Available via license: CC BY 4.0
Content may be subject to copyright.
Ecology a nd Evolution . 201 8 ;1–1 2 .
|
1
www.ecolevol.org
Received:29Januar y2017
|
Revised:3 0January2018
|
Accepted:9February2018
DOI:10.1002/ece3.3962
ORIGINAL RESEARCH
The role of nurse successional stages on species- specific
facilitation in drylands: Nurse traits and facilitation skills
Marina Fagundes1 | Wolfgang Weisser2 | Gislene Ganade1
Thisisanop enaccessarticleundert hetermsoft heCreat iveCommonsAttr ibutionLicense ,whichpe rmitsuse,dist ributionandreproductioninanymedium,
provide dtheoriginalworkisproper lycited.
©2018TheAuthors.Ecology an d Evolutionpu blishedbyJohnWiley&SonsLtd.
1RestorationEcologyResearchGroup,
Depar tmentofEcology,UniversidadeFeder al
doRioGrandedoNorte,Natal,RN ,Brazil
2TerrestrialEcolog yResearchGroup,
Depar tmentofEcologyandEcosystem
Managem ent,SchoolofLifeScience s
Weihenstephan,TechnicalUniversit yof
Munich,Freising,Germany
Correspondence
WolfgangWeisser,TerrestrialEcology
ResearchGroup,D epartmentofEcosystem
Managem ent,SchooloflifeSciences
Weihenstephan,TechnicalUniversit yof
Munich,Germany.
Email:wolfgang.weisser@tum.de
Funding information
BrazilianNationalResea rchCouncil
(CNPq);PV Egrant,Grant/AwardNumber:
400672/2018-7;CNPqgrant,Grant/
AwardNumber:30 8701/2013-5;Brazilian
CoordinationforPersonalImprovement
(CAPES);GermanAcademicResearch
Council,Grant /AwardNumber:54 417975.
Thisworkw assupportedbytheGerman
ResearchFoundat ion(DFG)andthe
TechnicalUniversityofMunichwithinthe
fundingprogrammeOpenAccessPublishing
Abstract
Plantestablishmentis achallengeinsemiaridenvironmentsdue tointenseandfre-
quentdroughtperiods.Thepresenceofneighboringtrees(nurses)canincreasethe
establishmentofseedlings(targets)byimprovingresourceavailabilityandmicrocli-
mate.Thenurseeffect,however,mightvarydependingonnurse-targetspeciescom-
binations but factors that predict this specificity are poorly known. We used a
multispeciesexperimenttoinvestigatethefacilitationpotentialoftreesfromarange
ofsuccessionalstages,focusing onhownursefunctionaltraitscanpredictspecies-
specific interaction outcomes. We conducted a factorial field experiment in a
Braziliansemiaridtropicalforestduringaseveredroughtperiod.Sixtypairsofinter-
actingtreespecies,20potentialnurses,andthreetargetswereused.Seedlingsofall
target s were planted both und er and far from the n urse canopy, in a random ized
block designreplicated five times. Target growth and survivalwere monitored for
275daysfromthebeginningofthedryseason,andinteractionoutcomeswerecal-
culated using th e Relative Interact ion Intensity (RII) ind ex. Nurse func tional traits
suchassuccessionalstage,height,wooddensity,andcanopydiameterwereusedas
explanatoryvariablestopredict RII values. The averageeffect of nursespecies on
targetplantswasingeneralpositive,thatis,seedlingsurvivalandgrowthincreased
underthenursecanopy.However,forgrowthpairwiseinteractionsweresignificantly
speciesspecific.SuccessionalstagewastheonlyfunctionaltraitexplainingRII val-
ues, with pioneer tree species being stronger facilitators than later successional
trees.However,theexplanationpowerofthisvariablewaslow,andpositive,nega-
tive,orneutralinteractionswerefoundamongnursetreesofallsuccessionalstages.
Becauseseedlingmortalityduringdroughtinsemiaridsystemsishigh,futurestudies
shouldinvestigatehownursetraitsrelatedtowaterusecouldinfluencenursefacili-
tationskills.
KEYWORDS
Caatinga,competition,degradedland,drought,pairwiseinteractions,positiveinteractions,
successionalstage,survival
2
|
FAGUNDE S Et Al.
1 | INTRODUCTION
Facilitat ion is an impor tant process t hat allows plant sp ecies to
resist severe climatic conditions (Bagousse-Pinguet etal., 2014;
Cavieresetal.,2014).Facilitationoccurswhenone plant species,
referre d to as a “nurse,” improve s the surv ival or growt h of an-
other “target”species,byexpanding its realized niche(Soliveres
etal., 2011), by amelioratingabiotic conditions(Jankju,2013)or
improvingresourceavailabilit y(Zou,Barnes,Archer,&McMur try,
2005).Nursespeciesper formimportantrolesinstructuringplant
communitiesataglobalscale(McIntire&Fajardo,2014;Soliveres
&Maestre,2014),andtheireffectsareoftenreportedinsemiarid
lands (Soliveres & Maestre, 2014).Indry lands,airtemperatures
andevapotranspirationfromtargetspeciesarelowerunderneath
the nurs e canopy (Ja nkju, 2013). Nurs e plants c an also allevia te
waterlimitationforthewholeplantcommunitybyperforminghy-
draulic l ift (Dawson, 1993; Pug naire, Armas , Valladares, & Le ps,
2003).
Nurseeffect s,however,mightvar yfrompositivetonegativede -
pending onthetarget species thatestablishes undernurse crown,
and this pr ocess is referred as a sp ecies-spe cific interact ion out-
come (Call away,1998; C allaway & Walker, 1997). Species-specific
interaction outcomeshavebeen found to occurina widerangeof
ecosys tems (Landero & Valien te-Banuet , 2010; Paterno, Siqu eira,
&Ganade, 2016;Poulos, Rayburn, &Schupp, 2014)andhavebeen
pointed out as a strongfactor modulatingseedling regenerationin
plant comm unities (Patern o etal., 2016). However, predicti ng the
outcome of nurse-targetinteractions canbe difficult,especially in
highdiversit yecosystemswheremultiplepairsofnurseandtarget
species areabletointeract.Nurseplantsmayalsohavepositiveef-
fectson target survival butnegativeor neutral effectson growth
(Gómez-Aparicio,2009;Paternoetal.,2016),makingtheinteraction
predictionsevenmorecomplex.Therefore,thereisanurgent need
toidentifythenursetraitsthatinfluencetargetperformance.These
factorshave rarelybeeninvestigatedbecause manipulativeexperi-
mentsconnectingmultiplespeciesarescarce.
