Identifying uncertainties in scenarios and models of socio-ecological systems in support of decision-making

There are many sources of uncertainty in scenarios and models socio-ecological systems, understanding these uncertainties is critical supporting informed decision-making about the management natural resources. Here, we review across steps needed to create scenarios, from narrative storylines representation human biological processes estimation scenario model parameters. We find that would benefit moving away “stylized” approaches do not consider a wide range direct drivers their dependency on indirect drivers. Indeed, greater focus social phenomena fundamental functioning nature human-dominated planet. no panacea for dealing with uncertainty, but several evaluating still routinely applied modeling, this becoming increasingly unacceptable. However, it important avoid an excuse inaction when facing environmental challenges. “The whole problem world fools fanatics always so certain themselves, wiser people full doubts.”1Russell B. History Western Philosophy. Simon & Schuster, 1945Google Scholar With phrase, Bertrand Russell highlights imperative embracing rather than fooling ourselves into thinking does exist. This holds especially true how understand world, including role humans systems. know systems complex. They non-linear, bifurcate, have feedbacks tipping points, all which makes future development inherently uncertain difficult predict. place can never know; cannot observe it, measure it. Yet, decision-makers challenged planning short- long-term strategies preserving biodiversity contributions people2IPBESThe IPBES regional assessment report ecosystem services Europe Central Asia.in: ). IPBES, 2018Google and, so, need anticipate what may hold. The scientific response challenge has been explore space plausible, unknown, futures.3IPBESThe methodological services. 2016Google Scenarios predictions, “a plausible often simplified description develop based coherent internally consistent set assumptions key driving forces relationships.”4Millennium Ecosystem AssessmentEcosystems Human Well-Being: Synthesis. Island Press, 2005Google commonly underpinned by qualitative descriptions (narrative storylines) underlying change, policy options,3IPBESThe Scholar,5Rounsevell M.D.A. Metzger M.J. Developing change assessment.Wiley Interdiscip. Rev. Clim. Change. 2010; 1: 606-619Google translated impacts biodiversity, services, complex using storyline simulation approach.3IPBESThe Hence, be qualitative, quantitative, or both. As such, invaluable tools guiding long-term, strategic policies prompt actions increase public awareness threats nature.6Acosta L.A. Wintle B.A. Benedek Z. Chhetri P.B. Heymans S.J. Onur A.C. Painter R.L. Razafimpahanana A. Shoyama K. Using inform decision making design implementation.in: Ferrier S. Ninan K.N. Leadley P. Alkemade R. Acosta Akçakaya H.R. Brotons L. Cheung W.W.L. Christensen V. Harhash K.A. 2016: Methodological Assessment Models Biodiversity Services. 35-81Google Due complexity also advances knowledge observation capacity, being developed increasing complexity, involving feedbacks, integrating multiple components ecosystem, physical environment societies. Examples include, land-use models,7Alexander Prestele Verburg P.H. Arneth Baranzelli C. Batista e Silva F. Brown Butler Calvin Dendoncker N. et al.Assessing land cover projections.Glob. Change Biol. 2017; 23: 767-781Crossref PubMed Scopus (62) Google agent-based models,8Brown Seo Rounsevell M. Societal breakdown as emergent property large-scale behavioural use change.Earth Syst. Dyn. 2019; 10: 809-845Google marine models,9Tittensor D.P. Eddy T.D. Lotze H.K. Galbraith E.D. W. Barange Blanchard J.L. Bopp Bryndum-Buchholz Büchner al.A protocol intercomparison fishery models: fish-MIP v1.0.Geoscientific Model. Dev. 2018; 11: 1421-1442Crossref (0) Scholar,10Travers Shin Y.J. Jennings Cury Towards end-to-end investigating effects climate fishing ecosystems.Prog. Oceanography. 2007; 75: 751-770Crossref trophic levels,11Harfoot M.B.J. Newbold T. Tittensor Emmott Hutton J. Lyutsarev Smith Scharlemann J.P.W. Purves D.W. Emergent global patterns structure function mechanistic general model.PLoS 2014; 12: e1001841https://doi.org/10.1371/journal.pbio.1001841Crossref dynamic vegetation models,12Pavlick Drewry D.T. Bohn Reu Kleidon Jena Diversity-Dynamic Global Vegetation Model (JeDi-DGVM): diverse approach representing terrestrial biogeography biogeochemistry plant functional trade-offs.Biogeosciences. 