Content uploaded by Adria Lopez-Baucells
Author content
All content in this area was uploaded by Adria Lopez-Baucells on Oct 08, 2017
Content may be subject to copyright.
Mammalian
Biology
80
(2015)
237–245
Contents
lists
available
at
ScienceDirect
Mammalian
Biology
jou
rn
al
hom
epage:
www.elsevier.com/locate/mambio
Original
Investigation
Pest
control
service
provided
by
bats
in
Mediterranean
rice
paddies:
linking
agroecosystems
structure
to
ecological
functions
Xavier
Puig-Montserrata,d,
Ignasi
Torrea,
Adrià
López-Baucellsa,
Emilio
Guerrierib,
Maurilia
M.
Montib,
Ruth
Ràfols-Garcíaa,
Xavier
Ferrerc,
David
Gisbertc,
Carles
Flaquera,∗
aGranollers
Museum
of
Natural
Sciences,
Granollers,
Catalonia,
Spain
bInstitute
for
Sustainable
Plant
Protection,
the
National
Research
Council
of
Italy,
Portici,
Italy
cADV
(Crop
Defence
Association)
–
Agrupació
de
Defensa
Vegetal
de
l’arrós
i
altres
cultius
al
delta
de
l’Ebre,
Deltebre,
Catalonia,
Spain
dGalanthus
Association,
Celrà,
Catalonia,
Spain
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
28
July
2014
Accepted
27
March
2015
Handled
by
Danilo
Russo
Available
online
3
April
2015
Keywords:
Rice
paddies
Pipistrellus
pygmaeus
Chilo
supressalis
Ecosystem
services
Integrated
pest
management
a
b
s
t
r
a
c
t
Pest
control
through
integrated
pest
management
systems
stands
as
a
very
convenient
sustainable
hazard-free
alternative
to
pesticides,
which
are
a
growing
global
concern
if
overused.
The
ability
of
the
soprano
pipistrelle
bat
(Pipistrellus
pygmaeus)
to
control
the
rice
borer
moth
(Chilo
supressalis),
which
constitutes
a
major
pest
of
rice
around
the
world,
was
studied
in
the
Ebre
Delta,
Northeastern
Iberia.
Evidence
was
found
on
the
ability
of
this
particular
bat
species
to
control
borer
infestations:
(a)
the
moth
was
consumed
during
at
least
the
last
two
peaks
of
the
moth
activity,
when
most
crop
damage
is
done;
(b)
the
activity
of
bats
significantly
increased
with
moth
abundance
in
the
rice
paddies;
(c)
the
pest
levels
have
declined
in
the
study
area
(Buda
Island,
Eastern
Ebre
Delta)
after
the
deployment
of
bat
boxes
and
their
subsequent
occupation
by
soprano
pipistrelles.
The
value
of
the
ecosystem
service
provided
by
bats
was
estimated
at
a
minimum
of
21D
per
hectare,
equivalent
to
the
avoided
pesticide
expenditure
alone.
We
suggest
that
this
natural
service
can
be
enhanced
by
providing
bat
populations
with
artificial
roosts
in
rice
paddies
were
some
key
ecosystem
features
are
present.
©
2015
Deutsche
Gesellschaft
für
Säugetierkunde.
Published
by
Elsevier
GmbH.
All
rights
reserved.
Introduction
“Any
genuine
improvement
in
the
human
condition
on
this
planet
must
be
concerned
with
rice”
(Heinrichs
and
Miller,
1991).
In
recent
decades,
both
rice
scientists
and
farmers
have
gained
experience
in
the
cultivation
of
rice
and
there
has
been
a
shift
from
a
primar-
ily
unilateral
approach
to
insect
pest
control,
relying
strongly
on
insecticides,
to
a
multilateral
approach
involving
a
combination
of
control
tactics
(Heinrichs
and
Miller,
1991).
Scientists
throughout
the
world
strive
to
develop
and
implement
strategies
to
control
rice
pests
more
effectively
and
economically,
to
improve
crop
pro-
ductivity
and
consequently
the
welfare
of
human
populations.
According
to
Food
and
Agriculture
Organization
of
the
United
Nations
(FAO)
statistics,
the
global
production
of
rice
increased
linearly
from
216
million
tonnes
(mt)
in
1961
to
722
mt
in
2011
(FAO).
New
strategies
to
control
insect
pests
include
the
use
of
one
insecticide
per
one
specific
target
species
combined
with
biological
∗Corresponding
author.
E-mail
address:
cflaquer@ajuntament.granollers.cat
(C.
Flaquer).
methods
(such
as
the
use
of
parasitoids).
However,
the
adoption
of
non-chemical
approaches
to
pest
control
is
not
evenly
distributed
around
the
world
and
occurs
mainly
in
those
regions
where
legal
constraints
limit
the
number
of
approved
chemical
products
(c.
a.
European
Union).
Over-use
of
pesticides
is
still
an
issue
to
be
addressed
(Normile,
2013;
Peng
et
al.,
2009),
particularly
in
those
regions
in
which
the
vast
majority
of
the
world’s
rice
production
is
concentrated,
and
the
use
of
pesticides
keeps
growing
(FAO).
Fifty
percent
of
the
insecticides
used
in
rice
fields
in
Asia
target
lepidopteran
insects
(Heong
et
al.,
1994).
In
1991
it
was
estimated
that
an
average
global
annual
yield
loss
of
10
million
tonnes
was
caused
by
just
three
moths:
the
striped
rice
borer
(Chilo
suppres-
salis),
the
yellow
stem
borer
(Scirpophaga
incertulas)
and
the
leaf
folder
(Cnaphalocrocis
medinalis)
(Herdt,
1991).
The
striped
rice
borer
is
an
Asian
moth
currently
present
in
paddies
worldwide.
The
adults
lay
their
eggs
on
the
stems
and
leaves
of
rice
plants,
and
the
larvae
bore
the
stems
to
feed
on
the
internal
tissues,
compromising
both
plant
growth
and
productivity,
sometimes
fatally.