Some authors have pointed out that nurse-target interaction
outcomes could be predicted based on nurse species’ ecological
strategies(Schöb,Armas,Guler,Prieto,&Pugnaire,2013;Soliveres,
Smit,&Maestre,2015).Forexample,pioneernursesinsemiaridsys-
temsmighthavea higher tolerancetoenvironmentalstressessuch
aslightintensityanddrought(Kitao,Lei,Koike,Tobita,&Maruyama,
2000),whichcouldaffectconditionsandresourcesprovidedtotheir
neighbors(Diaz&Cabido,20 01).Pioneernursesinanaridenviron-
mentcoulddepleteresourcesslowerthanlate-successionalnurses
by having st ress-tol erant featur es such as high woo d density, and
smallsize,whichwouldallowthemtoestablishinharshordegraded
areas(Grime,1977).Ontheotherhand,pioneernursescoulddeplete
resourcesfasterthanlate-successionalnursesbyexhibitingfeatures
related to highrelative growth ratesuch as low wood density and
large size, wh ich would gua rantee rapi d colonizati on in open gap s
(Kazakou,Vile, Shipley,Gallet, & Garnier, 2006).Therefore,nurses
from different successional stages could have different effects
on the sam e target speci es, a process th at could partia lly explain
species-specificinteractionoutcomes.
Theaimsof this studywere asfollows: (1)Totestthe extentto
which facilitation by nurse species occurs in a Brazilian semiarid
system usingamultispecies experiment.(2)To test whether nurse
successionalstage andmorphological traits canpredict facilitation
andspecies-specific interactionoutcome. Weexpectedfacilitation
to be frequ ent, althou gh other intera ction outcome s might occur.
Wealso expected that nurse successional stageand morphological
traitswould explain facilitation skills because to establish inharsh
semiari d areas, pionee r nurses might h ave evolved stress-toler ant
featuresthatreducetheirratesofresourceuptakeandconsequently
decreasetheircompetitiveability(Grime,1977).
2 | MATERIALS AND METHODS
2.1 | Study area
ThisstudywasconductedintheCaatingasemiaridtropicalforestof
Brazil.Thevegetationischaracterizedbystrongseasonalitywithan
average precipitation around700mm/year,restrictedto 4months
of rainy seas on, from Febr uary until J une when rain i s usually er-
ratic.Themeantemperatureis29°C,andsoiltemperaturecanreach
60°Cduringthedryseason( Velloso,Sampaio,&Pareyn,2002).Our
study sitewasa degradedareaonceusedforselective loggingand
cattlefarming.Landuseendedin1950,afterwhichforestrecovery
wasallowedtotakeplace.Foreststructurecomprisespioneer,early
and late-successional stage treesatthe same sitedue to selective
logging.Theexperimentalsiteisnowpartofthe“NationalForestof
Açu”protectedarea(FlorestaNacionaldeAçu,FLONA ,ICMBio,RN)
inNortheastBrazil(05°35′02,1″S–36°56′41,9″W).
2.2 | Species interaction experiment
Totesttheeffectofnurseontargetspecies,weconductedamulti-
speciesexperimentusing20nativenursetreespeciesandthreena-
tivetargettreespecies.Arangeofsuccessionalstrategieswasused
toselect nurse species. The strategies followeddefinition byMaia
(2012)andvariedfrompioneer(treesthatarethe firsttoestablish
inopen degraded areas), early-successional(trees thatestablish in
open degr aded areas ju st after pione er species have e stablishe d),
andlate-successionaltreespecies(treesthatrarelyestablishinopen
degradedareas).Alltreespecieswerecommonlypresentatthesite
(Table1).
We chose nurs e individua ls spread in a ra dius of 1km around
FLONA de Açu’s hea d office with dis tance betwe en nurses var y-
ingfrom2.5to1,200m.Selectionofnursetreeswasbasedonthe
following criteria:(1)tree trunklargerthan 10cmcircumferenceat
breastheightand;(2)isolatedindividualstoavoidneighborinterfer-
ence.Nursespeciesheightandcanopydiameterweresimilaramong
speciesfromdifferentsuccessionalstages,butsomevariationwithin
successional stages was allowed (Table1). Three target species,
Poincianella pyramidalis (Tul.) L.P Queiroz, Anadenanthera colubrina
|
3
FAGUNDES E t Al.
(Vell.)Brenan,andMyracrodruon urundeuvaAllemão,wereselected.
Targetselectionwas based onthefollowingcriteria: (1) all species
werenativeandwidespreadintheCaatingavegetation;(2)theyem-
bracedistinctsuccessionalstages;(3)theyoccurrednaturallyinthe
study area;and(4)theywere availableincommercialgreenhouses
insufficient numberstoconduct theexperiment.Targetindividuals
were6monthsoldandwere,onaverage,20cm±0.5tallatthestart
ofthe experiment. Speciesin Caatingahaveevolvedtohaveahigh
growthrate,becauseoftheshortrainyseason.Thus,itisrealisticfor
youngplantstoreach20cmheightduringtherainyseason.