2013; 4137-4177Crossref Scholar,13Prentice I.C. Bondeau Cramer Harrison S.P. Hickler Lucht Sitch Sykes M.T. Dynamic modeling: quantifying responses change.in: Terrestrial Ecosystems Changing World. Springer Berlin Heidelberg), 2007: 175-192https://doi.org/10.1007/978-3-540-32730-1_15Crossref state transition landscape models,14Daniel C.J. Frid Sleeter B.M. State-and-transition framework forecasting change.Methods Ecol. Evol. 2016; 7: 1413-1423https://doi.org/10.1111/2041-210X.12597Crossref niche-based species change.15Buisson Thuiller Casajus Lek Grenouillet G. Uncertainty ensemble distribution.Glob. 16: 1145-1157https://doi.org/10.1111/j.1365-2486.2009.02000.xCrossref strong developing comprehensive modeling empirical evidence,16Medlyn B.E. De Kauwe M.G. Zaehle Walker A.P. Duursma R.A. Luus Mishurov Pak Wang Y.-P. al.Using guide field experiments: priori predictions CO2 nutrient- water-limited native Eucalypt woodland.Glob. 22: 2834-2851Google Scholar,17Cury P.M. Planque Durant J.M. Fromentin Kramer-Schadt Stenseth N.C. Travers Grimm oceanography fisheries.Trends 2008; 338-346Abstract Full Text PDF but, until now, far less effort dedicated exploring within models, used quantify scenarios. Identifying achieving buy-in stakeholders, evidence-based decision-making, shift mindsets perception ecosystems, To influence analyses trigger appropriate responses, Intergovernmental Science-Policy Platform Services (IPBES) strongly encouraged warns “should care, taking account unpredictability associated model-based projections.”3IPBESThe A improving augment capacity among projections.18Pereira H.M. P.W. Proença Fernandez-Manjarrés J.F. Araújo M.B. Balvanera Biggs al.Scenarios 21st century.Science. 330: 1496-1501Crossref (1085) current context mean decisions made private sector organizations. In doing seek address some challenges raised Elsawah al.19Elsawah Hamilton S.H. Jakeman A.J. Rothman D. Schweizer Trutnevyte E. Carlsen H. Drakes Frame Fu al.Scenario socio-environmental analysis futures: recent efforts salient research agenda making.Sci. Total Environ. 2020; 729: 138393https://doi.org/10.1016/j.scitotenv.2020.138393Crossref relate such stakeholder engagement co-development linking geographical, sectoral, temporal scales, links between quantitative addressing surprises, consistency, communicating decision-making. aim undertake exhaustive evaluation types. Instead, examples very illustrate uncertainty. require models. wider ranging found IPBES3IPBESThe provide overview treated enhance confidence creation next generation novel both tackling implications setting out number potential solutions recommendations deal uncertainties. three categories uncertainty: (see Table 1) realms. chain useful decision-makers, storylines, parameters, initialization evaluation. Some differences worldviews, limits our others represent reliability input data spatial scales. Figure 1 shows types (from observational data, development, construction projections together decision-making.Table 1Sources systemsUncertainty sourcesDescriptionUncertainty typesScenario uncertaintyThe alternative worldviews quantification parameters conditional descriptions.Linguistic similar terms different things communities, e.g., pathways, ensembles, boundary conditions.Narratives imagining unknown futures (e.g., unknowns). relate, example, participatory arising internal consistency limitations.Scenario parameter subsequently Scenario follows interpretation values narratives, “high population growth” scenario.Model done.Structural (epistemic) choice models.Input variability baseline conditions initialize model, thematic classification, i.e., classes defined in, maps.Error propagation amplification (or dampening) transmission errors coupled meta-modeling (such cross-sectoral interactions).Decision uncertaintyCommunicating translating results studies decision-making.Data Selective information interpretation.Analyzing at relevant spatiotemporal selection scales simulated analyzed, granularity derived indicators level integration facets, merging subsets services).Decision-making tools. variety decision-supporting methods, multi-criteria analysis. Open table new tab Linguistic classified five distinct types: vagueness, dependence, ambiguity, indeterminacy theoretical terms, under-specificity.20Regan Colyvan Burgman M.A. taxonomy treatment ecology conservation biology.Ecol. Appl. 2002; 618-628Crossref Of these, ambiguity vagueness arguably occur most commonly, largely because terminology common language words. word “scenario” itself derives theater. communities sometimes attribute meanings same “precise” word, ambiguous. For “pathways” synonym “projections” “trajectories” (as shared socio-economic pathways),21O’Neill B.C. Kriegler Riahi Ebi K.L. Hallegatte Carter T.R. Mathur van Vuuren research: concept socioeconomic pathways.Climatic 122: 387-400Crossref alternatively describe time-dependent required achieve vision.2IPBESThe term one sense lead confusion if interpreted other sense. Vagueness relates statements insufficient precision. “population growth will over coming 50 years” tells us nothing actually looks like. Is doubling population, tripling, something else? These linguistic they