In
Europe
most
bats
are
insectivorous.
Bats
may
eat
up
to
80–100%
of
their
body
mass
in
insects
on
a
nightly
basis
(Kurta
et
al.,
1989),
and
during
the
last
decade
several
authors
have
drawn
http://dx.doi.org/10.1016/j.mambio.2015.03.008
1616-5047/©
2015
Deutsche
Gesellschaft
für
Säugetierkunde.
Published
by
Elsevier
GmbH.
All
rights
reserved.
238
X.
Puig-Montserrat
et
al.
/
Mammalian
Biology
80
(2015)
237–245
Fig.
1.
A
soprano
pipistrelle
(Pipistrellus
pygmaeus)
hunting
a
striped
rice
borer
(Chilo
supressalis),
in
the
Ebre
Delta.
Photographic
stroboscopic
effect
(series
of
short
or
instantaneous
samples).
Courtesy
of
Oriol
Massana
attention
to
the
important
contribution
that
bats
make
to
insect
pest
control
(Agosta
and
Morton,
2003;
Boyles
et
al.,
2011;
Cleveland
et
al.,
2006;
Ghanem
and
Voigt,
2012;
Kunz
et
al.,
2011;
Lee
and
McCracken,
2005;
Leelapaibul
et
al.,
2005;
McCracken
et
al.,
2012;
Whitaker,
1995;
Park,
2015).
The
referred
contribution
has
been
assessed
by
the
presence
of
pests
in
the
diet
of
wild
bat
populations
or
by
taking
a
step
further
and
accounting
for
the
economic
value
of
such
an
ecosystem
service.
Given
the
natural
complexity
of
ecological
systems
it
is
difficult
to
place
a
monetary
value
on
the
services
provided
by
bats,
a
fact
that
restricts
how
their
importance
is
understood
by
the
public
(Fisher
and
Turner,
2008).
Cleveland
et
al.
(2006)
estimated
the
economic
contribution
of
bats
to
the
cotton
dominated
agroecosystems
of
southern
Texas,
USA,
to
be
$12–$173
per
acre
each
year.
By
extrapolating
these
figures
to
the
whole
country,
Boyles
et
al.
(2011)
valued
bats’
economic
contribution
to
the
USA’s
agroecosystems
at
between
$3.7
and
$53
billion/year.
Even
if
the
actual
figures
were
lower,
given
the
evidence
gathered
so
far,
the
positive
impact
of
bats
on
this
aspect
of
the
economy
seems
to
be
beyond
dispute.
This
benefit
can
exceed
the
monetary
value
if
the
affected
crop
is
a
staple.
For
example,
the
pest
control
service
provided
by
wrinkle-lipped
bats’
(Tadarida
plicata)
is
responsible
for
securing
the
meals
of
26,152
(±15,817
SD)
people
each
year
in
Thailand
alone
(Wanger
et
al.,
2014).
All
the
aforementioned
authors
have
stressed
the
consequent
importance
of
protecting
bat
populations
if
the
ecological
service
they
provide
is
to
be
preserved.
Although
there
is
no
general
agreement
on
how
to
define
ecosys-
tem
services
(Wallace,
2008),
such
services
are
generally
regarded
as
ecosystem
outcomes
(e.g.
use
of
less
pesticides)
that
contribute
to
human
well-being
(Fisher
and
Turner,
2008;
Fisher
et
al.,
2009;
Nelson
et
al.,
2009;
Wallace,
2007).
Understanding
the
mechanisms
that
link
ecological
systems
to
human
well-being
is
a
fundamen-
tal
task
when
studying
ecosystem
services
(Haines-Young
and
Potschin,
2009).
From
a
functional
perspective,
the
human
benefit
(whether
monetary
or
not)
is
the
final
outcome
of
a
cascade
process
resulting
in
a
service
(Rollett
et
al.,
2008).
The
soprano
pipistrelle
(Pipistrellus
pygmaeus,
Fig.
1)
was
sepa-
rated
taxonomically
from
the
common
pipistrelle
bat
(Pipistrellus
pipistrellus)
on
1997
(Barratt
et
al.,
1997).
Previous
studies
on
diet
contain
mixed
data
on
diet
and
distribution
(e.g.
Vaughan,
1997).
Soprano
pipistrelle
is
a
common
European
bat
species
occurring
from
the
British
Isles
through
much
of
continental
Europe
(includ-
ing
the
islands
of
Corsica
and
Sardinia)
East
to
Western
Asia
Minor,
the
Caucasus
and
Siberia
(Dietz
et
al.,
2009).
It
is
more
abundant
in
lowland
areas
and
is
frequently
associated
with
freshwater
bodies
(rivers,
lakes,
wetlands,
etc.),
being
common
in
coastal
wetlands,
where
most
paddies
in
southern
Europe
occur.
The
few
unam-
biguous
studies
on
diet
indicate
that
they
feed
mainly
on
small
diptera,
though
they
include
a
wide
array
of
small-sized
aerial
insects
in
their
diet
and
is
regarded
as
an
opportunistic
species
(Bartonicka
et
al.,
2008b).
Following
our
success
in
improving
soprano
pipistrelle
populations
in
Northeastern
Iberia
in
a
mixed
landscape
with
both
wetlands
and
paddies
(Flaquer
et
al.,
2006),
we
investigated
the
ability
of
the
species
to
control
striped
rice
borer
populations.
Having
determined
the
potential
of
bats
to
control
rice
borer,
we
aimed
to
better
understand
under
which
environmental
conditions
soprano
pipistrelles’
populations
could
be
enhanced
to
further
promote
their
pest
control
ability.
Since
the
experiment
was
conducted
in
a
reduced
area,
we
describe
the
conditions
under
which
bat
populations
had
been
encouraged
in
the
rice
produc-
ing
landscape
of
study,
and
emphasise
the
economic
benefits
they
could
provide.
