ON HABITAT LOSS

ROAD IMPACT ASSESSMENT

ON HABITAT LOSS

IN LATIN AMERICA




Karolina Argote, Louis Reymondin, Carolina Navarrete, Denny
Grossman, Jerry Touval, Andy Jarvis

Decision and Policy Analysis Research Area (DAPA)

International Center for Tropical Agriculture (CIAT)

Conservation Biology Institute (CBI)

The Nature Conservancy (TNC)



















Spot the road?

Project outcomes


This project presents a habitat change monitoring methodology
that can be used
to identify environmental
impacts of road
construction, and support improved design of future projects
that would minimize
negative environmental
impacts.


The project has also helped understand the nature of
environmental impacts of road infrastructure projects in distinct
contexts across Latin America, demonstrating the importance of
policy and ecosystem specific safeguards.

Objectives of the project

•
Evaluate the
environmental impacts
of road
infrastructure in the past through monitoring
of natural habitat loss pre
-

and post
-

construction for
5 road projects
across Latin
America

•
Demonstrate how this can be integrated into a
decision support tool
–

DATABASIN

•
Identify entry points by which
ex ante
assessment
can provide
improved safeguards


Content

1.
Methodologies

1.1. Terra
-
I
–

Habitat Change Monitoring

2. Road Impact Results

2.1. The BR
-
364 highway in Brazil.

2.2. The IIRSA projects in Peru.

2.3. The Pan
-
American Highway in Panama.

2.4. The Santa Cruz
–

Puerto Suarez corridor in Bolivia.

2.5. The Trans
-
Chaco Highway in Paraguay.

3.
Methodology of the Future Deforestation
Scenarios and Results

4.
Carbon
-
Conservation Interface

5.
Conclusions and Recommendations


Photo by Alvaro
Gaviria

in Cartagena del
Chaira

Parques

Nacionales

Naturales

de Colombia

1.1. Methodologies:

Terra
-
i

Methodology

Terra
-
i

is a system that monitoring the habitat change in Latin America

using Neural Network and satellite data





We

therefore

try

to

learn

how

each

pixel

(
site
)

responds

to

climate
,

and

any

anomoly

corresponds

to

human

impact
.


Neural
-
network,

is

a

bio
-
inspired

technology

which

emulates

the

basic

mechanism

of

a

brain
.


It

allows

…


To

find

a

pattern

in

noisy

dataset


To

apply

these

patterns

to

new

dataset

INPUTS
:
Past

NDVI
(MODIS
MOD13Q1
)


Previous

Rainfall

(
TRMM
)


OUTPUT
:
16
day

predicted

NDVI

1.1. Methodologies:
Terra
-
i

The

Bottom
-
Line

•
250m
resolution

•
Latin

American
coverage

(
currently
)

•
Satellite

data
to

monitorithe

habitat

every

16
days

•
Identification

of
habitat

change

events

•
Habitat

loss

data online
to

visualize

and
download
.


www.terra
-
i.org



1.1. Methodologies:
Interpreting the maps

1.1. Methodologies:
Interpreting the graphs

Area of
habitat
lost

Buffer distance
from the road (km)

Geographic
footprint of
the road

When did the habitat
loss happen within each
buffer?

BR
-
364 Highway, Brazil

Date

Acre Segment: Construction 2002 to 2010

Rondonia Segment: Construction 1985 to 1997

2.1. Road Impact Results:
BR
-
364 Highway, Brazil

Study Area

Section

2
:

a

corridor

of

515
km

which

connects

the

town

of

Rio

Branco

in

the

state

of

Acre

and

Porto

Velho

in

the

state

of

Rondônia
.


2.1. Road Impact Results:
BR
-
364 Highway, Brazil

Study Area

The

road

Cruzeiro

do

Sul

–

Porto

Velho

was

analyzed

into

two

different

sections
.

Section

1
:

a

corridor

of

623
km

Cruzeiro

do

Sul

-

Rio

Branco

in

the

state

of

Acre,

Brazil
.


This

section

passes

through

a

large

biological

corridor

of

the

state

of

Acre

which

has

been

regulated

by

39

protected

areas

connected

to

each

other
.


