Content uploaded by Jorge Lozano
Author content
All content in this area was uploaded by Jorge Lozano on Jan 26, 2018
Content may be subject to copyright.
Land
Use
Policy
47
(2015)
130–144
Contents
lists
available
at
ScienceDirect
Land
Use
Policy
jo
ur
nal
ho
me
pag
e:
www.elsevier.com/locate/landusepol
Forgetting
fire:
Traditional
fire
knowledge
in
two
chestnut
forest
ecosystems
of
the
Iberian
Peninsula
and
its
implications
for
European
fire
management
policy
Francisco
Seijoa,∗,
James
D.A.
Millingtonb,
Robert
Grayc,
Verónica
Sanza,b,c,
Jorge
Lozanod,
Francisco
García-Serranoe,
Gabriel
Sangüesa-Barredaf,
Jesús
Julio
Camarerof
aMiddlebury
College
C.V.
Starr
School
in
Spain,
Madrid,
Spain
bDepartment
of
Geography,
King’s
College
London,
London,
UK
cRW
Gray
Consulting
Ltd,
Chilliwack,
BC,
Canada
dDepartamento
de
Ciencias
Naturales,
Sección
de
Biología
Básica
y
Aplicada,
Universidad
Técnica
Particular
de
Loja,
San
Cayetano
Alto,
C/París
s/n.,
Loja
1101608,
Ecuador
eSaint
Louis
University,
Madrid,
Spain
fInstituto
Pirenaico
de
Ecologia-CSICAvda.
Monta˜
nana,
1005.
50059
Zaragoza,
Spain
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
6
October
2014
Received
in
revised
form
25
February
2015
Accepted
14
March
2015
Keywords:
Fire
exclusion
policies
Traditional
ecological
knowledge
Traditional
fire
knowledge
Chestnut
forest
ecosystems
Fire
Paradox
a
b
s
t
r
a
c
t
Human
beings
have
used
fire
as
an
ecosystem
management
tool
for
thousands
of
years.
In
the
context
of
the
scientific
and
policy
debate
surrounding
potential
climate
change
adaptation
and
mitigation
strate-
gies,
the
importance
of
the
impact
of
relatively
recent
state
fire-exclusion
policies
on
fire
regimes
has
been
debated.
To
provide
empirical
evidence
to
this
ongoing
debate
we
examine
the
impacts
of
state
fire-exclusion
policies
in
the
chestnut
forest
ecosystems
of
two
geographically
neighbouring
municipal-
ities
in
central
Spain,
Casillas
and
Rozas
de
Puerto
Real.
Extending
the
concept
of
‘Traditional
Ecological
Knowledge’
to
include
the
use
of
fire
as
a
management
tool
as
‘Traditional
Fire
Knowledge’
(TFK),
we
take
a
mixed-methods
and
interdisciplinary
approach
to
argue
that
currently
observed
differences
between
the
municipalities
are
useful
for
considering
the
characteristics
of
“pre-industrial
anthropogenic
fire
regimes”
and
their
impact
on
chestnut
forest
ecosystems.
We
do
this
by
examining
how
responses
from
interviews
and
questionnaire
surveys
of
local
inhabitants
about
TFK
in
the
past
and
present
correspond
to
the
current
biophysical
landscape
state
and
recent
fire
activity
(based
on
data
from
dendrochrono-
logical
analysis,
aerial
photography
and
official
fire
statistics).
We
then
discuss
the
broader
implications
of
TFK
decline
for
future
fire
management
policies
across
Europe
particularly
in
light
of
the
published
results
of
the
EU
sponsored
Fire
Paradox
research
project.
In
locations
where
TFK-based
“pre-industrial
anthropogenic
fire
regimes”
still
exist,
ecosystem
management
strategies
for
adaptation
and
mitigation
to
climate
change
could
be
conceivably
implemented
at
a
minimal
economic
and
political
cost
to
the
state
by
local
communities
that
have
both
the
TFK
and
the
adequate
social,
economic
and
cultural
incentives
to
use
it.
©
2015
Elsevier
Ltd.
All
rights
reserved.
“This
universe,
which
is
the
same
for
all,
has
not
been
made
by
any
god
or
man,
but
it
always
has
been,
is,
and
will
be
an
ever-
living
fire,
kindling
itself
by
regular
measures
and
going
out
by
regular
measures”.
Heraclitus
∗Corresponding
author.
Tel.:
+34
913071825.
E-mail
address:
fseijo@middlebury.edu
(F.
Seijo).
Introduction
Contemporary
ecology
understands
that
many
ecosystems
can
be
considered
“fire
adapted”,
their
structure
and
function
being
partly
determined
by
the
fire
regimes
with
which
they
have
co-evolved
(Gill,
2002;
Pausas
and
Keeley,
2009).
Humans
have
often
played
a
long-standing
role
in
such
fire-adapted
ecosystems.
