1 23
Landscape Ecology
ISSN 0921-2973
Volume 34
Numbe 2
Landscape Ecol (2019) 34:389-402
DOI 10.1007/s10980-019-00779-x
Weak e ec s o a ming p ac ices
co esponding o ag icul u al g eening
measu es on a mland bi d di e si y in
bo eal landscapes
Johan Ek oos, Juha Tiainen, Tuomas
Seimola & I ina He zon
1 23
You a icle is published unde he C ea i e
Commons A ibu ion license which allows
use s o ead, copy, dis ibu e and make
de i a i e wo ks, as long as he au ho o
he o iginal wo k is ci ed. You may sel -
a chi e his a icle on you own websi e, an
ins i u ional eposi o y o unde ’s eposi o y
and make i publicly a ailable immedia ely.
RESEARCH ARTICLE
Weak e ec s o a ming p ac ices co esponding
o ag icul u al g eening measu es on a mland bi d
di e si y in bo eal landscapes
Johan Ek oos .Juha Tiainen .Tuomas Seimola .I ina He zon
Recei ed: 11 Ap il 2018 / Accep ed: 28 Janua y 2019 / Published online: 8 Feb ua y 2019
ÓThe Au ho (s) 2019
Abs ac
Con ex The cu en Common Ag icul u al Policy
(CAP) o he Eu opean Union includes h ee g eening
measu es, which a e pa ly in ended o bene i a m-
land biodi e si y. Howe e , he ela i e biodi e si y
e ec s o he g eening measu es, including join
e ec s o landscape con ex , a e no well unde s ood.
Objec i es We s udied he e ec s o inc easing c op
di e si y, p opo ions o p oduc ion g asslands and
allows, co esponding o CAP g eening measu es, on
open a mland bi d di e si y, whils con olling o he
e ec s o dis ance o o es s, ield edge densi y and
p opo ion o buil -up a eas.
Me hods We su eyed open a mland bi ds using
e i o y mapping in Sou he n Finland. We modelled
e ec s o g eening measu es and landscape s uc u e
on a mland bi ds (7642 e i o ies) using gene alised
linea mixed models.
Resul s Inc easing p opo ions o g asslands
inc eased a mland bi d species ichness and di e si y
in open a mland, whe eas inc easing p opo ions o
allows inc eased bi d di e si y. Inc easing c op
di e si y bene i ed indi idual species, bu no species
ichness o di e si y. Inc easing ield edge densi ies
consis en ly inc eased he species ichness o all
a mland species, in- ield nes e s and non-c op
nes e s, as well as o al a mland bi d di e si y. The
ela i e e ec o edge densi y was much s onge
compa ed o he h ee g eening measu es.
Conclusions Ou esul s show ha p omo ing al-
lows and g asslands, in pa icula g azed g asslands
and a ious ypes o semi-na u al g asslands, has he
highes po en ial o bene i a mland bi d di e si y.
Main aining o inc easing ield edge densi ies, cu -
en ly no suppo ed, seems o be o e en mo e bene i .
In open a mland, wi h li le o no ield edges, allows
and g asslands a e pa icula ly bene icial.
Keywo ds Ag i-en i onmen schemes Common
whi e h oa G eening unde Pilla I Meadow pipi 
Skyla k Whincha
Elec onic supplemen a y ma e ial The online e sion o
his a icle (h ps://doi.o g/10.1007/s10980-019-00779-x) con-
ains supplemen a y ma e ial, which is a ailable o au ho ized
use s.
J. Ek oos (&)
Cen e o En i onmen al and Clima e Resea ch, Lund
Uni e si y, 223 62 Lund, Sweden
e-mail: [email p o ec ed];
[email p o ec ed]
J. Tiainen T. Seimola
Na u al Resou ces Ins i u e Finland, P. O. Box 2,
00790 Helsinki, Finland
I. He zon
Depa men o Ag icul u al Sciences, 00140 Helsinki,
Finland
I. He zon
Helsinki Ins i u e o Sus ainabili y Science, HELSUS,
P. O. 20 Box 65, 00140 Helsinki, Finland
123
Landscape Ecol (2019) 34:389–402
h ps://doi.o g/10.1007/s10980-019-00779-x(0123456789().,- olV)(0123456789().,- olV)
In oduc ion
The Eu opean Union (EU) has se a a ge o s opping
biodi e si y declines wi hin he Union’s membe
s a es by 2020 (Eu opean Commission 2011). In
Eu opean a mland, he main policy app oach o
coun e ac widesp ead biodi e si y declines a e ag i-
en i onmen schemes unded unde Common Ag i-
cul u al Policy (CAP) (Pe’e e al. 2014). While some
a ge ed ag i-en i onmen schemes ha e been highly
success ul in e e sing declines o ed-lis ed species
(Pe kins e al. 2011), many widely adop ed ag i-
en i onmen schemes ha e been c i icised o no
being pa icula ly e ec i e (Kleijn e al. 2011). The
ecen CAP e o m in oduced so-called g eening
measu es o add ess challenges ela ed o clima e
change and he en i onmen , including he decline o
biodi e si y (Pe’e e al. 2014).
