A well-timed switch from local to global agreements accelerates climate change mitigation
Vadim A. Karatayev, Vítor V. Vasconcelos, Anne-Sophie Lafuite, Simon A. Levin, Chris T. Bauch, Madhur Anand
AA well-timed switch from local to global agreements acceleratesclimate change mitigation
Vadim A. Karatayev *, V´ıtor V. Vasconcelos , , , , , Anne-Sophie Lafuite , Simon A. Levin , , , ,Chris T. Bauch , Madhur Anand School of Environmental Sciences, University of Guelph, Guelph, Canada. PrincetonInstitute for International and Regional Studies, Princeton University, Princeton, NJ, USA. Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ,USA. Andlinger Center for Energy and the Environment, Princeton University, Princeton,NJ, USA. Princeton Environmental Institute, Princeton University, Princeton, NJ, USA. Informatics Institute, University of Amsterdam, Amsterdam, Netherlands. Resources forthe Future, Washington, DC, USA. Beijer Institute of Ecological Economics, Stockholm,Sweden. Department of Applied Mathematics, University of Waterloo, Waterloo, Canada.* - corresponding author. E-mail [email protected]
Abstract:
Recent attempts at cooperating on climate change mitigation highlight thelimited efficacy of large-scale agreements, when commitment to mitigation is costly andinitially rare. Bottom-up approaches using region-specific mitigation agreements promisegreater success, at the cost of slowing global adoption. Here, we show that a well-timed switchfrom regional to global negotiations dramatically accelerates climate mitigation comparedto using only local, only global, or both agreement types simultaneously. This highlightsthe scale-specific roles of mitigation incentives: local incentives capitalize on regional dif-ferences (e.g., where recent disasters incentivize mitigation) by committing early-adoptingregions, after which global agreements draw in late-adopting regions. We conclude thatglobal agreements are key to overcoming the expenses of mitigation and economic rivalryamong regions but should be attempted once regional agreements are common. Graduallyup-scaling efforts could likewise accelerate mitigation at smaller scales, for instance whencostly ecosystem restoration initially faces limited public and legislative support.Existing attempts to realize international cooperation on halting climate change highlightthat many ecosystem services are public goods that can incentivize beneficiaries to free-rideon others’ mitigation efforts [1]. The limited capacity of the atmosphere to absorb green-house gases before emissions severely affect climate globally is one public good that can bepreserved though investments to reduce emissions (‘mitigation’ hereafter) [2, 3]. Adoptingalternative investment opportunities with faster payoffs however can delay cooperation unlessit is reinforced by social norms [4] or short-term climate forecasts become severe [5]. Playerscan also delay mitigation to wait for others to develop mitigation technology and infras-tructure, or to wait for rival players to invest first and give up their competitive advantage[6]. At the same time, achieving climate change mitigation globally and rapidly is criticalto minimizing adaptation costs, and in preventing irreversible ecosystem degradation (e.g.,species loss, [7, 8]), degradation of human well-being, especially in developing countries thathave lower capacity for climate adaptation [9, 10], and preempting the collapse of favorable1 a r X i v : . [ n li n . AO ] J u l limate regimes when human impacts exceed environmental tipping points [11, 12].Advancing mitigation commitments requires incentives such as trade agreements or laws[13] (‘agreements’ hereafter) that are specific to the spatial scales of political bodies. Thefailure of global climate agreements to be effective has led to a focus on building blocksapproaches to global cooperation, in which an agreement grows from local setups [1, 14, 15].Game-theoretic work predicts that such localized mitigation agreements might succeed whereglobal cooperation fails by reducing the number of interacting players and the potential forfree-riding [16, 17, 18, 19]. However, the greater potential for bottom-up approaches toachieve mitigation locally can trade off with a much longer time to achieving mitigationglobally, as change happens more slowly in more distributed systems [20]. So, can societiesenhance both the probability and pace of desired outcomes, such as reducing greenhouse gasemissions, by a properly-timed switch from local- to large-scale mitigation agreements?To answer this question, we consider a system of L regional groups, representing coalitionsof countries with shared interests or geography (e.g., BRICS, AILAC, Western ClimateInitiative), formal unions (e.g., EU), or countries with partly-independent states (e.g., UnitedStates). Given that many benefits of climate mitigation require global cooperation andcurrently are politically debated, we focus on the impact of the (high) costs of mitigation.In each region, players can free-ride by not acting on the provision of the public good, orthey can contribute to providing the public good by mitigating emissions. Simultaneously,mitigators support agreements that, when sufficiently endorsed, punish free-riding throughfines, sanctions, or other penalties. Such sanctioning agreements can be local, affectingplayers only within a specific region, global, supported by and affecting all players acrossregions, or both [21]. We examine whether bottom-up regional (local) agreements, world-wide (global) agreements, or a dynamic combination of these two best promotes fast climatechange mitigation. System dynamics.