Anthropogenic
landscape
burning
is
believed
to
have
had
a
signif-
icant
ecological
impact
on
the
Earth
system
for
thousands
of
years
and
in
many
flammable
ecosystems
worldwide
it
has
become
a
key
ecological
process
conditioning
present
biodiversity
and
climate
http://dx.doi.org/10.1016/j.landusepol.2015.03.006
0264-8377/©
2015
Elsevier
Ltd.
All
rights
reserved.
F.
Seijo
et
al.
/
Land
Use
Policy
47
(2015)
130–144
131
(Stewart,
1957;
Ruddiman,
2003;
Bowman
et
al.,
2010).
However,
biomass
burning
has
also
been
recently
determined
to
be
an
impor-
tant
contributor
to
the
global
greenhouse
gas
emissions
causing
climate
change,
though
the
impact
of
anthropogenic
fires
on
net
emission
outputs
is
highly
uncertain
and
remains
widely
debated
within
the
scientific
community
(Hurteau
et
al.,
2008;
Fule,
2008;
Pausas
and
Fernández-Mu˜
noz,
2012;
Moritz
et
al.,
2013;
IPCC
AR5,
2014;
Gill
et
al.,
2014).
In
the
context
of
the
scientific
and
policy
debate
surround-
ing
potential
climate
change
adaptation
and
mitigation
strategies,
researchers
continue
to
discuss
the
importance
of
the
impact
of
state
fire
exclusion
policies
on
fire
regimes.
Fire
exclusion
poli-
cies
have
been
defined
as
the
attempt
to
exclude
all
types
of
landscape
fires
from
a
specified
area
(Scott,
2015).
One
of
the
first
contemporary
large-scale
attempts
at
implementing
a
state-
wide
fire
exclusion
policy
was
carried
out
by
the
United
States
throughout
the
20th
century
prompting
other
countries
receiv-
ing
its
technical
advice
and
forestry
aid
funds,
such
as
Spain,
to
follow
suit
(Donovan
and
Brown,
2007;
Seijo
and
Gray,
2012).
At
the
end
of
the
20th
century
it
has
become
apparent
that
the
effects
of
fire
exclusion
policies
on
fire
regimes
may
be
provoking
what
some
researchers
have
termed
a
“firefighting
trap”
(Collins
et
al.,
2013).
By
altering
historical
fire
regimes
and
landscape
fuel
structures,
state
fire
exclusion
policies
may
well
be
contributing
to
contemporary
“megafires”
that
seem
to
positively
feedback
with
anthropogenic
climate
change
as
well
as
spiralling
fire
suppression
costs
in
many
countries
(Millar
et
al.,
2007;
Seijo
and
Gray,
2012;
Pezzatti
et
al.,
2013;
Stephens
et
al.,
2014;
Fernandes
et
al.,
2014).
In
this
volatile
and
uncertain
scenario,
theoretical
concepts
such
as
“applied
historical
ecology”
and
“pre-industrial
anthropogenic
fire
regimes”
have
been
advanced
in
an
attempt
to
come
to
terms
with
the
role
that
historical
fire
patterns
(in
general)
and
traditional
anthropogenic
fire
practices
(in
particular)
should
or
should
not
play
as
a
baseline
for
informing
future
fire
management
decisions
(Swetnam
et
al.,
1999;
Keane
et
al.,
2009;
Seijo
and
Gray,
2012;
Pezzatti
et
al.,
2013;
Gill
et
al.,
2014;
Petty
et
al.,
2015).
Traditional
ecosystem
management
practices
are
reliant
on
“Traditional
Ecological
Knowledge”
(TEK),
defined
as,
“the
cumu-
lative
body
of
knowledge,
practice,
and
belief,
evolving
by
adaptive
processes
and
handed
down
in
generations
by
cultural
transmis-
sion,
about
relationships
of
living
beings
[including
humans]
with
one
another
and
with
their
environment”
(Berkes
et
al.,
2000:8).
A
variant
of
TEK
but
with
particular
regard
to
the
use
of
fire
as
a
man-
agement
tool,
“Traditional
Fire
Knowledge”
(TFK)
has
more
recently
been
defined
as,
“fire-related
knowledge,
beliefs,
and
practices
that
have
been
developed
and
applied
on
specific
landscapes
for
specific
purposes
by
long
time
inhabitants”
(Huffman,
2013:1).
Decline
of
the
use
of
TEK
and
TFK
can
lead
to
significant
changes
in
ecosystems.
Some
of
the
first
descriptions
of
these
impacts
appear
in
Omer
C.
Stewart’s
collection
of
essays
“Forgotten
Fires”
(Stewart,
1957),
which
has
inspired
this
article’s
title.
In
his
1950s
pioneering
work,
Stewart
identified
diverse
TFK-based
fire
uses
by
native
Americans
for
ecosystem
management
and
described
some
of
the
ecological
consequences
that
emerged
when
these
practices
were
“forgotten”.
