The Kunming-Montreal Global Biodiversity Framework is a worldwide plan to urgently address and reverse biodiversity loss, intending to achieve a harmonious relationship between humanity and nature by 2050. This paper seeks to contribute to operationalising the framework, specifically concerning biodiversity conservation and nature's contributions to people. Using a global analytical approach, we identify optimised areas for conservation, restoration and agriculture, considering food production, urban expansion, population growth, and climate change projections. By formulating scenarios for increasing natural areas enabled by improvements in agricultural productivity and trade, and considering local and global constraints on restoration actions, we analyse potential outcomes for biodiversity and people. Our findings demonstrate that an optimised spatial allocation of land use could substantially mitigate projected negative impacts and even surpass the current situation, leading to significant socio-environmental gains. However, the best global scenarios for nature and people require integrated planning that considers mitigating climate change, reducing human pressure on natural habitats, increasing trade, and changing human behaviour. Aligning efforts to protect and restore nature with broader sustainability goals through coordinated and transformative action is central to implement the Global Biodiversity Framework and delivery of a more sustainable future.
Biological Sciences - Article
Optimising global conservation, restoration, and agriculture for people and nature
https://doi.org/10.21203/rs.3.rs-3043578/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
In December 2022, Parties to the Convention on Biological Diversity (CBD) reached consensus on the Kunming-Montreal Global Biodiversity Framework (GBF) and accompanying measures to address biodiversity loss and promote the restoration of natural ecosystems1. The GBF includes four overarching global goals for 2050 and 23 targets to be achieved by 2030. For instance, Goal A of the framework summarises objectives related to biodiversity conservation, while Goal B aims to value, maintain, and enhance nature's contributions to people (NCP). The goals are supported by a larger number of targets, including Target 1, which aims to ensure integrated biodiversity-inclusive spatial planning; Target 2, to ensure explicit targets for restoration actions; Target 3, for protecting important sites for biodiversity and its associated contributions to people; Target 8, to minimise the impact of climate change on biodiversity and increase its resilience; Target 10, for sustainable management of production systems, and Target 11 to ensure the maintenance and improvement of NCP1. These goals and targets add precision to the existing 2030 Agenda for Sustainable Development, in particular its goals for agricultural sustainability (Goal 2), climate change response (Goal 13), and terrestrial biodiversity (Goal 15).
Here, we comprehensively assess three key management actions (preservation of remaining natural areas; restoration of natural ecosystems; and maintenance and expansion of agricultural lands) and their resulting impacts, intending to advance progress towards Goals A and B, as well as the effective execution of Targets 1, 2, and 3 supported by strategic actions related to Targets 8, 10, and 11 of the GBF. In this context, we address several key challenges: i) integrating multiple actions simultaneously; ii) assessing biodiversity and people's projected needs to evaluate outcomes of the actions; iii) considering multiple threats to biodiversity and NCP. Our analysis explores the feasibility of halting or even reversing the projected negative impacts on natural systems through spatial optimisation without undermining agricultural production. Reaching these significant socio-environmental gains is facilitated by sustainably increasing yields and international trade. As such, this research integrates multiple factors consistent with the transformative change mandated by the GBF.
Prioritising actions now while considering future needs
Using a spatial optimisation approach rooted in linear programming2, we identify key areas for strategic conservation, restoration, and agricultural expansion actions. Our approach (Fig. 1) provides crucial insights into the most effective strategies for balancing conservation goals with socioeconomic realities. To prioritise conservation and minimise land-use change, we focus on protecting natural areas to the greatest extent possible. Agricultural expansion is only considered when projected food production within a country cannot be met by current agricultural land, even after efforts to increase yields in a context of regionally-constricted trade regimes. This applies regardless of whether the land is currently used for agriculture or pasture, as we initially allow for conversion between agricultural land uses. Hence, our spatial analysis provides insights into the synergies and trade-offs between different interests and actions for national governments and other decision-makers who aim to implement the GBF.
We evaluate the likely impacts of modelled actions on a diverse range of critical metrics (Methods), including biodiversity (species extinction risk3,4, ecosystem structural integrity5, and ecoregion vulnerability), NCP (pollination6, coastal protection6, nutrient retention6, and climate change mitigation3), as well as socioeconomic considerations (opportunity costs3 and implementation cost of restoration3). Our metrics represent future (2050) biodiversity and human needs and were projected following a scenario of high population growth, urban and agricultural expansion, as well as regionalisation of trade blocs with only a small portion of agricultural goods traded globally - the Shared Socioeconomic Pathway number 3 (SSP3)7. While our aspirations for a sustainable future are ambitious, we have selected the SSP3 scenario as our baseline due to its projected high demand and impact on natural resources. This choice allows us to assess the differences achieved through spatial optimisation compared to this challenging scenario, providing valuable insights into potential strategies for mitigating the impacts of resource utilisation. To assess biodiversity and people’s needs by 2050, we consider the impacts of projected land use, climate, and human population density changes for the SSP3 scenario (Methods). Our NCP metrics also incorporate SSP3's precipitation, fertiliser load, and sea-level rise6,8. Opportunity costs were calculated with reference to 2050 agricultural production9, and all metrics include projected urban expansion (Methods and SM1 for limitations).
To achieve an increase in natural ecosystems by 2050 (one of the elements of GBF Goal A), we establish targets that aim for a net global increase of natural lands and limit optimisations (Methods). Specifically, we propose three global optimisation targets corresponding to 15% (3.7 Mkm2), 20% (5.0 Mkm2), and 30% (7.5 Mkm2) of the current agricultural land area (Methods; Fig. 1). If realised, these targets would result in an estimated increase of approximately 3.5%, 5%, and 7% in the world's natural areas, respectively.