As observed in other research [19, 22] and in reality [23], we findthat premature global-scale efforts to achieve cooperation on climate invariably fail (Fig.1a). Although exogenous events occasionally produce brief mitigation efforts within indi-vidual regions, at a global scale, mitigation never reaches the quorum required to institutesanctioning agreements (Fig. 1b.iii). In contrast, local sanctioning agreements can capitalizeon stochastically induced, regional mitigator dominance to establish mitigation as the normwithin individual regions. Eventually, this process occurs in every local group, translatingto global climate mitigation (Fig. 1b.ii).Once local agreements produce a sufficiently high global proportion of mitigators, how-ever, a shift to global agreements more than doubles the rate of mitigation spread (Fig.1b, 2a). This difference arises because enacting global sanctions once mitigation is commonpushes non-mitigating regions to cooperate on reducing emissions. Compared with localagreements only, switching to global agreements doubles the amount of accumulated miti-gation achieved by year 60 (Fig. 1a) and halves the time it takes for 80% of countries tocommit to mitigation (Fig. 2a). This benefit of switching to global agreements increaseswith mitigation cost, sanctioning cost, and the role of payoffs in strategy choice (see Sup-plementary Figure 1 for sensitivity analysis) because players quickly reconsider mitigation2ommitments that arise through chance events. High mitigator costs therefore decrease theoverall probability that chance events induce mitigation without the benefit of among-regionsanctions that occurs through global agreements once many regions commit to mitigation.Notably, the threshold required for a switch to be effective does not need to exceed a 50%majority of regions (Fig. 1a).Bootstrapping cooperation through local agreements may face difficulties. Investmentsin a shared technology, infrastructure, or public support create impediments to early motiva-tions. Typically, these investments are initially more costly and only as they are implementedthey become progressively more accessible, as more nearby players become mitigators. Thisoccurs through research and development of industries of trading partners and industri-ally related products [24], and economies of scale (we refer to these effects collectively aseconomies of scale). When there are substantial gains for scaling, even though initial costsare high, we observe that such economies of scale further increase the benefits of switchingto global agreements (19%, Fig. 2b) as inter-group interactions pull along non-mitigatingregions where mitigation costs remain high, making regional action even more important.A second major obstacle in climate change cooperation arises at the global scale fromgame-of-chicken dynamics among economically competing regions, for which investment inmitigation can reduce the competitiveness of regional economies through opportunity costs(Falkner et al. 2010). For instance, China maintains that industrialized countries shouldinvest in mitigation first so that other countries can catch up economically, while industrial-ized countries such as the United States wish to retain a competitive advantage. Accountingfor such inter-region rivalry, we find that group rivalry disproportionally slows down miti-gation through locally based agreements. As local agreements provide no incentive for pairsof intensely competing regions to commit to mitigation, a switch to global institutions ismore critical in pulling along holdout regions (Fig. 2c). Unlike economies of scale operatingwithin regions, the positive feedback by which group rivalry holds back mitigation arises atthe global scale. This distinction is highlighted by the fact that, with local agreements only,some regions experiencing strong competition never adopt mitigation. This highlights thatglobal agreements can play a key role in promoting not only the pace but also chance ofglobal climate cooperation.
