Seed bank, seed size and dispersal in moisture gradients of temporary pools in Southern France

HAL Id: hal-01681544

https://hal.archives-ouvertes.fr/hal-01681544

Submitted on 7 May 2018

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

of temporary pools in Southern France

Kristin Metzner, Sophie Gachet, Pauline Rocarpin, Arne Saatkamp

To cite this version:

Kristin Metzner, Sophie Gachet, Pauline Rocarpin, Arne Saatkamp. Seed bank, seed size and dispersal

in moisture gradients of temporary pools in Southern France. Basic and Applied Ecology, Elsevier,

2017, 21, pp.13-22. �10.1016/j.baae.2017.06.003�. �hal-01681544�

torforco-existenceinplantcommunities(Silvertown2004;

Silvertown,Araya, &Gowing 2015).Atrade-off between wateruseefficiencyandrelativegrowthrateseemstodrive nichepartitioningalongthe hydrologicalgradient(Angert, Huxman,Chesson,&Venable2009;Silvertownetal.2015). Niche partitioning and risk-reduction are both important mechanismsforpersistence, butwestill donotknowhow niche partitioningrelates torisk-reduction (Simons 2011). This is important because most plants grow separately in local gradients(e.g. García-Baquero, Silvertown,Gowing, &Valle2015)andperceivevaryingrisklevelsaccordingto theirnicheboundaries.

Soilseedbanksreduceriskswhenrainfallisunpredictable (Cohen1966)andthisstrategycanresultinhighabundance ofseedsinthesoil(Volis&Bohrer2013).Fordeserts,one singleglobalvarianceinrainfallissufficienttomodel evo-lutionofsoilseedbanks(Clauss&Venable2000;Venable 2007). However, insmall-scale gradients of moisture, the responseofplantstorainfallvariabilityislesssynchronous (García-Baqueroetal.2015;Tielbörger&Petru2008).Inthis caseofuncorrelatedlocalvariances,whenadjacentpatches aredifferent,seeddispersalistheoptimalrisk-reductiontrait (Nathan&Muller-Landau2000;Siewert&Tielbörger2010;

Snyder2006;Venable&Brown1988).Thissuggestsashift ofseeddispersaldistancesandseedbankabundancesin mois-ture gradients, which has never been detected yet. Large seeds increase survival of seedlings under drought (Baker 1972;Daws,Crabtree,Dalling,Mullins, &Burslem2008;

Leishman&Westoby1994)andhencedecreasetheriskof extinctionofaplant species.Soilseedbanksmaythusbe negativelycorrelatedwithseedsizeandpositivelycorrelated withmoisture.Sinceseedsizeisinawell-knowntrade-off withseednumber(Moles&Westoby2006),largeseedshave lowerabundanceinthesoilandlimiteddispersal(Bruun& Poschlod2006;Saatkamp,Affre,Dutoit,&Poschlod2009). Moreover,largeseedshavebeensuggestedarisk-reduction traitbecause theystore resourcestosurvive adverse times (Cohen1966;Leishman&Westoby1994;Philippi&Seger 1989;Siewert&Tielbörger 2010).Modelsrevealeda neg-ative correlation between seed size, seed banks and seed dispersalforlocallyhomogenoushabitats (Cohen&Levin 1987;Venable&Brown 1988).Altogether,largeseedsize should be associated with dry habitats and trade-off with soilseedbankabundanceandseeddispersalinthemoister habitats.

Beyondseedsize,plantsdifferinadaptationstohydrology leadingtovaryingwateruseefficiency,e.g.byC3,C4and CAMphytosynthesis(Keeley1998)andinthiswayenable temporarilycoexistingguilds(Angertetal.2009)or coexis-tanceinspatiallypartitionedhydrologicalniches(Ellenberg 1953;García-Baquero etal.2015;Silvertownetal.2015). Ithasbeenshownthatniche partitioninginlocalmoisture gradientscontributesimportantlytoplantdiversity(Bauder 2000;Violle etal.2010).Wetherefore expectthat effects ofmoisturenichesonrisk-reductiontraitsarestrongestfor moisturegradientsonshortspatialscalesandwhenmoisture

gradientsinterferewiththeextentofseeddispersalsuchas inMediterranean temporarypools (Deil2005).Temporary pools arealso subjectedtostrongyear-to-yearfluctuations inrainfall,increasingthe importanceof risk-reductionand theyarenotoriouslyrichinannualplants(Deil2005). Inter-estingly,year-to-yearchangesinsurfaceareunequalforwet anddryhabitatsinthehydrological gradient.Inwetyears, highandsteepareasremaindrywhilemanylowerpartsare flooded,whereasindryyearseventhelowestpointsremain dry.Thisimpliesthattheimportanceofglobalversuslocal temporal variance experienced by plants differsaccording to their moisture niche inour gradient, i.e. aquatic plants havenoescapeandbuilduphighlyabundantsoilseedbanks (Bauder 2005),whereas plantswithintermediate moisture nichesshouldshowhigherseeddispersal.

Buildingonthiscontext,wetestedthefollowing hypothe-ses on risk-reduction traits and moisture niches. (1) Abundance of seeds in soil banks increases towards the wetendofthemoisturegradient,whichisparticularly pro-nounced in aquatic plants. (2) Dispersal of seeds in local gradientsis highestfor specieswithintermediatemoisture niches,resultinginahigherspreadinthesoilseedbankfor intermediatespecies.(3)Seedsizeincreasesforspeciesfrom dryhabitats,reducingtheeffectsofdrought.(4)Seed disper-sal, seedbankabundanceandseedsize maybe negatively related,sincetheyarealternativestrategiesofriskreduction. Wequantifiedabundanceanddispersalofseedsinthesoil bankaswellasseedsizeforallvascularplantspeciesin15 Mediterraneantemporarypools.Wealsoquantifiedchanges ofwaterlevelovershortdistances,whichareknowntovary inter-annuallywithrainfallandtodrivesmall-scaleturnover intheseplantcommunities(Brock2011;Deil2005;Rhazi, Grillas,TanHam,&ElKhyari2001).