We assume that countries must align with the projected production of food and agricultural commodities as outlined in the SSP3 scenario. Accordingly, we imposed limitations on each country, specifying the minimum and maximum extents of agricultural expansion or restoration, respectively (Methods). This procedure safeguards countries' agricultural output in 2050. We have also created scenarios without these country restrictions, guaranteeing agricultural goods to every country through a more globalised trade as opposed to the regionalised trade blocs of the SSP3 scenario.
We also considered three yield intensification efforts across all landscapes through narrowing their yield gaps by 30%, 50%, and 70%. (Methods; Fig. 1). Consequently, countries facing the challenge of meeting projected agricultural demand within their borders by 2050 could reduce their need for agricultural expansion and conserve more natural land within certain simulations10. Moreover, when a certain yield threshold is reached, some countries can restore natural land rather than expand agricultural areas. However, it is important to carefully analyse the potential impacts of yield intensification, considering the approach chosen and starting point11. Negative externalities on biodiversity and NCP may arise due to increased use of fertilisers, pesticides, and other chemical inputs, depending on the context6. Nevertheless, the net global impact could be positive, as it may reduce overall agricultural area10. Dynamically assessing these externalities is complex and beyond the scope of this study. However, we recognize that agricultural intensification requires strategic investments and innovative approaches to minimise trade-offs and ensure food security10,12.
Last, we also subject the optimisation to constraints at the local level3 (Fig. 1). In landscapes where natural areas can be restored, we limited the restoration area to the available land after closing yield gaps (Methods). This approach simulates a situation where restoration decisions consider the need to maintain local agricultural production3. It also ensures that every landscape undergoing restoration retains some agricultural land for fertilisation (which affects water quality regulation) and/or pollination services (see Table S1 for results without this assumption).
Our analyses pioneer the holistic optimisation of the three key management actions and their associated outcomes for biodiversity and NCP. In doing so, we leverage on previous research that focused on fewer factors, such as previous assessments of conservation13–15 and restoration3 priorities, evaluations of future NCP outcomes6, and analyses of land-use change drivers and their impacts on biodiversity12. This study breaks new ground by integrating and optimising multiple factors simultaneously.
Optimised scenario outcomes
We devised multiple optimised scenarios in global restoration and yield gap closure strategies, considering local constraints and optimisation metrics (see Fig. 1). To streamline the analysis, here we focus on scenarios with local constraints and with three combinations of effort levels for achieving a net increase in natural lands and closing yield gaps: i) high effort − 30% net increase and 70% yield gap closure; ii) intermediate effort − 20% and 50%; and iii) low effort − 15% and 30%.
Results of individual metrics exhibit significant variability across scenarios (Fig. 2) as optimisations focused on biodiversity metrics produced lower gains for NCP and vice versa, resulting in shifts in optimised areas (Fig. 3a-b vs c-d). This disparity underscores the importance of joint optimisations that seek synergies between multiple objectives and integrate Goals A and B of the GBF (Fig. 3e-f). Furthermore, the incorporation of cost minimisation as an objective reduces the outcomes for several benefit metrics (Fig. 2b, d, f, and g; Figure S1), particularly for the ones with locally-based attributes like pollination, water quality regulation, and coastal protection, as these benefits cannot be outsourced to regions with lower costs. However, there is some synergy between cost minimisation and the ecosystem structural integrity metric (Fig. 2), as the restoration of cultivated grasslands generally incurs lower opportunity costs and a greater reduction in the human footprint compared to croplands5,16 (SM1). While recognising a time-lagged response between restoration actions and the recovery of species and ecological processes, we estimate the potential long-term benefits for biodiversity and NCP3.
We present our results as potential gains relative to the impacts of the SSP3 scenario projected between 2015 and 2050. However, for some metrics (i.e., ecoregion vulnerability and climate change mitigation), the impacts are completely mitigated and/or compensated with the optimised allocation of conservation, restoration, and agriculture. Therefore, the reported results correspond to the additional gain relative to the current (2015) situation. In this approach, the high-effort strategy represents a decisive redirection towards a sustainable future, as reflected in higher outcomes (Fig. 4). In some cases, this high-effort strategy has almost doubled the gains compared to those obtained with low effort (Fig. 4c, d, g). Overall, spatial optimisation considering various metrics, threats, and actions has the potential to produce remarkable results when integrated with goals for restoration and sustainable yield intensification that do not necessarily aim for the highest levels of achievement (e.g., 20% of global restoration target and 50% of yield gap closure; see Fig. 4a, b, e, f). This underscores the need to identify critical global areas for biodiversity and NCP preservation that can be prioritised for the three key management actions3,17 and the significance of spatial planning as proposed in Target 1 of the GBF.
Implications for each benefit metric
Our study reveals alarming projections based on a database of over 14,000 mammals, amphibians, birds, and reptiles3,4 combined with nine general circulation models (GCMs) for 2050. Under the SSP3 RCP7.0 scenario, species extinction risk could increase between 35–150%, depending on the specific GCM used and its associated climate change projections. The median estimate of a 60% increase in extinction risk underscores the urgent need for action to mitigate the consequences of climate change on global biodiversity. Many simulations using the different GCMs found that climate change impacts in the 2015–2050 period would be greater than the impacts of land-use change under the SSP3 scenario, with most estimates showing it could be three times greater (Table S2). This pattern is consistent with findings in the literature across taxa and biodiversity indicators such as species' geographic range, richness, and population size18,19. However, our findings reveal significant variation in the optimisation results across the nine GCMs (Table S3). This variation underscores the inherent uncertainties in predicting global climate change. To mitigate this, we have reported the median values from the nine GCMs for the species' extinction risk metric in each optimised scenario.