2.1. Road Impact Results:
BR
-
364 Highway, Brazil

2.1. Road Impact Results:
BR
-
364 Highway, Brazil

2.1. Road Impact Results:
BR
-
364 Highway, Brazil

Road impact

Comparing

the

two

segments

one

can

see

a

huge

difference

in

the

deforestation

rates

and

in

how

it

is

the

spatially

distributed
.


More
localised

footprint, and
and

lower overall deforestation
levels. Nevertheless, increase
in last 2 years.

Much higher deforestation
rates, and much BIGGER
footprint >50km due to
secondary roads etc.

2.1. Road Impact Results:
BR
-
364 Highway, Brazil

2.1. Road Impact Results:
BR
-
364 Highway, Brazil

Protected Areas

Bom

Futuro

and

Jaciparaná

are

the

two

protected

areas

most

affected

by

deforestation

in

Rondônia

and

are

located

next

to

the

analyzed

road,

within

a

buffer

area

of

20
km
.



Actually,

the

deforestation

rate

in

Bom

Futuro

has

been

of

10
,
188

hectares

per

year

(adding

up

to

76
,
406

hectares

converted

in

7
.
5

years)

whereas

it

has

been

of

6
,
337

hectares

per

year

in

Rio

Jaciparaná

(adding

up

to

47525

hectares

converted

in

7
.
5

years)
.


2.1. Road Impact Results:
BR
-
364 Highway, Brazil

Conclusions

•
Section

1
:

Acre
.

Habitat

loss

of

19
,
542

hectares

was

recorded

per

year

in

average

in

a

buffer

area

of

50
km

of

the

Cruzeiro

do

Sul

-
Rio

Branco

Segment


•
Section

2
:

Rondonia
.

Habitat

loss

of

79
,
783

hectares

per

year

within

a

same

buffer

size

around

the

Rio

Branco

-
Porto

Velho

Segment
.


•
Much

higher

in

the

segment

Rio

Branco
-
Porto

Velho

(in

Rondonia)

than

in

Cruzeiro

do

Sul
-
Rio

Branco

(In

Acre)

likely

due

to

the

conservation

policies

implemented

in

Acre

state
.

Note

fewer

secondary

roads,

and

greater

protection

from

National

Parks

in

segment

1
.

IIRSA Project, Peru

Date

Construction: 1998 to 2007

2.2. Road Impact Results:
IIRSA Projects, Peru

Study Area

The

analyzed

roads

have

a

total

length

of

1584
km

and

go

through

all

Peru

from

the

Pacific

coast

to

the

Acre

state

in

Brazil
.

The

road

was

split

into

three

different

sections

for

the

analysis
:

Section 1

Andean

Section 2

Piedemonte

Section 3

Amazon



Section

1
:

752
km
.

Paita

on

the

Pacific

coast

(Piura)

to

Tarapoto
.



Section

2
:

381
km
.

Tarapoto

-

Huanuco

(where

it

passes

2
km

away

from

Tingo

Maria

National

Park)
.



Section

3
:

451
km
.

Tingo

Maria

(
Huanuco
)

-

Cruzeiro

do

Sul

(Acre,

Brazil)
.

2.2. Road Impact Results:
IIRSA Projects, Peru

2.2. Road Impact Results:
IIRSA Projects, Peru

Road Impact

Section

1

(
Paita
-
Tarapoto
)

:

accumulated

loss

of

40
,
794

hectares

(
5
,
439

Ha/
yr
)
.


Most habitat
loss in first
10km (45%)

Significant
increase in
deforestation in
past 3
-
4 years

2.2. Road Impact Results:
IIRSA Projects, Peru

2.2. Road Impact Results:
IIRSA Projects, Peru

Road Impact

Section

2

(
Tarapoto
-
Tingo

Maria
)

:

accumulated

loss

of

30
,
763

Ha

(
4
,
102

Ha/
yr
)
.

Most

impacted

areas

are

located

in

a

buffer

of

30
km

from

the

road
.

Most habitat
loss in first
30km (88%)

Significant
increase in
deforestation in
past 3
-
4 years

2.2. Road Impact Results:
IIRSA Projects, Peru

2.2. Road Impact Results:
IIRSA Projects, Peru

Road Impact

Section

3

(
Tingo

Maria
-
Cruzeiro
)
:

accumulated

loss

of

58
,
900

hectares

(
7
,
853

Ha/year)
.

Most

impacted

areas

are

located

in

a

buffer

of

30
km

from

the

road
.