The g eening measu es include es ablishing eco-
logical ocus a eas o e a ce ain po ion o a a m
a ea, e en ion o pe manen g asslands and enhancing
c op di e si y (Eu opean Commission 2013). I has
been a gued ha he g eening measu es in hei
app o ed o m became less biodi e si y- iendly han
o iginally in ended (Pe’e e al. 2014) and hey a e no
based on solid e idence (Dicks e al. 2014). Fo
example, ecological ocus a eas we e o iginally sug-
ges ed o consis o allows o bu e s ips, bu la e
addi ional op ions, such as legumes unde con en-
ional managemen , we e app o ed as ecological ocus
a eas hough hei alue o biodi e si y can be
ques ioned (Pe’e e al. 2014). Re en ion ules o
pe manen g asslands became less s ic han o igi-
nally p oposed: a educ ion o up o 5% in hei ne
a ea a na ional o egional scales is pe mi ed (Pe’e
e al. 2014). Finally, he ou come o nume ous
exemp ions esul ed in ha hese measu es apply o
only 50% o EU a mland (ibid).
Because he biodi e si y e ec s o he cu en
g eening measu es a e la gely unknown bu he
impe a i e o imp o ing en i onmen al pe o mance
o he CAP emains s ong, he e is a clea need o
u he empi ical e idence (Dicks e al. 2014; Pe’e
e al. 2017). In addi ion, he added alue o manage-
men in e en ions o a mland biodi e si y depends
on landscape con ex (Ba a
´ y e al. 2011; Schepe e al.
2013). Implemen ing g eening measu es may he e-
o e ha e a s onge impac on a mland biodi e si y
in s uc u ally simple landscapes, whe e wildli e-
iendly managemen can c ea e a s onge ecological
con as be ween a eas wi h and wi hou ag i-en i on-
men schemes (Ba a
´ y e al. 2011). Mo eo e , bi ds
b eeding in open a mland a oid se lemen s and o he
buil -up a eas and o es edges, whe eas p edomi-
nan ly open ield bounda ies a e pa icula ly bene icial
non-c op habi a s uc u es (Vepsa
¨la
¨inen e al. 2010;
Tiainen and Seimola 2014). In his con ex , imple-
men ing g eening measu es can be expec ed o a ec
a mland bi ds di e en ly depending on he a ailabil-
i y o ield bounda ies and dis ance o o es s and
se lemen s ac oss ag icul u al landscapes. Fu he -
mo e, indi idual species can be expec ed o espond
di e en ly o g adien s in land-use in ensi y and
landscape s uc u e depending on con as ing ecolog-
ical equi emen s be ween species (Vepsa
¨la
¨inen e al.
2010; Picke and Si iwa dena 2011).
Bi d species b eeding in ields espond di ec ly o
changes in ield managemen p ac ices, pa icula ly in
open a mland cha ac e ised by la ge ields and low
p opo ions o non-c op habi a s. In con as , bi d
species b eeding in edge habi a s, e.g. in non-c op ield
bounda ies, bu eeding a leas pa ially in ields,
espond o ield managemen indi ec ly because o
e ec s o landscape complemen a ion and landscape
supplemen a ion (B o ons e al. 2005; Smi h e al.
2014; Jose sson e al. 2017). While a mland bi ds
bene i om allows ( an Buski k and Willi 2004;
He zon e al. 2011) and g asslands in ce eal-domi-
na ed a mland (Piha e al. 2007), i is less clea
whe he inc easing c op di e si y bene i s a mland
bi ds (Hi on e al. 2015; Jose sson e al. 2017).
Impo an ly, he ela i e e ec s o allows, g asslands
and c op di e si y on bi d assemblages a e i ually
unknown, in pa icula conside ing mode a ing e ec s
o s uc u al landscape a ibu es.