“Nurse” and “No nurse”treatment swerearranged in ablock con-
sisting of o ne nurse plant indi vidual and six ta rget plants, wit h one
individualtarget species in each treatment (Appendix1).Blocks were
replicatedfivetimesforeachof the20nursespecies,resultingin100
nursetreesintotal(100blocks)and600targetindividuals.Oncenurse
treeindividualswerechosen,a2m×2mplotwasmarkedaroundeach
nursetree.Allvegetationpresent,commonlyherbaceousspecies,was
weededbeforetargetplanting.Thesameweedingtreatmentwasper-
formed ina2m×2m“nonurse”treatmentplotthatwas locatedat a
distanceof2.5mfromthenurseplotandwasfreefromanyotherplant
canopy inf luence. Target saplings were placed a pproximately 40cm
fromthetrunkofeachnurseplantindividual.Wecountedthenumber
ofleavesoftarget seedlings beforeplanting. Immediatelyafterplant-
ing,eachtargetreceived 2Lof water.Therewasnofurtherirrigation,
but all targets werevisited twice duringthe firstweek after planting
andnoplantdiedduringthisperiod.
2.3 | Monitoring survival and growth
The experiment began on June 2014, toward the end of the rainy
season. We used the dry season because during the rainy sea-
son, nurseplants effectscan be maskedby highwater availability.
Growthand survival of targetswere recorded once a week in the
first2weeksandthenevery15daysfor85daysuntilAugust2014,
TABLE1 ListofCaatingatreespeciesusedinthenurse-targetinteractionexperimentandtheirsuccessionalstagebasedonMaia(2012).
Mean±1standarderrorofnursetraits:height,canopydiameter,andwooddensityweremeasuredusingthreeindividualsofeachnurse
species
Family Abbreviation Nurse species Successional stage Height (m)
Canopy
diameter (m) Wood density
Bixaceae C.vit Cochlospermum vitifolium Pioneer 6.66±0.33 4.7±0.19 0.35±0.05
Burseraceae C.lept Commiphora leptophloeos Pioneer 5.0±1.25 4.0 ± 0.81 0.33± 0.02
Combretaceae C.lep Combretum leprosum Pioneer 3.50±0.86 3.2± 0.40 0.75± 0.02
Euphorbiaceae C.bla Croton blanchetianus Pioneer 3.0± 0.28 2.6± 0.29 0.73±0.03
FabaceaeMimosoideae P.mon Pityrocarpa moniliformis Pioneer 6.83± 1.40 5.26±1. 26 0.76±0.03
Fabaceae—Papilionoideae A.cea Amburana cearensis Pioneer 6.9±2.05 8.21 ±1.66 0.60± 0.01
Fabaceae—Mimosoideae M.ten Mimosa tenuiflora Pioneer 5.16±0.60 6.68± 0.14 0.80 ± 0.01
Fabaceae—Mimosoideae P.sti Piptadenia stipulacea Pioneer 6.3±0.60 5.58± 1.40 0.77 ± 0.02
Apocynaceae A.pyr Aspidosperma pyrifolium Early-successional 6.83± 0.92 6.20±1.06 0.69± 0.04
Boraginaceae C.glaz Cordia glazioviana Early-successional 5.56±0.60 2.93± 0.29 0.64± 0.00
Capparaceae C.has Cynophalla hastata Early-successional 3.50± .0.28 3.6±0.62 0 .74 ± 0.01
Erythroxylaceae E.num Erythroxylum nummularia Early-successional 5.16± 0.88 1.82 ±0.73 0.84 ± 0.00
Fabaceae—
Caesalpinoideae
B.che Bauhinia cheilantha Early-successional 4.16±0.66 3.65± 0.92 0.79 ± 0.00
Fabaceae—
Caesalpinoideae
P.gar Poincianella gardneriana Early-successional 4.33±1.16 5.88± 0.94 0. 87 ± 0.02
Fabaceae—Mimosoideae A.col Anadenanthera colubrina Early-successional 5.80± 2.10 6.00±3.00 0.80 ±0.05
Fabaceae—
Caesalpinoideae
L.fer Libidibia ferrea Early-successional 4.53±0.03 7. 65 ± 0.81 0.77 ±0.35
Malvaceae P.mar Pseudobombax marginatum Late-successional 6.00±0.86 3.96±0.85 0.29 ± 0.01
Euphorbiaceae S.mac Sebastiania macrocarpa Late-successional 5.16± 0.88 1.82 ±0.73 0.75± 0.00
Bignoniaceae H.imp Handroanthus impetiginosus Late-successional 5.50±0.50 5.85±0.45 0.83±0.03
Anacardiaceae S.tub Spondias tuberosa Late-successional 7. 6 6± 0.72 16.11± 1 .19 0.57±0.03
Target species
Fabaceae—
Caesalpinoideae
P.pyr Poincianella pyramidalis Pioneer
Fabaceae—Mimosoideae A.col Anadenanthera colubrina Earlysuccessional
Anacardiaceae M. uru Myracrodruon urundeuva Late-successional
4
|
FAGUNDE S Et Al.
whenalltargetslosttheirleaves.Targetswerethenmonitoredonce
moreinMarch2015,1monthaf terthestartof thefollowingrainy
season.
Werecordedgrowthby countingthenumber ofleaves flushed
ateachinspection.Thenumberofleaveswasusedinsteadofheight,
becauseheightcanremainunchangedinthissemiaridsystemduring
earlygrowth,whenseedlingsallocatemostoftheirbiomasstoroots.
Leafflushing,ontheotherhand,isstrongly responsive toenviron-
mentalstress.Seedlingsunderstressfulconditionswouldlosetheir
leaves and avo id further fl ushing, but the y can quickly fl ush new
leaves once environmentalconditionsareimproved (Lima & Rodal,
2010).Due to thehigh numberof targetindividuals in the experi-
ment,wedidnotmarkleavestofollowtheirindividualfate.Foreach
survey,weusednumberofleavesproducedinrelationtotheinitial
numberofleavesregisteredatthebeginningoftheexperiment.We
thus cal culated the p ercentage of l eaves gained o r lost relati ve to
thenumber ofleaves that the targethadwhenplanted. Therefore,
themeasureofleavesgainedineachsurveywasusedasaproxyof
growththroughtime,wherebyleaflossindicatesastressresponse,
whereasleafflushing indicates lack of stress. Becausespecies re-
place the ir leaves regular ly,va lues <100 do not repre sent lack of
leaves,but,rather,thattherateofleafrenewalwassmallerthanthe
rateofleafloss.