Although
there
is
a
long
tradition
of
rice-fish
cul-
ture
aimed
both
at
controlling
pests
and
enhance
food
production
(Xie
et
al.,
2011),
to
our
knowledge
no
other
wild
local
vertebrate
populations
have
been
artificially
manipulated
for
such
a
purpose
and
no
previous
research
exists
on
enhancing
bat
populations
as
a
pest
biological
control
method
in
rice
paddies.
Material
and
methods
Study
area
We
carried
out
the
study
within
the
Ebre
Delta
Natural
Park
(0◦50E,
40◦42N,
Fig.
2),
Catalonia,
NE
Iberia,
which
is
one
of
the
biggest
deltas
in
Europe
(320
km2).
Approximately
65
km2are
wetlands,
240
km2are
crops
(203
km2of
which
are
paddies)
and
16
km2are
urban
areas
that
host
around
5000
inhabitants.
Mean
annual
temperatures
range
between
17
and
18 ◦C,
mean
annual
precipitation
between
500
and
550
mm,
and
mean
annual
evapo-
transpiration
between
855
and
997
mm,
in
what
is
essentially
a
semi-arid
Mediterranean
climate,
with
a
pronounced
summer-
drought.
Prior
to
this
study,
in
1999,
69
small
wooden
bat
boxes,
either
with
single
or
double
compartment
(Flaquer
et
al.,
2006),
were
erected
in
Buda
Island
Natural
Reserve
(12.1
km2,
Fig.
2),
a
defor-
ested
area
lacking
suitable
vertical
structures
for
the
bats
to
roost
X.
Puig-Montserrat
et
al.
/
Mammalian
Biology
80
(2015)
237–245
239
Fig.
2.
Situation
of
the
study
area
and
the
sampling
stations:
Small
black
dots
indicate
the
87
independent
bat
detector
stations
(3–5
min
duration)
performed
from
July
to
September
in
2007,
2009,
and
2010,
and
large
grey
dots
represent
the
location
of
the
striped
rice
borer
light-traps
monitored
from
May
to
September
2007–2010.
Shaded
areas
represent
urban
habitats.
in
(Flaquer
et
al.,
2005).
Bat
boxes
were
rapidly
occupied
(243
individuals
were
counted
one
year
after
boxes
were
deployed)
by
an
increasing
number
of
individuals,
reaching
a
maximum
of
approx-
imately
3500
individuals
in
2008
(Flaquer
et
al.,
2006).
Pest
control
systems
Following
Cleveland
et
al.
(2006)
we
can
estimate
the
value
of
the
service
provided
by
bats
in
two
ways:
1/assessing
the
economic
expenses
that
were
reduced/avoided
due
to
bat
predatory
activity,
and
2/assessing
the
value
of
the
crops
that
remained
undamaged.
The
latter
is
harder
to
assess
since
some
of
the
measurements
required
are
difficult
to
gather
in
the
wild
(e.g.
the
total
number
of
bats
hunting
per
unit
area
of
rice
fields
in
a
period
of
time).
Thus,
the
conservative
figure
that
is
most
easily
calculated
from
the
avail-
able
local
data
is
the
expenditure
on
chemical
treatments
that
was
avoided
in
the
2.9
km2of
paddies
in
Buda
Island
crops
between
2006
and
2012.
Pest
control
procedures
in
the
region
are
done
by
farmers
in
coordination
with
the
Catalan
Government
that
have
created
the
Plant
and
Crop
Protection
Association
in
the
Ebre
Delta.
These
pro-
cedures
consist
of
a
permanent
network
of
trapping
devices
lured
with
chemical
attractants
targeted
at
rice
borer
males.
These
are
deployed
at
a
rate
of
4.5
traps
per
hectare
all
over
the
203
km2
of
paddies,
totalling
approximately
91,300
traps
at
an
estimated
annual
expenditure
of
275,000D
.
In
addition
to
this
non-hazardous
biological
method,
a
moth
surveillance
programme
measuring
the
density
of
affected
rice
stems
is
done
throughout
the
area
dur-
ing
both
the
second
and
the
third
peak
of
the
rice
borer’s
activity.
Aerial
spraying
with
tebufenocide,
at
a
cost
of
21D
/ha,
is
applied
in
areas
where
the
presence
of
rice
borer
eggs
or
caterpillars
exceeds
0.85
rice
stems/m2during
the
second
peak
of
borer
activity,
or
2.15
rice
stems/m2during
the
third
and
final
peak.
The
effective-
ness
of
the
described
treatments,
measured
as
the
reduction
of
striped
rice
borer
moth
density,
is
estimated
to
average
around
70%
for
trapping
and
35%
for
aerial
spraying
(ADV
–
Crop
Defence
Association).
Rice
borer
consumption
by
bats
Chilo
suppressalis
species
–
specific
primers
for
28S-D2
and
mito-
chondrial
COI
genes
have
been
designed
in
order
to
determine
its
presence
in
the
soprano
pipistrelles
faecal
DNA.
DNA
extraction
from
rice
borer,
amplification
and
sequencing
The
5region
of
C.
suppressalis
cytochrome
oxidase
I
(COI)
gene
was
recovered
from
GeneBank
whilst
the
COI
3region
and
the
28S-D2
sequences
were
not
available.
To
obtain
these
sequences
amplifications
and
sequencing
of
28S-D2
and
the
3region
of
the
COI
gene
were
performed
on
rice
borer
DNA
by
using
two
different
set
of
universal
primers,
ND2F/ND2Rev
(Campbell
et
al.,
1993)
and
C1-J-2183/TL2-N-3014
(Simon
et
al.,
1994)
respectively.
DNA
extraction
from
the
borer
was
carried
out
by
homogenising
the
head,
one
forewing
and
one
leg
with
a
plastic
pestle,
and
treat-
ing
them
with
Chelex
resin/proteinaseK
(Vickerman
et
al.,
2004).
240
X.
Puig-Montserrat
et
al.
/
Mammalian
Biology
80
(2015)
237–245
The
PCR
cycling
program
for
28S-D2
primers
pair
was:
3
min
at
94 ◦C,
followed
by
35
cycles
of
45s
at
94 ◦C,
1
min
at
52 ◦C,
2
min
at
72 ◦C,
and
a
final
extension
of
7
min
at
72 ◦C.