Today,
the
gradual
abandonment
of
tra-
ditional
land
use
–
resulting
in
a
loss
of
both
TEK
and
TFK
–
has
been
recognized
as
one
of
the
main
structural
factors
leading
to
the
emergence
of
so-called
“Large
Wildland
Fires”
across
Mediter-
ranean
Type
Ecosystems
in
Europe
(Seijo
and
Gray,
2012;
Galiana
et
al.,
2013;
Pezzatti
et
al.,
2013;
Montiel,
2013;
Stephens
et
al.,
2014;
Fernandes
et
al.,
2014).
Much
of
the
literature
now
acknowl-
edges
that
socio-economic
and
political
drivers
are
at
the
core
of
this
change
(Seijo,
2005;
Seijo
and
Gray,
2012;
Pezzatti
et
al.,
2013;
Fernandes
et
al.,
2014).
However,
little
attention
has
been
paid
to
the
exact
mechanisms
by
which
state
fire
exclusion
policies
–
which
have
been
shown
to
be
ecologically,
economically
and
politically
undesirable
(Seijo
and
Gray,
2012;
Montiel,
2013)
–
and
rural
devel-
opment
policies
have
impacted
TEK
and
TFK.
For
example,
in
the
Iberian
Peninsula
these
policies
often
set
the
stage
to
the
enclosure
of
large
tracts
of
land
for
new
industrial
era
uses
(e.g.
afforesta-
tions,
conservation
areas,
recreational
hunting
estates,
etc.)
and
the
prohibition
of
traditional
land
use
practices
such
as
extensive
ani-
mal
husbandry
and
swidden
agriculture
(Fernandes
et
al.,
2014)
These
changes
shifted
rural
economies
away
from
approaches
that
required
the
use
of
TFK-based
practices
and
therefore
contributed
to
rural
abandonment
(Seijo
and
Gray,
2012;
Stephens
et
al.,
2014).
It
is
important,
therefore,
to
re-evaluate
the
fire
management
potential
of
TFK-based
practices,
particularly
since
continent-wide
European
Union
funded
research
projects
such
as
Fire
Paradox
are
calling
for
a
reform
of
present
fire
suppression
based
management
strategies
and
advocating
for
the
promulgation
of
new
European
legislation
on
the
matter
in
the
form
of
a
“Fire
Framework
Directive”
(Montiel,
2013).
As
an
evidence-based
contribution
to
this
ongoing
debate,
in
this
study
we
examine
the
current
biophysical
attributes
of
two
adjacent
sweet
chestnut
forest
ecosystems
of
the
Iberian
Peninsula
and
local
inhabitants’
perspectives
on
pre-industrial
anthropogenic
burning
within
them.
The
present
existence
of
chestnut
forest
ecosystems
throughout
Europe
was
only
made
possible
by
cen-
turies
of
intense
management
by
local
communities
(Conedera
et
al.,
2004;
Conedera
and
Krebs,
2009).
In
fact,
the
chestnut
for-
est
ecosystems
of
the
study
sites
we
consider
in
this
paper
–
the
municipalities
of
Casillas
and
Rozas
de
Puerto
Real
in
the
foothills
of
the
mountains
of
Gredos,
central
Spain
–
can
be
theoretically
described
as
coupled
human–natural
systems
because
of
the
histor-
ically
verified,
prolonged
and
intense
interaction
between
human
and
natural
system
variables
in
them
(Liu
et
al.,
2007).
Commu-
nities
in
this
region
have
managed
their
chestnut
forests
with
a
sophisticated
ecosystem
management
toolkit
that
exemplifies
TEK
and
TFK.
Through
time
these
communities
actively
participated
in
the
design
of
their
chestnut
forest
ecosystems
through
terracing,
grafting,
pruning,
careful
tree
species
selection
and
burning
in
what
can
be
most
aptly
described
as
a
pre-industrial
effort
at
large-scale
environmental
engineering
(Martin
et
al.,
2010).
Changes
in
the
use
of
TEK
in
coupled
human–natural
chest-
nut
forest
ecosystem
management
have
been
known
to
result
in
substantial
transformations
in
both
their
structure
and
function
as
natural
succession
processes
resume
unaltered
(e.g.
Mazzoleni
et
al.,
2004;
Romero-Calcerrada
and
Perry,
2004;
Millington
et
al.,
2007;
Millington
et
al.,
2009).
Such
change
in
forest
stands
for-
merly
dominated
by
chestnut
trees
has
been
observed
in
Corsica,
for
example,
with
the
encroachment
of
mixed
and
closed-canopy
stands
dominated
by
Holm
oak
(Quercus
ilex
L.)
and
Cluster
pine
(Pinus
pinaster
Ait.)
following
abandonment
(San
Roman
et
al.,
2013).
In
Bulgaria,
in
the
absence
of
traditional
management,
chest-
nut
forests
have
apparently
become
increasingly
vulnerable
to
pest
disturbances
such
as
chestnut
blight
(Zlatanov
et
al.,
2013),
and
in
Switzerland
the
loss
of
ecologically
valuable
old
growth
“giant”
chestnut
trees
is
feared
–
as
well
as
the
emergence
of
significant
fire
regime
changes
–
as
the
anthropogenic
silvicultural
practices
of
the
past
fade
away
(Krebs
et
al.,
2012;
Pezzatti
et
al.,
2013).