We also checked f or survival of t arget plants at eac h inspec-
tion. Whenanindividuallost all itsleaves, we tested for mortality
byscratchingthebarkcarefullytocheckwhetheritstissuewasstill
greenorfresh.Thesurvivalresponsevariablerepresentedthenum-
ber of days a given target was abletosurviveafter being planted.
The maxi mum surviv al days were set by th e total durati on of the
experiment,275days.
2.4 | Nurse trait measurements
Wecollectednursetraitsfromthreeindividualsofeachnursespe-
cies.Foreachindividual,weestimatedheightand canopydiameter
and measured wood density. Canopy diameter represented the
average length of two perpendicular axes thatwere placed onthe
treecrown facingnorth and south.We measuredwooddensity by
samplingonebranchfromeachtree,removingitsbarkandapplying
thewaterdisplacementmethodperformedbyPérez-Harguindeguy
etal.(2013).
To calculate the ef fect of a nurs e species on a tar get species
growthandsurvival,wecalculatedthepairwiseRelativeInteraction
Intensity—RII(Armas,Ordiales,&Pugnaire,20 04):
where Bw represents target performance under the nurse, and Bo
representstargetperformanceinthe“nonurse”plot.TheRIIvalues
rangefrom−1to+1;wherebynegative valuesindicate competitive
interactions(negativeeffec tofnurseontarget)andpositivevalues
indicatefacilit ation(positiveeffectofnurseontarget).Forsurvival,
wecalculated one RII-valueforeachof the threetargetspecies,in
eachblock.Forgrowth,thesamecalculationswereperformedsepa-
ratelyforeachmeasurementrecordedthroughtime.
2.5 | Statistical analysis
AllanalyseswereperformedinR(w ww.r-project.org,RCoreTeam,
2015,version3.2.0)followingtheZuur,Ieno,Walker,Saveliev,and
Smith(2009)protocol.Tounderstandthefacilitationeffectofdif-
ferentnursetreespeciesontargetplants,weappliedtwogeneral-
izedlinearmixedmodels(GLMM)oneusing survivalandtheother
usinggrowthastheresponsevariable.TheGLMM usedthe“lmer”
functioninthe“lme4”package(Batesetal.,2014),andtheexplana-
tory va riables wer e nurse spec ies, targ et species , and their inte r-
action. Significance was established by log-likelihood ratio tests
removing eachvariable from the fullmodeltocalculate itsoverall
eff ect.Weuse danormalerro rdistributionforallan alysis(Crawley,
2007).
Tote st whether n urse attr ibutes can pr edict nurs e facilitat ion
effects,weper formedaLinearModel Selection, followingCrawley
(2007). T he variables g rowth and sur vival were used a s response
variablesandnursesuccessionalstage,height,canopydiameter,and
wood densit yasexplanatory variables. For grow th, repeated mea-
surement sovertimewereincludedasarandomfactornestedwithin
blockstocorrectfortemporalpseudo-replication.Forsurvival,there
werenorepeatedmeasurementsovertime,andonlyblockwascon-
sideredarandomfactor.
3 | RESULTS
3.1 | Facilitation effect of nurses on targets
As expected, facilitation was common in the Caatinga semiarid
tropicalforest.Forsurvival,18nursesshowedpositiveaverageef-
fects,thatis,facilitation,andtwonursesshowednegativeaverage
effects,thatis,competition(Figure1a).Theaverageincreaseinsur-
vivalacross all nurse plantswas8daysfor the targetA. colubrina,
7daysforthetargetM. urundeuvaand18daysforthetargetP. py-
ramidallis,thatis,2.9%,2.6%,and6.5%,respectively(Appendix2).
Whenonlypositivenurse-targetinteractionswereconsidered(i.e.,
wherethepresenceof a nurse increased averagetargetsurvival
across the five replicates), the average increase in survival was
35days(12%)forA. colubrina,20days(7.2%)forM. urundeuva, and
36days (13%) for P. pyramidallis. It is imp ortant to hig hlight that
despite the positiveef fect ofnurseson targets, few targetplants
survived throughout thedry season. After 275days, only18tar-
get individuals survived under nurse canopies and eightwithout
anurse.
Forgrowth,theaveragenurseeffect(averageRIIacrossalltarget
species and replicates) waspositive in 14of 20nurse species.One
nurseshowed,onaverage,aneutraleffect(RII=0),andfivenurses
had,onaverage, anegativeeffect on targetgrowth (Figure1b).All
targetspecieswereabletoflushnewleavesinboth“nurse”and“no
nurse”treatments(Appendix3).
RII
=
B
w−
B
o
B
w
+B
o
,
|
5
FAGUNDES E t Al.
3.2 | Species- specific relationship
Forsurvival, we found no species specificity(Table2).Despite the
factthatfewnursesexertedaconsistentnegativeorpositiveeffect
ontargets,therewasnosignificantinteractionbetweennursesand
targetspecies(χ2=44.804,df=38,p=.207,Figure2a).Onlythree
ofthe20nurses,namelyPityrocarpa moniliformis, Erythroxylum num-
mularia, and Mimosa tenuiflora,facilitatedall targets, and no nurse
hadnegativeeffects(competition)onalltargets.
Withrespecttogrowth,nurse-targetinteractionswerestrongly
speciesspecific(Table2,χ2=144.93,df=38,p=<.001).Fewnurse
speciesexertedaconsistentpositiveornegativeeffectonalltarget
species (Figure2b). Fromthe 20 nurse species, only Handroanthus
impetiginosus exerted positive effects on all target species, and
only Sebastiania macrocarpaexerted negativeeffects onalltarget s
species.Allothernursespeciesexertedbothpositiveandnegative
effects on target species. Moreover, target species also showed
different responseswhen interacting with different nurse species
(Figure2b).
Netnurseeffectalsovariedbetweentargetgrowthandsurvival.
Positiveeffectsonsur vivalbutnegativeeffectsongrow thwere,for
example , found for the n urse S. macrocarpawhenpairedwithtar-
getA. colubrina,andthecombinationsP. moniliformis–M. urundeuva,
and Pseudobombax marginatum–P. pyramidalis. Positive effects on
growth but negativeeffectsonsurvivalwerefoundforthecombi-
nationsPoincianella gardneriana–A. colubrina and Piptadenia stipula-
cea–M. urundeuva(Figure2).