For
the
COI
gene
the
PCR
cycles
were
the
same
as
described
for
28S-D2
except
that
the
annealing
temperature
was
lowered
to
48 ◦C.
All
the
amplifications
were
carried
out
in
a
40
l
reactions
using
4
l
of
DNA
template,
1×
buffer
(Promega),
0.2
mM
of
each
dNTP,
10
pmol
of
each
primer
and
0.6
units
of
GoTaq
DNA
polymerase
(Promega),
and
checked
on
1%
agarose
gel
stained
with
ethidium
bromide.
The
amplicons
were
directly
sequenced
using
the
ABI
Prism
BigDye
Terminator
Cycle
Sequencing
Ready
ReactionKit
(PE
Applied
Biosystems),
on
the
ABI
PRISM
310
DNA
Sequencer.
Primers
specific
for
C.
suppressalis
28S
and
COI
shorter
internal
portions
were
designed
(see
supplementary
material
for
PCR
primers,
length
of
amplicons
and
amplifications
conditions)
with
the
software
Primer-blast
at
NCBI
(Rozen
and
Skaletsky,
2000).
It
uses
Primer3
to
design
PCR
primers
and
then
submits
them
to
BLAST
search
against
user-selected
database.
In
this
study
the
database
was
made
for
all
Lepidoptera
other
than
C.
suppressalis
known
sequences
of
the
same
regions.
The
results
are
then
automatically
analysed
to
avoid
primer
pairs
that
can
cause
amplification
of
targets
other
than
the
input
template.
Bat
droppings
collection
During
2008
and
2009
we
analysed
80
bat
droppings
(30
+
50,
respectively)
to
determine
whether
the
pest
under
study
was
being
consumed
by
the
soprano
pipistrelles.
All
the
droppings
were
col-
lected
from
50
bat
boxes
in
Buda
Island
Natural
Reserve.
In
2008,
all
samples
were
collected
during
June,
thus
corresponding
to
the
second
peak
of
the
Rice
borer,
while
in
2009
all
droppings
were
col-
lected
in
August,
during
the
third
and
highest
peak
using
the
same
method.
In
all
cases
the
droppings
were
preserved
in
alcohol
70%.
DNA
extraction
from
bat
droppings
Droppings
were
air
dried
on
filter
paper
to
remove
as
much
ethanol
as
possible,
then
placed
into
eppendorf
tubes
containing
150
l
of
extraction
buffer
(10
mM
tris
HCl
pH
8.0,
1
mM
EDTA,
1%
Nonidet
P-40,
200
g/ml
Proteinase
K)
vortexed
briefly,
incubated
overnight
at
55 ◦C,
and
2
h
at
−20 ◦C.
Samples
were
then
subjected
to
a
second
round
of
lysis
by
adding
50
l
of
extraction
buffer,
and
incubation
at
55 ◦C
for
2
h.
Proteinase
K
was
inactivated
heating
the
sample
10
min,
and
DNA
recovered
after
centrifugation
of
10
min
at
14,000
rpm.
The
pellet
containing
the
dropping
debris
was
utilised
for
microscopy
analysis.
Each
sample
was
split
in
two,
and
one
of
the
portions
was
saved
for
later
analysis.
As
positive
control
a
small
fragment
of
an
antenna
and
a
tarsum
of
the
borer
(size
of
fragments
similar
to
that
retrieved
in
bats’
droppings)
were
added
to
two
dif-
ferent
faeces
samples
prior
to
perform
DNA
extraction
(to
be
sure
that
there
are
not
compounds
inhibiting
the
PCR
in
the
faeces).
As
negative
control,
DNA
was
extracted
with
the
same
protocol
described
above,
from
fragments
retrieved
in
bat
droppings
and
identified
by
microscopy
analysis
as
“not
Lepidoptera”
specimens;
they
were
two
legs,
a
head,
and
head
with
antenna.
Morphological
analysis
of
insects
fragments
found
in
bat
droppings
Before
running
morphological
analysis,
two
slides
(one
male
and
one
female)
of
the
borer
were
prepared
following
Noyes
(1982)
(mounting
processes
used
for
parasitoid
insects)
for
identifying
some
key
morphological
characters
that
could
help
in
identifying
putative
fragments
in
the
bat
droppings.
Morphological
analysis
of
the
droppings
was
performed
on
all
samples
after
DNA
extrac-
tion,
by
crumbling
gently
the
droppings
in
absolute
ethanol
and
by
examining
at
stereo-binocular
(30×)
the
single
fragments
to
look
for
those
that
could
belong
to
C.
supressalis.
Putative
fragments
were
slide
mounted
and
compared
with
homolog
parts
on
the
slides
(male
and
female)
of
C.
supressalis.
Activity
patterns
In
southern
European
paddies,
three
generations
of
rice
borer
occur
in
a
year,
resulting
in
three
moth
peaks
during
the
rice
crop
growing
season
from
May
to
September
(Fig.
4).
To
ascertain
the
flight
activity
patterns
of
the
striped
rice
borer
populations,
includ-
ing
both
males
and
females,
we
distributed
a
total
of
eight
light
traps
throughout
the
Ebre
Delta,
and
monitored
each
light
trap
continuously
from
mid
May
to
late
September
during
four
years
(2007–2010).
We
assessed
moth
activity
by
counting
the
individ-
uals
caught
in
traps
every
2–3
days.
In
order
to
ensure
that
data
on
rice
borer
moth
activity
was
comparable
to
bat
activity
data,
we
pooled
the
counts
in
ten
day
intervals.
We
quantified
total
bat
activity
as
passes
per
minute
by
means
of
acoustic
surveys
and
aimed
to
count
in
situ
the
number
of
bat
passes
around
53
kHz
(the
fundamental
frequency
most
used
by
soprano
pipistrelles
in
the
area)
using
the
heterodyne
system.