Importance of switching agreement scales.
Given that both local and global agree-ments are critical to promoting mitigation, does a concerted strategy attempting agreementsat both scales simultaneously improve outcomes or does it risk diluting limited resources?Attempts to form sanctioning agreements can involve constitutive costs – resources spenteven if agreements fail to establish, such as lobbying for a carbon tax bill that ultimately fails.Likewise when enacting timely and effective laws is not possible, sanctioning might initiallydepend on a few audacious players that refuse cooperation with non-mitigators at substan-tial cost to themselves, even if sanctions ultimately fail to promote mitigation. For instance,trade sanctions by Western European nations against all nations lacking aggressive climatemitigation commitments might disproportionally weaken Western European economies whenglobal mitigation commitment is rare.We find that under high constitutive sanctioning costs mitigators must first focus all3fforts locally, for instance refusing trade only with non-mitigators who are in their group(Fig. 3). Thereafter, we find that the switch from local to global mitigation incentives isbest done earlier (i.e., lower switching threshold T ) when agreements are easier to establish(i.e., require less support or smaller quorums to effectively sanction defectors) or globalagreements have greater efficacy (Supplementary Figure 1), but premature shifts to globalagreements can delay mitigation at low levels for extended periods (Fig. 1b.iii). For lowconstitutive sanctioning costs, efforts to establish local and global sanctioning agreementssimultaneously do little to affect outcomes (Fig. 3) because global agreements initially failand local agreements eventually become redundant. However, a more graduated shift wherelocal and global agreement attempts overlap might be more beneficial when early globaldialogue establishes a framework for later negotiations, when sanctions are more effectiveamong neighbors than among regions, or when uncertainties exist in players’ commitmentsto mitigation or in the quorum needed to establish a global agreement. Agreements have scale-specific roles.
We find that a timely shift from local to globalagreements consistently accelerates and establishes global climate mitigation over agreementsfocused on a single scale. Building on analyses of cooperation dynamics near steady-state[19], our focus on transient dynamics highlights that global agreements play a key rolein pulling along late-adopting regions once mitigation is sufficiently common. This roleincreases as economies of scale, among-group rivalry, or high costs (Supplementary Figure1) make mitigation less likely to arise through chance events such as popular movementsor natural disasters. Global agreements investigated here can also complement a regimecomplex approach to mitigation, wherein groups can form over different mitigation targetsand address different issues that collectively tackle climate change [25, 26]. Overlap ingroup membership can accelerate mitigation globally by emphasizing the marginal gains tocooperation when information about the outcomes of action is missing [21]. Therefore thecurrent reality that players commit to multiple mitigation targets and political coalitions,some of which enjoy a high membership, could further incentivize mitigation and reduce theextent of sanctioning that falls upon local and global agreements.Our multi-scale model also underscores the importance of local agreements in establish-ing mitigation during the early phases of conservation efforts. First, local agreements cancapitalize on increases in mitigation that arise from local variation but have little impacton mitigation at the global scale, as seen in lower benefits of switching as stochasticity be-comes more prevalent in strategy choice (see SM for description of the mitigation behaviorfor varying group sizes, rate of exogenous events and impact of payoff). This aspect can beespecially critical as climate disasters create strong but region-specific momenta for climatemitigation [27, 28]. For instance, whilst recent wildfires in Australia reinforce climate changeconcerns globally, they could create an especially strong case for changing the domestic poli-cies in Australia towards mitigation. Second, local agreements ‘lock-in’ mitigation as thenorm in early-adopting regions during early adoption phases when global agreements fail.For this reason, local agreements help mitigation establish and spread when rare whereasglobal agreements fail [19].
Implications for ecosystem restoration.