3.3 | Effects of nurse traits on facilitation skills
Nurse successional stagesignificantly explained RII values for tar-
get growt h (F=3.53, df =2; 2,382, p-val ue=.029, r2=.09), while
the other variables height, canopydiameter, and wood density did
notimprovethemodel. Facilitationoftargetgrowth wasmorefre-
quent and intenseinaverage for nurses from pioneer successional
stages thanfor other successional stages (Figure3).Differences in
effectsizewere,nevertheless,relativelysmall,and therewas con-
siderablevariationinRIIwithinasinglesuccessionalstage(Figure3).
For the target survival model, neither nurse successional stage
FIGURE1 Averageeffec tsof20nursespeciesonthreetarget
speciesgrowth(a)andsurvival(b),measuredusingtheRIIindex.
Negativevaluesindicatecompetitiveinteractions(negativeeffect
ofnurseontarget,i.e.,growthorsurvivalislowerunderthe
nursecanopythanoutsidethenursecanopy)andpositivevalues
indicatefacilitation(positiveeffectofnurseontarget,i.e.,growth
orsurvivalishigherunderthenursecanopythanoutsidethenurse
canopy).Eachbarrepresentstheaverageeffectofonenurse
speciesacrossthreetargetspeciesreplicated15times,errorbars
represent1standarderror.Thecompletenameofallspeciescanbe
foundonTable1.RII,RelativeInteractionIntensity
TABLE2 Tableoflinearmixed-effectmodelsofnurseeffec ton
targetsurvivalandgrowth.Theexperimentconsistsof20Caatinga
nursetreesandthreetargetplantspeciesreplicatedfivetimes.
RelativeInteractionIntensit yindex—RII(Armasetal.,200 4)usedas
responsevariableswascalculatedbasedontargetsurvival(number
ofsurvivaldays)andtargetgrowth(proportionofleavesgained
throughtime).Theexplanatoryvariables(fixedfactors)arenurse
species,targetspecies,andtheirinteractions.Forgrowth
measurements,timewasnestedinplotasarandomfactor
Log- likelihood χ2df p Value
Survival
Completemodel −16. 9195
Nurse×target −5.4827 44.804 38 .2079
Nurseef fect −1 0.185 54.21 57 .5804
Targeteffect −6.6636 4 7.1 6 6 40 .2029
Growth
Completemodel −1, 574. 5
Nurse×target −1 , 6 47. 0 14 4.93 38 <.0 01
Nurseef fect −1 , 67 2 .9 196.84 57 <.0 01
Targeteffect −1, 6 5 4.2 1 59. 2 8 40 <.0 01
RII,RelativeInteractionIntensity.
6
|
FAGUNDE S Et Al.
(F=0.162, df =2; 263, p-value=.928), nurse height (F=0.892,
df =1;263,p-value=.269),canopy diameter(F=0.362, df =1; 263,
p-va lue=.547), and woo d density had ( F=0.0002 , df =1; 263, p-
value=.964) a sign ificant effe ct on RII values . Additionally, the re
was no match between nurse successional stage and target suc-
cessionalstageexplainingRII values,sonursespecies were able to
bothfacilitateandcompetewithtargetsfromallsuccessionalstages
(Figure2).
4 | DISCUSSION
An impor tant novelt y of this work is that n urse tree suc cessional
stages can par tially explainfacilitationskillsinthistropicaldryfor-
est,wherenursepioneertreespresentedastrongestpositiveeffect
ontargetsthanlate-successionalspecies.Nursesuccessionalstage,
therefore,could partiallyexplainspecies-specific interactions. Our
resultsalsocorroborate threeprocessthatarefrequentlyreported
intheliterature:(1)Facilitationisawidespreadprocessinharshen-
vironme nts (Flores & J urado, 20 03; He, Ber tness, & Alt ieri, 2013;
Soliveres&Maestre,2014);(2)Species-specificinteractionoutcomes
arecommonforsemiaridbiomes(Landero&Valiente-Banuet,2010;
Paternoetal.,2016;Wright,Schnit zer,&Reich,2014);and(3)Nurse
positiveeffectsarestrongeronsurvivalthangrowth,ageneralpat-
ternfoundindifferentecosystemtypes(Bertoncello,Oliveira,Hool,
&Martini,2016;Ganade&Brown,20 02;Gómez-Aparicio,20 09).
4.1 | Pioneers nurse effect
The reason why successional stages could partially explain tree
facilit ation skills m ight be related to t he fact that p ioneer spec ies
have evolved s tress-tol erance chara cteristi cs to estab lish in harsh
arid ecosystems (Grime, 1977). In this case,pioneerspecies would
deplete resources slower,makingsoilmoistureavailable for longer
periods, which would benefit target species establishing under
their crow ns. However, there was n o evidence that n urse specie s
had speci fic charac teristi cs of stress-toler ant species su ch as high
wooddensity,low height,andsmall crown size. Additionally,these
traitsdidnotdif ferbetweensuccessionalstagesnordidtheyinflu-
ence faci litation. Even a key n urse trai t such as crown size, w hich
creates the microclimateameliorationfor targetspecies was of lit-
tle impor tance forpredicting facilitation (Soliveres, 2014; Zhang&
Zhao, 2014).Althoughcommon morpho-functionalplant traits can
be used to in dicate compet itive and stre ss-tolera nce strategie s in
semiaridlands(Graff&Aguiar,2017), they might notbeenough to
elucidatethefullcomplexityofnursefacilitationmechanismsindry
forest s. Future stud ies should conte mplate how physio logical and
morpho logical tr aits stro ngly related to w ater use such as ro oting
architecture or specific leafarea could influence nurse facilitation,
species-specificinteractions.