We
used
Pet-
tersson
D240X
bat
detectors
(Pettersson
Elektronik
AB,
Sweden)
with
heterodyne
and
time
expansion
(10×)
systems,
and
a
digital
recorder
(Edirol
R9)
to
randomly
record
time
expanded
samples
of
the
echolocation
and
social
calls
to
verify
the
field
identifications
in
the
laboratory
(BatSound
Software,
Pettersson
Elektronik).
All
the
field
work
was
undertaken
by
the
same
researcher
to
prevent
dif-
ferences
in
the
identification
skills
of
individuals
to
interfere
in
the
results
(Limpens,
2004).
We
collected
data
at
randomly
selected
points
in
three
different
years;
survey
effort
varied
year
to
year
according
to
the
budget
available.
Each
point
(totally
87)
was
sur-
veyed
only
one
year
but
in
different
occasions
in
order
to
have
correlated
data
with
moth
activity
from
the
same
year.
During
2007
we
counted
bat
passes
at
40
points
(32
in
rice
paddies
and
8
in
reed
beds
Phragmites
spp,
as
a
control).
Each
point
survey
lasted
5
min
and
was
located
more
than
100
m
apart
from
the
nearest
station
to
avoid
pseudoreplication.
In
order
to
observe
differences
in
bat
foraging
activity
during
the
striped
rice
borer
flight
period,
we
sur-
veyed
each
point
eight
times
from
early
July
to
mid
September
every
10
days.
During
each
survey
we
began
the
sampling
pro-
cess
from
a
different
point
to
avoid
time
biases,
and
all
stations
were
surveyed
on
the
same
night
within
a
3
h
period.
The
same
approach
was
performed
in
2009
with
20
stations
(17
in
rice
and
3
in
reed
beds)
and
four
temporal
replications
(from
late
August
to
mid
September),
centring
the
sampling
on
the
period
when
the
third
and
highest
moth
peak
occurs.
Finally,
in
2010
the
acoustic
survey
consisted
of
3-min
samples
at
27
points
(22
in
rice
and
5
in
reed
beds)
with
six
temporal
replications,
from
August
to
mid
September.
A
total
of
87
acoustic
stations
(71
in
rice)
produced
2486
min
of
recordings
(82%
in
rice
paddies).
All
stations
were
sam-
pled
under
similar
climatic
conditions,
avoiding
sampling
in
windy
and/or
rainy
days.
Rice
borer
activity
patterns
between
the
four
years
were
ana-
lysed
by
means
of
simple
correlations,
thus
making
all
pair-wise
comparisons
for
the
average
number
of
moths
counted
in
the
eight
light-traps
per
sampling
session
(a
total
of
18
sessions
from
May
to
September).
We
used
differences
between
the
mean
activity
of
bats
in
rice
paddies
and
in
control
areas
(reed
beds)
to
reveal
the
relationship
between
bat
and
rice
borer
moth
activity.
The
former
was
consid-
ered
to
be
the
dependent
variable
that
would
respond
to
the
rice
borer
population
levels.
To
examine
bat
activity
in
relation
to
moth
activity
while
accounting
for
the
effects
of
habitat,
sampling
location
or
time
of
survey,
we
used
Generalised
Linear
Mixed
Models
(GLMMs,
Bolker
X.
Puig-Montserrat
et
al.
/
Mammalian
Biology
80
(2015)
237–245
241
et
al.,
2009).
According
to
Gotelli
and
Ellison
(2004)
we
assumed
bat
activity,
which
measures
the
number
of
occurrences
of
bat
passes
in
a
fixed
interval
of
time,
to
be
Poisson
distributed.
We
introduced
moth
counts
and
month
of
sampling
in
the
model
as
independent
continuous
variables.
The
model
also
accounted
for
the
effects
of
habitat
as
an
independent
(fixed)
factor.
Since
some
sampling
loca-
tions
differed
among
years
we
introduced
the
former
as
a
random
effect
nested
within
the
sampling
year.
Prey-predator
associations
alone
constitute
unreliable
evidence
of
the
ability
of
a
particular
predator
to
control
its
prey
popula-
tions,
since
e.g.
the
predator
may
only
feed
on
certain
prey
when
the
latter
is
very
abundant,
and
thus
the
predatory
pressure
may
not
have
any
effect
on
its
populations
(Arditi
and
Dacorogna,
1988).
With
the
aim
of
examining
at
which
level
of
rice
borer
pest
activ-
ity
the
bats
were
significantly
attracted
to
the
rice
paddies,
we
arranged
the
data
to
meet
the
needs
of
an
analysis
of
variance
by
generating
two
datasets:
1/grouping
moth
activity
into
four
cate-
gories
using
the
quartiles
of
its
density
distribution,
and
2/grouping
the
same
variable
into
two
binary
categories
(Rice
borer
moths
present
or
absent).
Given
the
non-normality
of
the
distribution
of
the
dependent
variable
(p
<
0.01
in
the
Kolmogorov–Smirnov
test
of
normality
in
both
raw
and
log-transformed
data),
we
used
a
non-
parametric
approach
to
test
for
differences
in
bat
activity
between
the
groups
described.
We
used
the
Kruskal–Wallis
by
Ranks
Test
in
the
first
case
(four
categories
of
the
predictor
available),
and
the
Mann–Whitney
U
Test
for
the
second
one.
We
performed
post
hoc
multiple
comparisons
of
mean
ranks
when
results
were
significant
to
test
for
differences
between
all
pairs
of
groups.
Statistica
7.0
(StatSoft
Inc.,
Tulsa,
USA)
and
R
3.0.1
(R
Development
Core
Team,
2014)
were
used
to
compute
the
analysis.
Habitat
quantification
within
the
study
area
We
used
land
cover
maps
produced
by
the
Catalan
Govern-
ment
(CREAF,
2009)
to
quantify
the
key
habitats
present
within
soprano
pipistrelle
territories
around
the
occupied
bat
boxes
in
Buda
Island.