In
an
effort
to
restore
ecosystem
structure
and
process
in
aban-
doned
chestnut
forest
ecosystems
a
debate
is
thereby
emerging
concerning
the
appropriate
role
of
traditional
pre-industrial
era
burning
in
the
ecological
restoration
of
these
ecosystems.
Some
researchers
advocate
for
continued
use
of
TFK-based
practices
or
surrogate
prescribed
burning
(Grove
and
Rackham,
2003;
Seijo
and
Gray,
2012;
Fernandes
et
al.,
2013)
in
contrast
to
others
who
argue
it
should
be
limited
to
certain
specific
sites
where
the
goal
is
the
conservation
of
locally
endangered
species
associated
via
coppiced
(or
abandoned)
chestnut
stand
communities
(Grund
et
al.,
2005;
Moretti
et
al.,
2006,
2008;
Pezzatti
et
al.,
2013).
132
F.
Seijo
et
al.
/
Land
Use
Policy
47
(2015)
130–144
Fig.
1.
Study
area
location.
Casillas
is
located
within
Comunidad
de
Castilla
y
Leon
and
Rozas
within
Comunidad
de
Madrid,
both
in
central
Spain.
Here,
we
take
a
mixed-methods
and
interdisciplinary
approach
to
explore
these
issues,
integrating
both
quantitative
and
quali-
tative
methods,
to
capture
and
understand
the
complexities
and
feedbacks
between
coupled
human
and
natural
systems
in
our
sites
(Bryman,
2006;
Liu
et
al.,
2007).
We
compare
contemporary
fire
regimes,
landscape
characteristics
and
use
of
TEK-based
land-
scape
burning
practices
in
two
Spanish
municipalities
and
argue
that
currently
observed
differences
are
useful
for
considering
the
characteristics
of
“pre-industrial
anthropogenic
fire
regimes”
and
their
impact
on
chestnut
forest
ecosystems.
We
do
this
by
exam-
ining
how
responses
from
interviews
and
questionnaire
surveys
of
local
inhabitants
about
TFK
in
the
past
and
present
correspond
to
the
current
biophysical
landscape
state
and
recent
fire
activity.
We
then
discuss
the
broader
implications
of
TFK
decline
for
fire
management
policies
across
Europe.
Methods
Study
species
and
site
selection
The
sweet
chestnut
(Castanea
sativa
Mill.)
is
a
deciduous,
hard-
wood
angiosperm
tree
species
belonging
to
the
Fagaceae
family.
It
has
been
widely
cultivated
throughout
the
temperate
world,
particularly
across
the
Mediterranean
Basin
in
areas
with
abun-
dant
precipitation
and
its
geographical
range
is
closely
associated
with
the
activities
of
pre-industrial
traditional
agrarian
societies
(Conedera
et
al.,
2004).
Currently,
sweet
chestnut
forests
are
mainly
concentrated
in
southern
Europe
(France,
Italy,
Spain,
Portugal
and
Switzerland)
and
Turkey
where
there
is
a
long
tradition
of
their
cul-
tivation
as
groves
for
nuts
and
wood
production
(Conedera
et
al.,
2004).
The
species
itself
seems
to
be
native
to
the
Iberian
Penin-
sula
with
pre-Holocene
glacial
refugia
having
been
identified
in
primary
and
secondary
foci
located
in
N
and
W
Spain
and
N
Portugal
regions
(Krebs
et
al.,
2004;
Postigo-Mijarra
et
al.,
2010).
The
chest-
nut
forest
ecosystems
of
our
study
area
are
located
in
the
foothills
of
the
mountains
of
Gredos
(central
Spain).
These
forests,
in
all
like-
lihood,
originated
in
the
13th
and
14th
centuries
as
a
result
of
the
anthropogenic
diffusion
of
sweet
chestnut
after
the
“Reconquista”
(Reconquest)
and
the
subsequent
“repoblación”
(re-population)
of
these
territories
with
social
groups
originating
from
Northern
Spain,
though
further
palynologic
and
genetic
research
would
be
needed
to
confirm
this
hypothesis.
Our
data
collection
was
conducted
in
the
municipalities
of
Casil-
las,
autonomous
community
of
Castilla
y
León,
and
Rozas
de
Puerto
Real
(hereafter
abbreviated
as
Rozas),
autonomous
community
of
Madrid
(Fig.
1)
from
June
1st,
2012
to
May
31st,
2013.
These
munic-
ipalities
are
contiguous
geographically
but
separated
by
a
political
boundary
between
autonomous
communities
in
the
Spanish
state
(Fig.
1).
Site
selection
was
informed
by
a
previously
articulated
conceptualization
of
“pre-industrial
anthropogenic
fire
regimes”
in
Mediterranean
Type
Ecosystems
(MTEs),
which
hypothesizes
that
uneven
economic
and
political
development
processes
have
driven,
to
varying
degrees,
the
fire
regime
changes
taking
place
at
present
in
many
MTEs
(Seijo
and
Gray,
2012).