When considering the optimised distribution of three management actions, implementing a high-effort strategy could decrease the projected impacts on extinction risk caused by the SSP3 2015–2050 scenario by 67–88%, depending on the chosen metrics driving optimisations (Fig. 4a). These results emphasise the urgent need for a coordinated approach between biodiversity and climate conventions. Additional efforts to mitigate the projected impacts of the SSP3-RCP7.0 scenario by 2050 have the potential to reduce species loss substantially. Moreover, shifts in species range to track suitable climates20 should be considered in the landscape planning process and facilitated by improving connectivity and habitat quality. Suppose species can reach all habitats with suitable climate conditions within their current or neighbouring ecoregions21. In such a case, SSP3 impacts on their suitable habitat can be mitigated and/or compensated, and species' current extinction risk could be reduced by 9–19% in a high-effort strategy (Figure S2). This result again underlines the importance of Target 1 of the GBF. It shows the crucial role that spatial optimisation can play in safeguarding biodiversity, especially in a scenario with substantial climate change impacts.
We reached similar conclusions for ecosystems as the structural integrity of more than 400 Mha was reduced to a low or very low level (< 0.33 on a scale of 0 to 1)5 in the SSP3 scenario compared to the current situation. The total effect of this pessimistic scenario is equivalent to a loss of almost 10% of ecosystem structural integrity globally. Population growth projected under the SSP3 scenario had a greater impact than land-use change per se (Table S4), as high population density significantly increases the human footprint on natural ecosystems16. Under a high-effort strategy, the impact of the SSP3 scenario on ecosystems' structural integrity is reduced by 21–30% with spatial optimisation (Fig. 4b) - the net area of natural ecosystems whose structural integrity is reduced to low levels would be only 284–322 Mha. Additionally, the vulnerability of ecoregions would increase by nearly 8% due to the projected land-use change of the SSP3 scenario in 2050. In a high-effort strategy, spatial optimisation of the three management actions could mitigate and/or compensate for these projected trends and go even further, reducing the current vulnerability of ecoregions by 40–48% (Fig. 4c). Therefore, decisions regarding conservation, restoration, and agriculture should incorporate indicators of ecosystem structural integrity and vulnerability5 rather than just area per se.
The impacts of the SSP3 2050 scenario due to pollutants not retained by vegetation reaching waterways, coastal hazards, and crop losses from insufficient wild pollination have been demonstrated by a previous study6. It shows a disproportionate share of the impacts of developing countries, especially in Africa and South Asia. In the high-effort strategy and depending on the metrics used in the planning process, the 2015–2050 impact of SSP3 on coastal protection could be reduced by 16–58% (Fig. 4e). In addition, the spatial optimisation of conservation, restoration, and agriculture could benefit rural people around the world (Fig. 3) by generating net improvements in water quality regulation of 16–19%, even when considering increased stressors in SSP3 RCP 7.0 of heavier precipitation and fertiliser loads (Fig. 4d). The outcomes of this high-effort strategy also show an additional 1.4–1.9 billion people whose nutritional needs could be fully met by pollination-dependent crop production6 (Fig. 4f) - though, in reality, this production would benefit a far larger number by partially contributing to their nutritional needs.
Our estimates indicate that the SSP3 scenario projects a release of almost 90 GtCO2 into the atmosphere due to land use change. However, looking at the long-term potential of carbon stocks after restoration, this pattern shifts towards net sequestration (up to 303 GtCO2) in our optimised scenarios. When the three key management actions are allocated to only minimise costs, this value drops to 226 GtCO2 and 109 GtCO2 in the low-effort case (Fig. 4g). However, caution should be exercised in interpreting these findings, as the timeframes for carbon dioxide sequestration and the realisation of its benefits vary significantly from the timeframe of emissions. Detailed results for sequestration and emissions for each scenario are presented in Table S5, warranting careful analysis.
The importance of the agricultural sector and trade
Under scenarios which allow trade globalisation, all countries are able to meet agricultural demands through combinations of reduced yield gaps and increased imports (Fig. 3). However, if trade is restricted to regional blocs, despite the simulated high-yield growth, several countries, particularly in Africa, would still need to convert natural lands to meet their agricultural demands per the SSP3 scenario (Fig. 3). Our modelled scenarios optimise this necessary land conversion for agriculture by minimising adverse impacts on biodiversity and NCP metrics. Additionally, our analysis prohibits land-swapping practices (convert in one place, restore in another) within countries, prioritising the conservation of natural areas before any allocation of land-use changes. This approach is grounded on the premise that safeguarding remaining natural ecosystems is the foremost action in preserving biodiversity and NCP provision17,22,23.
Global environmental agendas should go beyond agricultural intensification to avert the anticipated conversions of natural ecosystems for agriculture, secure food production and create opportunities for restoration. International trade and other behavioural changes are vital in bending the curve of biodiversity loss and maintaining or improving NCP12. When not constraining our optimisations at the country level (Fig. 3), outcomes for all biodiversity and NCP metrics could be significantly enhanced under all three levels of effort while still leaving sufficient land for agricultural production (Table S6), suggesting a crucial role for international cooperation in achieving nature-positive transformations. This also highlights the need to engage with the agricultural sector, to incorporate environmental sustainability standards, such as zero-conversion criteria, into trade agreements and supply chains, to strengthen multilateral development goals that link nature and trade, and to seek behavioural change on the demand side to support biodiversity conservation and the provision of NCP.
A significant challenge lies in meeting the increasing demand for agricultural production while supporting biodiversity conservation and recognizing the role of nature in sustaining human livelihoods, especially under changing pressures. We show that a comprehensive spatial planning agenda considering simultaneous actions and multiple indicators is essential to address this challenge, identify solutions and achieve positive outcomes aligned with Goals A and B of the GBF.
Our study demonstrates that it is possible to substantially reduce and even halt biodiversity and natural capital loss within the evaluated timeframe. However, the relative importance of significant threats is expected to evolve. Addressing these different threats, such as mitigating climate change, reducing human pressures on natural habitats, increasing trade, and promoting behavioural changes, requires simultaneously concerted efforts. This calls for engagement and commitment from multiple actors across diverse sectors.