Materials

and

methods

Study

sites

We studied three temporary wetland areas in Southern France, Feuillane (43◦2814N, 4◦5241E), Evenos (43◦1240N, 5◦5110E) and Plaine des Maures (43◦2112N, 6◦2606E). At Feuillane, the mean annual temperature is15.2±0.68◦Candthemeanannualrainfall is455±158mm(2004–2013,Istres).Thissitehasalluvial soilswithtemporarypoolsinwinter.

Evenoshasameanannualtemperatureof16.4±0.47◦C and amean annualrainfall of 585±168mm (2004–2013, Toulon).Alayeroftertiarybasaltisthesubstratefor tempo-rarypools.

Plaine des Maures has a mean annual temperature of 15.4±0.56◦Candameanannualrainfallof747±299mm (2004–2013,LeLuc).Permiansandstonesformshallowsoils and temporary streamlets andpools. Thissite is of prime interest forthe conservationof temporarypools inEurope

andharbors27 rare or endagered plant (Grillas, Gauthier, Yavercovski,&Perennou2004).

Sampling

of

established

vegetation

In eachof the threestudy siteswe selectedfive tempo-rarypoolsandateachpoolatransectwaslaidoutfromthe deepestpointatthepoolbottomtoadrypointinthe surround-ingarea.Alongeachtransect,weestablishedfivesampling plots(1–5mapart).Eachplotconsistedof20smallquadrates (10×10cm),whichwerearrangedatrightangleswiththe transect.Presence/absenceofallvascularplantswasrecorded ineachofthesmallquadrates,resultinginatotalcensusarea of0.2m2perplot.Foranalysis,weonlyusedaplant’s fre-quencyinthesmallquadratesatplotlevel.Wegathereddata fromApriltoJune2012forFeuillaneandEvenosandinJune 2013forPlainedesMaures.Nomenclatureofplantnames followsTison,Jauzein,andMichaud(2014).

Soil

seed

bank

sampling

Aroundeachplotwetook13randomlyplacedsoilcores. Coreshadadiameterof6cmandadepthof10cm.Before sampling,weremovedvegetationandlitter.Wemixedsoil samplesforeach plot.Themixedsoilsampleswere trans-portedandstoredat3–5◦Cuntiltheemergenceexperiment. WesampledsoilinAprilandJune 2012forFeuillaneand EvenosandinJune 2013for PlainedesMaures.Sampling inspringavoidedrecentlydispersedseedsand,hence, repre-sentsthepersistentsoilseedbank(Walck,Baskin,Baskin,& Hidayati2005).However,someseedsofCardaminehirsuta, PolypogonmaritimusandErophila vernadispersed before sampling.

Wequantifiedseeds inthe soilusing theseedling emer-gence method (ter Heerdt 1996). We concentrated soil samples using sievesof 4mm and0.1mm.The fine sieve retained eventhe smallest expectedseeds (Juncus). These reducedsampleswerespreadasa0.5-cmlayerintraysfilled withvermiculiteandcultivatedinanunheatedgreenhouseat AixMarseilleUniversity.Weinclinedthetraysinwater-filled panelswithonesidewater-saturatedandtheotherside5cm abovethewaterline. Wepooledalldata,sincewedidnot detectstrongdifferencesinemergencebetweenbothsides, andsincebothaquaticandxerophyticplantsgerminatedwell inthissetting.WecultivatedsoilsamplesfromFeuillaneand EvenosfromJulytoSeptember2012andforPlainedes Mau-res,fromNovember2013toJune2014.Wecountedemerged seedlingseveryweekandremovedidentifiedplants. Uniden-tifiedseedlingswerecultivateduntilidentification.

Seed

size

2Wemeasuredseedsizeasdryweightofthreereplicates of10or50seedsoffiveindividualsperspeciesin2012and 2013.

Measurement

and

simulation

of

water

levels

in

temporary

pools

Wemeasuredwaterlevels(L)atfivedatesinspring2013 andspring2014foreachofthe15studiedtemporarypools. Wethencompileddataondailyprecipitation(P)anddaily potential evapo-transpiration(E)for theperiod 2011–2014 foreachsite.WeusedthesimplewaterbudgetfromBrooks (2004) to simulate water levels (in cm) for each pool on a daily basis.This model usesthree parameters (a, b and c), whichwe fitted by maximizing the R2 of the relation L[t+1]=(L[t]+a×P[t+1]−b×E[t+1])×cinarepeated linear regression. Wethen used the maximum waterlevel obtainedforeachpool.Wemeasuredrelativetopographical heightsof thedifferentplotsof alltemporarypools witha water balance. The model for water level was runfor the deepestpointofeachpoolandwethensubstractedthe topo-graphicalheighttoreachthewaterlevelvalueforeachplot. Thisyieldednegativewaterlevelsfordryhabitats.Weused thesewaterlevelsasasurrogateforthemoisturegradientin temporarypoolsandwecalculatedthemeanandthemedian waterlevelforeachspecies.

Statistical

analyses

Wealsoused the positiononan axisof Canonical Cor-respondenceAnalysis(CCA)asanalternativemeasureofa species’moistureniche(TerBraak1986).Weperformeda CCAonaplotbyspeciesmatrix,andtransectpositionand siteasenvironmentdatausingthedefaultsettingsofthecca modulein vegan/R(Oksanen etal.2007).Thisforced the firstaxisparalleltothemoisturegradient.Wethencompared a species’ mean or median position on this axis from the established vegetation withits occurrence inthe soilseed bank.