Casillas
and
Rozas
have
differing
economic
conditions
(Table
1),
but
relatively
similar
mountainous
MTEs.
Casillas
has
lower
family
income
and
munic-
ipality
expenditure
than
Rozas,
with
a
smaller
proportion
of
the
local
population
employed
in
the
service
sector
and
a
greater
pro-
portion
in
agriculture.
By
selecting
these
sites
we
have
endeavoured
to
ensure
that
the
biophysical
variables
underlying
any
differences
in
fire
regimes
could
be
held
(as
far
as
possible
in
a
natural
non-
laboratory
setting)
constant
so
as
to
better
highlight
the
ecological
effects
of
different
fire
management
recently.
This
is
a
proven
methodological
approach
that
has
yielded
interesting
findings
on
changing
fire
regimes
in
other
MTEs
(Minnich,
1983).
Biophysical
data
To
assess
the
current
biophysical
state
of
the
municipalities’
forests
and
their
fire
regimes
we
used
National
Forest
Inventory
data
(IFN3,
2007),
fire
reports
from
autonomous
communities,
dendrochronological
sampling
and
military
aerial
photographs
from
2011.
Individual
fire
reports
for
the
two
municipalities
were
received
from
the
regional
autonomous
governments
of
Madrid
and
Castilla
y
Leon
for
the
period
1984–2012.
In
this
study,
we
have
only
quantified
the
fire
regime
attributes
that
can
be
inferred
from
official
Spanish
government
fire
statistics
for
the
selected
sites.
These
include
information
on
fire
regime
characteristics,
F.
Seijo
et
al.
/
Land
Use
Policy
47
(2015)
130–144
133
Table
1
Development
level
indicators
for
Casillas
and
Rozas.
Income
levels,
employment
sectors
and
other
indicators
were
compiled
to
determine
the
uneven
stages
of
development
present
in
both
study
sites.
Indicator
Casillas
Rozas
Year
Source
Territory
(km2) 11.96 30.15 2011
Caja
Espa˜
na
Population
(no.
of
inhabitants)
841
466
2011
Caja
Espa˜
na
GDP
per
capita
(%
of
EU
27
average)*98.80
135.80
2012
Eurostat
Family
income
(D
per
capita)
16,290
23,929
2013
AIS
Municipality
expenditure
(D
per
capita)
1245
3442
2011
Caja
Espa˜
na
Municipal
property
tax
(D
per
capita)
181
237
2011
Caja
Espa˜
na
Occupational
sectors
(%
of
population)
2011
Caja
Espa˜
na
Agriculture
6.2 2.1
Industry
0
1
Construction
65.2
19.6
Services
28.6
77.3
Unemployed
24
16.2
Regional
government
fire
suppression
costs
(D
per
hectare
of
territorial
jurisdiction)
14
75
2006
ASEMFO
*Data
are
for
the
respective
autonomous
communities.
frequency,
size,
season
and
causality.
Comparable
figures
for
both
municipalities
were
only
available
for
the
period
1984–2009.
Aerial
photographs
were
received
from
the
“Centro
Geografico
del
Ejercito”
of
the
Spanish
army.
We
used
dendrochronology
to
estimate
the
age
of
chestnut
stems
apparently
not
affected
by
recent
fires.
We
took
cores
at
1.3
m
from
dominant
trees
(n
=
11
in
Rozas,
n
=
18
in
Casillas)
randomly
selected
in
each
site
and
not
presenting
visible
‘catfaces’
(cavities
at
the
foot
of
the
trunk
with
darkened
burn
markings),
using
a
Pressler
increment
borer.
We
sampled
non-fire
scarred
trees
distributed
as
broadly
as
possible
across
each
of
the
two
study
areas.
All
wood
samples
were
air
dried,
sanded
using
several
papers
of
successively
finer
grains
until
tree-rings
were
clearly
visible
and
then
visually
cross-dated.
The
sampling
area
was
approximately
from
1
to
3
ha
in
each
study
site.
We
interpreted
forest
stand
structure
(i.e.
primarily
canopy
cover)
in
the
two
selected
study
sites/municipalities
using
his-
torical
aerial
photographs
(see
Fig.
3).
Aerial
photographs
were
obtained
from
the
Spanish
army’s
geographical
services
and
included
color
photographs
for
2011
for
both
Rozas
and
Casillas.
To
analyze
the
data,
we
placed
a
grid
over
the
existing
cartography
for
the
two
municipalities
and
numbered
each
grid
cell
(Nowak
et
al.,
1996).
Using
a
random
number
generator
74
grid
cells
(plots)
were
selected
for
analysis
(Rozas
n
=
27;
Casillas
n
=
47).
Only
complete
grid
cells
were
analyzed
while
plots
crossing
municipal
boundaries
were
discarded.
Grid
cells
including
buildings,
roads,
orchards
and
other
man-made
infrastructures
were
also
discarded.