Furthermore, effective conservation and restoration actions must be rooted in participatory and inclusive processes at regional to local scales, with global spatial optimisation providing relevant inputs for decision-making (refer to SM1 for detailed discussions). Combining these efforts with socio-economic metrics assessing human vulnerability is imperative, as identifying who will benefit and bear the costs of different management decisions and addressing equity concerns to ensure a sustainable and equitable future development.
- Convention on Biological Diversity. Kunming-Montreal Global Biodiversity Framework. https://www.cbd.int/doc/c/e6d3/cd1d/daf663719a03902a9b116c34/cop-15-l-25-en.pdf (2022). (accessed 25 January 2023)
- Beyer, H. L., Dujardin, Y., Watts, M. E. & Possingham, H. P. Solving conservation planning problems with integer linear programming. Ecol. Model. 328, 14–22 (2016).
- Strassburg, B. B. N. et al. Global priority areas for ecosystem restoration. Nature 586, 724–729 (2020).
- IUCN. The IUCN Red List of Threatened Species. Version 2022-2. https://www.iucnredlist.org/. (accessed 10 December 2022)
- Beyer, H. L., Venter, O., Grantham, H. S. & Watson, J. E. M. Substantial losses in ecoregion intactness highlight urgency of globally coordinated action. Conserv. Lett. 13, 1–9 (2020).
- Chaplin-Kramer, R. et al. Global modeling of nature’s contributions to people. Science 366, 255–258 (2019).
- Popp, A. et al. Land-use futures in the shared socio-economic pathways. Glob. Environ. Change 42, 331–345 (2017).
- Jones, B. & O’Neill, B. C. Spatially explicit global population scenarios consistent with the Shared Socioeconomic Pathways. Environ. Res. Lett. 11, (2016).
- Robinson, S. et al. International Model for Policy Analysis of Agricultural Commodities and Trade (IMPACT) version 3.1. International Food Policy Research Institute (IFPRI) 128 https://www.ifpri.org/publication/international-model-policy-analysis-agricultural-commodities-and-trade-impact-model-0 (2015).
- Folberth, C. et al. The global cropland-sparing potential of high-yield farming. Nat. Sustain. 3, 281–289 (2020).
- Pretty, J. et al. Global assessment of agricultural system redesign for sustainable intensification. Nat. Sustain. 1, 441–446 (2018).
- Leclère, D. et al. Bending the curve of terrestrial biodiversity needs an integrated strategy. Nature 585, 551–556 (2020).
- Montesino Pouzols, F. et al. Global protected area expansion is compromised by projected land-use and parochialism. Nature 516, 383–386 (2014).
- Brooks, T. M. et al. Global Biodiversity Conservation Priorities. Science 313, 58–61 (2006).
- Naidoo, R. et al. Global mapping of ecosystem services and conservation priorities. Proc. Natl. Acad. Sci. U. S. A. 105, 9495–500 (2008).
- Sanderson, E. W. et al. The Human Footprint and the Last of the Wild. BioScience 52, 891 (2002).
- Chaplin-Kramer, R. et al. Mapping the planet’s critical natural assets. Nat. Ecol. Evol. 7, 51–61 (2023).
- Newbold, T. Future effects of climate and land-use change on terrestrial vertebrate community diversity under different scenarios. Proc. R. Soc. B Biol. Sci. 285, (2018).
- Warren, R., Price, J., Graham, E., Forstenhaeusler, N. & VanDerWal, J. The projected effect on insects, vertebrates, and plants of limiting global warming to 1.5°C rather than 2°C. Science 360, 791–795 (2018).
- Hannah, L. et al. 30% land conservation and climate action reduces tropical extinction risk by more than 50%. Ecography 43, 943–953 (2020).
- Feng, X. et al. How deregulation, drought and increasing fire impact Amazonian biodiversity. Nature 597, 516–521 (2021).
- Di Marco, M., Ferrier, S., Harwood, T. D., Hoskins, A. J. & Watson, J. E. M. Wilderness areas halve the extinction risk of terrestrial biodiversity. Nature 573, 582–585 (2019).
- Maron, M., Simmonds, J. S. & Watson, J. E. M. Bold nature retention targets are essential for the global environment agenda. Nat. Ecol. Evol. 2, 1194–1195 (2018).
Acknowledgements:
We are very grateful for the support provided by the whole IIS team in the preparation of this manuscript, in particular Breno Valle, Bruna Pavani and Luis Martinez. We especially thank Piero Visconti, Valerie Kapos and Lera Miles for the precious discussions and Hawthorne Beyer for all the help in generating the ecosystem structural integrity metric. We acknowledge inputs from the Secretariat of the Convention of Biological Diversity and funding from the UK Research and Innovation’s Global Challenges Research Fund under the Trade, Development, and the Environment Hub project (ES/S008160/1) - https://tradehub.earth. The views expressed in this publication do not necessarily reflect those of IUCN.
Author contributions:
B.B.N.S., G.T.D, J.M.K., R.C., E.L., D.R., L.G.O., and A.I. conceived the study. G.T.D and J.M.K. coordinated the development of the spatial planning approach, led the analyses, and wrote the first version of the paper. G.T.D, J.M.K, R.C., E.L., S.M., D.R., L.G.O., and A.I developed part of the metrics and led the multicriteria modelling. R.C-K and R.S developed the input datasets for the Nature Contributions to People. T.M.B, B.J. E., X.F., L.H., C.M and P.R provided the necessary input data for calculating species extinction risk. T.M.B., R.C-K, S.L.H, D.L., M.O, P.R, R.L., J.W. and B.B.N.S. contributed to discussing policy applications. F.G. led the construction of the figures. G.T.D, J.M.K, R.C., E.L., S.M., D.R., L.G.O., A.I, R.L., T.M.B., R.C-K, S.L.H, D.L, M.O, P.R, J.W. and B.B.N.S., analysed the results. All authors provided input into subsequent versions of the manuscript.
