When Do Humans Co-Produce Ecosystem Services? New Matrix Tells Assessors Where to Look

Ecosystem services have long been framed as gifts of nature, delivered to humanity by soils, forests, wetlands and oceans. A growing body of research now insists on a more complicated truth: most of the benefits people actually obtain from ecosystems are produced jointly by natural capital and human inputs such as labour, machinery, knowledge, institutions and money. Crop yields depend on pollinators and climate alongside fertilisers, tractors and agronomic expertise. Flood regulation combines ecological functions with engineering, maintenance and governance. Even nature-based recreation is mediated by access infrastructure, interpretation and social practice. Yet despite broad conceptual recognition of this co-production, ecosystem assessments still struggle with a basic operational question: when is it actually necessary to make these human contributions explicit, and how should they be assessed?

A new study published in One Ecosystem offers a decision-oriented answer. Rather than attempting to measure or causally decompose co-production in every conceivable context, an international team working under the EU Horizon SELINA project developed a transparent screening process to determine where explicit co-production assessment is warranted and to guide the selection of fit-for-purpose methods. The work combines a structured synthesis of the conceptual and methodological literature with an expert-elicited relevance matrix spanning ecosystem types and services, illustrated through five European case studies ranging from Cypriot mountain terraces to Slovenian commons.

The conceptual foundation rests on a capital-based framing. Natural capital is understood as the interacting biotic and abiotic components of ecosystems, including their structures, processes, functions and condition, that underpin services. Anthropogenic inputs serve as an umbrella term for four further capitals: human capital covering knowledge, skills and labour; social capital covering networks, trust, institutions and collective capacities; physical capital covering infrastructure, equipment and technology; and financial capital covering the monetary resources that enable investment and maintenance. Co-production, in this framing, is the joint contribution of natural and anthropogenic inputs to the generation of realised ecosystem service flows, which may shape ecosystem potential or mobilise, access, use or transform realised flows.

A critical analytical distinction underpins the entire approach: the difference between ecosystem potential and realised flow. Ecosystem potential refers to the ability of an ecosystem, given its condition, to generate a service before it is mobilised or used, essentially a hypothetical maximum yield. Realised flow refers to the actual contribution accessed, experienced or consumed by people after relevant anthropogenic inputs have been applied. The authors argue that conflating these two can misattribute outcomes to ecosystems and obscure the human inputs on which realised flows depend. High crop yields may reflect intensive fertiliser and pesticide use rather than good ecosystem condition. Flood regulation may result from engineered infrastructure and governance arrangements as much as ecological function. Failure to make these distinctions weakens both ecological and economic accountability and the equity of benefit distribution.

The empirical core of the study is an expert-elicited relevance matrix. Sixteen experts from the SELINA co-production working group scored the relevance of anthropogenic inputs for realised service flows across 231 combinations of twelve ecosystem types, following the European MAES classification, and 23 ecosystem services, following the SEEA-EA reference list. Forty-five structurally implausible combinations were excluded during pre-screening. Each expert assigned a relevance score on a six-point scale, from zero for not relevant to five for very high relevance, alongside a confidence score and a declared level of expertise for each ecosystem type. After independent scoring, experts reviewed aggregated panel results in a consensus round following Delphi logic, and could revise their judgements where appropriate.

Aggregation followed a deliberately careful two-stage procedure. Mean relevance scores were converted into three verdict categories using fixed cut-points: combinations scoring above 3.5 were classified as Relevant, those below 1.5 as Not relevant, and those in between as Intermediate. Trust in each verdict was assessed independently using two dimensions: evidence, calculated as the geometric mean of normalised expertise and confidence, and agreement, represented by the standard deviation of relevance scores. These were crossed in a lookup table to assign one of five trust tiers, from Very low to Very high. Relevance scores were not weighted by self-reported confidence, a practice the matrix-method literature warns can amplify overconfidence bias. A knowledge-gap rule, applied leniently, retained all twelve ecosystem types.