We
reclassified
land
use
into
ecologically
meaningful
categories
based
on
the
habitat
selection
studies
available
for
this
species
(Bartonicka
et
al.,
2008a;
Boughey
et
al.,
2011;
Davidson-
Watts
et
al.,
2006),
maintaining
those
categories
relevant
to
the
present
research:
(a)
riparian
forests;
(b)
marshes
and
wetlands;
(c)
river
(water
surface
and
unforested
riversides);
(d)
rice
paddies;
(e)
herbaceous
vegetation;
(f)
urban
and
suburban;
and
(g)
others.
We
calculated
a
2500
m
buffer
(19.63
km2)
around
the
area
with
the
occupied
bat
boxes
existing
on
Buda
Island,
as
an
approximation
of
the
reported
foraging
areas
for
breeding
colonies
(Davidson-
Watts
et
al.,
2006;
Nicholls
and
Racey,
2006).
To
test
for
differences
in
the
available
habitat
between
Buda
Island
and
the
rest
of
the
Delta
we
estimated
habitat
composition
around
69
points
that
were
randomly
distributed
across
the
whole
Ebre
Delta
and
compared
the
mean
coverage
of
each
habitat
category
in
the
random
sam-
ple
with
its
mean
coverage
in
the
buffer
around
the
69
bat
boxes
with
a
Monte
Carlo
randomised
analysis
(Gotelli
and
Ellison,
2004).
Geographical
data
extraction,
analysis
and
representation
was
run
under
R
3.0.1
(R
Development
Core
Team,
2014).
Service
valuation
and
Minimum
Safe
Unit
We
estimated
the
value
of
the
service
provided
by
bats
using
the
avoided-cost
approach,
which
accounts
for
the
expenditure
pre-
vented
by
a
service,
following
the
method
of
Cleveland
et
al.
(2006).
We
included
only
those
figures
that
were
readily
measurable
(cost
of
the
avoided
treatments)
and
avoided
accounting
for
less
measur-
able
economical
benefits
(undamaged
crops
or
avoided
social
cost
of
pesticides)
that
would
increase
the
uncertainty
of
the
result
and
the
number
of
assumptions
to
be
made.
Table
1
GLMM
Poisson
regression
results
for
the
relationship
between
bat
activity
and:
density
of
the
moth
Chilo
supressalis,
habitat,
and
month
of
sampling
(N
=
543).
Effect
GLMM
estimate
Standard
error
Z
value
P
value
Intercept
0.900
0.354
2.544
0.011
Density
of
rice
borer
0.011
0.001
8.243
<0.0001
Habitat
–
reed
−0.942
0.222
−4.230
<0.0001
Month
−0.13511
0.042
0.048
0.962
Random
effects:
Sampling
location/year
estimated
variance
±
SD
=
0.28
±
0.53.
Year
estimated
variance
±
SD
=
0.08
±
0.29.
Implementing
the
concept
of
Minimum
Safe
Unit
(Luck
et
al.,
2009)
defined
as
the
minimum
amount
of
an
element
or
set
of
elements
of
the
ecosystem
necessary
to
ensure
a
particular
function
that
brings
about
a
desired
service,
we
estimated
the
number
of
soprano
pipistrelles
per
hectare
needed
to
maintain
the
striped
rice
borer
populations
below
two
different
thresholds
that
are
used
locally
by
farmers
to
trigger
the
pest
treatments:
no
treatment
needed
(<3500
moths/ha)
and
aerial
treatment
needed
(>8500
moths/ha).
Results
Rice
borer
predation
During
summer
inspections
of
the
roosts,
bats
were
frequently
observed
hunting
rice
borer
moths.
In
2008,
six
out
of
thirty
bat
droppings
(20%)
were
positive
for
the
presence
of
striped
rice
borer
during
the
second
peak
of
borer
activity
(mid
July).
In
2009,
the
screening
was
conducted
with
samples
from
the
third
peak
(last
week
of
August
to
mid
September)
and
resulted
in
25
out
of
50
positive
results
(50%).
Temporal
distribution
of
the
average
num-
ber
of
moths
trapped
per
every
ten
days
in
the
eight
light
traps
were
significantly
correlated
among
years
(2007–2010,
mean
correlation
coefficients
for
six
pair-wise
comparisons:
r
=
0.76,
range
0.67–0.86,
all
p
<
0.05;
n
=
18
sampling
periods),
suggesting
that
moth
dynam-
ics
was
similar
among
the
study
years,
showing
lows
and
peaks
at
similar
dates.
Bat
activity
was
strongly
associated
to
foraging
events
since
bat
passes
were
highly
correlated
with
hunting
buzzes
in
the
three
study
years
(year
2007:
r
=
0.80,
p
<
0.0001,
n
=
320;
year
2009:
r
=
0.90,
p
<
0.0001,
n
=
80;
year
2010:
r
=
0.93,
p
<
0.0001,
n
=
162).
Between
July
and
September
(data
pooled
from
the
three
years
of
sampling)
the
bat
passes
per
minute
in
rice
paddies
averaged
2.9
SD
±
3.86
(range
0
to
22,
n
=
456).
In
the
control
areas,
mean
bat
passes
per
minute
were
1.18
SD
±
2.38
(range
0
to
11,
n
=
106).
Bats
significantly
increased
their
activity
when
rice
borer
den-
sity
increased
(Table
1,
GLMM:
|z|
=
8.243,
p
<
0.0001),
and
showed
a
lower
activity
within
the
reed
beds
(Table
1,
GLMM:
|z|
=
4.230,
p
<
0.0001).
Bat
activity
was
significantly
higher
in
the
rice
paddies
than
in
the
reed
bed
control
plots
provided
the
rice
borer
moths
were
present
at
any
level
(U
=
8450.0;
p
<
0.001,
df
=
541).
Con-
versely
no
such
difference
was
found
when
moths
were
absent
from
light
trap
catches
(U
=
35.0,
p
>
0.5,
df
=
18;
Fig.
3).
Bat
activity
did
not
vary
significantly
among
the
four
quartiles
of
moth
den-
sity
in
the
reed
beds
(K3,80 =
6.848,
p
>
0.05),
however
there
was
a
significant
increase
in
bat
activity
with
rising
levels
of
moth
activ-
ity
in
the
rice
paddies
(H3,463 =
36.433,
p
<
0.0001).