In an icipa ion o esea ch speci ically ocused a
he g eening measu es implemen a ion and i s bene-
i s, his s udy ocuses a he ela i e bene i s o he
ield ypes ha exis ed be o e he policy e o m bu
ha co espond o he g eening measu es in hei
unc ional ole o a mland bi ds. To his end, in ou
s udy (i) ecological ocus a eas a e ep esen ed by
allows o a ious kinds (legume c ops a e no
included due o hei low occu ence), (ii) pe manen
g asslands a e indica ed by all g asslands excep
o a ional silage leys (see me hods), and (iii) c op
di e si y is calcula ed based on se en main c op ypes.
We ocus on open a med landscapes in a bo eal zone
123
390 Landscape Ecol (2019) 34:389–402
o s udy he ela i e e ec s o h ee measu es co e-
sponding o he g eening measu es on a mland bi d
di e si y and on he abundance o he mos common
species, while also es ing o in e ac i e e ec s
be ween he g eening measu es and landscape a i-
ables. We use an ex ensi e da a-se collec ed in
Sou he n Finland (Fig. 1), and we explici ly consid-
e ed all bi d species b eeding in open a mland
habi a s. We expec ed g asslands and allows o
bene i a mland bi d di e si y o a la ge ex en han
c op di e si y (Jose sson e al. 2017), and because we
ocus on bi d species b eeding in open a mland we
also expec ed s onge e ec s u he away om o es
edges (Piha e al. 2007; W e enbe g e al. 2010) and in
landscapes cha ac e ised by low sha es o buil -up
a eas (Vepsa
¨la
¨inen e al. 2010). Finally, because o
la ge ecological con as s (Kleijn e al. 2011), we
expec ed s onge e ec s o g eening measu es in
a mland wi h low a ailabili y o non-c op ield
bounda ies.
Ma e ials and me hods
S udy a ea
We used bi d da a collec ed du ing 2009–2011 in 47
su ey a eas si ua ed wi hin an a ea o 400 9150 km
ac oss Sou he n Finland (Fig. 1; Suppo ing ma e ial
S1). Finland is di ided in o h ee zones wi h di e en
le els o ag icul u al subsidies, and ou s udy a eas
we e si ua ed in zones A and B, con aining 21% and
26% o Finland’s u ilised ag icul u al a ea, espec-
i ely. In Finland, he g eening measu es a e only
Fig. 1 S udy a ea and he g eening policy zona ion in sou he n
Finland. The con inuous line delinea es he sou he n ag icul u al
suppo a eas (aand b) whe e h ee c op plan s a e demanded in
a ms la ge han 30 ha, and no he n a ea C whe e only wo
c op plan s is equi ed. Sou h o he dashed line (a), a ms la ge
han 15 ha a e expec ed o ound ecological ocus a eas
cons i u ing 5% o hei ield a ea. Do s show he loca ion o
s udy a eas in hei ue size. Land a ea is indica ed by ligh g ey
shading and ag icul u al ields wi h da k g ey shading
123
Landscape Ecol (2019) 34:389–402 391

applied o he h ee sou he n-mos p o inces while he
es a e exemp ed due o a high o es co e . The
na ionally app o ed measu es include di e si ica ion
o cul i a ion wi h a leas h ee c op plan s in a ms
o e 30 ha o wo c op plan s in a ms wi h 10–30 ha
in sou he n Finland (zones A and B, see Fig. 1) o wo
c op plan s in a ms o e 10 ha u he no h (zone C),
and an ecological ocus a ea o 5% o ield a ea in
a ms o e 15 ha in zone A (Finlex 2017). The a ea o
semi-na u al g asslands o o e i e yea s old cul i-
a ed g asslands a e o be e ained wi hin 95% o a
na ional e e ence alue based on he su ace a ea in
2015 (132,000 ha). The equi emen o 5% ecological
ocus a eas can be achie ed h ough pe manen
g assland, di e en kinds o allows, sho - e m cop-
pice and legume c ops in a ms la ge han 15 ha.
Finnish a mland consis s o mosaic landscapes
whe e ag icul u al land is concen a ed o pa ches o
a mland su ounded by o es o o he land use ypes.
