Strategic approaches to restoring ecosystems can triple conservation gains and halve costs 1
Bernardo B. N. Strassburg
1,2,3
, Hawthorne Beyer
4
, Renato Crouzeilles
1,2,3
, Alvaro Iribarrem
1,2
, 2
Felipe Barros
2
, Marinez Ferreira de Siqueira
5
, Andrea Sánchez-Tapia
5
, Andrew Balmford
6
, 3
Jerônimo Boelsums Barreto Sansevero
7
, Pedro Henrique Santin Brancalion
8
, Eben 4
North Broadbent
9
, Robin Chazdon
2,10,11
, Ary Oliveira Filho
12,
Toby Gardner
2,13
, Ascelin 5
Gordon
14
, Agnieszka Latawiec
1,2,15,16
, Rafael Loyola
17
, Jean Paul Metzger
18
, Morena Mills
19
, 6
Hugh P. Possingham
20,21
, Ricardo Ribeiro Rodrigues
22
, Carlos Alberto de Mattos Scaramuzza
23
, 7
Fabio Rubio Scarano
3,24
, Leandro Tambosi
25
, Maria Uriarte
26
8
9
1 – Rio Conservation and Sustainability Science Centre, Department of Geography and the 10
Environment, Pontifícia Universidade Católica, 22453900, Rio de Janeiro, Brazil 11
2 – International Institute for Sustainability, Estrada Dona Castorina 124, 22460-320, Rio de 12
Janeiro, Brazil 13
3 – Programa de Pós Graduacão em Ecologia, Universidade Federal do Rio de Janeiro, 21941-14
590, Rio de Janeiro, Brazil 15
4 – ARC Centre of Excellence for Environmental Decisions, School of Biological Sciences, 16
University of Queensland, St. Lucia, 4072, Queensland, Australia 17
5 – Botanical Garden Research Institute of Rio de Janeiro, Rua Jardim Botânico 1008, 22470-18
180 Rio de Janeiro, Brazil 19
6 – Department of Zoology, University of Cambridge, Downing Street, Cambridge, CB2 20
3EJ, UK 21
7 – Department of Environmental Science, Instituto de Florestas, Federal Rural University of Rio 22
de Janeiro, Rodovia BR 465, Km 07, s/n - Rural, Seropédica, 23890-000, Brazil 23
8 – Departamento de Ciências Florestais – Esalq/USP, Av. Pádua Dias 11, 13.418-900, 24
Piracicaba, SP, Brazil 25
9 – Spatial Ecology and Conservation Lab, School of Forest Resources and Conservation, 26
University of Florida, Gainesville, Florida, USA, 32611 27
10 – Department of Ecology and Evolutionary Biology, University of Connecticut, 75 N. 28
Eagleville Road, Unit 3043, Storrs, CT 06269-3043, U.S.A. 29
11 – World Resources Institute, Global Restoration Initiative, 10 G Street, NE Suite 800, 30
Washington, D.C. 20002 - USA 31
12 – Department of Botanic, Federal University of Minas Gerais, Av. Antônio Carlos, 6627, 32
31270-901, Belo Horizonte, MG, Brazil 33
13 – Stockholm Environment Institute, Linnégatan 87D, 115 23 Stockholm, Sweden 34
14 – School of Global Urban and Social Studies, RMIT University, GPO Box 2476 Melbourne, 35
Australia 3001 36
15 – Institute of Agricultural Engineering and Informatics, Faculty of Production and Power 37
Engineering, University of Agriculture in Kraków, Balicka 116B, 30-149 Kraków, Poland 38
16 – School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, United 39
Kingdom 40
17 – Laboratório de Biogeografia da Conservação, Departamento de Ecologia, Universidade 41
Federal de Goiás, Goiânia, GO, Brazil 42
18 – Department of Ecology, Institute of Biosciences, University of São Paulo, Rua do Matão, 43
321, Travessa 14, 05508-900, São Paulo, SP, Brazil 44
19 – Faculty of Natural Sciences, Department of Life Sciences, Silwood Park, Imperial College 45
London, London SW7 2AZ, UK 46
20 – The Nature Conservancy, 4245 Fairfax Drive, Arlington, VA 22203, USA 47
21 – The University of Queensland, St Lucia, QD, 4072, Australia 48
22 – Department of Biological Science, Escola Superior de Agricultura Luiz de Queiroz, 49
University of São Paulo, Avenida Pádua Dias n. 11, Centro, 13418900 Piracicaba, SP, Brazil 50
23 – Department of Ecosystems Conservation, Brazilian Ministry of the Environment (MMA). 51
Ed. Marie Prendi Cruz, SEPN 505 Norte, Bloco "B" 4 º andar sala 416, 70.730-542, Brasília, 52
Brazil 53
24 – The Brazilian Foundation for Sustainable Development, R. Eng. Álvaro Niemeyer, 76 - São 54
Conrado, Rio de Janeiro - RJ, 22610-180, Brazil 55
25 – Centro de Engenharia, Modelagem e Ciências Sociais Aplicadas, Federal University of 56
ABC, Av. dos Estados, 5001, Bairro Santa Terezinha, Santo André, SP, Brazil 57
26 – Department of Ecology, Evolution and Environmental Biology, Columbia University, 116th 58
St & Broadway, New York, NY 10027, USA 59
International commitments for ecosystem restoration add-up to one-quarter of the world´s 60
arable land
1
. Fulfilling them would ease global challenges such as climate change
2
and 61
biodiversity decline
3
, but could displace food production
4
and impose financial costs on 62
farmers
5
. Here we show a novel restoration prioritization approach capable of revealing 63
these synergies and trade-offs, incorporating for the first time ecological and economic 64
efficiencies of scale and modelling specific policy options. We show, for an actual large-65
scale restoration target of the Atlantic Forest hotspot, that our approach can deliver an 66
eightfold increase in cost-effectiveness for biodiversity conservation compared to a baseline 67
of non-systematic restoration. A compromise solution avoids 26% of the biome’s current 68
extinction debt of 2864 plant and animal species (an increase of 257% compared to the 69
baseline), and sequesters 1 billion tonnes of CO
2
Eq (a 105% increase) while reducing costs 70
by US$ 28 billion (a 57% decrease). Seizing similar opportunities elsewhere would offer 71
substantial contributions to some of humankind´s greatest challenges 72