Grid
cells
for
the
municipality
of
Casillas
covered
5.2
ha
while
grid
cells
for
Rozas
covered
7.3
ha.
To
visually
estimate
foliage
cover
in
both
sites
we
used
a
standardized
comparison
chart
in
order
to
determine
the
canopy
cover
for
each
grid
cell.
To
describe
the
structure
of
each
grid
cell,
we
developed
a
stand
structure
code
with
the
following
characterizations:
no
canopy;
open,
mixed-canopy
stand;
closed
mixed-canopy
stand;
closed
and
small-canopy
stand;
open
and
small-canopy
stand;
open
and
large-canopy
stand;
and
closed
and
large-canopy
stand
(see
Fig.
3).
Human
activity
and
perceptions
data
Between
September
2012
and
May
2013
we
interviewed
and
carried
out
a
survey
among
54
randomly
selected
respondents
in
the
municipalities
(n
=
29
Casillas;
n
=
25
Rozas).
The
respondents
were
asked
to
identify
their
individual
preferences
to
a
series
of
questions
on
a
Likert
scale
while,
simultaneously,
the
qualitative
justifications
for
these
responses
were
recorded
by
the
interview-
ers.
The
survey
itself
was
divided
into
four
sections:
land
use,
land
tenure,
fire
use
and
demographic
related
questions.
As
in
other
sim-
ilar
survey-
or
interview-based
studies
this
design
was
conditioned
by
the
research
questions
(Mistry,
1998;
Fernández-Giménez
and
Fillat,
2012).
The
surveys/interviews
were
administered
and
com-
pleted
face-to-face
with
respondents
by
the
principal
investigator
and
other
collaborators
since,
as
has
been
the
case
in
other
research
studies
on
TEK
taking
place
in
Spain,
many
respondents
were
functionally
illiterate
or
had
problems
reading,
understanding
or
answering
the
written
questions
on
their
own
(Otero-Rozas
et
al.,
2013).
The
questionnaire
itself
contained
83
semantic
differential
scale
and
rank
order
type
questions,
38
multiple
choice
and
13
open-ended
questions.
Questions
examined
in
this
paper
refer
to
inhabitants’
per-
spectives
on
fire
causes,
fire
sizes,
attitudes
towards
fire
and
fire
use
(see
Table
2).
To
test
for
possible
differences
in
answers
between
groups
of
respondents
(e.g.
between
municipalities),
Fig.
2.
“Traditional”
vs
“abandoned”
chestnut
forest
stand
structure.
Traditionally
man-
aged
chestnut
forest
groves
in
Casillas
(a,
photo
taken
26/11/2012)
Rozas
(b,
photo
taken
31/10/2013)
Note
how
trees
in
(a)
exhibit
an
open
stand
structure
compared
to
younger
saplings
re-sprouting
from
root
systems
in
(b).
134
F.
Seijo
et
al.
/
Land
Use
Policy
47
(2015)
130–144
Fig.
3.
Example
aerial
imagery
used
to
evaluate
landscape
structure
in
the
municipalities.
(a)
Open
large
canopy,
(b)
open
medium
canopy,
(c)
closed
small
canopy,
(d)
closed
mixed
canopy.
Summary
of
the
proportions
of
different
structure
types
in
the
landscapes
is
shown
in
Table
3.
non-parametric
Mann–Whitney
U
tests
were
used
to
compare
median
values
of
coded
responses
(all
results
are
reported
at
the
95%
confidence
level).
Results
Biophysical
characteristics
Forest
cover
and
land
use
results
Based
on
National
Forest
Inventory
data
(IFN3,
2007),
forests
within
the
municipal
territory
of
Rozas
de
Puerto
Real
are
com-
posed
mainly
of
six
tree
species,
namely:
C.
sativa,
Quercus
pyrenaica
Willd.,
P.
pinaster,
Pinus
pinea
L.,
Quercus
ilex
and
Fraxinus
angusti-
folia
Vahl.
In
Rozas
40.7%
of
rural
lands
are
forested
while
55.8%
form
pastures
and
shrubland
(including
‘dehesa’).
A
‘dehesa’
is
an
extensive
area
which
is
generally,
but
not
always,
enclosed,
with
low
densities
of
old
growth
trees
that
allows
multiple
pas-
ture
and
arable
farming
practices
in
the
spaces
between
(e.g.
see
Millington
et
al.,
2007).
The
most
important
agricultural
crop
in
Rozas
is
wine
which
covers
3.2%
(Caja
Espa˜
na,
2012)
and
the
most
extensive
tree
species
is
C.
sativa
which
is
estimated
to
occupy
39%
of
Rozas
forested
surface
(442
ha).
In
Casillas,
IFN3
data
identifies
C.
sativa,
P.
pinaster,
Q.
pyrenaica,
Pinus
sylvestris
L.
and,
to
a
lesser
extent,
Pinus
nigra
J.
F.
Arn.
as
the
main
tree
species.
In
Casillas
52.9%
of
rural
lands
are
forested,
while
42.1%
are
pasture
and
shrubland
and
4.8%
are
agricultural
land.