Forcomparisonsbetweenseedbankandstanding vege-tation weused box-plots of waterlevelsor first CCAaxis positions(Fig.2)andt-orU-testsforcomparisonofmeans andBartlett’stestforvariances.

Inordertotesttheeffectofmoisturenicheandseedsize ontheabundanceofseedsinthesoil,weusedazero-inflated negativebinomialmodeltoaccountforover-dispersioninour data(Zuur,Ieno,Walker,Saveliev,&Smith2009).Wedid amodelselection basedonlikelihood-ratiotestsand AIC-valuesasdescribedinZuuretal.,2009.Theoptimalmodel forthecountpartwasthefullmodelwithseedsize,gradient positionandtheirinteraction.Forthepresence-absencepart, onlyseedsizewasretainedaspredictor.

Wemeasuredspreadofseedsinthesoilalongthegradient intermsofwaterlevelsorCCA-axisusingtheinterquartile range(IQR)ofgradientpositionsofallseedsofaspecies, IQRhasbeenproventobeindependentofsamplesizeand canbeusedtocomparespeciesinoursettingsincenumber of occurrences vary; sinceIQRcanbesensible tooutliers

(B) (A) ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● -20 0 10 20 30 1 10 100 1000 10000 * * * ****** * * * * ************* ** * *** 0.0 0.2 0.4 0.6 0.8 1.0 Pr obab ty o f fa s e a bs enc es 0.01 0.1 1 10 * * * * * ** **** * * ** * *************** * * * * ** * * * * * * * * * **** * ** ** ****** **************** ***** ** -10

Fig1. Abundanceofaspeciesseedsinthesoilonitsmeansimulatedwaterlevel.(A)Observed(dots,starsforzeros)andfittedcounts accordingtoazero-inflatednegativebinomialmodelusingseedsizeascovariate(abundance,straightline,p=0.0060,z=2.74;seedsize p<0.0001,z=−4.124).(B)Probabilityoffalsezeroaccordingtozero-inflatedmodel(seedsize,p=0.0487,z=1.972).Seetextfordetails.

we also calculatedRousseeuw and Croux’s (1993)robust measureofscaleasanalternative.

Wetestedtherelationbetweenmoisturenicheandspread byusing a linearmodel of IQRon medianseed positions inthe gradient.Additionally, we wanted toknowwhether anon-linearmodelwasmoreappropriate.Wethereforealso usedgeneralizedadditivemodels(GAM)withtwodegreesof freedomtocomparewiththelinearmodel.Wealsorepeated allanalysesusingCCAaxisinplaceofsimulatedwaterlevels. Tostudytherelationshipbetweenseedsizeandmoisture niche,wefittedalinearregressionoflog-transformedseed sizeagainst themedian andmean of waterlevelsfor each speciesandof positionsonthe firstCCAaxisand alterna-tivelycomparedittoageneralizedadditivemodel(GAM).

All analyseswereperformed in R(R-Core-Team 2014) usingthepackagesveganforCCA;psclandlmtestfor zero-inflatedmodelsandmcgvforGAM.

Results

Theabundanceof seedsinthe soilseedbankwas posi-tivelyrelatedtoincreasingmoistureinthehabitatsofaplant (linear model of log-transformed abundances only, t=2.2, p=0.0285).Thisrelationwas evenclearerwhen absences wereincluded inazero-inflatedmodelwithseedsizeas a covariate (Fig.1).Thisindicates that species occurring in moist areas have higher abundances of seeds in the soil, independentlyof seed size (Fig. 1A).In thismodel, there wasaweakinteractionbetweenseedsizeandgradient posi-tion,indicatingthattheeffectofmoisturenichewasstronger for large than for small seeds. The correlation between plantpositionalongthegradientandabundanceinsoilseed bankremainedsignificantwhenmedian(z=2.6,p=0.0092) insteadofmeanpositionswereusedorwhenitsmaximum abundance(z=2.040,p=0.0413)insteadoftotalcountswas used.

Moreover,theprobabilityofanadultplanttobedetectedin thesoilseedbankdecreaseswithincreasingseedsizeas

illus-tratedbytheincreaseoffalseabsencesinthe zero-inflated model (Fig.1B).Thisrelationbetweenabsencefrom seed bankandseedsizewasalsofoundinalogisticregressionon presence–absencedataonly(z=3.174,p=0.0015).

For46species,wehadatleastthreeoccurrencesinboth datasets,sowecomparedtheirpositionsalongthemoisture gradientbetweenstandingvegetationandsoilseedbank visu-allyusingboxplots(Fig.2).Thisrevealedthatmostspecies wererestrictedtoaparticularrangeofwaterlevelsasadult plants. Whensortedbythemedianpositionof adultplants (Fig.2),thesoilseedbankofaplanttendedtobeondrier partsatthemoistgradientendandonmoisterpartsatthedry gradientendcomparedtoadults.

Using data on mean niche position and spread (inter-quartile range) in the soil seed bank of 78 species, we found anon-linear relationship (GAM analysis with seed sizeandabundanceinthesoilseedbankas(non-significant) covariates),withspreadbeinglargeratintermediategradient positions(n=78,F=3.374,p=0.0398,Fig.3).Thiswas con-firmedwhenweusedtheRosseuw–Crouxindexinsteadof theinter-quartilerange(GAM,n=78,F=4.024,p=0.0221) orwhenweusedmedianpositionsinsteadofmeanpositions (GAM,n=78,F=3.509,p=0.0352)andneitherabundances of seeds in the soil (F=0.661, p=0.5195) nor seed size (F=0.022, p=0.9782) had an effect on spread. A linear model on the same data did not show any trend (df=74, t=−0.465,p=0.644).