While we focus on mitigating climatechange on the international stage, the benefit of leveraging the scale-specific roles of agree-4ents found here can also accelerate environmental mitigation at smaller scales. In additionto climate negotiations, mitigating the impacts of climate disasters and reducing greenhousegas emissions will increasingly require restoring ecosystems over large spatial scales [29]. Theurgency of ecosystem-level mitigation rises further as concentrated human impacts can beslow to change [30] and ongoing ecological feedbacks are accelerating ecosystem degradation[31]. As with climate change mitigation, ecosystem restoration can be costly when localeconomies must limit resource extraction or reorient towards recreation. Initial public sup-port can therefore be limited to a few local communities where traditional values, publicoutreach, or successful restoration projects establish conservation as a self-sustaining socialnorm [32, 33].Building on this, our results suggest that environmental agencies first focus limited re-sources locally rather than in many places at once, prioritizing communities where restorationhas lower costs or greater public support. Establishing ‘buy-in’ among local communitiescan in turn accelerate and sustain ecosystem restoration [34, 35, 36]. Ecological feedbackloops can further increase importance of focusing efforts by creating threshold effects, revers-ing meager restoration efforts while increasing resilience of restored locations to disturbance[37]. After a string of local successes garners greater public support, a focus shift towardsenacting state-level legislation and concerted restoration efforts in all remaining locationscould ultimately boost the resources, public support, and efficiency of restoration projects[38, 29]. While most successful conservation movements begin with activism and culminatein legislation, our results underscore the importance of focusing efforts on one scale at a timeand carefully considering when to shift from local to regional restoration and public outreach.Continually worsening climate will eventually force countries to reduce greenhouse gasemissions, with current debates centering on the urgency and best course of action to achieveglobal mitigation. We have shown one strategy that can greatly accelerate cooperation tomitigate climate change in the near future by leveraging local and global efforts.
Methods
We consider L regions. Each region, j , contains N players, of which M j pay a substantialcost c to reduce emissions while all others do not mitigate, for a mitigator proportion m j = M j N − . When mitigators reach a majority, consensus establishes agreements that sanctionnon-mitigators through fines or economic sanctions by an amount p NM . Mitigators alsoincur a substantial cost of sanctioning ( p M < p NM ), for instance when sanctions prohibitlucrative trade opportunities, a fraction f of which is incurred even if non-mitigators are notsanctioned when sanctioning coalitions, institutions, or bills fail to establish.We consider a combination of local and global efforts where mitigators initially form onlylocal agreements until mitigation proportion m G exceeds a threshold T L , and then shift theirefforts towards global agreements once global mitigation proportion m G exceeds a threshold T G ≤ T L (when these are the same we write T ). We consider four types of agreements: i)coordinated flip from local to global agreements, 0 < T <
1; ii) purely local agreements,5 = 1; iii) purely global agreements, T = 0; and iv) temporary coexistence of local andglobal agreements, 0 < T G < T L <
1. Global agreements might additionally differ insanctioning efficacy by a factor E . At each time step, one randomly chosen player decides toreconsider their climate strategy with probability κ , the social learning rate. Players tend tochoose strategies with greater payoffs to a degree β , which reflects the importance of financialconsiderations in climate policy. With probability µ , however, chance events supersede anypayoff differences and players choose strategies at random.Throughout, we use a decreasing sigmoid function Θ( x ) = (1 + exp( hx )) − to modelsmooth transitions from local to global agreements ( h = 20, 0 ≤ x ≤ b = 0 and overlapping throughout the mitigationprocess for T = 0 and b >
