Presented by Julia Naime (Federal University of Minas Gerais (UFMG)) at "Identifying effective policy interventions for different deforestation dynamics" on 4 May 2023
2. Introduction
• The mantra ‘Context matters!’ has been repeatedly stated for why the impact of forest
conservation measures varies greatly across locations.
• Impact evaluations highlight how context affect policy impacts.
• Yet, there is no synthesis of what factors hinder or improve the impacts of conservation
policies, i.e., how impact depends on context.
• Need middle-range theories (Meyfroidt et al. 2018): A ‘sweet spot’ between local, highly
contextual studies and swiping generalizations.
• The Global Comparative Study on REDD+ (GCS REDD+) aims to fill this gap by
developing and operationalizing the diagnostics approach to address the key question:
Which policy measures are likely to work where to reduce deforestation, in a given
context (and why)?
3. Context matters!
Deforestation diagnostics tells you how it matters
Policies
P1 P2 P3
Archetypes
(context)
A1
A2
A3
Three tasks:
1. Define archetypes (describe the symptoms or disease): (x1, x2, x3, …) => A (rows)
2. Develop a policy typology (columns)
3. Review of evidence (cells)
5. Niki de Sy (WUR), Arild Angelsen (NMBU), Julia Naime (CIFOR-ICRAF), Malte
Ladewig (NMBU), Valentina Robiglio (CIFOR-ICRAF), Karla Vergara (CIFOR-
ICRAF), Martin Herold (PIK)
Forest change archetypes of Tropical
Moist Forests
6. Introduction
• Despite the increasing availability of data, the
understanding of recurrent deforestation patterns and
underlying configurations of drivers remains poor (Pendrill
et al., 2022).
• Only recently, research started addressing this knowledge
gap through archetype analysis(Seitz et al., 2019;
Oberlack et al., 2019; Eisenack et al., 2021, Buchadas et
al., 2022).
• Our study adds to this recent literature by:
• Looking at forest dynamics.
• Looking at current forest state.
• Looking at deforestation related land use conversions
(i.e. proximate drivers)
Deforestation map from MapBiomass
7. An example of archetype analysis (Buchadas et al. 2022, Nature Sustainability)
• Use multiple variables to describe deforestation
frontiers: % of deforested área, speed of
deforestation or deforestation rates,
fragmentation rate.
• Indetify 5 archetypes in tropical dry woodlands
1. Inactive frontiers
2. Consolidated frontiers
3. Fragmented frontiers
4. Rampant frontier
5. Looming frontiers
• Could be useful for:
1. Non prescriptive recommedations of
where to implement which policies
2. Faciliate comparative studies across
regions.
8. Objectives
Three overall objectives of our study:
1. Identify broad archetypes of forest state and change in tropical moist
forest landscapes over the last 2 decades (2000-2021);
2. Assess the proximate drivers, i.e., conversion from forest to other land
use, in deforestation fronts and
3. Assess the deforestation risk in tropical moist forest landscapes with high
remaining forest cover and low historical deforestation.
Evaluating these archetypical patterns presents relevant entry points for policy
formulation.
9. Methods
Our analytical framework consists of
four steps :
1) derive metrics to describe forest
state and forest change,
2) build thematic typologies, and
use them to
3) develop archetypes,
4) with some archetypes further
developed into nested archetypes.
10. Data
• Forest extent, forest state and forest
change retrieved from the tropical moist
forest (TMF) dataset (Vancutsem et al,
2021).
• We selected the TMF dataset because
• It provides detailed spatial (30m
resolution)
• Detailed temporal (annual) information
on long-term forest cover changes.
• Detailed information on forest
degradation and post-deforestation
recovery.
https://forobs.jrc.ec.europa.eu/TMF/explorer.php
https://forobs.jrc.ec.europa.eu/TMF/resources.php
11. Metrics Typologies
• All metrics were aggregated from
30m resolution pixels to 5km
resolution “landscapes”.
Metric Low Medium High
Forest cover 2000 & 2021 (% of
land)
0-25 25-75 75-100
Deforestation 2000-2021 (% of
land)
0-5 5-20 20-100
Forest degradation (% of forest) 0-20 20-65 65-100
Inactive Old Active Emerging
Activeness
Annual 3y loss
rates < 1%
Annual 3y loss
rate > 1%, only
before 2015
Annual 3y loss rate >
1%, before & after
2015
Annual 3y loss rate > 1%,
only after 2015
13. Results 1a:Typologies in Brazil
Deforestation Severity
Forest cover 2000 (% of land)
Low (<25%)
Medium (25%
to 75%)
High
(>75%)
Total
forest
loss
(%
land)
2000-2021
Low (<5%) 41 3 35
Medium
(5%-20%)
4 5 4
High
(>20%)
0 4 5
14. Results 1b:Typologies in Brazil
Current forest state
Forest cover 2021 (% of land)
Low
(<25%)
Medium (25%
to 75%)
High
(>75%)
Forest
degradation
2021
(%
forest)
Low (<20%) 1 4 38
Medium
(20%-65%)
13 8 2
High (>65%) 35 1 0
15. Results 1c:Typologies in Brazil
Hotspots
Forest cover 2000 (% of land)
Low (<25%)
Medium (25%
to 75%) High (>75%)
Average
annual
deforestation
rate
Gradual (<1%) 42 4 35
Old (>1% before
2015)
2 4 3
Persistent (>1%
before and after
2015)
0 3 5
Emerging (>1%
after 2015)
1 0 1
16. Results 2: Archetypes
• We defined two broad archetypes where
deforestation has been historically low:
core landscapes and moderate
landscapes.