Most habitat
loss in first
30km (81%)

No apparent
increase in
deforestation
during or after
road
construction

2.2. Road Impact Results:
IIRSA Projects, Peru

2.2. Road Impact Results:
IIRSA Projects, Peru

Conclusions

Section

1
:

Andes
.

Footprint

more

localised

(<
10
km),

25
%

increase

in

habitat

loss

post
-
project

versus

pre
-
project
.

Section

2
:

Piedemonte
.

Larger

footprint

(
10
-
20
km),

and

>

doubling

of

deforestation

after

road

finalization
.

Section

3
:

Tingo

Mario
-
Cruzeiro
.

High

baseline

levels

of

deforestation

in

the

region,

but

no

increase

since

road

project

(major

sections

of

road

still

not

complete)
.

Pan
-
American Highway, Panama

Date

Construction: 1985 to 1990

2.3. Road Impact Results:
Pan
-
American Highway, Panamá

Study Area

The

Pan
-
American

Highway

is

located

in

the

Darien

province

in

Panama

at

the

eastern

end

of

the

country

and

its

length

is

approximately

262
km,

i
n

a

30
km

of

buffer

around

the

road

are

located

more

than

10

protected

areas

with

important

ecological

functions
.

2.3. Road Impact Results:
Pan
-
American Highway, Panamá

Habitat Change Monitoring

Methodology

i.
Reclassify

the

Land

Cover

Maps

of

1992

and

2000

using

ArcGIS

software

in

Vegetation

and

non

vegetation

maps
.

ii.
Generated

the

deforestation

map

of

1992
-
2000
.

iii.
Applied

the

Terra
-
I

Methodology

to

monitoring

the

habitat

change

between

2004

to

2010
.

iv.
Analyze

the

road

impact

in

buffers

areas

in

10
,

20
,

30
,

40

and

50
km

of

the

road
.



MAIN

INPUTS



For

generated

deforestation

maps

before

2000
:



A

dataset

of

land

cover

produced

by

the

Forest

Information

System

Project,

the

National

Environmental

Authority

(ANAM)

for

1992

and

2000
.



For

generated

deforestation

maps

between

2004

and

2010
:

Terra
-
I

dataset
.


2.3. Road Impact Results:
Pan
-
American Highway, Panamá

2.3. Road Impact Results:
Pan
-
American Highway, Panamá


The habitat loss is greater the


closest it’s to the road.





Vast majority of habitat change
occurred in the 1990’s directly
after road construction.
Deforestation 2004
-
2011 < 10%
of 1990’s levels.

Road impact

2.3. Road Impact Results:
Pan
-
American Highway, Panamá

Conclusions

•
Between

1992

and

2000

there

was

an

alarming

loss

of

7
%

of

the

total

national

forest

cover

in

Panama

which

is

equivalent

to

497
,
306

hectares
.

This

deforestation

is

localized

mostly

in

the

provinces

of

Panama

and

Darien

and

close

to

the

road
.


•
The

impact

occurs

mainly

in

the

direct

influence

area

of

the

road

(
0

to

10
km)
.

•
The

Darien

province

lost

24
%

of

its

forests,

and

Panama

23
%
.

Most

of

this

deforestation

occurred

in

Mixed

Cative

forest

in

order

to

create

new

cropland

areas
.

Santa Cruz
-
Puerto Suarez, Bolivia

Date

Construction: 2000 to 2011

2.4. Road Impact Results:
Santa Cruz
-
Puerto Suarez Corridor, Bolivia

Study Area

The

corridor

Santa

Cruz
-
Puerto

Suarez

is

located

in

the

South

East

of

Bolivia
.

Its

length

is

approximately

636
km

and

connects

the

towns

of

Santa

Cruz

de

la

Sierra

and

Puerto

Suarez

located

on

the

border

with

the

state

of

Mato

Grosso

do

Sul

in

Brazil
.

In

the

area

one

can

see

four

easily

distinguishable

types

of

ecoregions
:

Pantanal
,

Dry

Chaco,

Chiquitano

Dry

Forest

and

Cerrado

2.4. Road Impact Results:
Santa Cruz
-
Puerto Suarez Corridor, Bolivia

2.4. Road Impact Results:
Santa Cruz
-
Puerto Suarez Corridor, Bolivia

Road Impact

•
Road

still

under

construction
.