1. Thus, the amount of punishment attempted by mitigators is δ ( m j ) = Θ( m G − T − b ) + Θ( T − m G ) and reflects a switch from local to global agreementattempts as m G exceeds T . The amount of punishment realized by successfully establishedagreements is then∆( m j ) = Θ( m G − T − b )Θ( p T − m j ) + Θ( T − m G )Θ( p T − m G ) E. (1)The total payoffs for strategy s can depend on global ( m G = ( LN ) − (cid:80) j M j ) and local( m j = M j N − ) mitigation frequency:Π NM ( m j ) = − p NM ∆( m j ) (2)Π M ( m j ) = − c − p M ( f δ ( m j ) + (1 − f )∆( m j )) . (3)At each model iteration, a randomly chosen player in group j with strategy s switches tostrategy k (with n k,j denoting the number of players with strategy k ) with probabilityPr s → k,j = κ (cid:18) (1 − µ ) n k,j N − β (Π s ( m j ) − Π k ( m j ))) + µ (cid:19) . (4)To model economies of scale, we set initial mitigation costs c that decline by a fraction r L past m j = r T using the updated mitigation cost c ( m j ) = c (1 − r L Θ( r T − m j )), where c = c ( r T + (1 − r T )(1 − r L )) − ensures that (cid:82) m j =0 c ( m j ) = c in our base model.To account for the effects of among-region economic competition, we model reduced pay-offs for regions where players commit to climate mitigation in leu of alternative investmentsto bolster their economies. Among-group interactions in the matrix A are 0 for a i = j while a i (cid:54) = j terms are drawn from a multivariate uniform distribution with rivalry reciprocity R = cor ( a i,j , a j,i ) >
0. Given group-level payoffs as P ( (cid:126)m ) = ( (cid:126)m Π M ( (cid:126)m )+(1 − (cid:126)m )Π NM ( (cid:126)m ))( A + I L ),we normalize the rows of A by their sum to obtain A N and arrive at the fitness of group j with n mitigators λ j,n = P ( (cid:126)m (cid:48) ) · ( A N − I L ) j , where (cid:126)m (cid:48) i = j = n and (cid:126)m i otherwise. We thenincorporate the consequences of individual choice for group-level payoffs using the updatedpayoffs Π Gs ( m j ) = Π s ( m j )+ αG s /
2, where G NM = λ j,m j − N − − λ j,m j , G M = λ j,m j + N − − λ j,m j ,6nd α scales individual payoff of group competitiveness relative to punishment and mitigationcosts. With this formulation mitigation adoption slows when intense competition happensamong pairs of non-mitigating regions. Acknowledgments
We thank Easton White and Mikaela Provost for feedback that improved the manuscript.This research was supported by the Natural Sciences and Engineering Research Council andthe New Frontiers Research Fund (to M.A. and C.T.B.).
Author Contributions
V.A.K. conceived the study, V.A.K., C.T.B., V.V.V., and M.A. designed and analyzedthe model, and all authors participated in writing the manuscript.
Competing Interests
The authors declare no competing interests.
References [1] Ostrom, E. E., Dietz, T. E., Dolˇsak, N. E., Stern, P. C., Stonich, S. E., and Weber,E. U.
The drama of the commons.
National Academy Press, (2002).[2] Dolsak, N. and Ostrom, E.
The Commons in the New Millenium. Challenges andAdaptation . (2003).[3] Tavoni, A. and Levin, S. Managing the climate commons at the nexus of ecology,behaviour and economics.
Nature Climate Change (12), 1057–1063 (2014).[4] Levin, S. A. Public goods in relation to competition, cooperation, and spite. Proceedingsof the National Academy of Sciences (3), 10838–10845 (2014).[5] Bury, T. M., Bauch, C. T., and Anand, M. Charting pathways to climate changemitigation in a coupled socio-climate model.
PLoS Computational Biology (6) (2019).[6] Falkner, R., Stephan, H., and Vogler, J. International Climate Policy after Copenhagen:Towards a ‘Building Blocks’ Approach. Global Policy (3), 252–262 (2010).[7] Ceballos, G., Ehrlich, P. R., and Raven, P. H. Vertebrates on the brink as indicatorsof biological annihilation and the sixth mass extinction. Proceedings of the NationalAcademy of Sciences (2020).[8] Trisos, C. H., Merow, C., and Pigot, A. L. The projected timing of abrupt ecologicaldisruption from climate change.