The
chest-
nut
forests
of
Casillas
occupy
approximately
24%
of
all
forested
land
(155
ha),
though
this
figure
may
be,
in
fact,
larger
since
it
only
accounts
for
chestnut
trees
employed
for
nut
production.
Other
chestnuts
are
classified
as
timber
wood
and
may
consti-
tute
a
large
proportion
of
the
remaining
483
ha
of
the
forested
surface.
Chestnut
forest
groves
in
Rozas
and
Casillas
exhibit
subtly
dif-
ferent
stand
structures
as
can
be
appreciated
in
Fig.
2.
A
significant
proportion
of
chestnut
forest
patches
in
both
Rozas
and
Casillas
are
composed
of
young
saplings
re-sprouting
from
root
systems.
These
chestnut
trees
grow
in
closed
canopy,
coppiced
stand
forma-
tions.
Results
from
our
dendrochronological
sampling
indicate
that
Table
2
Semantic
differential
scale
questions.
Questions
were
grouped
by
topic
and
responses
coded
for
analysis
as
shown.
Topic
Question
Coding
Fire
cause
On
a
scale
from
‘none’
to
‘all’,
how
many
fires
were
started
in
this
municipality
by
the
following
causes?
[accidental/intentional/natural]
in
the
[past/present]
None
=
0
All
=
10
Fire
size
On
a
scale
from
‘small’
to
‘large’
how
large
in
area
were
the
biggest
fires
in
this
municipality
in
the
[past/present]?
Small
=
0
Large
=
10
Attitudes
to
fire
On
a
scale
from
‘strongly
disagree’
to
‘Strongly
Agree’,
how
do
you
agree
with
each
of
these
statements
about
fire?
[fires
are
.
.
.
bad/good/destructive/necessary/useful]
Strongly
disagree
=
0
Strongly
agree
=
10
Fire
Use
On
a
scale
from
‘Not
Useful’
to
the
‘Best
method’,
how
useful
is
fire
for
the
following
purposes?
[clearing
land
for
cultivation/clearing
land
for
livestock
pasture/improving
soil
for
cultivation/improving
livestock
pastures/eliminating
shrubs
and
weeds
underneath
or
near
useful
trees/eliminating
shrubs
and
weeds
in
understory
of
forests/improving
habitat
for
wildlife
(e.g.
rabbit,
birds)]
Not
useful
=
0
Best
method
=
10
F.
Seijo
et
al.
/
Land
Use
Policy
47
(2015)
130–144
135
Table
3
Landscape
cover
types
in
Casillas
and
Rozas.
Percentages
of
different
landscape
cover
types
derived
from
aerial
photo
analysis
(e.g.
Fig.
3)
for
2011
are
similar
between
municipalities
except
‘open,
no
canopy’
and
to
a
lesser
extent
‘closed,
small
canopy’.
Landscape
cover
type
Casillas
Rozas
ONC
(open,
no
canopy)
21
7
OLC
(open,
large
canopy)
17
26
OSC
(open,
small
canopy)
0
0
OMC
(open,
mixed
canopy)
15
15
CLC
(closed,
large
canopy)
0
0
CSC
(closed,
small
canopy) 13
19
CMC
(closed,
mixed
canopy) 34
33
tree
stems
in
chestnut
groves
are
on
average
younger
in
Rozas
than
in
Casillas,
with
median
age
64
years
and
96
years,
respectively.
Furthermore,
survey
respondents
claimed
that
the
young
chest-
nut
forests
of
Rozas
are
apparently
the
result
of
salvage
logging
implemented
after
the
“large
fire”
event
that
took
place
in
1985.
Many
chestnut
stems
in
Rozas,
therefore,
seem
to
be
of
the
same
age
cohort
though
there
are
also
disperse
patches
with
what
seem
like
older
chestnut
trees
and
younger
individual
saplings
seemingly
planted
from
seed
and
managed
for
timber
rather
than
chestnut
production.
In
Casillas
chestnut
groves
seem
to
be
slightly
older
overall
with
trees
growing
from
single
stems
though
younger
C.
sativa
and
Q.
pyrenaica
saplings
are
increasingly
encroaching
into
these
patch
types.
Many
of
Casillas’
older
chestnut
trees
are
reg-
ularly
pruned,
grafted
and
planted
in
front
of
old
terraces
to
both
contain
erosion
and
favour
chestnut
production.
Aerial
photo
analysis
(see
Fig.
3)
reveals
that
current
landscape
structure
in
the
municipalities
is
similar.
The
main
differences
that
do
exist
are
the
proportion
of
the
landscape
with
‘open,
no
canopy’
patches
(greater
in
Casillas),
and
‘open,
large
canopy’
patches
(greater
in
Rozas;
Table
3).
In
the
former
case
this
may
be
the
result
of
Casillas
greater
elevation,
slope
inclination
and
granite
outcrops
which
impede
trees
from
growing
in
some
areas.