This initiative was developed in continuous interaction with the Co-Chairs of the Open-Ended Working Group (OEWG) and the Secretariat of the Convention of Biological Diversity (CBD) in light of the Kunming-Montreal Global Biodiversity Framework (GBF)1. By providing integrated and internally consistent modelling of land-use scenarios, the main objective was to evaluate the optimal areas to leverage conservation, restoration and agriculture considering the maximisation of desired benefits in the year 2050. In other words, we aimed to model human efforts towards a sustainable future, already accounting for projected needs in 2050. These needs included not only the safeguarding of biodiversity and nature's contributions to people (NCP), but also the projected agricultural and urban expansion (here, based on the assumptions of the pessimistic SSP3 scenario). To achieve this goal, we first had to identify the metrics representing 2050 needs and the future land use context. Each metric calculated in this work had its own set of spatially explicit databases, and some of these data were common to two or more metrics. Next, we simulated global efforts to increase the world's natural areas and to sustainably increase agricultural yields, assessing optimal areas for the three actions at once in each effort context. Hence, we divided this methods section into three main components:
1) Description of common databases;
2) Description of each biodiversity, NCP, and costs metrics calculation and specific data;
3) Description of the optimisation, its constraints, effort context, and scenarios.
Common database
Current land use/cover maps
We assembled a map representing the current distribution of land use/cover classes as a starting point for optimisation and to define the areas feasible for restoration, conservation, and agricultural expansion. We used the European Space Agency Climate Change Initiative (ESA-CCI)2 data from 2015 as a baseline, with the reclassification and adjustments according to the first global priority areas restoration analysis3–5. The resulting maps corresponded to the proportions of a grid area of nine land use/cover classes: forest, wetlands, deserts, shrublands, natural grasslands, cultivated grasslands, croplands, urban areas, and others (water bodies and ice lands). All the analyses and maps were constructed using a roughly 5 km spatial resolution, and each grid cell (pixel) was defined as a planning unit. We considered as feasible for restoration actions, any proportion of a planning unit area with cultivated grasslands and/or croplands classes3 (see SM1 for limitations). Conservation and agricultural expansion actions could only occur in the proportion of the planning unit area corresponding to natural classes (forest, wetlands, deserts, shrublands, natural grasslands) in the current land use/cover maps (refer to SM1 for discussion of the role of each natural class). No restoration, conservation, or agricultural expansion actions were permitted in current urban areas or those expected to have urban expansion according to the baseline land use/cover maps described below.
Shared Socioeconomic Pathway 3 land use/cover maps.
The Shared Socioeconomic Pathway 3 (SSP3) was translated into land-use scenarios for each year between 2015 and 21006. Since this set of maps had different land-use classes and were constructed in different ways than ESA-CCI maps, we developed an approach to formulate our 2050 land-use maps that accounted for the SSP3 agricultural and total urban area. First, for each planning unit, we calculated the difference between croplands, cultivated grasslands, and urban areas separately between the years 2050 and 2015 using the state maps available on the Land Use Harmonization website7. Therefore, we ended up with projected maps of cropland, cultivated grasslands, and urban areas expansion. In each planning unit, we added the values of these projected expansions to their respective current land-use class, adjusting proportionally those in which the sum of these classes exceeded 100% of its area. The other land-use/cover classes were constructed according to the remaining percentage of the planning unit cell. At this stage, we used the current land-cover maps to distribute the missing classes. In cases where there were no natural classes in the planning unit in the current land-use map, we adopted the proportion of the other classes presented in the original vegetation map for that respective planning unit.
Original land cover maps.
Restoration optimisation should consider several ecosystem types simultaneously, restoring areas to their original type to ensure the maintenance of local and regional biodiversity and NCP provision3. Accordingly, we followed the methods of previous work3 and assembled a set of land cover maps representing the proportions of the planning unit area covered by each original vegetation type, using the same natural classes of our current land use/cover maps. We used ESA CCI land use/cover maps of 1992, ecoregions’ information, and boundaries8 to delineate the extent of original vegetation types and their corresponding proportions.
Yield gaps
To consider the total area devoted to agricultural production in a future scenario and to generate new scenarios where different levels of agricultural yield improvements are projected, raised the need for a map that represents the yield distribution and spatial variation. Hence, we followed a previous work3, and used the yield ratio of croplands and cultivated grasslands from the GAEZ database9 and methods of ref10. The difference between a situation where the yield is at its maximum potential (yield ratio equals 1) and the current yield ratio results in the yield gap for each planning unit.
Optimisation metrics
For the construction of spatially explicit metrics, it was necessary to search and evaluate consistent global databases, indicators, or data generated by models disseminated and accepted by the academic community. We ended up with three metrics to assess biodiversity needs (species extinction risk, ecosystem structural integrity, and ecoregion vulnerability), four metrics representing the NCP (pollination, coastal protection, water quality regulation, and climate change mitigation in terms of carbon sequestration and emissions), and two metrics considering the associated costs of restoration and conservation actions (opportunity and implementation costs) for 2050 scenario. For more information on the calculation and uncertainties, refer to SM1.
Species extinction risk
The extinction risk metric (ER) is inversely related to the ratio between the amount of remaining natural area that serves as habitat for a given species (Ar) and the total area of habitat originally present within its range (Ao) before any human action (Equation 1)3,11. Therefore, the higher this proportion, the lower the risk of extinction. The z factor is a constant equal to 0.25. Through the derivative of this extinction risk function, we calculated marginal values representing the increase/decrease of this metric after restoration/agricultural expansion actions. The sum of these marginal values for all species in each planning unit formed the spatially explicit layer included in the optimisation.
