Save First,
Restore Second
Comparing the true cost and benefit of three ways to save the Amazon’s forests
SBR Project ID
#NE01
Date of project
2010–2020
Location
Pará, Brazil
Type of project
Conservation Effectiveness Study
Categories
Lead institution and funders
Tropical forests that have never been cleared are vital for protecting wildlife, slowing climate change and supporting the people who live in and around them. Yet forest loss remains high despite many policies and investments. In the Amazon, most efforts go into stopping deforestation, where forest is cut down to make way for pasture and crops, and increasingly into restoring forest on cleared land.
Far less attention goes to forest that is damaged but left standing. Fires spread into it from farmland, where slash-and-burn clearing is still common, and loggers take trees out of it. Together, deforestation and this damage are turning tropical forest regions from places that absorb carbon into places that release it.
Stopping deforestation, preventing fire and logging damage and restoring forest each have their own laws and programmes. What has been missing is a direct comparison of how much each one does for wildlife and for the carbon stored in trees, and at what cost.
This study makes that comparison in Santarém and Paragominas, two regions of Pará state in Brazil that together cover more than three million hectaresA hectare is an area of 100 by 100 metres – roughly one and a half football pitches.. Old-growth forest (forest that has never been felled) and secondary forest (trees growing back on land that was once cleared) lie here alongside pasture and large mechanised soy farms, so all three strategies can be compared in the same place. Using ten years of records from both regions, the researchers modelled what each strategy would have achieved and what it would have cost.
3 Million+
Hectares
of Amazon landscape across Santarém and Paragominas were analysed in this study.
158,980
Hectares
of forest were cut down in the two regions between 2010 and 2020.
273,558
Hectares
of standing forest were burned or logged over the same ten years.
Follow this research
Get an email whenever we publish new findings on this project.
AIMS
What we want to know
We have two main questions.
The researchers compared how much of the landscape’s bird and tree life each approach protects or brings back, and how much carbon stays stored in trees and other plants – first for each approach on its own, and then in combination. The second question adds cost.
The costs include the farming income landowners give up, plus the cost of restoring forest and preventing fires. The question is whether the approach that protects the most also gives the most for the money spent.
IMPORTANCE
Why it matters
Preventing fire and logging damage is already part of international climate schemes such as REDD+A United Nations framework that pays countries for reducing emissions from deforestation and forest damage. Opens the UN climate change website. and of Brazil’s current plan against deforestation in the Amazon. But until now there has been little evidence comparing it with the other two strategies on the same land and at their real cost, or showing whether any single strategy is enough to reverse the losses.
The answer shows where conservation money and enforcement are likely to protect the most, and what part restoring forest should play alongside protecting the forest that remains.
Every strategy costs money, so choosing between them decides how much forest, wildlife and carbon a conservation budget can save.
PROJECT
So we set up a study
Fully stopping deforestation or fire and logging damage, or restoring every hectare of forest the law requires, would be extremely difficult to achieve across a whole region. So the researchers used counterfactual modellingA way of comparing what actually happened with what would have happened if different choices had been made. instead. They started from satellite maps of what happened across Santarém and Paragominas between 2010 and 2020: forest cleared, burned, logged and growing back. Then they built alternative versions of that decade, one for each strategy, as if it had been fully put into practice. These are modelled scenarios, not policies that were carried out.
Strategy 1
Stopping deforestation
The model removes all forest clearing recorded between 2010 and 2020, so the forest stays standing.
158,980 hectares kept as forestStrategy 2
Preventing fire and logging damage
The model prevents every fire and logging event recorded inside standing forest during the decade, so the forest stays undamaged.
273,558 hectares kept undamagedStrategy 3
Restoring forest
The model adds new secondary forestYounger forest that grows back on land that was once cleared, either on its own or after trees are planted. on private land wherever the law already requires landowners to restore forest. It starts next to existing forest, where trees tend to regrow more easily.
80,705 hectares of forest regrownWhat we measured
Biodiversity
How likely each of 160 bird species and 426 tree species is to be found in each part of the landscape. Species that live in only a small area count for more.
Above-ground carbon
The carbon held in trees and other plants above the ground, measured in tonnes per hectare.
Cost
The farming income landowners give up, plus the cost of the work itself: restoring forest, and building and maintaining firebreaksStrips of cleared ground that stop a fire spreading from farmland into forest.. The researchers estimated lost farming income from interviews with the owners of 499 rural properties.
RESULTS
What we have found so far
Both questions now have answers. Here is what the comparison found.
Question 1Which strategy does more for biodiversity and stored carbon?
At a glance
How much of the biodiversity and carbon lost between 2010 and 2020 each strategy would have prevented or made up for.
Stopping deforestation
Preventing fire and logging damage
Forest restoration
Strategies combined
How much more biodiversity and carbon the landscape would have had than in 2010.
Preventing deforestation and fire and logging damage
All three strategies
Stronger together
No single strategy would have stopped the losses on its own. The landscape would only have ended the decade with more biodiversity and carbon than in 2010 if both deforestation and fire and logging damage had been prevented. Adding forest restoration would have given the biggest gains.
When the researchers modelled the strategies in combination, together they saved more than their separate effects added up: about a fifth more biodiversity, and between a tenth and a quarter more carbon.
Question 2Once cost is counted, which strategy gives the best value?
Best value for biodiversity
Preventing fire and logging damage protected the most biodiversity for the money spent – about three times as much as stopping deforestation.
Best value for carbon
Stopping deforestation kept the most carbon for the money spent, followed by preventing fire and logging damage.
Restoring forest costs most
Restoring forest was the least cost-effective strategy for both biodiversity and carbon, even when the researchers assumed its costs were 95% lower than estimated.