Thisworkshowsthatnursesuccessionalstagecouldplayarole
in complex species-specific interaction outcomes, which are fre-
quentlyunpredictable(Anthelme,Meneses,Valero,Pozo,&Dangles,
FIGURE2 Effectof20nursespecies
ongrowthofthethreedifferenttarget
speciescalculatedusingtheRelative
InteractionIndex(RII).Nursesofall
successionalstagescanaffectpositively
ornegativelytargetplants.Growth
ismeasuredaspercentageofleaves
producedduringtheexperiment.Bars
representtheaveragenurseeffecton
performanceofeachtarget,toSurvival
(a)andGrowth(b)varyingfrom−1
(competition)to1(facilitation)foreach
nurse-targetcombination.Errorbars
represent±1standarderror.Thecomplete
nameofallspeciescanbefoundon
Table1
|
7
FAGUNDES E t Al.
2017).Thismightbeduetodifferencesinthewaynursesfromdis-
tinct su ccessional stag es alter conditions a nd available reso urces
for the same target species (Diaz & Cabido, 2001). Interaction
outcomes might also depend on how nurse strategies combine
with different target needs (Holmgren, Gomez-Aparicio, Quero, &
Vallarades, 2012; Paterno etal., 2016;Woods&Miriti,2016).For
example,targetsthataremo repronetowaterstressaremorelikely
to be facil itated by nurs es that mainta in water in the sy stem, for
example,by per forminghydraulic lift or presenting high water use
effic iency. In our work, t here was no pre dictable m atch between
nurse and targetsuccessional stages.Additionally,theexplanator y
power of the modelwas not strong, and alterations to interaction
outcomes werefoundforallnurse successional stages. Therefore,
the role of nursesuccessionalstage on nurse performanceshould
beconsideredwithcaution.Targetandnursemorpho-physiological
traitsthattogetherpredicttheoutcomeofaparticularinteraction
matchshouldbeinvestigatedinthefuturetorefineinteractionout-
comepredictions.
4.2 | Tropical dry forest dynamics
OurresultsemphasizethattheBrazilianCaatingaisaharshenvi-
ronm e n t w h e r e d r o u ghtis a s t r o n g f orces h a p i n g p l a n trecruitment .
Despitegenerallypositivenurseeffects onsurvival(Bertoncello
etal.,2016;Heetal.,2013),drought wasstillthestrongestforce
limitingregeneration(Jankju,2013).Itisimportanttounderstand
that wet sea sons can be ver y erratic in C aatinga, a nd seedling s
haveto reach acertainrootsize, andaminimumamountofstor-
agetobeabletokeepthemselvesalive throughdr yperiodsuntil
thenextrainarrives.Therefore,anyprocessthatpromoteshigher
probabilityofsurvivalandgrowthshouldinfluencetheseedlings’
chance to pe rsist until wa ter becomes ava ilable. Our re sults re-
inforcetheimportanceofnursespeciesina biome intenselylim-
itedbywatersupplyinwhichtheunpredictabilityofrainstrongly
influences seedling recruitment (Holmgren & Scheffer, 2001).
Futureunders tandingofthemechanismst hatdefineagoodnurse
intropicalsemiaridlandsmightrevealkeyfactorstocombatland
degradationanddesertificationandimproveprogramsofrestora-
tionandlandmanagement.
ACKNOWLEDGMENTS
We are thankful to the Brazilian National Research Council
(CNPq) for fi nancial suppor t to conduct the f ield work, provi d-
ing W.W.W. with a PVE grant (400672/2018-7) and providing
G.G.withaPQgrant(308701/2013-5).Wearethankfultothe
BrazilianCo ordin ationforPer sonalImprovem ent(CAPES)forpro -
vidingscholarshiptoM.F.Further supportofW.W.W.(54417975
oftheGerman AcademicResearch Council(DAAD) was given by
theTUMBRA, “anet workforusingecologicalanalysis todeepen
our understanding of the relationship between biodiversity and
sustainablelanduse.”
CONFLICT OF INTEREST
Nonedeclared.
AUTHOR CONTRIBUTIONS
AuthorFM,GG,andWWconceivedtheideasledthewritingofthe
manuscript;FMand GGdesignedmethodology,collected thedata,
andanalyzed thedata;GGfundedfieldwork;W Wfundedinterna-
tionalexchangetrips.Allauthorscontributedcritically tothe drafts
andgavefinalapprovalforpublication.
DATA ACCESSIBILITY
Species-specificinteractionsvalues:availablethroughDryad(http://
datadr yad.org/)afteracceptanceofthemanuscript.
FIGURE3 Averageeffec tofnursespeciesontargetgrowth
calculatedusingRIIforeachsuccessionalstage.Allsuccessional
stagespresentspecies-specificinteractionsandpotentialto
facilitation.Pioneernursespresentinaveragehigherfacilitative
effects.Errorbarsrepresent±1standarderror.Thecomplete
nameofallspeciescanbefoundonfoundonTable1.RII,Relative
InteractionIntensity
8
|
FAGUNDE S Et Al.
ORCID
Marina Fagundes http://orcid.org/0000-0002-9358-9488
REFERENCES
Anthel me, F.,M eneses, R ., Valero, N., Pozo, P., & Dangle s, O. (2017).
Finenurse variations explaindiscrepanciesinthestress-interaction
relationship in alpine regions. Oikos, 126, 1173–1183. ht tps://doi.
org /10.1111/oik .0 4248
Armas,C ., Ordiales, R., & Pugnaire, F.(2004).Measuringplant interac-
tions:Anewcomparativeindex.Ecology,85,2682–2686.https://doi.
org/10.1890/03-0650
Bagousse-Pinguet, Y., Xiao, S., Brooker,R. W., Gross, N., Liancour t, P.,
Straile , D., & Michale t, R. (2014). Facili tation di splaces hot spots of
diversity and allowscommunitiestopersistinheavily stressed and
disturbed environments. Journal of Vegetation Science, 25, 66–76.
https://doi. org/10.1111/j vs.120 64
Bates, D., Maechler, M., Bolker, B., Walker, S., Christensen, B., &
Singmann, H. … Green, P. (2014). lme4:Linear mixed-ef fect s mod-
els using Eigen and S4. R package version 1.1–7. Retrieved from
cran.r-project.org/web/packages/lme4/index.html
Bertoncello,R.,Oliveira,A.,Hool,K.,&Martini,A.M.Z. (2016).Cluster
planting facilitates survival but not growth in early development
of restor ed tropic al forest . Basic and A pplied Ecology, 17,4 89–496.