considerations outcomes decision-makers. Recent technology provides means minimize building ontologies, formal definitions concepts relationships domain interest, synonyms equivalents closely related domains. While domain-specific ontologies exist facilitate mining sharing,22Madin J.S. Bowers Schildhauer M.P. Jones Advancing ecological ontologies.Trends 159-168Abstract knowledge, there widely accessible controlled vocabulary thesaurus standardizing meaning basic case Panel Climate (IPCC).23Sleeman Finin Halem Ontology-grounded topic science research.arXiv. (arXiv:1807.10965v2)Google first step storylines.5Rounsevell trajectories (and interactions) change. Socio-economic changes conditions, where assumed of, societal consumption industrialization.24van O’Neill Edmonds Kram matrix architecture.Climatic 373-386Crossref arise mental models,25Metzger Schröter Leemans spatially explicit vulnerability service Europe.Reg. 8: 91-107Google well difficulty historical analogs sufficient possible futures.26Maier Guillaume J.H.A. Delden Riddell G.A. Haasnoot Kwakkel J.H. An future, deep robustness adaptation: fit together?.Environ. Softw. 81: 154-164Crossref (201) Scholar,27Trutnevyte Guivarch Lempert Strachan Reinvigorating technique expand consideration.Climatic 135: 373-379Crossref (58) affects “plausibility” whether causal reflect real-world developer. particular “black swans,” shocks surprises system, events unexpected low probability occurring, high impact.28Taleb N.N. Black Swan: Impact Highly Improbable. Random house, 400Google swans imagine, unprecedented historically. way clearly document underpin narrative, communicate reporting study.29Metzger den Heiligenberg H.A.R.M. Pérez-Soba Hardiman P.S. How personal judgment influences development: example rural Europe.Ecol. Soc. 15: 5http://www.ecologyandsociety.org/vol15/iss2/art5/Google Most supply side resource crop production fish harvesting), say little demand patterns, dietary preferences) economic institutional transformations implicitly underlie storylines. Although diets, consequences vegetarian vegan,30Henry R.C. Alexander Rabin Anthoni transitions safeguarding biodiversity.Glob. 58: 101956Crossref Scholar,31Vuuren Van Stehfest Gernaat D.E.H.J. Den Berg Bijl D.L. Boer H.S. Daioglou Doelman J.C. Edelenbosch O.Y. Harmsen al.The negative emission technologies.Nat. 391-397https://doi.org/10.1038/s41558-018-0119-8Crossref today situation.32Brown Holman I. Achievement Paris goals unlikely due time lags system.Nat. 9: 203-208https://doi.org/10.1038/s41558-019-0400-5Crossref explicit. Existing ecosystems narrow drivers, change,33Lotze Barrier Bianchi al.Global reveal ocean biomass declines change.Proc. Natl. Acad. Sci. 116: 12907-12912Crossref (95) short-term interventions protected areas effort). Moreover, consideration seafood international trade, expanding scope interventions, interactions fishing, management.34Gaines S.D. Costello Owashi Mangin Bone Molinos J.G. Burden Dennis Halpern B.S. Kappel C.V. al.Improved fisheries could offset change.Sci. Adv. 4: 1-9Google Scholar, 35Dueri Guillotreau Jiménez-Toribio Oliveros Ramos Maury O. Food security, profitability? Projecting skipjack tuna under various strategies.Glob. 41: 1-12Google 36Maury Campling Arrizabalaga Aumont Merino Squires Goujon al.From pathways (SSPs) oceanic system (OSPs): policy-relevant fisheries.Glob. 45: 203-216Google longer tradition multiple, cross-scale storylines.37Harrison P.A. Dunford R.W. I.P. impact modelling needs include interactions.Nat. 6: 885-890Google overreliance driver, accounting invasive alien species, trade wild air water pollution.2IPBESThe Furthermore, failure indirect, interactions.37Harrison Participatory approaches, co-created add richness diversity strengthen link models,38Kok Bärlund Flörke Gramberger Sendzimir Stuch Zellmer European strengthening stories models.Climatic 128: 187-200Google highly dependent individual stakeholders extent tacit knowledge. Stakeholder mapping exercises38Kok maximize resolving problem. realm, planning.39Planque Mullon Arneberg Eide J.J. Hoel A.H. Niiranen Ottersen Sandø A.B. method social-ecological systems.Fish Fish. 20: 434-451Google Scholar,40Gopnik Fieseler Cantral McClellan Pendleton Crowder Coming table: early planning.Mar. Pol. 2012; 36: 1139-1149Google Simulation specific indicators. requires translation inputs, introduce additional uncertainties.5Rounsevell draw distinction here “scenario uncertainty” “model uncertainty.” narratives values, itself, vary high, medium, storyline. general, themselves inputs. refers functions modeled processes, rate constant often, always, biophysical depends independent scenario. uses best-guess estimates uncertain, analogs. majority process estimating themselves. few exceptions “credible” ranges,41Pedde Kok Onigkeit Bridging simulations scenarios.Reg. 19: 655-666Crossref (16) probabilistic me