Some

direct

impacts

especially

close

to

Santa

Cruz
.

•
Major

indirect

impacts

of

fires

originating

from

“
slash

and

burn
”

practices,

especially

in

2010
.

2.4. Road Impact Results:
Santa Cruz
-
Puerto Suarez Corridor, Bolivia

Conclusions

•
Too

early

to

say

what

direct

impacts

are

until

road

fully

connects

Bolivia

with

Brazil

•
Nevertheless,

clear

indirect

impact

through

fires

originating

from

slash

and

burn

activity,

especially

in

the

region

of

Santa

Cruz

Trans
-
Chaco Highway, Paraguay

Date

Construction: 2002 to 2006

2.5. Road Impact Results:
The Trans
-
Chaco Highway, Paraguay

Study Area

The

trans
-
Chaco

highway

length

approximately

736
km,

extending

from

the

boundaries

between

Bolivia

and

Paraguay

near

the

military

post

Mayor

Infante

Rivarola

in

the

department

of

Boqueron

until

it

intersects

with

the

9
th

Road

which

runs

through

the

Dry

Chaco

in

Boqueron

continuing

through

the

department

of

Presidente

Hayes

across

the

humid

Chaco

region

up

to

the

Asuncion

metropolitan

area

in

the

Central

Department
.

2.5. Road Impact Results:
The Trans
-
Chaco Highway, Paraguay

Habitat Change Monitoring

Methodology

i.
Classify

the

Landsat
-
4

satellite

images

using

the

k
-
Means

Algorithm

.

ii.
Generated

the

deforestation

map

of

2000
-
2004
.

iii.
Applied

the

Terra
-
I

Methodology

to

monitoring

the

habitat

change

between

2004

to

2011
.

iv.
Analyze

the

road

impact

in

buffers

areas

in

10
,

20
,

30
,

40

and

50
km

of

the

road
.



MAIN

INPUTS



For

generated

deforestation

maps

between

2000

and

2004
:



D
ataset

from

the

high

spatial

resolution

satellite

Landsat

4

Thematic

Mapper

in

the

Dry

Chaco

ecoregion
.




For

generated

deforestation

maps

between

2004

and

2010
:

Terra
-
I

dataset
.


2.5. Road Impact Results:
The Trans
-
Chaco Highway, Paraguay

2.5. Road Impact Results:
The Trans
-
Chaco Highway, Paraguay

2.5. Road Impact Results:
The Trans
-
Chaco Highway, Paraguay

Road Impact

•
Total

of

650
,
000

Hectares

lost

in

50
km

buffer

since

2004

•
Massive

increase

since

2007

(project

completion)

•
Large

footprint,

covering

>
50
km

from

road


2.5. Road Impact Results:
The Trans
-
Chaco Highway, Paraguay

Conclusions

•
Very

high

levels

of

deforestation

pre
-

and

post
-

road

construction

•
But

>

300
%

increase

in

deforestation

rates

since

road

finished,

with

a

footprint

that

likely

goes

beyond

50
km

buffer

Methodologies:
Future Deforestation Scenarios

Steps…

1)
Train neural network to recognize pixels that are likely to be deforested

2)
Create maps of potential deforestation

3)
Select random pixels from the potential maps

4)
Repeat 2 and 3

INPUTS



The distance to the nearest road,

to the nearest river,

to the nearest city (> 1000 people)



The elevation



The base map (what was already deforested before the
analysis with terra
-
I (from human classification of the
terra
-
i

clusters))



The detection from the terra
-
I model.

This methodology

is still
under

development, nevertheless its
implementation as proof of concept
in Brazil generated good results in
areas with clear patterns of
deforestation.

Currently, the tool only takes into account topographic variables, but the idea is in the future is
to include others inputs such as administrative information (protected areas, country…), social
information (small farmers, industrial exploitation, community managed forest…) and
ecosystems in order to improve the estimation.

3. Future Deforestation Scenarios:
Applied in BR
-
364 Road,
Brasil


Applied




The

study

area

was

the

area

around

the

road

Rio

Branco
-
Porto

Velho,

in

Brazil

from

the

1
st

of

January

2004

to

the

10
th

of

June

2011
.



Constant

rate

to

10
’
000

hectares

per

16

days

period

(the

average

rate

recorded

by

Terra
-
i

in

this

area)




Sampled

10
’
000

pixels

to

train

the

neural

network
.