This work evaluated a broad set of terrestrial vertebrate species, including mammals, birds, amphibians, and reptiles. To delimit the habitat range of each species and consider the climatic effects on its distribution, we used the intersection between IUCN Red List database12 and ten climate envelopes developed through species distribution models described in ref13. The climate envelopes corresponded to maps containing binary information (suitable/not suitable) for each species, using a threshold of 5% training omission rate13. We had 9 maps generated for each species considering different general circulation models (BCC-CSM2-MR, CNRM-CM6-1, CNRM-ESM2-1, CanESM5, GFDL-ESM4, IPSL-CM6A-LR, MIROC-ES2L, MIROC6 and MRI-ESM2-0), all of them with the mean climate conditions of 2040-2060 period in the SSP3 RCP 7.0. Additionally, we used one map representing the mean climate conditions of the 1970-2000 period as a baseline climate.
With the intersection, we limited species migration due to future climate within their own IUCN Red List range. The use of different datasets restricted the final number of species we could use in our analysis. We had 14,000 species ranges with information in all datasets. To represent the conditions where landscapes are more susceptible to species movement so that they can track suitable climate, we also performed all analyses using climate envelopes as species ranges without intersecting with IUCN Red List data. We use the resultant range maps representing future ranges to assess the remaining habitat area for each species (Ar). On the other hand, we used one map representing a baseline climate to calculate the habitat area originally present for each species (Ao). In both cases (Ar and Ao), we performed an area of habitat analysis (AOH)14 considering the land cover classes that served as habitat for each species according to the IUCN Red List database.
Ecosystem structural integrity
As a proxy for ecosystem structural integrity, we chose the intactness index proposed by ref15, which is sensitive to changes in habitat area, quality, and fragmentation. Following this reference, we adopted a generic measure of habitat quality to assess global patterns and changes in intactness: the Human Footprint Index (HFI)16. One of the parameters that compose the HFI is a population density layer. So, we modified the index values to account for the population growth projected for the SSP3 scenario and its impact on ecosystem intactness. We used the data available in ref17 and followed the methods described in ref16, in which the human influence factor of the total HFI increased linearly from 0 to 10 for densities between 0 and 10 persons per km2, and the score above 10 persons per km2 was held constant at 10. For example, a planning unit with a population density of 3 people per km2 in the current scenario and more than 10 people per km2 in the SSP3 scenario would have the correspondent HFI increased by a score of seven.
To calculate intactness, we used an exponential transformation to scale this new HFI to create a habitat quality measure in which quality is 1 when HFI is 0, and quality is 20% when HFI is 4. We also considered the habitat amount and quality of each planning unit's neighbours (within a 15km radius)15. Finally, to include intactness as an explicit layer in the optimisations, we quantified the expected change in this metric after restoration/agricultural expansion actions using the derivative of the intactness function.
Ecoregion vulnerability
Reducing the area of an ecoregion decreases the quantity of resources available, its capacity to sustain species populations, and the ecological processes that maintain biodiversity and NCP provision18,19. We have developed a metric that reflects the idea that the vulnerability of an ecoregion is a function of the area of natural habitat that remains within its boundaries and the potential area of the ecoregion if all available area is restored to natural habitat. This function follows the same relationship as the metric for species extinction risk3, with its derivative representing the marginal gain/loss in vulnerability after agricultural expansion/restoration actions. We used the ‘Terrestrial Ecoregions of the World’ maps6 as a proxy for ecoregion boundaries.
Climate change mitigation
We estimated the potential and marginal values of carbon stock in above- and below-ground biomass20 and soil21 terrestrial pools, following the steps of a previous global analysis3. This approach uses the original vegetation maps, ecoregion boundaries, and four different maps of carbon stocks to estimate the marginal values of carbon sequestered per planning unit depending on the proportions of natural land cover classes being restored. We used current land cover (instead of original vegetation) and the same carbon stock maps to estimate the marginal value of carbon emissions that would result from converting natural ecosystems for agriculture.
Nature's contribution to people
Besides climate change mitigation, we included three additional, locally-occurring NCP: water quality regulation (through the evaluation of avoided nutrient export), pollination (through the potential contribution of nearby habitat to provide wild pollinators for crop production), and coastal protection (through the evaluation of coastal risk reduced by coastal habitat). We first assessed and mapped the biophysical values of each contribution (also called "potential ecosystem service"), following the methods of ref22. In terms of avoided nutrient export, the biophysical model for the water quality regulation considers as inputs the projected 2050 RCP 7.0 precipitation, run-off, and fertiliser application rates for agriculture. For pollination, we used the marginal values of an equivalent-people-fed metric based on the nutrition provided by wild pollinators on each planning unit with agricultural land. It is calculated according to pollination habitat sufficiency and pollination-dependent nutrient production. The biophysical model for coastal risk reduction is the difference in coastal risk with and without habitat, based on a coastal risk index calculated from the protective distance and risk attenuation from marine and terrestrial habitats, coastal topographic relief (digital elevation model), wind, wave, geomorphology, and a projected 2050 RCP 7.0 sea level rise.
To generate the input layers for the optimisations, marginal values for each potential service were calculated, indicating how much service would be gained/lost after the restoration/expansion of agriculture in the respective planning unit. We used potential vegetation land-cover maps and an "all natural ecosystems converted to agricultural use" map as inputs for each NCP model, corresponding to the restoration and agricultural expansion scenarios, respectively. The difference between the values of the services using these land-cover maps as inputs and the "current maps" correspond to marginal values. For the pollination model, when using this potential vegetation map, we assumed the value of pollination-dependent agriculture according to the current crop mix, even though a fully restored planning unit would contain no agriculture.