Wealsoanalyzednichespreadforeachspeciesinthe stand-ingvegetationwithatleastthreeoccurrences,measuredby the inter-quartile range inwater levels (Fig.3).Wefound thatitdidnotchangewithmeanpositioninalinearmodel (df=96,t=0.1, p=0.92),nor inaGAM(n=98,F=0.01, p=0.92).

Whenwecomparedinter-quartilerangesbetweenstanding vegetation and soil seed bank, we found no overall sig-nificant difference (t-test,t=1.6, p=0.1022, N=80).This result did not change when a robust measure of spread (Rosseuw–Croux index) was used or when only species occurring in both data sets were used in a paired t-test

40 20 0 20 40 Hypericumperforatum Tuberarialignosa Airopsistenella Radiola linoides Bellis annua Airacupaniana Chaetonychia cymosa Anagallis arvensis Galium parisiense Moenchiaerecta Crepissancta Isolepissetacea Trifolium campestre Cicendiafiliformis Juncuscapitatus Dittrichiaviscosa Sisymbrellaaspera Gaudiniafragilis Agrostis tenerrima Teucrium scordium Vulpiabromoides Polygonumaviculare Juncustenageia Juncus bufonius Ranunculusrevelieri Saginasubulata Mentha pulegium Lythrum borysthenicum Lythrum thymifolium Filagogallica Bromushordaceus Euphorbiamaculata Ranunculuspaludosus Juncuspygmaeus Polypogonmaritimus Poaannua Anagallis minima Callitrichebrutia Juncusarticulatus Ranunculussardous Ranunculuspeltatus Centauriumerythraea Laurentiamichelii Plantagocoronopus Lythrum hyssopifolia Ranunculusophioglossifolius

seedbank adult plants

Fig.2. Distributionsofseedsintheseedbank(grey)andadultplants(white)growingalongtransectsrangingfromthebottomoftemporary poolstopermanentlydryplacesinthevicinity.Speciesaresortedbygrowingconditionsofadultplants,withwetconditionsatthetop. Negativevaluesofsimulatedmaximumwaterlevelsindicatedryconditions.

(t=−0.4,df=40,p-value=0.6727).Thenon-lineareffectof ahigherspreadatintermediategradientpositionswaseven stronger when the moisture relatedCCA-axis 1 was used insteadofwaterlevels(analysisnotshown).

Whenwe comparedwaterlevelsfor seedsfromthe soil comparedtoadultplantswithinspecies,onlytwospecies,

Ranunculus paludosus (t=−2.3, df=8, p=0.0572) and

Lythrum hyssopifolia (t=−2.0,df=15,p=0.0524) tended tohaveadultsinsignificantlymoisterconditionsthanseeds inthesoilseedbank.

Additionally,comparingpresenceinthesoilseedbankto thestandingvegetationinthedistributionalongthegradient foreachspeciesseparately,aBartletttestofhomogeneityof variancesrevealedthatLythrumborysthenicum(K2=4.441,

100 200 300 400 500 600 -3 -2 -1 0 1 2 ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●

Fig.3. Seedspreadinsoilalongthegradient(inter-quartilerange) onmeanwaterlevelofaspeciesandfittedlinefromageneralized additivemodel(n=78,F=3.374,p=0.0398)withseedsizeand seedabundance as covariates.Inter-quartile rangeissquare-root transformedandwaterleveltransformedto(x+50)1.5_tolessen

left-skew.

p=0.0351),Juncustenageia (K2=8.5911, p=0.0034)and

Moenchiaerecta(K2=4.5512,p=0.0329)hadsignificantly greater variance in the soil seed bank and only Mentha pulegium(K2=3.8831,p=0.0488)hadgreaterspreadinthe standingvegetation.

Using CCA axis coordinates instead of water levels to study positions of seeds from the soil comparedto adults revealedthatR.paludosusseeds(t=−4.3,df=8,p=0.0023) werefoundindrierconditionsthantheadultplantsinthe veg-etation.AsimilardifferencewasfoundforTuberariaguttata

(t=2.4,df=12,p=0.0354).

Using the CCA-axis, nine species (Callitriche brutia, Isolepis setacea,Juncusarticulatus,Juncusbufonius, Jun-custenageia,Lythrumborysthenicum,Poaannua,Polypogon maritimus,Ranunculuspeltatus,Saginasubulata)showeda largerspreadinthesoilseedbankcompared tothe stand-ingvegetation(Bartlett’stestofhomogeneityof variances, K2=4.2–23.4,p<0.05).Only Juncuspygmaeus (K2=4.2, p=0.0348)and Plantagocoronopus (K2=3.1, p=0.0796) hadhigherspreadinthestandingvegetation.

Seedsizeofaspeciesdecreasedsignificantly(Fig.4)with increasingly moist niches of adults as measured by their medianpositioninthegradient.Analyzingthedatasetwith non-linear regressionmethods(GAMR2=0.14,F=8.699, p=0.0003)onlyslightlyincreased R2comparedtothe lin-ear model andrevealed again a generallydecreasing seed size withincreasingmoisture. The speciesposition onthe moisture-relatedCCA-axiswasalsolinearilyrelatedtoseed size (p<0.0001,t=4.3, R2=0.17). Analyzingthis dataas community-weighted means of seed size yielded a much weakerrelationship(p=0.049,marginalR2=0.02).

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

_●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

_●

●

●

●

●

●

●

●

●

●

●

●

●

Fig.4. Seedsize(weightonlog-scale)andmeansimulatedwater levelsfor95speciesoftemporarypools.Theregressionlinewas cal-culatedfromalinearmodelonlog-transformeddata(p=0.00172, t=−3.2, R2=0.10). Only species with at least 3 records are included.SeeMaterialsandmethodssectionfordetailsofwater levelsimulation.