UNEXPLORED
Unexplored questions
Like all good research, this study answered some questions and raised others. Explore the topics below to see what we still need to know.
Tap or click a pin to open each one.
FAQ
Frequently asked questions
Does this mean restoring forest isn’t worth doing?
It is still worth doing, but it works best alongside protecting the forest that remains. On its own, restoring forest made up for only a small share of what was lost. Combined with the other two strategies it added gains of its own, and on farms with little forest left it did more for biodiversity than stopping deforestation.
Why does preventing fire and logging damage matter so much?
Because fire and logging damage forest that is still standing, harming wildlife and releasing carbon without the forest being cleared. In this study, preventing that damage protected more biodiversity than any other single strategy, and more for the money spent.
Is this specific to the Brazilian Amazon, or does it apply elsewhere?
The data come from two regions of Pará state. The researchers say their findings are likely to apply to other tropical regions where farmland is expanding into forest, though this has not yet been tested with local data from those places.
TIMELINE
The long journey
Fieldwork begins
Teams from the Sustainable Amazon Network start surveying forest and farmland across Santarém and Paragominas – recording birds, identifying trees and measuring the carbon they hold. Over the same period they interview the owners of 499 rural properties about how their land is used.
Fieldwork complete
The last field surveys are finished. Together with the property interviews, they give the team a detailed picture of both landscapes at the start of the decade – the starting point every scenario is measured against.
A decade of change on record
Satellite maps now cover ten years of deforestation, fires, logging and regrowth across both regions – the changes the models start from.
Published in Science
The findings are published in the journal ScienceRead the original research paper: Miranda et al., Science, 17 September 2026..
IMPACT
What’s happened
Here’s some of what we’ve published and contributed to. We’ll keep adding to this as the work develops.
Published in Science
Data and code openly available
Future
Hopes for the future
Through this research, we hope to:
see protection from fire and logging treated as a main conservation strategy, with the same enforcement and funding that stopping deforestation and restoring forest already receive.
help shape schemes that reward landowners for protecting forest elsewhere in the tropics, once similar comparisons can be run with local data.
TEAM
Who are we?
Lead Investigator
Dr Leonardo S. Miranda
Senior Research Associate
Lancaster University, UK
Co-investigators
James R. Thomson
Arthur Rylah Institute for Environmental Research, Australia
Joice Ferreira
EMBRAPA Amazônia Oriental, Brazil
Toby A. Gardner
Stockholm Environment Institute, Sweden
Erika Berenguer
Lancaster University, UK
University of Oxford, UK
Alexander C. Lees
Manchester Metropolitan University, UK
Cornell University, USA
Ralph Mac Nally
University of Canberra, Australia
Luiz E. O. C. Aragão
National Institute for Space Research, Brazil
University of Exeter, UK
Pedro H. S. Brancalion
University of São Paulo, Brazil
re.green, Brazil
Silvio F. B. Ferraz
University of São Paulo, Brazil
Rachael D. Garrett
University of Cambridge, UK
Paulo G. Molin
University of São Paulo, Brazil
Federal University of São Carlos, Brazil
Nárgila G. Moura
MCTI/Museu Paraense Emílio Goeldi, Brazil
Sâmia S. Nunes
Vale Institute of Technology, Brazil
Luke Parry
Lancaster University, UK
Juliana M. Silveira
Lancaster University, UK
Ima C. G. Vieira
MCTI/Museu Paraense Emílio Goeldi, Brazil
Cecilia Viana
Lancaster University, UK
Jos Barlow
Lancaster University, UK
UK
Lancaster University – Lancaster Environment Centre
Lancaster, UK
University of Oxford – Environmental Change Institute
Oxford, UK
Manchester Metropolitan University – School of Biological and Chemical Sciences
Manchester, UK
University of Cambridge
Cambridge, UK
University of Exeter – Department of Geography
Exeter, UK
Brazil
EMBRAPA Amazônia Oriental
Belém, Brazil
MCTI/Museu Paraense Emílio Goeldi
Belém, Brazil
Vale Institute of Technology
Belém, Brazil
National Institute for Space Research
São José dos Campos, Brazil
University of São Paulo – Department of Forest Sciences
Piracicaba, Brazil
Federal University of São Carlos
Brazil
re.green
Rio de Janeiro, Brazil
Australia
Arthur Rylah Institute for Environmental Research
Heidelberg, Victoria, Australia
University of Canberra – Centre for Conservation Ecology and Genomics
Canberra, Australia
Sweden
Stockholm Environment Institute
Stockholm, Sweden
USA
Cornell University – Cornell Lab of Ornithology
Ithaca, NY, USA
Funders
Instituto Nacional de Ciência e Tecnologia (CNPq)
Brazil
FAPESPA
Brazil
Empresa Brasileira de Pesquisa Agropecuária (Embrapa)
Brazil
Natural Environment Research Council (NERC)
UK
Darwin Initiative
UK
Global Centre on Biodiversity for Climate (GCBC)
UK
The Nature Conservancy
US
BNP Paribas Foundation
France
Lancaster University Strategic Priorities and Impact Acceleration Funds
UK
Ramboll Foundation
Denmark
CONTACT
Get involved
Questions about this project? Get in touch with the team below.
Research team
Principal Investigator
Dr Leonardo S. Miranda
Lancaster Environment Centre, Lancaster University
Stay connected
Leonardo S. Miranda on LinkedInMore about this project
The project on the Global Centre on Biodiversity for Climate websiteFollow this research
Get an email whenever we publish new findings on this project.
Join the conversation
Should governments spend as much on preventing fires and logging in standing forest as they do on stopping forest being cleared?
If you were in charge of a conservation budget, would you spend it on protecting existing forest or on restoring forest that has been lost – and why?