Then we delineated the beneficiary part of the contribution. The “realised contribution” is the biophysical values (marginal values) multiplied by the number of people benefitting from the NCP metrics described in ref23. In brief, the coastal risk reduction was multiplied by the number of people within the protective distance of that habitat, water quality regulation was multiplied by the number of people downstream, and realised pollination was calculated as the end benefit to an indeterminate beneficiary for pollination - equivalent number of people fed by wild pollination-derived crop production (number of people whose nutritional needs could be fully met by pollination-dependent crop production)22. The realised contribution maps served as inputs for the optimisation.
Opportunity costs
We estimated foregone benefits of agricultural commodities, for croplands and cultivated grasslands separately, as a proxy for conservation and restoration opportunity costs (see SM1 for limitations). For croplands, we obtained equilibrium-price estimates for several commodities from the extended data of the IMPACT model24. The methods used in this analysis followed previous work3, except that the SSP3 scenario was here chosen for the selection of equilibrium prices. The result of the selected methodology was the net present value (NPV) of one tonne of each commodity, which we used to convert the current productivity maps from GAEZ9 of each crop in each planning unit from a produced quantity per area to production value per area. Last, we summed all production values per area per planning unit to obtain an opportunity cost layer for cropland.
For cultivated grassland, we used the stocking rate data from the ‘Gridded Livestock of the World v2.0’25, and values of animal yield per country from the IMPACT model to convert the stocking rate from head per hectare to tonnes of produced beef per hectare per year and then to a NPV of production per area. We followed the same procedures and assumptions as ref3, resulting in an opportunity cost layer for cultivated grasslands.
To obtain the combined opportunity cost layer of restoration, we used the average of the values for croplands and cultivated grassland, weighted by their proportion in the planning unit. The proportion of each land use was obtained from current land use maps for units with land available for restoration. Opportunity costs of conservation were estimated using the same economic benefits of agricultural commodities and accounting for projected agricultural expansion. For planning units that had agricultural land on current land use maps, we assumed that the new agricultural area would follow the current ratio of cropland and cultivated grassland. For planning units that did not have agricultural land, opportunity costs were calculated dynamically in our model runs by using the ratio of the cropland and/or cultivated grassland that each country needed to expand in each scenario.
Implementation costs of restoration
Restoration optimisation needs to account for the costs of implementing this action, which could vary depending on socioeconomic and environmental conditions. We used the same approach design by ref3 to account for countries' variation of implementation costs of restoration actions. These cost values are derived from a basis of US$2,148 per hectare for Brazil26, and adjusted to all countries, considering their relative costs of agricultural labour27 and input (fertiliser28).
Problem Construction
Definition of targets and optimisation constraints
The post-2020 GBF urges for a worldwide effort to increase natural areas, with the best outcomes for biodiversity and NCP. To aid the discussions of targets and goals of this framework, we had set area-based constraints in our optimisations accounting for a global net increase of natural ecosystems (Figure 1). In other words, the optimisation of lands for restoration and agriculture would cease after a predetermined net increase in natural areas. With different area-based constraints, we could assess the resulting trade-offs for biodiversity and NCP. Three global net targets were used to build simulated scenarios: the amount of the area corresponding to 15%, 20%, and 30% of current agricultural lands (respectively ~3.7 Mkm2, ~5.0 Mkm2 and ~7.5 Mkm2). This would correspond to ~3.5%, ~5%, and ~7% increase in the world's natural areas, respectively.
We aimed to optimise restoration, conservation, and agriculture assuming that countries fulfil projected demands for food and commodity production through yield gap closure. Therefore, we introduced additional constraints on the country level corresponding to the minimum and maximum area that must be, respectively, converted to agriculture or restored to natural ecosystems. We calculated these constraints for each agricultural land use class: cropland and cultivated grassland (Figure 1). We took the land use data of the SSP3 scenario as a baseline to set the country-level constraints according to a three-step procedure (Figure 1):
1 - For each country, we calculated the difference between the agricultural area in the SSP3 scenario and the current scenario. This was done separately for croplands and cultivated grasslands to consider possible changes in dietary habits. Therefore, a country could fall into one of the following three options:
a) If a country has more area in the SSP3 scenario than at present, this country will need to expand the agricultural land to at least the area corresponding to this difference. For this group of countries, we optimised agricultural expansion for the areas with the smallest impact on our benefit metrics and with the highest agricultural returns. We restricted the areas that the algorithm could select for agricultural expansion to those planning units that had the corresponding class of agricultural land use in the SSP3 and/or current land use/cover maps.
b) Countries with a higher current agricultural area than in the SSP3 scenario were allowed to restore up to the agricultural area corresponding to this difference. We cost-effectively optimised restoration actions for this group of countries accounting for impacts on biodiversity and NCP.
c) There was a particular case where a country could restore one class of agricultural land use but had to expand the other. In these cases, we assumed that the additional area needed for one agricultural use was allocated in the spared area of the other. This assumption resulted in one constraint being zero and the other being the difference between the two initial constraints. The allocation occurred in places with the highest values of suitability based on GAEZ9 suitability maps considering climate change impacts estimated by Hadley CM3 A2. This model was the closest to the impacts estimated in the SSP3 scenario. The suitability maps were rescaled to vary between zero and 1.
2 - We simulated a global effort to close different fractions of the yield gap, with countries increasing their agricultural yields at a higher rate than in the SSP3 scenario (see SM1 for limitations). We assumed a yield gap closure within these simulations by 30%, 50% and 70%. Consequently, countries that needed agricultural expansion could reduce this need and conserve more natural ecosystems. After a certain fraction of the yield gap closing, these countries could restore natural ecosystems. Countries able to restore areas without the need for agricultural intensification would be able to restore even more after the simulation of the yield gap closing.