There was a significant negative relation between log-transformed seed size and log-transformed abundance in the soilseedbank withaslopeof−0.3±0.1(t=−2.480, p=0.0154).Withgradientpositionasacovariate,therewas norelationbetweenseedsizeor seedabundanceand mea-suresofseedspreadinthesoil.

Discussion

Seed bank abundance and probability of detection of plant species in the seedbank by the seedling emergence methodincreasedcontinuouslywithincreasinglymoist habi-tats,extendingpreviousstudiesthatreportedgenerallyhigh levels of seed bank for ephemeral wetlands (Brock 2011;

Cross etal. 2015; Faist & Collinge 2015; Leck & Brock 2000; Rhazi etal.2001). The effectof moisture niche on seedbankisstatisticallyindependentfromseedsizeinour dataandisthusnottriggeredbyincreasingseedsizedueto drought adaptation(Leishman&Westoby1994).Thehigh levelofriskofaquaticspeciessuggestedbytheirabundant soilseedbanksunderlinespreviousstudiesshowingthatthese species rely heavily on soilseed banks(Bonis, Lepart,& Grillas1995;Brock2011).Increasingsoilseedbankswith higher moisture mightbe interpreted as aconsequence of gravitational movementofseeds invernalpools.However, moistureisthemainfactorthatchangesconsistentlyinour transectseveninflatorveryruggedareas.Moreover, abun-dance of soilseedbankhas beenanalyzed for position of adult plants,not forseedpositions,andspecies show

spe-cificlowerlimitsofseeddistributions(Figs.1Aand2).We thereforethinkthat moisture nichesof adultplants trigger this change in the soil seed bank more than gravitational movement.Higher seeddensitiesfor thebottomcompared tothemarginsofCalifornianvernalpoolsconfirmthis ten-dencyforotherregions(Faist&Collinge2015)andpointtoa relationbetweenmoisturenicheandseedbanking.Plantscan achievehighabundanceofseedsinthesoilbyproducinghigh numbersofsmallseeds,increasedlevelsofdormancy,light andfluctuating temperature requirements (Saatkampet al. 2009;Saatkamp,Poschlod,&Venable2014).Forplantsof Mediterraneantemporary pools detailed laboratory studies (Carta2016;Carta,Bedini,Müller,&Probert2013)indicate lightanddiurnallyfluctuatingtemperaturerequirementsas mechanismsforthebuildupofsoilseedbanksandthatanoxia preventsgermination.Experimentalstudiestryingtoexhaust seedbanksofAustraliantemporarypoolsshowedthatmany effectivewateringcyclesormorethantwelveyearsof succes-siveexposuretogerminationseasonsareneededtodeplete soilseedbankstoalowlevel,indicatingthatdormancyand specificgermination conditions maintain an ungerminated butviableseedreservoirinthesoiloftemporarypools(Brock 2011;Crossetal.2015).Togetherwithdatapresentedhere thissuggeststhatagradientofincreasingseedbankingbuilt uponadaptivetraitssuch as germinationrequirements and reducedseedsizeexistsfromthesurroundingstotheinterior oftemporarypools.

Seedsrecordedinthesoilseedbankshowedhigherspread along the gradient for species with intermediate moisture requirements.Thiswasindependentofsoilseedabundance orseedsize,whichconfirmsthatspreadmeasureswere inde-pendent of sample size. Weacknowledge that small-scale dispersalforindividualplantscanonlybequantifiedbyusing seedtrapsaround isolated individuals duringthe dispersal season(Bonis, Lepart,& Grillas1995), butwe think that spreadofseedstrappedbysoilisanindicatorofsmall-scale dispersal.Higherspreadof theseeds inthesoilseedbank for intermediate species can also result from year-to-year shiftingof above ground plantpopulations. However, itis difficultto quantify how much thisshifting contributes to seedspreadinourdatacomparedtodifferentdispersal dis-tances.Nevertheless, whenpopulationsmove accordingto moving environmental conditions, all species should have similarspreadsofseedsinthesoilincludingaquaticor xero-phytic species, butthis isnot the casein ourdata, where wedetectedahigherspreadofintermediatespecies,while controllingforseedproduction.Wesuggestthatamodelling approachandinsitudispersal dataare neededto quantify dispersalinmoisturegradients.

Evidently,species onthe moistordrygradientend can-notmovetobetterpatchesonalocalscale,sinceinextreme yearstherearenomoisterordriersitesandadultplantswould die fromdrought or flooding.Plantsmust thushave other adaptationsforlocalpersistencewithoutmovementsuchas aseedbankoralargeseedsize.Thesemightbecombined withadaptationsforlongdistancedispersalinsteadoflocal

scaledispersalsuchashooks(Fischer,Poschlod,&Beinlich 1996;Römermann,Tackenberg,&Poschlod2005)or adap-tationsforlongdistancedispersalbywind(Cheptou,Carrue, Rouifed,&Cantarel2008).Theeffectivelongdistance dis-persalforaquaticspeciesbywaterandbirdsisalongknown fact(Darwin1859;Figuerola2002).But,shortdistance dis-persalofthesespeciesisverydifficulttoassess.

Speciesshiftedtheirmeangradientpositionsbetweensoil seedbankandstandingvegetation.Thetwospecieswitha significantshift,R.paludosusandL.hyssopifolia,havehigh moisturerequirementsandoccupiedamoisternicheasplants inthevegetationthanasseedsinthesoilseedbank.Evidence foralargerspreadinthesoilseedbankforindividualspecies isstronger,threeoffourspeciesforwhichtherewasa signif-icantdifferenceinspreadaboveandbelowgroundhadlarger spreadinthesoilseedbankandallofthesehad intermedi-atepositions.Theseresultssuggestthatmaintainingahigh spreadoftheseedbankinlocalgradientsforannualplants is amechanismtoenhancelocal populationpersistence at intermediategradientpositions.