• We then distinguished four archetypes
labeled “deforestation fronts”. We
describe them considering a temporal
dimension: Old, Persistent, Recent and
Gradual.
17. Results 2: Archetypes in Brazil
Archetype Percentage
Core forest landscapes 34
Moderate forest landscapes 3
Past deforestation fronts 9
Persistent deforestation fronts 8
Recent deforestation fronts 2
Gradual deforestation fronts 4
Sparse forest landscapes 37
Landscapes with no forest 10
18. • We further described
• Core Landscapes and Moderate
landscapes according to their
deforestation risks.
• Deforestation fronts according to
their dominant land use following
deforestation.
Results 3: Nested Archetypes
19. Results 3: Nested archetypes of the deforestation fronts
• Information on forest conversion to tree plantation, forest
regrowth and water are directly derived from the TMF dataset.
• Most of the remaining forest to land use conversions were
derived from the 2021 ESA Worldcover dataset (Zanaga et al.,
2022).
• We overlaid the grassland pixels from the 2021 ESA Worldcover
dataset with the 2019 land use map of Wrinkler et al. (2019) to
distinguish between managed (i.e. pastures) and unmanaged
grassland.
Name Colour
Tree plantation
Cropland/Agriculture
Forest regrowth – cropland
Pasture
Shrubland - grassland
Forest regrowth – grassland
Urban
Shrubland
Forest regrowth
Water
20. • Deforestation risk was defined based on accessibility and
agricultural suitability
• As a proxy for accessibility we used a 1 km resolution
dataset on travel time to cities (Weiss et al., 2018).
• Agricultural suitability was derived from the GAEZ 3.0
dataset (IIASA/FAO, 2012).
• Agro-climatic potential for low input, rainfed agriculture of
the main commodity crops.
Results 3: Nested archetypes of the core and moderate forests
Agricultural suitability (0 to 100)
Accessibility
Low
suitability
Medium
suitabilit
y
High
suitabili
ty
Low
access
Medium
access
High
access
21. Conclusions and further research
• Geographically defining different archetypes of tropical forest deforestation can help to:
• Identify in which contexts different conservation measures are implemented and their
effects (ex-post analysis).
• Identify the most appropriate measures and public policies for the archetype/context
(targeting, ex-ante), depending on the cause of deforestation and the risk of deforestation.
• Generate and facilitate comparative studies across regions and countries (the REDD+
ECG includes Brazil, Democratic Republic of Congo, Peru, and Indonesia).
• In practice, archetypes could also help to:
• Help understand why certain measures are more effective in certain areas and not in
others.
• Jurisdictional approaches: identifying different patterns of deforestation in specific
provinces.
• Benefit sharing: stock and flow mechanisms.
22. cifor.org | worldagroforestry.org | globallandscapesforum.org | resilientlandscapes.org
The Center for International Forestry Research (CIFOR) and World Agroforestry (ICRAF) envision a more equitable world where forestry and
landscapes enhance the environment and well-being for all. CIFOR–ICRAF are CGIAR Research Centers.
cifor.org/gcs
THANK
YOU
23. Usefulness
• Policy makers and practitioners
• a tool for identifying the most robust policy interventions, by “identifying elements of individual
problems that are significant from a problem-solving perspective” (Young 2002, 176)
• learning across similar locations, also across national borders. The archetypes classification defines the
relevant locations to learn from
• will not replace local, context-specific insights, rather it builds on this knowledge and aims to
systematize and aggregate it.
• For case researchers
• identify critical factors for understanding individual cases
• locate them within the broader universe of tropical forest contexts.
• For generalists and meta-analysts
• help in the aggregation of studies by providing more structured case-data, and thus improve future
assessments of “what would work where and why?”.
Editor's Notes
Change for map biomass
Forest state (undisturbed forest, degraded forest) and forest transition dynamics (forest regrowth, deforested land, conversion to plantations, conversion to water, afforestation, and changes within mangroves) from the initial observation period to the end of 2021, as well as the timing (dates and duration), recurrence, and intensity of each disturbance.
The TMF dataset includes the tropical rain forest and the tropical moist deciduous forests in the “tropical rainforest,” “tropical moist forest,” “tropical mountain system,” and “tropical dry forest” Global Ecological Zones (FAO, 2012).
So, 2500 has
Hotspots: These landscapes describe areas where deforestation in 2000-2021 has been high or medium, or where deforestation rate is higher than 1%, and therefore classified as hotspots.