The size o hese a mland pa ches a y om a ew o
se e al hund eds o hec a es. Because o his mosaic
s uc u e, a mland pa ches con ain well-delinea ed
local communi ies o a mland bi ds. In his s udy, he
47 su ey a eas we e delinea ed based on he ex en o
indi idual a mland pa ches, i.e. a med a eas su -
ounded by o es ed a eas. Because he a mland
pa ches a ied in size, he su ey a eas also a ied in
size (mean ±SD = 234.2 ±263.04, ha, min = 80,
max = 1675; Suppo ing ma e ial S1). Su ey a eas
ei he cons i u ed an en i e a mland pa ch o subse s
o la ge a mland pa ches, in which se e al su ey
a eas we e delinea ed o co e he a mland pa ches.
As a as possible we used da a collec ed in 2010, bu
when he 2010 da a we e no a ailable we used da a
om 2009 o 2011. The o al a ea su eyed was
12,300 ha, o which 10,572 ha ep esen ed cul i a ed
land.
Field su eys and da a p epa a ion
We su eyed a mland bi ds using a e i o y mapping
me hod wi h h ee su ey ounds du ing ea ly May o
mid-June. The e i o y mapping was unde aken by a
eam o expe ienced ield o ni hologis s, whe e each
eam membe su eyed sligh ly o e 100 ha a mland
du ing one mo ning. Beginning a sun ise and ending
oughly be o e noon, all a mland habi a s wi hin he
su ey a ea we e ho oughly sea ched o a mland
bi ds, which we e ma ked on isi maps paying
pa icula ca e o simul aneous obse a ions on bi ds
o neighbou ing e i o ies. Based on he h ee isi
maps we in e p e ed he posi ion o indi idual bi d
e i o ies, which we e subsequen ly ep esen ed as
poin objec s in a GIS laye .
We used o icial digi ized block maps (In eg a ed
Adminis a ion and Con ol Sys em da abase), sup-
plemen ed wi h da a on wi hin-block bounda ies o
di e en c ops based on ield no es and ae ial
pho og aphs. The maps we e u he supplemen ed
wi h digi ized i e s, majo di ches, oads, o es s and
di e en open, bushy o wooded isle s, as well as
a ms eads and o he buil -up a eas. These da a we e
combined in a digi ized ec o map con aining
spa ially explici , geo e e enced da a on all c ops
cul i a ed in ou s udy landscapes. Du ing ield
su eys, we no ed all sp ing-sown ce eals on one
hand and a ious ypes o sown g asslands on he
o he . C op ypes included (i) non-pe manen , sown
leys o silage and (ii) pas u es on a able land, o en
e ained o some yea s as a pa o c op o a ion, (iii)
sp ing-sown ce eals, (i ) au umn-sown ce eals,
( ) sp ing-sown dico s (oilseed apes, b oad bean
e c.), ( i) au umn-sown oilseed ape and ca away, ( ii)
allows, and ( iii) s ubble ields (i.e. no- ill sp ing-
sown c ops, including bo h ce eals and oilseed ape),
which oge he comp ised all a able land wi hin each
sampling uni . No e ha silage leys and pas u es on
a able land a e di e en as habi a s, he o me being
in in ensi e cul i a ion and he la e ep esen ing less
in ensi e habi a o bi ds, exis ing o se e al yea s
hough no necessa ily pe manen in a s ic sense.
Using c op ypes a he han sepa a ing be ween all
c ops be e e lec s unc ional habi a ypes o
a mland bi ds (Hi on e al. 2015).
We he ea e es ablished ci cula sampling uni s
wi h a adius o 200 m (12.56 ha) ac oss all su ey
a eas. The numbe o sampling uni s was maximized
wi hin he su ey a eas gi en h ee cons ain s; (i) a
leas 50% o he plo had o consis o open a mland,
(ii) he sampling uni s we e exclusi e, i.e. no spa ial
o e lap among he sampling uni s was allowed, and
(iii) he cen oid o a sampling uni had o be on an
ac i ely a med ield pa cel. We disca ded all sam-
pling uni s con aining abandoned a mland, which
ei he consis ed o g assy o bushy o me ields.
The ea e we coun ed he numbe o e i o ies o he
20 bi d species b eeding in open a mland in his s udy
a ea (see below o de ails) o each sampling uni ,
123
392 Landscape Ecol (2019) 34:389–402
along wi h in o ma ion on land co e da a. Following
his p ocedu e we ob ained 657 sampling uni s,
co e ing some 8200 ha o a mland ac oss sou he n
Finland. The selec ed sampling uni s included obse -
a ions o 7624 indi idual bi d e i o ies o he
selec ed 20 a mland bi d species (Suppo ing ma e ial
S2).