Plant species growing in dry parts showed consistently higherseedsizethanspeciesfromintermediateormoistplots. Thisfindingsupportstheadaptivevalueoflargeseedsindry environments(Baker1972;Dawsetal.2008;Leishman& Westoby 1994). Interestingly,plantswithlargeseeds have smallerinterannualvariabilityinsurvival(Metzetal.,2010;

Pake&Venable1996).Herewesuggestthatriskreduction traitsarerelatedtothenichepositionofaspeciesin small-scale moisture gradients.Plantsin dryparts, seemto rely moreonconservativebethedgingvialargerseedsize.Inour data,specieswithseedslighterthan1mgonlyoccurredon plotswaterloggedduringtheobservationyear(Fig.4). Proba-bly,manylargeseedswithoutimpermeabletegumentswould alsodieinprolongedperiodsofflooding,andthuscontribute tothesize-hydrologyrelation.

Wefoundno evidenceforatrade-off betweendispersal andseedbankwiththeindicesandatthescalewestudied. Includingseedsizeandgradientposition,bothpotential con-foundingfactorswithsignificanteffectondispersalandseed bankabundance,didnotchangethispicture.However,our datasupporttheideathatabundanceof seedsinthesoilis linkedtoseedsizeviatheseedsize-numbertrade-off, includ-ingahigherdetectabilityoflightandmorenumerousseeds (Saatkampetal.2014;Thompson,Band,&Hodgson1993). Ratherthanbeinginatrade-off,seedsizeandseedbank were bothpositively correlated tomoisture. Thissuggests thatseedbanksorheavyseedsarenotalternativebet hedg-ingstrategiesatagivenmoisturecondition(Pake&Venable 1996;VenableandBrown1988)butthatthemaindifferences inthesetraitsaretriggeredbydifferencesinaplant’s mois-tureniche.Moreover,ourfindingsdonotsustaintheideaofa trade-offbetweendispersalandseedbankthathasbeen sug-gestedbytheoreticalworks(e.g.Venable&Lawlor1980). Sincebothdispersal(Bruun&Poschlod2006)andseedbanks (Saatkampetal.2009)dependonhighseedproduction,seed productionincreasestheprobabilityofaseedtoreach

dis-tantplacesortimes.Dispersalandseedbanksaretherefore positivelyrelatedinsteadofbeinginatrade-off.Moreover, ahigherproductivityisknownforlight-seededplantsdueto theseednumber-masstrade-off(Jakobsson&Eriksson2000;

Moles,Falster,Leishman,&Westoby2004),soonewayto increaseseednumberadaptivelyisviadecreasingseedsize (Shipley&Dion1992).Possibly,strongerspatialand tem-poralvariabilityinmoisture bothselect forincreased seed number,loweringseedsizeatthemoistgradientend.

Probably,ourresultsonrisk-reductionmayalsobefound inotherrainfall-relatedlocalmoisturegradientssinceeven outside the context of temporary pools, small-scale mois-turegradientsmovefromoneyeartoanother.Thedifferent niche positions and adaptations of plants to drought and water-logged soils are, therefore, just oneexplanation for differencesinthetemporalvariancebetweenspecieswithin communities(Angertetal.2009;Venable2007).

OurdatasustaintheviewofChildsetal.(2010)that diver-sifying bet hedging should be favored in situations when escape inspace or time is possible,which isthe casefor persistentsoilseedbanksofaquaticspeciesinourdata.

Acknowledgments

We are grateful to Dominique Guicheteau (Réserve

NaturellePlainedes Maures)andSOMECA foraccess to study siteandDREALPACA andCNPNfor permits.We thankPatrickGrillasandAlexandreMillonforhelpful dis-cussionsandDanielPavonandManonBatistaforfieldwork andFrauke Behrendt (University of Brighton) for English languagerevision.SG benefits fromIKEAgranttoIMBE 13/8/2009andASfromRégionProvenceAlpesCôted’Azur

grant“Gévoclé”.

References

Angert,A.L.,Huxman,T.E.,Chesson,P.,&Venable,D.L.(2009).

Functionaltradeoffsdeterminespeciescoexistenceviathe

stor-ageeffect.Proceedingsofthe NationalAcademyofSciences,

106,11641–11645.

Baker,H.G.(1972).Seedweightinrelationtoenvironmental

con-ditionsinCalifornia.Ecology,53,997–1010.

Bauder,E.T.(2000).Inundationeffectsonsmall-scaleplant

distri-butionsinSanDiego,Californiavernalpools.AquaticEcology,

34,43–61.

Bauder,E.T. (2005).The effectsof anunpredictable

precipita-tionregimeonvernalpoolhydrology.FreshwaterBiology,50,

2129–2135.

Bonis,A.,Lepart,J.,&Grillas,P.(1995).Seedbankdynamicsand

coexistenceofannualmacrophytesinatemporaryandvariable

habitat.Oikos,74,81–92.

Brock,M.A.(2011).PersistenceofseedbanksinAustralian

tem-porarywetlands.FreshwaterBiology,56,1312–1327.

Brooks,D.R.(2004).Weather-relatedeffectsonwoodlandvernal

poolhydrologyandhydroperiod.Wetlands,24,104–114.

Bruun,H.H.,&Poschlod,P.(2006).Whyaresmallseedsdispersed

throughanimalguts:Largenumbersorseedsizeperse?Oikos,

113,402–411.

Bullock,J.M.,Shea,K.,&Skarpaas,O.(2006).Measuringplant

dispersal:An introductionto fieldmethodsand experimental

design.PlantEcology,186,217–234.

Bulmer, M. G. (1984). Delayed germination of seeds: Cohen’s

modelrevisited.TheoreticalPopulationBiology,26,367–377.

Carta, A.(2016). Seedregenerationin Mediterranean temporary

ponds: Germination ecophysiology and vegetationprocesses.