3 - We also created scenarios subjecting the optimisation analysis to constraints in the planning unit level3. For planning units within countries with restoration targets in certain simulated scenarios, we have limited the maximum area that could be restored to the area spared after closing the yield gaps.
We also optimised scenarios without considering country borders and their specific needs of agricultural expansion, representing a situation with higher trade between countries. Thus, we had only the global and local constraints for restoration optimisations. In addition, we calculated the impact of the SSP3 scenario in each of the optimisation metrics using the projected land use change for comparison purposes.
Scenarios and optimisation
For each of the three global targets and three yield gap closure simulations, we produced seven scenario variants along the following metric combinations: all metrics included (benefits and cost scenario), biodiversity and NCP metrics (benefits scenario), biodiversity and cost metrics, with the NCP and cost metrics, the three biodiversity metrics only, the four NCP metrics only and the two cost metrics. Our algorithm optimises spatial solutions considering multiple criteria at once. It is based on linear programming29, aiming to select areas for restoration and agriculture (see SM1 for more details and limitations). For each planning unit, the algorithm first calculates the cost-effectiveness of restoration or agricultural expansion actions, using a subset or all the metrics depending on the scenario's purposes. The area required for both actions is then calculated based on the respective objective function and its scenario constraints.
References
1. Convention on Biological Diversity. Kunming-Montreal Global biodiversity framework. https://www.cbd.int/doc/c/e6d3/cd1d/daf663719a03902a9b116c34/cop-15-l-25-en.pdf (2022). (accessed 25 January 2023)
2. European Space Agency. Climate Change Initiative (ESA CCI). https://www.esa-landcover-cci.org/?q=node/158. (accessed 13 February 2019)
3. Strassburg, B. B. N. et al. Global priority areas for ecosystem restoration. Nature 586, 724–729 (2020).
4. Strassburg, B. B. N. et al. Reply to: The risks of overstating the climate benefits of ecosystem restoration. Nature 609, E4–E6 (2022).
5. Strassburg, B. B. N. et al. Reply to: Restoration prioritization must be informed by marginalized people. Nature 607, E7–E9 (2022).
6. Hurtt, G. C. et al. Harmonization of global land use change and management for the period 850-2100 (LUH2) for CMIP6. Geosci. Model Dev. 13, 5425–5464 (2020).
7. Land-Use Harmonization (LUH2). https://luh.umd.edu/. (accessed 01 June 2020)
8. Olson, D. M. et al. Terrestrial ecoregions of the world: A new map of life on Earth. BioScience 51, 933–938 (2001).
9. IIASA & FAO. Global Agro-ecological Zones v3.0. https://www.gaez.iiasa.ac.at/. (accessed 01 June 2020)
10. Thomas, C. D. et al. Extinction risk from climate change. Nature 427, 145–148 (2004).
11. Hannah, L. et al. 30% land conservation and climate action reduces tropical extinction risk by more than 50%. Ecography 43, 943–953 (2020).
12. IUCN. The IUCN Red List of Threatened Species. Version 2022-1. https://www.iucnredlist.org/. (accessed 10 December 2022)
13. Feng, X. et al. How deregulation, drought and increasing fire impact Amazonian biodiversity. Nature 597, 516–521 (2021).
14. Brooks, T. M. et al. Measuring Terrestrial Area of Habitat (AOH) and Its Utility for the IUCN Red List. Trends Ecol. Evol. 34, 977–986 (2019).
15. Beyer, H. L., Venter, O., Grantham, H. S. & Watson, J. E. M. Substantial losses in ecoregion intactness highlight urgency of globally coordinated action. Conserv. Lett. 13, 1–9 (2020).
16. Sanderson, E. W. et al. The Human Footprint and the Last of the Wild. BioScience 52, 891 (2002).
17. Jones, B. & O’Neill, B. C. Spatially explicit global population scenarios consistent with the Shared Socioeconomic Pathways. Environ. Res. Lett. 11, (2016).
18. Hoekstra, J. M., Boucher, T. M., Ricketts, T. H. & Roberts, C. Confronting a biome crisis: Global disparities of habitat loss and protection. Ecol. Lett. 8, 23–29 (2005).
19. Watson, J. E. M. et al. Persistent Disparities between Recent Rates of Habitat Conversion and Protection and Implications for Future Global Conservation Targets. Conserv. Lett. 9, 413–421 (2016).
20. Erb, K. H. et al. Unexpectedly large impact of forest management and grazing on global vegetation biomass. Nature 553, 73–76 (2018).
21. Sanderman, J., Hengl, T. & Fiske, G. J. Soil carbon debt of 12,000 years of human land use. Proc. Natl. Acad. Sci. U. S. A. 114, 9575–9580 (2017).
22. Chaplin-Kramer, R. et al. Global modeling of nature’s contributions to people. Science 366, 255–258 (2019).
24. Robinson, S. et al. International Model for Policy Analysis of Agricultural Commodities and Trade (IMPACT) version 3.1. International Food Policy Research Institute (IFPRI) 128 https://www.ifpri.org/publication/international-model-policy-analysis-agricultural-commodities-and-trade-impact-model-0 (2015).
25. Robinson, T. P. et al. Mapping the global distribution of livestock. PLoS ONE 9, (2014).
26. Strassburg, B. B. N. et al. Strategic approaches to restoring ecosystems can triple conservation gains and halve costs. Nat. Ecol. Evol. 3, (2019).
27. International Labour Organization. ILOSTAT database. 1–5 https://ilostat.ilo.org/data (2021).
28. United Nations Statistics Division. UN Comtrade Database. https://comtrade.un.org/.
29. Beyer, H. L., Dujardin, Y., Watts, M. E. & Possingham, H. P. Solving conservation planning problems with integer linear programming. Ecol. Model. 328, 14–22 (2016).
There is NO Competing Interest.
- IISGBFprojectSupplementaryMaterialOK.pdf
Supplementary Information 1