Calcula ion o esponse a iables
In his s udy we selec ed all open- a mland bi d
species, belonging o wo ecological g oups: (i) 12
species b eeding on a able land and along open ield
bounda ies (he ea e e med ield nes e s), and (ii) 8
bi d species b eeding p ima ily amongs bushes and
highe he b ege a ion in non-c op habi a s, such as
ield bounda ies and o he edge habi a s (he ea e
e med non-c op nes e s; see Suppo ing ma e ial S2).
Thus we explici ly ocused on bi d species b eeding in
open a mland and no on a mland species b eeding in
o es edges o a ms eads (Jose sson e al. 2017). We
classi ied a mland bi ds in o hese wo g oups based
on ea lie published classi ica ions de eloped o
Finnish condi ions (Tiainen and Pakkala 2001). We
used o al species ichness, he species ichness o ield
nes e s and non-c op nes e s, as well as he di e si y o
a mland bi ds [using he in e se Simpson’s index (1/
D)], as communi y esponse a iables. In addi ion, we
analysed abundances o ou mos common indi idual
species: he ield b eede s skyla k (3936 e i o ies)
and he meadow pipi (686 e i o ies), and he non-
c op nes e s common whi e h oa (924 e i o ies) and
wincha (443 e i o ies).
Calcula ion o landscape a iables
We calcula ed he ollowing p edic o s co esponding
o he cu en g eening measu es: (i) he p opo ion o
allows wi hin he sampling uni s as a p oxy o
ecological ocus a eas; (ii) c op ype di e si y (see
below o de ails) wi hin he sampling uni s, and (iii)
he p opo ion o g asslands (all ypes o g asslands
excep o o a ional silage leys, i.e. c op ype (i) lis ed
in he sec ion Field su eys and da a p epa a ion
abo e) wi hin he sampling uni s as a p oxy o
pe manen and longe - e m g asslands. Fallows a e
comp ised o en i onmen al allows (i.e., a non-
p oduc i e ield se aside o a leas 2 yea s unde
an ag i-en i onmen scheme, Toi onen e al. 2013),
o he long- e m allows (combined on 421.0 ha in
o al, p esen in 41% o all sampling uni s), and
o a ional allows wi h s ubble o ba e-g ound allows
(in 8% o all sampling uni s o alling 100.2 ha). The
wo allow ypes we e combined because o he
ela i ely low sample size o o a ional allows.
We included wo g oups o adjus ing landscape
a iables in his s udy (Table 1). Fi s , in o de o
desc ibe he landscape s uc u e in e ms o non-c op
landscape cha ac e is ics, we measu ed he ollowing
p edic o s: (i) dis ance o he nea es o es om he
cen oid o each sampling uni ( ollowing Piha e al.
(2007); (ii) he p opo ion o buil -up habi a s ( ol-
lowing De ic o and Jigue 2007), including all buil -
up a eas and human se lemen s, bu in addi ion also
small isle s wi h ees o bushes ound p ima ily close
o oads, se lemen and ba ns; and (iii) an index o
ield edge densi y desc ibing he ela i e amoun o
non-c op ield bounda ies in he sampling uni s. We
chose o measu e dis ance o o es s ins ead o
p opo ion o es s wi hin he bu e s because he
o me will mo e accu a ely desc ibe he landscape
con ex o all sampling uni s, including hose which
had no o es s wi hin 200 me es om he cen oid
(40% o all 657 sampling uni s). We combined isle s
wi h ees o bushes wi h buil -up a eas because
accoun ing o isle s by hemsel es would no ha e
been s a is ically easible, as hey cons i u ed a iny
ac ion o o al land co e . We calcula ed he ela i e
amoun o ield edge densi y by di iding he numbe o
blocks in e sec ing wi h he sampling uni s wi h he
a ea o a able land wi hin espec i e sampling uni s.
Thus, low alues indica ed a low ela i e densi y o
ield bounda ies and high alues a high ela i e densi y
o ield bounda ies.
We cons uc ed ou measu e o c op di e si y using
majo c op ypes lis ed abo e ins ead o all a ailable
c op classes lis ed in he block da abase o s ipula ed in
he na ional egula ion on g eening. A classi ica ion
in o majo c op ypes by hei sowing iming and
axonomy has been shown o be ecologically mo e
ele an o a mland bi ds as compa ed o a mo e
de ailed dis inc ion be ween c ops as i is implemen ed
in g eening (Hi on e al. 2015). We de ined c op ype
di e si y as he in e se Simpson’s index ollowing
Palmu e al. (2014), calcula ed on he p opo ions o
se en g oups o c op ypes eco ded spa ially explic-
i ly o each sampling uni . De ined in his way ou
measu e o c op di e si y was highly co ela ed wi h
123
Landscape Ecol (2019) 34:389–402 393
c op ype ichness (
S
= 0.74, P 0.0001), and
equa ed o c op ype ichness only i c ops had equal
p opo ions wi hin each sampling uni .