Hydrobiologia,782,23–35.

Carta,A.,Bedini,G.,Müller,K.J.,&Probert,R.J.(2013).

Compar-ativeseeddormancyandgerminationofeightannualspeciesof

ephemeralwetlandvegetationinaMediterraneanclimate.Plant

Ecology,214,339–349.

Cheptou,P.O.,Carrue,O.,Rouifed,S.,&Cantarel,A.(2008).Rapid

evolutionofseeddispersalinanurbanenvironmentintheweed

Crepissancta.ProceedingsoftheNationalAcademyofSciences,

USA,105,3796–3796.

Childs,D.Z.,Metcalf,C.J.E.,&Rees,M.(2010).Evolutionary

bet-hedgingintherealworld:Empiricalevidenceandchallenges

revealedbyplants.ProceedingsoftheRoyalSocietyB:

Biolog-icalSciences,2777,3055–3064.

Clauss,M.J.,&Venable,D.L.(2000).Seedgerminationindesert

annuals:Anempiricaltestofadaptivebethedging.American

Naturalist,155,168–186.

Cohen,D.(1966).Optimizingreproductioninarandomlyvarying

environment.JournalofTheoreticalBiology,12,119–129.

Cohen, D., &Levin, S.A.(1987). Theinteractionbetween

dis-persal anddormancystrategiesin varyingandheterogeneous

environments. In Mathematical topics in population biology,

morphogenesisandneurosciences.pp.110–122.Springer.

Cross,A.T.,Turner,S.R.,Renton,M.,Baskin,J.M.,Dixon,K.W., &Merritt,D.J.(2015).Seeddormancyandpersistentsediment

seedbanksofephemeralfreshwaterrockpoolsintheAustralian

monsoontropics.AnnalsofBotany,115,847–859.

Darwin,C.(1859).Ontheoriginof speciesbymeans ofnatural selection,orpreservationof favoredracesinthe strugglefor life.London:Murray.

Daws, M.I., Crabtree,L.M.,Dalling,J.W., Mullins,C. E., & Burslem, D.F.R. P.(2008). Germinationresponses towater

potentialin neotropicalpioneerssuggestlarge-seededspecies

takemorerisks.AnnalsofBotany,102,945–951.

Deil,U.(2005).Areviewonhabitats,planttraitsandvegetationof

ephemeralwetlands—Aglobalperspective.Phytocoenologia,

35,533–706.

Ellenberg,H.(1953).PhysiologischesundökologischesVerhalten

derselbenPflanzenarten.Berichte derDeutschenBotanischen

Gesellschaft,65,351–362.

Faist,A.M.,&Collinge,S.K.(2015).Seedbankcompositionvaries

alonginvasionandinundationgradientsinvernalpoolwetlands.

PlantEcology,216,553–564.

Figuerola,J.(2002).Dispersalofaquaticorganismsbywaterbirds:

Areviewofpastresearchandprioritiesforfuturestudies.

Fresh-waterBiology,47,483–494.

Fischer, S.F.,Poschlod,P.,& Beinlich,B. (1996).Experimental

studiesonthedispersalofplantsandanimalsonsheepin

García-Baquero,G.,Silvertown,J.,Gowing,D.J.,&Valle,C.J. (2015).Dissectingthehydrologicalniche:Soilmoisture,space

andlifespan.JournalofVegetationScience,27,219–226.

Grillas,P.,Gauthier,P.,Yavercovski,N.,&Perennou,C.(2004).

Mediterraneantemporarypools;Vol.1—Issuesrelatingto con-servation,functioningandmanagement.Stationbiologiquede

laTourduValat.

Harrison,R.D.(2000).RepercussionsofElNino:Droughtcauses

extinctionandthebreakdownofmutualisminBorneo.

Proceed-ingsoftheRoyalSocietyB,267,911–915.

Hooftman,D.A.,Edwards,B.,&Bullock,J.M.(2016).Reductions

inconnectivityandhabitatqualitydrivelocalextinctionsina

plantdiversityhotspot.Ecography,39,583–592.

Inouye,B.D.(2000).Theecologicalandevolutionarysignificance

offrostinthe context ofclimatechange.Ecology Letters,3,

457–463.

Jakobsson,A.,&Eriksson,O.(2000).Acomparativestudyofseed

number,seedsize:seedlingsizeandrecruitmentin grassland

plants.Oikos,88,494–502.

Keeley,J.E.(1998).CAMphotosynthesisin submergedaquatic

plants.TheBotanicalReview,64,121–175.

Leck,M.A.,&Brock,M.A.(2000).Ecologicalandevolutionary

trendsinwetlands:EvidencefromseedsandseedbanksinNew

South Wales,Australia and NewJersey, USA.Plant Species

Biology,15,97–112.

Leishman,M. R.,& Westoby, M.(1994). Therole ofseedsize

inseedlingestablishmentindrysoilconditions—

Experimen-tal evidence fromsemi-arid species. Journalof Ecology, 82,

249–258.

Metz,J.,Liancourt,P.,Kigel,J.,Harel,D.,Sternberg,M.,& Tiel-börger,K.(2010).Plantsurvivalinrelationtoseedsizealong

environmentalgradients:Along-termstudyfromsemi-aridand

Mediterraneanannualplantcommunities.JournalofEcology,

98,697–704.

Moles,A.T.,Falster,D.S.,Leishman,M.R.,&Westoby,M.(2004).

Small-seededspeciesproducemoreseedspersquaremetreof

canopyperyear,butnotperindividualperlifetime.Journalof

Ecology,92,384–396.

Moles,A.T.,&Westoby,M.(2006).Seedsizeandplantstrategy

acrossthewholelifecycle.Oikos,113,91–105.

Nathan,R.,&Muller-Landau,H.C.(2000).Spatialpatternsofseed

dispersal,theirdeterminantsandconsequencesforrecruitment.