In
terms
of
the
greater
abundance
of
‘open
large
canopy’
patch
types
in
Rozas
this
is
likely
the
consequence
of
the
presence
of
large
“dehesa”
type
Q.
ilex
patches
which
sustain
sizeable
cattle
and
horse
herds.
In
Casil-
las
livestock
is
composed
mainly
of
sheep
and
goats
with
some
cattle
grazing
on
the
treeless
mountainous
pastures
and
shrublands
during
the
summer
months.
Patches
of
C.
sativa
are
difficult,
if
not
impossible,
to
discrim-
inate
purely
from
an
aerial
photo
analysis
since
this
species
can
be
easily
confused
with
other
deciduous
species
common
to
both
municipalities
(such
as
Q.
pyrenaica,
F.
angustifolia)
and
even
from
evergreen
pine
species.
Structure
types
in
our
analysis
therefore
include
all
tree
species
types
mentioned
in
the
IFN3
data.
Although
the
aerial
photo
analysis
sheds
little
light
on
the
quantitative
extent
of
differently
managed
chestnut
forest
patches
in
both
municipal-
ities
it
does
help
us
interpret
differences
between
burned
area
and
fire
size
differences
in
the
municipalities
(see
the
“Individual
fire
reports”
section).
Individual
fire
reports
Official
statistics
for
1984–2009
indicate
that
76.9%
of
fires
in
Casillas
were
classified
as
“surface”
fires,
11.5%
as
“crown”
fires
and
11.5%
as
“mixed-severity”
fires.
In
Rozas
all
incidents
were
classified
as
“surface”
fires.
The
total
number
of
recorded
fire
incidents
during
1984–2009
in
Casillas
was
52
whereas
in
Rozas
it
was
31.
Fire
incidence
in
Casillas
peaked
in
1989,
1995
and
2009
with
8,
4
and
10
recorded
fire
incidents.
In
Rozas
fire
incidence
peaked
in
1989,
1999
and
2005
with
3,
3
and
4
recorded
fire
incidents.
Median
burned
areas
are
similar
(Rozas
0.41
ha,
Casillas
0.42
ha)
but
once
differences
in
territorial
size
are
accounted
for,
burnt
surface
per
year
was
larger
in
Rozas
than
in
Casillas
by
a
factor
of
10
(2.12
ha
km2yr−1compared
to
Fig.
4.
Comparison
of
frequency
of
burning
between
municipalities
by
season.
The
majority
of
fires
in
Casillas
occur
outside
summer
months
(characteristic
for
TFK
practices
in
this
area)
whereas
in
Rozas
the
majority
of
fires
occur
in
summer
months.
0.22
ha
km2yr−1).
Rozas
experienced
a
1257
ha
“large
fire”
(offi-
cial
statistical
definition:
>500
ha)
in
1985,
whereas
in
Casillas
the
largest
fire
during
the
study
period
burned
20.7
ha
in
1989.
In
both
municipalities
the
fire
season
peaked
in
the
summer
months
(JJA)
but
in
Casillas
fires
were
more
evenly
spread
out
through-
out
the
calendar
(Fig.
4).
In
Casillas
fires
in
spring
(MAM)
and
autumn
(SON)
months
accounted
for
a
greater
proportion
(53%)
of
fire
events
than
in
summer
months
(47%).
This
is
in
contrast
to
Rozas
where
summer
months
accounted
for
the
vast
majority
of
events
(71%)
with
spring
and
autumn
months
contributing
far
less
(29%).
Finally,
77%
of
fires
in
Casillas
had
a
verified
anthro-
pogenic
origin,
21%
were
caused
by
unknown
factors
and
2%
by
lightning.
In
Rozas
58%
had
an
unknown
origin,
39%
were
anthro-
pogenic
and
3%
were
ignited
by
lightning.
In
many
cases
fires
classified
as
“unknown”
in
both
municipalities
probably
had
an
anthropogenic
cause
but
if
forest
agents
are
unable
to
verify
that
this
is
the
case
they
must
classify
it
as
such
in
the
official
report.
Perceptions
and
traditional
fire
knowledge
Survey
on
perceptions
of
fire
There
is
a
statistically
significant
difference
between
municipal-
ities
regarding
perspectives
on
causes
of
past
fires,
but
not
for
fires
in
the
present
(Table
4).
Differences
seem
to
be
driven
mainly
by
views
of
inhabitants
of
Rozas,
many
of
whom
disagree
or
disagree
strongly
that
past
fires
were
accidental
and
agree
or
strongly
agree
that
past
fires
were
intentional
(Fig.
5a).
In
contrast,
opinion
on
the
importance
of
accidental
or
intentional
causes
of
past
fires
in
Casil-
las
is
divided.
Inhabitants
of
both
municipalities
predominantly
disagree
that
the
cause
of
past
fires
was
natural
(inhabitants
of
Rozas
holding
views
more
strongly
than
Casillas,
Fig.
5a).
Regarding
present
fires,
respondents
in
both
municipalities
predominantly
agree
that
most
fires
at
present
are