Ou g assland a iable included a a ie y o
g asslands, including uly pe manen g asslands and
sown bu g azed g asslands on a able land bu no
sho - o a ional silage leys, which we e no g azed and
ypically kep o one o 2 yea s. Ro a ional g azed
g asslands a e ypically kep o se e al yea s, and
hey co e ed 173.3 ha in o al (p esen in 16% o all
sampling uni s). In Finland, only 1.4% o he ield a ea
has been classi ied as pe manen g assland o he
whole coun y (Bascou 2012). Pe manen g assland
consis ed o h ee di e en land-use ypes as speci ied
in he In eg a ed Adminis a ion and Con ol Sys em
da abase: semi-na u al pe manen g azed pas u es;
semi-na u al pe manen g azed pas u es on we lands,
and semi-na u al g asslands cha ac e ised by he b- and
g ass-domina ed ege a ion, bu no cu en ly g azed
o mown. Toge he , hese pe manen g asslands we e
ound in 15% o he 659 sampling uni s, co e ing a
o al a ea o 110.6 ha.
S a is ical me hods
We ound a signi ican mode a e co ela ion be ween
edge densi y and c op ype di e si y (
P
= 0.40). To
accoun o collinea i y be ween hese p edic o s we
eg essed c op ype di e si y agains edge densi y and
used he esiduals o his eg ession (G aham 2003) o
measu e c op ype di e si y, while accoun ing o ield
size. Thus, all p edic o a iables showed su icien ly
low co ela ions wi h each o he (
P
B0.30; G aham
2003; Zuu e al. 2009). We log- ans o med all
p edic o s o imp o e linea i y and he ea e scaled
he p edic o s o ze o mean and uni a iance. Scaling
he p edic o s allowed us o assess he indi idual and
join e ec s o p edic o s gi en he a e age o o he
included p edic o s. We he ea e cons uc ed indi-
idual s a is ical models o o al species ichness,
species di e si y (in e se Simpson’s index) and
abundance o he mos common species by i s
including he h ee adjus ing landscape a iables
(dis ance o o es s, p opo ion o buil -up a eas wi hin
he ci cula sampling uni s, and a mland edge
densi y) and he h ee a iables co esponding o
g eening measu es (c op di e si y and he p opo ions
o allows and g asslands wi hin he sampling uni s).
We i s de ined ull models conside ing all wo-way
in e ac ions be ween he h ee adjus ing landscape
a iables and he h ee a iables co esponding o
g eening measu es, and he ea e we emo ed all non-
signi ican in e ac ions one by one o simpli y he
models. All models included he su ey a ea iden i y
nes ed wi hin a a iable desc ibing egional iden i y
(Suppo ing Ma e ial S1), o con ol o non-indepen-
dence be ween sampling uni s wi hin he same su ey
Table 1 Summa y
s a is ics o explana o y
a iables and esponse
a iables based on he 657
sampling uni s (ci cula
uni s wi h a adius o
200 m)
*See Suppo ing Ma e ial
S2 o ull species lis ,
including numbe s o
obse ed e i o ies pe
species
Va iable Mean ±SD Range F equency (n, %)
Explana o y a iables
Dis ance o o es s (m) 219.9 ±172.2 0.00–1064.0 655 (99.7)
P opo ion buil -up a eas 0.06 ±0.07 0.00–0.41 576 (87.7)
Edge densi y 0.04 ±0.02 0.01–0.17 657 (100.0)
P opo ion allows 0.07 ±0.15 0.00–0.80 294 (44.7)
C op di e si y 1.92 ±0.73 1.00–5.36 657 (100.0)
P opo ion g asslands 0.04 ±0.10 0.00–0.67 164 (25.0)
Response a iables*
Species ichness 4.2 ±1.9 1.0–13.0 657 (100.0)
Species di e si y 2.7 ±1.2 1.0–7.2 657 (100.0)
Richness ield nes e s 2.3 ±1.2 0.0–7.0 630 (95.9)
Richness non-c op species 1.8 ±1.3 0.0–7.0 562 (85.5)
Skyla k 6.0 ±4.8 0.0–28.0 621 (94.5)
Meadow pipi 1.0 ±1.5 0.0–18.0 329 (50.1)
Common whi e h oa 1.4 ±1.4 0.0–7.0 465 (70.8)
Whincha 0.7 ±0.9 0.0–7.0 300 (45.7)
123
394 Landscape Ecol (2019) 34:389–402
a ea, and o la ge -scale au oco ela ion be ween
egionally clus e ed su ey a eas (Zuu e al. 2009).