TrendsinEcology&Evolution,15,278–285.

Oksanen,J.,Kindt,R.,Legendre,P.,O’Hara,B.,Simpson,G.L., Solymos,P.,etal.(2007).Vegan:Communityecologypackage.

Rpackageversion2.2–0.

Pake,C.E.,&Venable,D.L.(1996).Seedbanksindesertannuals:

Implicationsforpersistenceandcoexistenceinvariable

environ-ments.Ecology,77,1427–1436.

Philippi,T.,&Seger,J.(1989).Hedgingone’sevolutionarybets,

revisited.TrendsinEcology&Evolution,4,41–44.

R-Core-Team.(2014).R:Alanguageandenvironmentfor statis-ticalcomputing.Vienna,Austria:RFoundationforStatistical Computing.

Rees,M.(1994).Delayedgerminationofseeds:Alookattheeffects

ofadultlongevity,thetimingofreproduction,andpopulation

age/stagestructure.AmericanNaturalist,144,43–64.

Rhazi,L.,Grillas,P.,TanHam,L.,&ElKhyari,D.(2001).The

seedbankand thebetweenyears dynamicsofthe vegetation

ofaMediterraneantemporary pool(NWMorocco).Ecologia

Mediterranea,27,69–88.

Römermann, C., Tackenberg, O., & Poschlod, P. (2005). How

to predict attachment potential of seeds to sheep and

cat-tle coat from simple morphological seed traits. Oikos, 110,

219–230.

Rousseeuw,P.J.,&Croux,C.(1993).Alternativestothemedian

absolutedeviation.JournaloftheAmericanStatistical

Associa-tion,88,1273–1283.

Saatkamp,A.,Affre,L.,Dutoit,T.,&Poschlod,P.(2009).Theseed

banklongevityindexrevisited:Limitedreliabilityevidentfrom

aburialexperimentand databaseanalyses.AnnalsofBotany,

104,715–724.

Saatkamp,A.,Poschlod,P.,&Venable,D.L.(2014).Thefunctional

roleofsoilseedbanksinnaturalcommunities.InR.S.Gallagher

(Ed.),Seeds—Theecologyofregenerationinplantcommunities

(pp.263–294).Wallingford:CABI.

Shipley, B.,& Dion, J.(1992). TheAllometry of Seed

Produc-tion in Herbaceous Angiosperms. American Naturalist, 139,

467–483.

Siewert,W.,&Tielbörger,K.(2010).Dispersal-dormancy

relation-shipsin annual plants:Puttingmodel predictionsto the test.

AmericanNaturalist,176,490–500.

Silvertown,J.(2004). Plantcoexistenceand theniche.Trendsin Ecology&Evolution,19,605–611.

Silvertown,J.,Araya,Y.,&Gowing,D.(2015).Hydrologicalniches

interrestrialplantcommunities:Areview.JournalofEcology,

103,93–108.

Simons, A. M. (2011). Modes of response to environmental

change and the elusive empirical evidence for bet hedging.

Proceedingsofthe RoyalSocietyB-BiologicalSciences,278,

1601–1609.

Snyder,R.E.(2006).Multipleriskreductionmechanisms:Can

dor-mancysubstitutefordispersal?EcologyLetters,9,1106–1114.

TerBraak,C.J.(1986).Canonicalcorrespondenceanalysis:Anew

eigenvectortechniqueformultivariatedirectgradientanalysis.

Ecology,67,1167–1179.

ter Heerdt,G.N.J.(1996). Animproved methodforseed-bank

analysis:Seedlingemergenceafterremovingthesoilbysieving.

FunctionalEcology,10,144–151.

Thompson,K.,Band,S.R.,&Hodgson,J.G.(1993).Seedsize

and shapepredict persistencein soil.Functional Ecology, 7,

236–241.

Tielbörger,K.,&Petru,M.(2008).Germinationbehaviourofannual

plantsunderchangingclimaticconditions:Separatinglocaland

regionalenvironmentaleffects.Oecologia,155,717–728.

Tielbörger, K., & Valleriani, A. (2005). Canseeds predict their

future?Germinationstrategiesofdensity-regulateddesert

annu-als.Oikos,111,235–244.

Tison,J.M.,Jauzein,P.,&Michaud,H.(2014).FloredelaFrance méditerranéennecontinentale.Turriers:NaturaliaPublications.

Venable,D.L.(2007).Bethedgingin aguildofdesertannuals.

Ecology,88,1086–1090.

Venable,D.L.,&Brown,J.S.(1988).Theselectiveinteraction

ofdispersal,dormancyandseedsizeasadaptationsfor

reduc-ingrisks invariableenvironments.American Naturalist,131,

360–384.

Venable,D.L.,&Lawlor,L.(1980).Delayedgerminationand

dis-persalindesertannuals:Escapeinspaceandtime.Oecologia,

Violle,C.,Bonis,A.,Plantegenest,M.,Cudennec,C.,Damgaard, C., Marion, B., et al. (2010). Plant functional traits capture

speciesrichnessvariationsalongafloodinggradient.Oikos,120,

389–398.

Volis, S., & Bohrer, G. (2013). Joint evolution of seed traits

alonganariditygradient:Seedsizeanddormancyarenottwo

substitutable evolutionary traits in temporally heterogeneous

environment.NewPhytologist,197,655–667.

Walck,J.L.,Baskin,J.M.,Baskin,C.C.,&Hidayati,S.N.(2005).

Defining transient and persistent seed banks in species with

pronouncedseasonaldormancyandgerminationpatterns.Seed

ScienceResearch,15,189–196.

Zuur,A.F.,Ieno,E.N.,Walker,N.J.,Saveliev,A.A.,&Smith,G. M.(2009).Mixedeffects,modelsandextensionsinecologywith