https://doi.org/10.1016/j.baae.2016.04.006
Callaway,R.M.(1998).Arepositiveinteractionsspecies-specific?Oikos,
82,202–207.https://doi.org/10.23 07/3546931
Callaway,R. M.,&Walker,L . R. (1997). Competition andfacilitation: A
syntheticapproachtointeractionsinplantcommunities.Ecology,78,
1958–1965.https ://doi.or g/10.1890/0 012-9658 (1997) 078[1958: CA
FASA]2.0.CO;2
Cavieres, L. A ., Brooker,R. W., But terfield, B.J., Cook, B. J., Kikvidze,
Z., Lortie, C. J.,…C allaway, R. M.(2014).Facilitative plant interac-
tionsandclimatesimultaneouslydrivealpineplantdiversity.Ecology
Letter s,17,193–202.https://doi.org/10.1111/ele.12217
Crawley,M.J.(2007).The R book.WestSussex,UK:JohnWileyandSons
Ltd. .ht tps://doi.org/10.1002 /9780470515075
Dawson,T.E.(1993).Hydraulicliftandwaterusebyplants:Implications
for water balance, performance and plant–plant interactions.
Oecologia,95,565–574.https://doi.org/10.1007/BF00317442
Diaz,S.,&Cabido,M.(2001). Vive ladifference:Plantfunctionaldiver-
sitymatterstoecosystemprocesses.Trends in Ecolog y & Evolution,16,
646–655.https://doi.org/10.1016/S0169-5347(01)02283-2
Flores,J.,&Jurado,E.(20 03).Arenurse-protégéinterac tionsmorecom-
mon among plants from arid environments? Journal of Vegetation
Science, 14, 911–916. htt ps: //doi.org /10.1111 /j.1654 -1103.20 03.
tb02225.x
Ganade,G.,&Brown,V.K.(2002).Successioninoldpasturesofcentral
Amazonia:Roleofsoilfertilit yandplantlitter.Ecology,83,743–754.
https://doi.org/10.1890/0012-9658(2002)083[0743:SIOPOC]2.0.
CO;2
Gómez-Ap aricio, L . (2009). Th e role of plant i nteract ions in the re sto-
ration of d egraded ecosystems: A meta-analysis across life-forms
and ecosystems. Journal of Ecology, 97, 1202–1214. https://doi.
org /10.1111/j.1365 -2745. 20 09.01573 .x
Graff,P.,& Aguiar,M.(2017).Dospecies’ str ategies andtype ofstress
predictnetpositive effects inanaridecosystem?Ecology,98, 794–
806.https://doi.org/10.1002/ecy.1703
Grime, J . P. (1977). Evidence f or the exist ence of three p rimar y strate-
giesinplantsanditsrelevancetoecologicalandevolutionarytheory.
American Naturalist,111,1 16 9–1 19 4 . ht t p s : //d o i . o rg / 1 0. 1 0 86 / 2 83 2 4 4
He, Q., Ber tness, M . D., & Altieri, A . H. (2013). Glo bal shift s towards
positivespecies interactions with increasingenvironmental stress.
Ecology Letters,16,695–706.https://doi.org/10.1111/ele.12080
Holmgre n, M., Gomez-Ap aricio, L., Qu ero, L., & Vallara des, F. (2012).
Non-lineareffects of droughtunder shade: Reconciling physiologi-
calandecologicalmodelsinplantcommunities.Oecologia,169,293–
305.https://doi.org/10.1007/s004 42-011-2196-5
Holmgren,M., &Scheffer,M.(2001). El Niño as a windowof opportu-
nity fortherestorationof degradedaridecosystems.Ecosystems,4,
151–159.https://doi.org/10.1007/s100210000065
Jankju,M.(2013).Roleof nurse shrubsin restorationofan aridr ange-
land: Effects of microclimate on grass establishment. Journal
of Arid Environments, 89, 103–109. https://doi.org/10.1016/j.
jaridenv.2012.09.008
Kazakou , E., Vile, D., Shi pley, B., Gallet, C ., & Garnier, E. (20 06). Co-
variationsinlit terdeco mposition,leaftraitsandp lantgrow thi nspe-
ciesfromMediterraneanold-fieldsuccession.Functional Ecology,20,
21–30.https://doi.org/10.1111/j.1365-2435.2006.01080.x
Kitao, M., Lei, T., Koike, T., Tobita, H., & Maruyama, Y. (200 0).
Susceptibility to photoinhibition of three deciduous broadleaf
tree species withdifferent successional traits raised under various
light regimes. Plant Cell and Environment, 23, 81–89. https://doi.
org/10.1046/j.1365-3040.2000.00528.x
Lande ro, J. P. C., & Valient e-Banuet, A . (2010). Species-spe cificity of
nurseplantsforthees tablishm ent,survivorship,andgrowthofaco-
lumnarcactus.A merican Journal of Botany, 97,1289–1295.https://
doi.org/10.3732/ajb.1000088
Lima, A.L.A.,&Rodal,M.J.N.(2010).Phenologyand wooddensit yof
plantsgrowinginthesemi-aridregionofnortheasternBrazil.Journal
of Arid Environments, 74, 1363–1373. http s://doi.org/10.1016/j.
jaridenv.2010.05.009
Maia,G.N.(2012).Caatinga: ár vores e arbustos e suas utilidades(2nded.).
Fortaleza,Brazil:PrintcolorGráficaeEditora.
McIntire, E. J. B., & Fajardo, A. (2014). Facilitation as a ubiquitous
driver of biodiversity. New Phytologist, 201, 403–416. https://doi.
org /10.1111/nph.12478
Paterno,G.,Siqueira,FilhoJ.,& Ganade, G. (2016).Species-specificfa-
cilitation,ontogeneticshiftsandconsequencesforplantcommunity
succession. Journal of Vegetation Science, 27, 606–615. http s://doi.
org /10.1111/jvs.12382
Pérez-Har guindeguy, N., Díaz, S., Gar nier, E., Lavorel, S., Poorte r, H .,
Jaure-guiberr y, P., … Cornelisse n, J. H. C. (2013). New handbook
forstandardisedmeasurementofplantfunctionaltraitsworldwide.