Ecosystem restoration can provide multiple benefits to people and help to achieve multiple 73
Sustainable Development Goals
6-8
, including climate change mitigation and nature conservation. 74
For these reasons, 47 countries have collectively committed to have 150 and 350 million hectares 75
of degraded lands under restoration by 2020 and 2030, respectively, and have included major 76
restoration targets in national pledges to the Paris Climate Agreement
1
. Restoration, however, 77
has both direct costs – those required for implementation and maintenance – and indirect costs, 78
including the loss of revenues from foregone agricultural production. Crucially, these restoration 79
costs and benefits present trade-offs and synergies that vary greatly across space
9-11
. In the 80
context of safeguarding existing habitats there has been considerable progress in understanding 81
some of these trade-offs
3
, with the field of Systematic Conservation Planning (SCP) providing 82
methods for spatial prioritisation that maximize benefits while minimizing costs
12
. Despite some 83
recent efforts
9-10,13
, applications of comprehensive SCP approaches to complex large-scale 84
restoration problems with multiple objectives remain sparse. 85
Here we present a novel restoration prioritization approach based on Linear Programming (LP) 86
to solve customized complex restoration problems at large scales. We apply this approach to 87
solve a problem of global significance that will inform restoration policy and practice at a 88
national scale in the Brazilian Atlantic Forest hotspot
14-15
, a highly deforested and fragmented 89
region poised to undergo one of the biggest large-scale restoration efforts
16
. We identify exact 90
cost-effective solutions that consider multiple benefits, costs and policy scenarios and 91
investigate: i) trade-offs in benefits and costs across different scenarios, and ii) the impacts of 92
increasing the size of restoration projects. LP can find exact solutions that can perform at least 93
30% better than mainstream SCP software
17
. It can also be more fully customized, allowing the 94
incorporation of complex aspects of restoration relevant to particular socioecological contexts. In 95
this application, we aimed at maximising restoration benefits for biodiversity conservation and 96
climate change mitigation while reducing restoration and opportunity costs. 97
We divided the biome into 1.3 million planning units of 1 km
2
. For biodiversity conservation, 98
benefit was measured as the reduction in projected extinctions owing to habitat restoration
18
. We 99
gathered and analysed species occurrence data in the Atlantic Forest and, following data 100
cleaning, identification of endemism by specialists and model selection (Methods), generated 101
potential species occurrence models for 785 species of plants, birds and amphibians endemic to 102
the Atlantic Forest, representing the best set of biodiversity data currently available for this 103
biome. We then calculated the marginal contribution of each hectare restored to reducing each 104
species’ extinction probability, based on a function
11,19
derived from the species-area 105
relationship. The benefit of habitat restoration to each species is dynamic in that the value of 106
restoring additional habitat for that species diminishes as the total area of habitat increases. Our 107
approach explicitly accounts for this effect, though for visualisation purposes we can aggregate 108
the restoration value of each planning unit across all species, thereby generating a biodiversity 109
conservation benefits surface (Extended Data Fig. 1). Our species data confirmed the severity of 110
the biodiversity crisis underway in the Atlantic Rainforest, with an estimated 27-32% of the 111
biome’s endemic species currently committed to extinction (2,621-3,107 plants and animals, see 112
Methods). For climate change mitigation, benefit was measured as the potential aboveground 113
carbon sequestration in the first 20 years following habitat restoration
20
. We produced the 114
climate change mitigation surface (Extended Data Fig. 2a) by applying and extending a recently 115
published empirical model of the carbon sequestration potential of restoration
20
to the whole of 116
the Atlantic Rainforest. Restoration implementation costs, including maintenance and 117
monitoring, were estimated based on a survey with restoration companies active in the Atlantic 118
Rainforest, spatially adjusted by a proxy for natural regeneration potential based on a recently 119
published model for ecological uncertainty of tropical forest restoration success
21
(Methods). 120
Opportunity costs, a measure of potential conflict with agricultural production, were estimated 121
based on land acquisition costs and spatial distributions of agriculture and pasturelands
22
. A 122
restoration costs surface (Extended Data Fig. 2b) was built based on these two costs (hereafter 123
referred to as total cost).
124
We also introduced advances regarding the impacts that the scale of a restoration project has on 125
its costs and benefits. Costs per unit area restored reduce with increasing area of the restoration 126
project, so we modelled these economies of scale using field evidence on how unitary costs fall 127
as projects grow (Methods and Extended Data Fig. 3). The size of the restoration project also 128