Bats
increased
significantly
their
activity
in
the
rice
fields
when
a
threshold
of
two
moths
per
light
trap
per
day
was
reached
(z
<
2.98,
p
<
0.05)
as
revealed
by
the
comparisons
between
groups
(Fig.
3).
Fig.
4
depicts
the
idealised
yearly
activity
cycle
of
the
rice
borer
and
the
thresh-
old
at
which
bats
significantly
increase
their
foraging
activity
in
the
rice
fields:
even
the
lowest
first
generation
of
the
moth
attracts
bats.
242
X.
Puig-Montserrat
et
al.
/
Mammalian
Biology
80
(2015)
237–245
Fig.
3.
Relationship
between
bats
and
striped
rice
borer
activity.
Black
boxes
represent
bat
passes
(±SD
and
95%
confidence
interval)
in
rice
paddies
and
white
boxes
bat
passes
in
reed-beds
considered
as
a
control
habitat
(***p
<
0.001;
**p
<
0.01).
Rice
borer
activity
was
grouped
according
to
the
four
quartiles
of
its
abundance
counted
at
the
light-traps.
Fig.
4.
Idealised
timing
of
the
activity
cycle
of
the
striped
rice
borer
moth
(Chilo
supressalis)
according
to
the
ADV
Delta
de
l’Ebre
-
Ebre
Delta
Crop
Defence
Association
(2005–2013)
and
activity
cycle
of
the
soprano
pipistrelle
(Pipistrellus
pygmaeus).
The
dotted
line
indicates
the
0.4
moths
per
trap
threshold
at
which
a
significant
difference
between
bat
activity
on
controls
and
paddies
appears
(as
shown
in
Fig.
3).
The
period
during
which
the
paddies
are
flooded
is
also
shown.
Interestingly,
the
experiment
of
deploying
bat
boxes
on
Buda
Island
reported
an
important
increase
of
the
bat
population
in
the
area
over
a
ten
year
period
(r
=
0.95,
p
<
0.001,
n
=
10,
Fig.
5).
Bats
started
occupying
boxes
in
the
year
2000
(243
individuals
were
counted
one
year
after
bat
box
installation
in
1999),
and
reached
a
maximum
count
of
3500
ind.
in
2007
and
2008
(the
last
years
Fig.
5.
Temporal
patterns
of
striped
rice
borer
damage
(rice
stems
infested/m2dur-
ing
the
first
peak
of
the
borer:
r
=
−0.73,
p
<
0.01,
n
=
10)
and
bat
densities
(ind./ha:
r
=
0.95,
p
<
0.001,
n
=
10)
on
Buda
Island
since
the
bat
boxes
were
installed.
with
a
census).
During
the
same
period
a
significant
decrease
in
the
number
of
rice
stems/m2infested
by
the
rice
borer
was
detected
in
the
area
(r
=
−0.73,
p
<
0.01,
n
=
10,
Fig.
5).
The
bat
density
in
the
boxes
was
negatively
correlated
to
rice
damage
over
a
ten
year
period
(r
=
−0.64,
p
<
0.05,
n
=
10).
Habitat
composition
around
bat
boxes
Wetlands,
rivers,
riparian
forests
and
herbaceous
vegetation
were
significantly
more
available
inside
the
2.5
km
buffer
sur-
rounding
the
Buda
Island
breeding
colonies
than
around
the
random
points
that
were
scattered
across
the
whole
Ebre
Delta
(|z|
>
3.50;
p
<
0.001),
whereas
there
was
no
such
difference
regarding
the
coverage
of
cultivated
land
(rice
paddies),
urban
structures
or
other
land
cover
(|z|
>
3.50;
p
>
0.09).
Around
the
occu-
pied
bat
box
stations
(n
=
69)
the
mean
coverage
of
riparian
forest
was
around
0.5%,
rivers
accounted
for
between
6–7%
and
urban
structures
ranged
between
6
and
10%
of
the
available
area.
Rice
paddies
were
the
dominant
habitat,
covering
between
50
and
64%
of
the
land,
followed
by
marshes
from
11%
to
24%.
Service
valuation
We
estimated
the
cost
of
every
chemical
treatment
is
21D
per
hectare,
i.e.
roughly
6.000D
in
less
than
3
km2.
According
to
our
X.
Puig-Montserrat
et
al.
/
Mammalian
Biology
80
(2015)
237–245
243
Fig.
6.
Estimate
of
the
soprano
pipistrelle
Service
Providing
Unit
required
to
keep
pest
density
below
the
target
threshold
during
the
second
peak
of
rice
borer
activity.
The
Service
Providing
Unit
is
expressed
as
the
number
of
adult
individuals
per
hectare.
Where
(1)
Source:
data
collected
by
the
authors
during
the
study;
(2)
Lowest
estimation
from
Kurta
et
al.
(1989);
(3)
Conservative
untested
assumption
(4)
Female
borer
population
is
proportional
to
number
of
stems
damaged
since
they
lay
the
eggs
on
the
rice
stems,
the
effect
of
reducing
the
male
borer
population
on
the
next
generation
of
larvae
is
uncertain
and
therefore
is
not
considered
in
this
model;
(5)
Source:
thresholds
used
by
the
ADV
(Ebre
Delta
Crop
Defense
Association
–
2013).
estimates
(see
Fig.
6)
between
9
and
16
bats
per
hectare
are
needed
to
reduce
the
number
of
stems
affected
by
the
striped
rice
borer
during
their
second
generation
to
less
than
0.85
affected
stems/m2
(aerial
treatment
threshold).
Between
42
and
67
bats
per
hectare
would
be
needed
to
further
lower
the
moth
density
below
any
treat-
ment
triggering
threshold
(0.35
affected
stems/m2).
Assuming
that
the
population
of
bats
in
Buda
Island
is
represented
by
the
colonised
bat
boxes
(around
3500
bats),
we
can
consider
that
during
the
study
period
there
were
around
12
hunting
bats
per
hectare
in
this
area.