To al species ichness o a mland bi ds was
analysed using gene alised linea mixed models wi h
Poisson e o dis ibu ions as implemen ed in he
unc ion glme () a ailable in he lib a y lme4 (Ba es
e al. 2015), whe eas species di e si y was analysed
using linea mixed models using he unc ion lme() in
he lib a y nlme (Pinhei o e al. 2015). Finally, while
analysing he abundances o he mos common a m-
land bi ds we e alua ed ou al e na i e e o s uc u es
(Poisson, ze o-in la ed Poisson, nega i e binomial and
ze o-in la ed nega i e binomial) by compa ing model
AIC:s using he lib a y glmmADMB (Fou nie e al.
2012). Following his p ocedu e, he abundance o
skyla ks, meadow pipi s, and whincha s was modelled
using a nega i e binomial dis ibu ion, whe eas he
common whi e h oa was modelled using ze o-in la ed
Poisson e o dis ibu ions. We e i ied model assump-
ions by isual examina ions o model esiduals, and by
con i ming ha model esiduals we e no spa ially
au oco ela ed using co elog ams as implemen ed in
he lib a y nc (Bjo ns ad 2016).
Resul s
Communi y-le el e ec s
The p opo ion o g asslands had a consis en ly
posi i e e ec on all measu es o species ichness
and di e si y (Table 2, Fig. 2a–b). Howe e , he
posi i e e ec o inc easing p opo ions o g asslands
o o al species ichness, o al species di e si y and
species ichness o open b eede s inc eased wi h
dec easing p opo ions o buil -up a eas. Gi en he
mean p opo ion o buil -up a eas in ou sample uni s
(0.06), an inc easing p opo ion o g asslands
inc eased he o al species ichness, o al species
di e si y and species ichness o open b eede s
(Suppo ing Ma e ial S3-4). Subdi iding he da a
based on he median in o high and low p opo ions
in buil -up a eas (mean = 0.11 in high and 0.02 in low)
showed ha o e all species ichness and ichness o
open b eede s signi ican ly inc eased wi h inc easing
p opo ions o g asslands gi en low p opo ions in
buil -up a eas (slope C1.09, PB0.008; Suppo ing
Ma e ial S3), bu no gi en high p opo ions (slope
C0.39, PB0.092; Suppo ing Ma e ial S3). Species
di e si y signi ican ly inc eased unde bo h low and
high p opo ions o buil -up a eas based on he abo e
subdi ision (Suppo ing Ma e ial S3). He e, he ela-
ionship only became non-signi ican as p opo ions in
buil -up a eas exceeded 15-20% (Suppo ing Ma e ial
S4). An inc easing p opo ion o allows signi ican ly
inc eased bi d species di e si y (Fig. 2c) and ichness
o edge species, bu no o al species ichness o
ichness o in- ield b eede s. Inc easing c op di e si y
had no e ec on any di e si y componen (Table 2).
To al species ichness and di e si y, and ichness o
ield nes e s and non-c op nes e s all consis en ly
inc eased wi h inc easing ield edge densi y, which
had he s onges e ec ou o he main e m p edic o s
(Table 2, Fig. 2d). In con as , inc easing p opo ions
o buil -up a eas had di e en e ec s on he ou
di e si y componen s. Species di e si y and ichness
o non-c op nes e s signi ican ly inc eased wi h
inc easing p opo ions o buil a eas, whe eas ichness
o open nes e s signi ican ly declined and o al species
ichness was una ec ed (Table 2). Finally, inc easing
dis ance o o es s consis en ly inc eased all di e si y
componen s. The p opo ion o buil -up a eas did no
a ec o e all species ichness, whe eas ield nes e s
signi ican ly dec eased and non-c op nes e s signi i-
can ly inc eased wi h inc easing p opo ions o buil -
up a eas (Table 2).