Nature (7804), 496+ (2020).79] Adger, W. N., Huq, S., Brown, K., Conway, D., and Hulme, M. Adaptation to climatechange in the developing world.
Progress in development studies (3), 179–195 (2003).[10] Mertz, O., Halsnæs, K., Olesen, J. E., and Rasmussen, K. Adaptation to climate changein developing countries. Environmental management (5), 743–752 (2009).[11] Lenton, T. M., Held, H., Kriegler, E., Hall, J. W., Lucht, W., Rahmstorf, S., andSchellnhuber, H. J. Tipping elements in the Earth’s climate system. Proceedings of theNational Academy of Sciences (6), 1786–1793 (2008).[12] Lenton, T. M., Rockstroem, J., Gaffney, O., Rahmstorf, S., Richardson, K., Steffen,W., and Schellnhuber, H. J. Climate tipping points - too risky to bet against.
Nature (7784), 592–595 (2019).[13] Fischer, C. and Newell, R. G. Environmental and technology policies for climate miti-gation.
Journal of Environmental Economics and Management (2), 142–162 (2008).[14] Stewart, R. B., Oppenheimer, M., and Rudyk, B. Building blocks for global climateprotection. Stanford Environmental Law Journal , 341 (2013).[15] Cole, D. H. Advantages of a polycentric approach to climate change policy. NatureClimate Change (2), 114–118 (2015).[16] Ostrom, E. Governing the commons: The evolution of institutions for collective action .Cambridge university press, (1990).[17] Ostrom, E. Polycentric systems for coping with collective action and global environ-mental change.
Global Environmental Change-Human and Policy Dimensions (4, SI),550–557 (2010).[18] Santos, F. C. and Pacheco, J. M. Risk of collective failure provides an escape from thetragedy of the commons. Proceedings of the National Academy of Sciences (26),10421–10425 (2011).[19] Vasconcelos, V. V., Santos, F. C., and Pacheco, J. M. A bottom-up institutional ap-proach to cooperative governance of risky commons.
Nature Climate Change (9),797–801 (2013).[20] Dietz, T., Ostrom, E., and Stern, P. C. The struggle to govern the commons. Science (5652), 1907–1912 (2003).[21] Vasconcelos, V. V., Hannam, P. M., Levin, S. A., and Pacheco, J. M. Coalition-structured governance improves cooperation to provide public goods.
Scientific Reports (1), 1–10 (2020).[22] Dannenberg, A. and Gallier, C. The choice of institutions to solve cooperation problems:a survey of experimental research. Experimental Economics (2019).823] Walker, B., Barrett, S., Polasky, S., Galaz, V., Folke, C., Engstrom, G., Ackerman, F.,Arrow, K., Carpenter, S., Chopra, K., Daily, G., Ehrlich, P., Hughes, T., Kautsky, N.,Levin, S., Maler, K.-G., Shogren, J., Vincent, J., Xepapadeas, T., and de Zeeuw, A.Looming Global-Scale Failures and Missing Institutions.
Science (5946), 1345–1346(2009).[24] Hidalgo, C. A., Klinger, B., Barabasi, A.-L., and Hausmann, R. The product spaceconditions the development of nations.
Science (5837), 482–487 (2007).[25] Keohane, R. O. and Victor, D. G. The Regime Complex for Climate Change.
Perspec-tives on Politics (1), 7–23 (2011).[26] Hannam, P. M., Vasconcelos, V. V., Levin, S. A., and Pacheco, J. M. Incompletecooperation and co-benefits: deepening climate cooperation with a proliferation of smallagreements. Climate Change (1), 65–79 (2017).[27] Weber, E. U. What shapes perceptions of climate change?