Discussion
Any
biological
pest
control
method
capable
of
diminishing
or
even
preventing
the
use
of
chemical
treatments
is
to
be
encouraged
where
there
is
a
concern
for
environmental
and
human
health.
Our
results
suggest
that
soprano
pipistrelle
bats
provide
natural
control
of
a
pest,
the
striped
rice
borer
moth
in
the
Ebre
Delta.
Evidence
of
this
control
is
based
on
two
related
results:
(1)
soprano
pipistrelle
bats
preyed
upon
striped
rice
borer
moth,
even
during
the
second
(lower)
peak
of
borer
activity;
(2)
activity
of
bats
tracked
the
pest
moth
abundance
in
rice
paddies.
Besides,
since
2006
borer
density
has
been
below
the
threshold
for
aerial
spraying
(i.e.
below
0.85
stems
hosting
either
larvae
or
eggs
per
square
metre
during
the
moths’
second
generation).
A
significant
decline
in
the
abundance
of
infested
plants
in
the
Buda
Island
could
be
partially
explained
by
the
presence
of
a
bat
population
(in
bat
boxes)
that
is
large
enough
to
reduce
rice
borer
populations,
but
this
topic
should
be
studied
in
depth.
Though
larvae
infestation
surveys
were
conducted
systemati-
cally
over
the
whole
Ebre
Delta
during
the
period
covered
in
this
study
that
trend
was
not
reported
anywhere
else
in
the
Ebre
Delta
(ADV
–
Crop
Defense
Association-data).
The
results
obtained
from
this
study
provide
further
evidence
that
soprano
pipistrelles
habit-
ually
prey
on
the
pest
and
move
into
the
paddies
even
when
few
moths
are
available
(Fig.
3),
responding
to
striped
rice
borer
emer-
gence
by
foraging
more
on
the
rice
paddies
when
the
aerial
phase
of
the
pest
is
present.
Bats
increase
their
activity
on
the
rice
fields
well
before
the
third
and
highest
peak.
This
observed
opportunistic
feeding
behaviour
seems
typical
of
this
species
(Bartonicka
et
al.,
2008b).
From
the
evidence
gathered,
the
ability
of
soprano
pipistrelles
to
control
rice
borer
population
levels
stands
as
the
most
par-
simonious
explanation
for
the
decrease
of
the
pest
in
the
Buda
Island
rice
paddies,
where
there
are
no
less
than
3500
bats,
cor-
responding
to
a
density
of
12
bats
per
hectare
of
paddies
within
the
range
of
influence
of
bat
boxes
(Flaquer
et
al.,
2006).
So,
we
have
tried
to
approximate
the
density
of
bats
per
hectare
required
to
keep
the
stripped
rice
borer
population
below
the
threshold
for
aerial
treatment
in
the
study
area
(Fig.
4).
When
properly
located,
e.g.
selecting
the
appropriate
support
and
exposure
(Flaquer
et
al.,
2014),
bat
boxes
are
readily
accepted
as
roosting
places
by
bats
in
the
area
(Flaquer
et
al.
2006),
and
can
be
made
at
minimal
cost
by
schoolchildren
on
environmental
awareness
programmes.
Even
when
ready-made
boxes
are
bought
by
farmers,
each
wooden
roost
costs
around
25D
and
lasts
more
than
10
years
without
any
main-
tenance.
In
contrast,
the
cost
of
pheromone
traps
can
be
around
15D
/ha
per
year
(3D
/trap
×
5
traps/ha),
while
spraying
one
hectare
once
a
year
represents
about
21D
/ha
per
year.
In
practice,
annual
costs
of
chemical
spraying
are
often
higher
because
a
second
treat-
ment
is
usually
made
if
the
third
moth
peak
is
severe.
Therefore,
the
estimates
on
both
chemical
treatments
are
6–8
fold
higher
than
the
installation
of
bat
boxes
accounting
for
a
minimum
of
12
bats
per
box.
In
other
areas
of
the
world
with
dense
human
populations
bat
roosts
are
known
to
contribute
to
the
reduction
of
pest
damage
on
crops,
e.g.
in
the
North
American
continent
(Boyles
et
al.,
2011;
Cleveland
et
al.,
2006;
McCracken
et
al.,
2012)
or
Asia
(Wanger
et
al.,
2014).
In
both
these
areas
the
molossids
(family
Molossidae)
involved
in
pest
control
gathered
in
huge
colonies
of
up
to
millions
of
individuals,
and
foraged
at
a
large
regional
scale.
Molossids
are
fast
flying
species
compared
to
many
vespertilionids,
including
the
pipistrelles,
which
fly
at
significantly
lower
speeds
(Norberg
and
Rayner,
1987).
Consequently
the
potential
area
for
pest
con-
trol
by
soprano
pipistrelle
colonies
is
significantly
smaller,
a
fact
that
brings
about
new
management
challenges.
Finding
out
which
habitat
features
best
encourage
the
presence
of
soprano
pipistrelles
would
allow
local
management
strategies
to
establish
populations
in
target
agroecosystems,
to
deliver
the
pest
control
ecosystem
service.
Habitat
availability
around
selected
bat
boxes
of
Buda
Island
is
consistent
with
the
preferences
for
the
species
described
in
other
studies
from
central
and
eastern
Europe
(Bartonicka
et
al.,
2008a;
Boughey
et
al.,
2011;
Davidson-Watts
et
al.,
2006),
show-
ing
a
reliance
on
flooded
habitats
(e.g.
marshes
and
rivers)
and
broadleaved
forests
(e.g.
riparian
woodlands).
The
rapid
colonisa-
tion
of
the
bat
boxes
that
were
installed
in
the
area
in
1999
and
the
subsequent
growth
of
the
local
bat
population
suggests
that
the
limiting
factor
for
this
bat
population
was
the
existence
of
suit-
able
roosts
(Flaquer
et
al.,
2006),
given
the
lack
of
vertical
structures
that
could
provide
shelter.
The