The results revealed a striking pattern. Intermediate was the most frequent verdict, accounting for 125 of the 231 combinations, compared with 65 Relevant and 41 Not relevant. This suggests that anthropogenic inputs carry at least some relevance for most assessed combinations, but that detailed disentangling is not always necessary. Relevance was highest in Urban, with a mean score of 3.92, and Cropland, at 3.68, which together accounted for 35 of the 65 Relevant combinations. Cultural and appreciation services recorded the highest scores across the board, with education, scientific and research services topping the list at 3.95, followed by recreation-related services and ecosystem and species appreciation. By contrast, the lowest relevance was found for Open Ocean, Shelf systems and Sparsely Vegetated Land, and for broad-scale atmospheric regulating services such as rainfall pattern regulation, air filtration and global climate regulation.

Five case studies illustrate how co-production enters assessment practice at different chain stages. In the Troodos Mountains of Cyprus, a cost-benefit analysis compared traditional drystone viticultural terraces with modern mechanically constructed alternatives, finding a net present value of €48,456 per hectare for the modern system against a negative value for the traditional system, while acknowledging that cultural landscape values and regulating services remain only partly captured in monetary indicators. In northern German croplands, an enhanced indicator framework separated ecosystem potential, represented by soil quality ratings, from anthropogenic inputs such as fertilisers, pesticides and irrigation, showing that realised flows and farm income depend on management intensity, market access and capital investment as much as on ecological potential.

The remaining cases address contrasting decision problems. In the Peloponnese, Greece, a spatial decision-support approach prioritised active restoration versus natural recovery in fire-affected forests, using transparent criteria based on fire-adaptive capacity, fire history and slope steepness, and treating human intervention as a targeted, evidence-based input. In Trento, Italy, high-resolution urban modelling combined the Storm Water Management Model for pluvial flooding with the SOLWEIG model for thermal comfort, demonstrating that in cities anthropogenic inputs are simultaneously pressures and enabling conditions: soil sealing reduces regulating capacity, while tree planting, desealing and drainage design enhance realised flows. In Triglav National Park, Slovenia, participatory assessment and downscaled MODIS satellite data showed that protection status alone does not determine realised service supply, since forage provision depended on land-use structure and commons governance rather than protection status.

The overarching implication is one of proportionality. Explicit co-production assessment should be prioritised where anthropogenic inputs materially shape realised flows, where attribution to ecosystem potential alone would be misleading, and where decision-making needs justify the analytical effort. In highly managed systems such as urban and cropland environments, and for cultural services where access, meaning and institutions may be constitutive of the service itself, detailed assessment adds clear value. In lower relevance or low-trust combinations, a qualitative statement, sensitivity analysis or transparent boundary choice may be more appropriate than a full co-production assessment. No single method can operationalise co-production across all contexts; indicator-based, spatially explicit, econometric, process-based, participatory and governance-oriented approaches provide complementary entry points depending on the focal service, chain stage, available data and decision problem. The study’s contribution lies not in causal attribution for each combination, but in a transparent decision logic that links relevance screening, capital identification and method selection, reducing the risk of misattributing realised ecosystem service flows and supporting more accountable, equitable integrated ecosystem assessments.

Subject of Research: Ecology

Article Title: Operationalising ecosystem service co-production: A decision-oriented approach for integrated ecosystem assessments

Article References: Zoumides, C., Bank, E., Aabid, S., Balfors, B., Bernardo, F., Brander, L., Dimopoulos, P., Geneletti, D., Giannakis, E., Šmid Hribar, M., Inácio, M., Kato-Huerta, J., Kokkoris, I., Kretsch, C., Longo, A., Mörtberg, U., Pereira, P., Ribeiro, D., Theodorou, C., & Burkhard, B. (2026). Operationalising ecosystem service co-production: A decision-oriented approach for integrated ecosystem assessments. One Ecosystem, 11, Article e206925. https://doi.org/10.3897/oneeco.11.e206925

Image Credits: AI Generated

DOI: 10.3897/oneeco.11.e206925

Keywords: ecosystem services, co-production, integrated ecosystem assessment, expert elicitation, One Ecosystem, SELINA, Operationalising, ecosystem, service, decision-oriented, approach, integrated