(
7000

unchanged

and

3000

deforested)


This first implementation of this methodology gave encouraging results as by
comparing this result with the actual detected deforestation one can see that the
general patterns resulting from the simulation are convincing and quite similar to the
real events. However, various improvements could be instigated.


INPUTS



The distance to the nearest road,

to the nearest river,

to the nearest city (> 1000 people)



The elevation



The base map (what was already deforested before the
analysis with terra
-
I (from human classification of the
terra
-
i

clusters))



The detection from the terra
-
I model.

Potential deforestation at T=0

Potential deforestation at
T=150

Predicted deforestation

Actual
deforestation (Terra
-
i
)

Base map

PROOF OF CONCEPT

3. Future Deforestation Scenarios:
Applied in BR
-
364 Road,
Brasil

4. Carbon
-
Conservation Interface

Methodology

4. Carbon
-
Conservation Interface

Inputs



Critical

Natural

Habitat

(species,

Protected

Areas,

Terrestrial

Ecosystems)



Natural

Habitat

(terrestrial

Ecosystems)



Other

Conservation

Value

(CI

Hotspots,

TNC

Portfolio

Sites)



Carbon

4. Carbon
-
Conservation Interface

Output

Final

Index

Value

Combined

Carbon

and

Biodiversity
.

A synthesis of findings





•
Roads clearly a significant driver of deforestation
and land
-
use change, demonstrated in all 5
projects studies. Impacts are both direct and
indirect.

•
A road makes a different “footprint” depending
on the ecosystem (Andes <10km, Amazon ~50km,
Chaco >50km).

Lessons learned

•
Other factors such as secondary roads and rivers are
important drivers of habitat change and roads open access to
them.

•
As a Monitoring Tool Terra
-
I is only useful to analyze projects
after 2004. In the cases of Acre
-
Rondonia
, Peru and Paraguay
we had three cases where the full power of Terra
-
I could be
shown.

•
Local, national and international policies are clearly important
contexts that should be taken into account during and after
road construction as they have a clear link to land
-
use change
and can contribute to mitigation or exacerbation of the road
project
encironmental

impact.



Policy Recommendations

•
Regional

and

national

environmental

policies

in

place

can

significantly

reduce

the

number

of

hectares

deforested

during

and

after

the

road

construction

project
.

The

most

outstanding

case

can

be

found

in

Brazil

where

Rondonia

has

higher

deforestation

rates

compared

to

Acre
.


•
Most

of

the

protected

areas

affected

directly

or

indirectly

by

the

road

construction,

were

established

after

the

roads

where

built
.

In

cases

were

critical

ecosystems

are

endangered,

policy

makers

and

development

planners

should

consider

for

the

future,

well

in

advance,

critical

natural

habitats

for

conservation

and

biodiversity

hotspots
.


•
It’s

expected

that

infrastructure

allows

the

expansion

of

economic

activities
.

In

this

sense,

national

and

regional

policies

and

incentives

to

promote

sustainable

and

environmentally

friendly

agricultural

practices

are

also

important
.

In

the

case

of

the

slash

and

burn

method

in

Bolivia

and

Peru

causing

multiple

forest

fires,

more

national

policies

and

programs

to

promote

more

sustainable

practices

should

be

in

place,

such

as

the

Slash

and

Mulch

Agroforestry

Systems
.

It’s

also

key

to

have

increased

productivity
.

Land

policy

and

law

enforcement

are

also

relevant

in

terms

of

reducing

the

negative

environmental

impact

of

road

infrastructure
.





Thank you!

Acknowledgements


This

Consultancy

Project

was

conducted

by

the

International

Center

for

Tropical

Agriculture

(CIAT),

the

Nature

Conservancy

(TNC)
,

and

the

Conservation

Biology

Institute

(CBI)

for

the

Environmental

and

Social

Safeguards

Unit

of

the

Inter
-
American

Development

Bank
.

This

project

was

supported

with

funds

from

the

German

Federal

Bundesministerium

fuer

wirtschaftliche

Zusammenarbeit

und

Entwicklung

(BMZ)

in

the

framework

of

a

cooperation

program

between

the

Inter
-
American

Development

Bank

(IDB)

and

the

Deutsche

Gesellschaft

fuer

Internationale

Zusammenarbeit

(GIZ)
.