Species-speci ic e ec s
The h ee g eening a iables had con as ing e ec s on
he abundances o he ou mos abundan bi d species
(Table 3). Fi s , an inc easing p opo ion o g asslands
signi ican ly inc eased he abundances o meadow
pipi s and common whi e h oa s, independen ly o
adjus ing landscape cha ac e is ics, whe eas whincha
abundance inc eased wi h inc easing p opo ions o
g asslands in combina ion wi h high dis ances o
o es s (Suppo ing Ma e ial S3). Skyla k abundance
was no a ec ed by an inc easing p opo ion o
g asslands (Table 3).
The e ec s o inc easing p opo ions o allows
we e signi ican only in in e ac ions wi h landscape
a iables. On he one hand, inc easing p opo ions o
allows a away om o es s signi ican ly inc eased
skyla k abundance while close o o es s hey did no
(Suppo ing Ma e ial S3-4). This in e ac i e e ec
was no signi ican o he o he h ee species
(Table 3). On he o he hand, inc easing ield edge
123
Landscape Ecol (2019) 34:389–402 395
Pinhei o J, Ba es D, DebRoy S, Sa ka D, R Co e Team (2015)
_nlme: linea and nonlinea mixed e ec s models_. R
package e sion 3.1-121
Schepe J, Holzschuh A, Kuussaa i M, Po s SG, Rundlo
¨ M,
Smi h HG, Kleijn D (2013) En i onmen al ac o s d i ing
he e ec i eness o Eu opean ag i-en i onmen al mea-
su es in mi iga ing pollina o loss—a me a-analysis. Ecol
Le 16:912–920
Smi h HG, Bi kho e K, Clough Y, Ek oos J, Olsson O, Rundlo
¨
M (2014) Beyond dispe sal: he ole o animal mo emen
in mode n ag icul u al landscapes. In: Hansson LA,
A
˚kesson S (eds) Animal mo emen ac oss scales. Ox o d
Uni e si y P ess, Ox o d, pp 51–70
So
¨de be g T (2016) Fo
¨ g o
¨ningen i p ak iken. Jo db uks e ke
Rapp 8:1–60 (In Swedish)
Tiainen J, Pakkala T (2001) Bi ds. In: Pi ka
¨nen M, Tiainen J
(Eds.), Biodi e si y o ag icul u al landscapes in Finland.
Bi dLi e Finland Conse a ion Se ies, ol 3, pp 33–50
Tiainen J, Seimola T (2014) Densi y a ia ion among habi a s o
sou h Finnish a mland. Linnu - uosiki ja 2013:70–77 (in
Finnish wi h English summa y)
Toi onen M, He zon I, Helenius J (2013) En i onmen al al-
lows: no el ege a ion ype in a bo eal landscape. Biol
Conse 159:355–366
Toi onen M, He zon I, Kuussaa i M (2016) Communi y com-
posi ion o bu e lies and bumblebees in allows: niche
b ead h and dispe sal capaci y modi y esponses o allow
ype and landscape. J Insec Conse 20:23–34
an Buski k J, Willi Y (2004) Enhancemen o a mland bio-
di e si y wi hin se -aside land. Conse Biol 18:987–994
Vasseu C, Joannon A, A i on S, Bu el F, Meyna d JM, Baud y
J (2013) The c opping sys ems mosaic: how does he hid-
den he e ogenei y o ag icul u al landscapes d i e a h o-
pod popula ions? Ag ic Ecosys En i on 166:3–14
Vepsa
¨la
¨inen V, Tiainen J, Holopainen J, Piha M, Seimola T
(2010) Imp o emen s in he Finnish ag i-en i onmen
scheme a e needed o suppo ich a mland a i auna. Ann
Zool Fennici 47:287–305
Wilson JD, Whi ingham MJ, B adbu y RB (2005) The man-
agemen o c op s uc u e: a gene al app oach o e e sing
he impac s o ag icul u al in ensi ica ion on bi ds? Ibis
147:453–463
W e enbe g J, Pa
¨ T, Be g A
˚(2010) Changes in local species
ichness o a mland bi ds in ela ion o land-use changes
and landscape s uc u e. Biol Conse 143:375–381
Zingg S, G enz J, Humbe J-Y (2018) Landscape-scale e ec s
o land use in ensi y on bi ds and bu e lies. Ag ic Ecosys
En i on 267:119–128
Zuu AF, Ieno EN, Walke NJ, Sa elie AA, Smi h G (2009)
Mixed e ec s models and ex ensions in ecology wi h R.
Sp inge , New Yo k
Publishe ’s No e Sp inge Na u e emains neu al wi h
ega d o ju isdic ional claims in published maps and
ins i u ional a ilia ions.
123
402 Landscape Ecol (2019) 34:389–402