Australian Journal of Botany, 61, 167–234. h ttps: //do i.org /10.1071/
BT12225
Poulos,J.,Rayburn,A.,&Schupp,E.(2014).Simultaneous,independent,
andadditiveeffectsofshrubfacilitationandunderstorycompetition
on the sur vival of a nat ive forb (Penstemon palmeri). Plant Ecology,
215,417–426.https://doi.org/10.1007/s11258-014-0312-4
Pugnaire,F.,Armas,C., Valladares, F.,& Leps,J.(2003). Soil as a medi-
atorinplant-plant interactionsina semi-aridcommunity.Journal of
Vegetation Science,15,85–92.
RCoreTeam(2015)R: A language and envidonment for statistical comput-
ing.Vienna,Austria:RFoundationforStatisticalComputing.
Schöb, C ., Armas, C., Guler, M., Prieto, I., & Pugnaire, F. (2013).
Variability in functional traits mediates plant interactions along
stress gradients. Journal of Ecolog y, 101, 753–762. https://doi.
org /10.1111/1365-2745.12062
Soliveres, S. (2014). Functional traits determine plant co-occurrence
more than environmentor evolutionary relatedness in globaldr y-
lands. Perspectives in Plant Ecology, Evolution and Systematics, 16 ,
164–173.https://doi.org/10.1016/j.ppees.2014.05.001
Soliveres, S., Eldridge, D.J.,Maestre,F.T.,Bowker,M.A ., Tighe,M.,&
Escudero,A.(2011).Microhabitatameliorationandreducedcompe-
titionamongunderstoreyplantsasdriversoffacilitationacrossen-
vironmental gradients: Towards a unif ying framework. Perspectives
in Plant Ecology, Evolution an d Systematic s,13, 247–258.https://doi.
org/10.1016/j.ppees.2011.06.001
|
9
FAGUNDES E t Al.
APPENDIX 1
Scheme of nurse- target interaction experiment
Illustrationofnurse-targetinteractionexperiment.Eachblockconsistsoftwoplotslocatedaround andfarfromanurseplant.Nursetreat-
mentplot(a),consisting ofonesapling ofeachtarget speciesplantedbelow the nursecanopy.Controltreatmentplot(b) consistingofone
saplingofeachtargetspeciesplanted2.5mfarfromanycanopyinfluence.Theexperimentconsistedof20nursetreespeciesreplicatedfive
times,andthreetargettreespecies,withatotalof100blocks.
Soliveres,S.,&Maestre,F.T.(2014).Plant-plantinteractions,environmen-
tal gradients and plant diversity: A global synthesisof community-
levelstudies.Per spective s in Plant Ecology, Evolution and Systematics,
16,154–163.https://doi.org/10.1016/j.ppees.2014.04.001
Soliver es, S., Smi t, C., & Mae stre, F. T.(2 015). Moving for ward on fa-
cilitationresearch:Responsetochangingenvironmentsand effects
on the dive rsity, functio ning and evolution of plant comm unities.
Biological Reviews,90,297–313.ht tp s://doi .org/10.1111/ br v.12110
Velloso,A .,Sampaio,E.,&Pareyn,F.(2002).Ecorregioes Propostas para o
Bioma Caatinga.76p.Recife,Brazil: InstitutodeConservaçaoambi-
ental,TheNatureConser vancydoBrasil.
Woods, N., & Miriti, M. (2016). Ubiquitous germination among com-
mon pere nnial species i n response to fa cilitated and un facilitated
microhabitats.Journal of Arid Environments,124, 72–79. ht tps://doi.
org/10.1016/j.jaridenv.2015.07.012
Wright,A.,Schnitzer,S.,&Reich,P.(2014).Livingclosetoyourneighbors:
Theimportanceofbothcompetitionandfacilitationinplantcommu-
nities.Ecology,95,2213–223.https://doi.org/10.1890/13-1855.1
Zhang, G ., & Zhao, W.(2014).Species-specific trait s determine shrub-
annual interactionsduringagrowingseason. Journal of Arid Land,7,
40 3–413.
Zou, C. B., Barnes, P. W., Archer, S., & McMur try, C. R. (2005). Soil
moisture redistribution as a mechanism of facilitation in savanna
tree-shrubclusters.Oecologia,145 ,32–40.https://doi.org/10.1007/
s00442-005-0110-8
Zuur,F.,Ieno,N.,Walker,N.,Saveliev,A.,&Smith,G.M.(2009)Statistics
for biolog y and health . Mixed effect m odels and ex tensions in ecol ogy with
R.NewYork:Springer.https://doi.org/10.1007/978-0-387-87458-6
How to cite this article:FagundesM,WeisserW,GanadeG .
Theroleofnursesuccessionalstagesonspecies-specific
facilitationindr ylands:Nursetraitsandfacilitationskills.Ecol
Evol. 2018;00:1–12. https://doi.org/10.1002/ece3.3962
10
|
FAGUNDE S Et Al.
APPENDIX 2
Average number of survival days for each targets species under the nurse and far from nurse
Meannumberofsur vival daysforeachtargetspeciesunder nurse (light bars)andnonurse(darkbars)treatments. Error bars representthe
standarddeviation.Eachnurse-targetinteractionwasreplicatedfivetimes.
|
11
FAGUNDES E t Al.
APPENDIX 3
Percentage of leaves gained/lost during the time of experiment of all targets under and far from nurse
Percentageofleavesgained/lostduringthetimeofexperimentofalltargetsunderandfarfromnurse.Targetsgrowthmeasuredasapercent-
ageofleavesgained/lostthroughtime.A. columbrina(pink),M. urundeva(green),andP. pyramidalis(blue).Solidlinesrepresenttargetperfor-
manceunderthenurse,anddashedlinesrepresenttargetperformancefarfromnurse.
12
|
FAGUNDE S Et Al.
APPENDIX3 : Continued.