation for their effectiveness. Sustainability-transitions research is particularly valuable for analyzing these regime actors and governance struggles (Geels and Schot, 2007; Markard et al., 2012; Hodson and Marvin, 2010).

These differences do not invalidate comparison, but they limit what shared complex-system properties can establish. Each maps onto a component of the urban succession framework: openness is represented through network context, agency and policy enter through opportunity creation and intervention, power and distribution enter through target definition and evaluation, and innovation changes the accessible activity pool. Shared classification therefore motivates comparison; it does not establish mechanism equivalence or prescribe a single endpoint.

Activity availability illustrates this boundary. The pool of potential economic activities is analogous to the species pool and dispersal at the level of the constraint, not the mechanism: seeds exercise no judgment about their destination, whereas economic arrival depends on decisions by founders, corporations, workers, and investors. Availability is therefore partly endogenous to the city because wages, amenities, institutions, tax treatment, and the existing portfolio affect whether agents with the relevant capabilities choose to act there. The mode of arrival also matters. As Hirschman (1958) emphasizes, an activity contributes to development through backward and forward linkages that induce further local investment. Locally rooted firms may purchase, hire, train, and make strategic decisions locally, whereas externally governed branches may rely more heavily on corporate networks elsewhere. Activity availability should therefore be assessed not only by whether an activity arrives, but also by whether its establishment creates local linkages and capabilities that alter subsequent entry.

9 Conclusion

9.1 Contribution

Sustainable urban transformation is often framed either through high-level targets and aggregated regularities or through place-specific strategies. The former identify desired directions and important patterns but offer limited guidance for intervention in a particular city; the latter preserve local context but are difficult to generalize or transfer. Existing middle-range traditions identify mechanisms of cumulative diversification, linkage formation, and technological fit, but these insights remain fragmented and incompletely operationalized. Ecological succession offers a framework for expanding and enriching this middle range of development theory by connecting broad patterns to a causal diagnosis of opportunity, arrival, performance, interaction, history, feedback, disturbance, and network context.

9.2 Boundary conditions

Shared complex-system properties make disciplined comparison between ecosystems and economies plausible, but the transfer is informative only where translated mechanisms can be observed and tested under economic agency. Urban economic succession is therefore not a claim that cities reproduce ecological trajectories or converge on a universal climax. It is a causal ordering in which each condition constrains the next without determining it, and in which compositional change includes entry, persistence, transformation, absorption, contraction, and exit. The framework adds value only if this ordering explains outcomes beyond contemporaneous capabilities, aggregate composition, and existing economic regularities.

9.3 Policy implications

The practical contribution is a disciplined way to diagnose stalled or harmful transitions, identify binding constraints, sequence interventions, and learn cumulatively across cities without erasing the importance of place. For sustainable cities, decarbonization, resilience, circularity, and inclusion are not single interventions or universal stages. They are conditional transformations of urban economic portfolios, networks, and capabilities. Policy should therefore begin by identifying whether the binding constraint lies in opportunity, access, local fit, interactions, historical legacy, strategic resistance, or extra-local network context, and then evaluate whether the selected intervention actually changes that mechanism.

Generative AI statement

The author(s) declared that Generative AI was used in the creation of this manuscript. During preparation and revision of this manuscript, the authors used OpenAI ChatGPT (GPT-5.6 Pro, OpenAI, San Francisco, CA, USA; accessed 24 August 2026) to assist with manuscript condensation, language editing, figure design, figure-file preparation, and LaTeX coding. Both figures were created specifically for this manuscript under the authors' direction. Figure 1 uses author-controlled, editable LaTeX/TikZ source, whereas Figure 2 is supplied as an author-controlled 300-dpi PNG image. Neither figure incorporates any third-party image, logo, proprietary diagram, or other copyrighted artwork. The authors reviewed and edited all output and take full responsibility for the content of the publication.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Philip Boland, Queen's University Belfast, United Kingdom

Reviewed by: Luis Manuel Cerdá Suárez, Universidad Internacional de La Rioja (UNIR), Spain

Horacio Samaniego, Universidad Austral de Chile, Chile