Wiley InterdisciplinaryReviews-Climate Change (3), 332–342 (2010).[28] Beckage, B., Gross, L. J., Lacasse, K., Carr, E., Metcalf, S. S., Winter, J. M., Howe,P. D., Fefferman, N., Franck, T., Zia, A., Kinzig, A., and Hoffman, F. M. Linking modelsof human behaviour and climate alters projected climate change. Nature Climate Change (1), 79+ (2018).[29] Aronson, J. and Alexander, S. Ecosystem Restoration is Now a Global Priority: Timeto Roll up our Sleeves. Restoration Ecology (3), 293–296 (2013).[30] Tekwa, E. W., Fenichel, E. P., Levin, S. A., and Pinsky, M. L. Path-dependent in-stitutions drive alternative stable states in conservation. Proceedings of the NationalAcademy of Sciences (2), 689–694 (2019).[31] Flores, B. M., Holmgren, M., Xu, C., Van Nes, E. H., Jakovac, C. C., Mesquita, R. C.,and Scheffer, M. Floodplains as an achilles’ heel of amazonian forest resilience.
Pro-ceedings of the National Academy of Sciences (17), 4442–4446 (2017).[32] Nyborg, K., Anderies, J. M., Dannenberg, A., Lindahl, T., Schill, C., Schlueter, M.,Adger, W. N., Arrow, K. J., Barrett, S., Chapin, F. S., Crepin, A.-S., Daily, G., Ehrlich,P., Folke, C., Jager, W., Kautsky, N., Levin, S. A., Madsen, O. J., Polasky, S., Scheffer,M., Walker, B., Weber, E. U., Wilen, J., Xepapadeas, A., and de Zeeuw, A. Socialnorms as solutions.
Science (6308, SI), 42–43 (2016).[33] Reyes-Garcia, V., Fernandez-Llamazares, A., McElwee, P., Molnar, Z., Ollerer, K.,Wilson, S. J., and Brondizio, E. S. The contributions of Indigenous Peoples and localcommunities to ecological restoration.
Restoration Ecology (1), 3–8 (2019).[34] Madzudzo, E., HaBarad, J., and Matose, F. Outcomes of community engagement incommunity-based natural resource management programmes. (2006).935] Morrison, J., Loucks, C., Long, B., and Wikramanayake, E. Landscape-scale spatialplanning at WWF: a variety of approaches. Oryx (4), 499–507 (2009).[36] Cetas, E. R. and Yasue, M. A systematic review of motivational values and conservationsuccess in and around protected areas. Conservation Biology (1), 203–212 (2017).[37] Suding, K. N., Gross, K. L., and Houseman, G. R. Alternative states and positivefeedbacks in restoration ecology. Trends in Ecology & Evolution (1), 46–53 (2004).[38] Neeson, T. M., Ferris, M. C., Diebel, M. W., Doran, P. J., O’Hanley, J. R., and McIn-tyre, P. B. Enhancing ecosystem restoration efficiency through spatial and temporalcoordination. Proceedings of the National Academy of Sciences (19), 6236–6241(2015).Figure 1: Mitigation is best achieved through a timely switch from local to global agreements,on account of scale-specific mechanisms. (A) Time series of the proportion of system-wideclimate mitigation over 10 replicate model runs with either global agreements only (red),local agreements only (gray), or a switch from local to global agreements (blue lines) oncemitigation is sufficiently common (threshold T = 0 . T ). Simulations show 8 regions with 18 players in each and no among-regionrivalry or economies of scale; all other parameters values listed in Supplementary Figure 1.10igure 2: Benefit of switching from local to global agreements increases when economiesof scale and group rivalry affect mitigation. Values denote the factor by which switchingoutperforms local-only approaches (i.e., 2=200% increase) in mitigation adoption rate (teal)and the total amount of mitigation achieved by time step 625 (red). Distributions giverelative improvements in rates and total mitigation found within each of 100 replicate modelruns, and vertical dashed lines denote mean improvement in each metric. Note that switchingoutperformed local-only strategies in all cases (i.e., all values > T = 0 .
5; all other parameters values listed in Supplementary Figure 1.Figure 3: Switching agreement scale (black, b = 0) outperforms simultaneous local and globalagreement attempts to promote mitigation (blue, b = 1 .