Topsoil
removal
in
degraded
rich
fens:
Can
we
force
an
ecosystem
reset?
Willem-Jan
Emsens
a,*
,
Camiel
J.S.
Aggenbach
a,b,
Alfons
J.P.
Smolders
c,d,
Rudy
van
Diggelen
aaEcosystemManagementResearchGroup,DepartmentofBiology,UniversityofAntwerp,Universiteitsplein1C,2610Wilrijk,Belgium b
KWRWatercycleResearchInstitute,P.O.Box1072,3430BBNieuwegein,TheNetherlands
c
DepartmentofAquaticEcologyandEnvironmentalBiology,InstituteforWetlandandWaterResearch,RadboudUniversityNijmegen,Heyendaalseweg1351, NL-6525EDNijmegen,TheNetherlands
d
B-WAREResearchCentre,Toernooiveld1,6525EDNijmegen,TheNetherlands
ARTICLE INFO Articlehistory:
Received26September2014
Receivedinrevisedform12January2015 Accepted17January2015 Availableonlinexxx Keywords: Alternativestate Drainage Eutrophication Peatlandrestoration Richfen Topsoilremoval ABSTRACT
Global land-use intensification and drainage has altered the biogeochemical properties of many peatlands, and concomitant eutrophication has led to a loss of low-competitive fen species. We investigatedthehypothesisthatremovalofadegradedandeutrophiedtoppeatlayer,therebyexposing anunderlyingpeatlayer,canimproveconditionsforrichfenrestoration.Westudiedthelong-term (3–18years)effectsofpasttopsoilremovalinsixrichfensinWesternEuropebycomparingtopsoil removalplotswith(untouched)controlplots.Overall,topsoilremovalplotswerecharacterizedbylower bulkdensitiesandsoilnutrientpoolsofPandKCl-extractableNH4+,whileorganicmattercontentsand soilC:Nratioswerehigher.PorewaterconcentrationsofNO3andNH4+alsodecreasedinthetopsoil removal plots, while concentrations of base cations (Ca2+, Mg2+, Na+, K+) and HCO
3 increased. Furthermore,lowernutrientlevelsappearedtorestrictherbbiomassproductioninthetopsoilremoval plots,sothatoptimizedlightconditionsledtotheestablishmentoflight-demandingtargetspeciesanda significantincreaseinbryophytecover.Multivariateanalysisrevealedthatmostvariationinvegetation assemblywasduetohighergroundwaterlevelsinthetopsoilremovalplots,closelyfollowedbyahigher relativelightintensity(RLI)atsurfacelevel,lowerporewaternutrient(NH4+)concentrations,andhigher concentrationsofbasecations.Weconclude thattopsoilremovalcanbeaneffectivemechanismto “reset”adegradedpeatlandtoitsinitialstateofnutrientlimitation,basesaturationandhighavailability oflight,therebyimprovingtheconservationprospectsofendangeredrichfencommunities.
ã2015ElsevierB.V.Allrightsreserved.
1.Introduction
Theglobalintensificationoflanduse,asharpincreaseintheuse ofartificialfertilizer,and anincreasein anthropogenicnitrogen deposition is compromising the functioning of many nutrient-limited ecosystems, both in the aquatic and (semi-) terrestrial environment(Matsonetal.,1997;Phoenixetal.,2006).Asmany endangered species are adapted to nutrient-poor habitats, eutrophication is considered a major threat to global species diversity(Wassenetal.,2005;Hautieretal.,2009).
Groundwater-fedpeatlands(henceforth“richfens”)are partic-ularlyvulnerabletoland-useintensificationand eutrophication. Typically,pristinerichfensarecharacterizedbycontinuouslywet,
base-richandmesotrophicconditionsandaredominatedby low-competitive small sedges and brown mosses (Grootjans et al., 2006;MalsonandRydin,2007).InEurope,however,mostofthe pristinerichfenshavedisappearedduetolandusechange,while degradationof theremainingfensleadstothegradual replace-mentoftypicalfenspecieswithgeneralwetlandspecies(Kooijman 1992;Lamersetal.,2014).Therefore,boththeconservationand restorationofthemanydegradedfensisatoppriorityforthe long-term protection of this habitat type together with its typical species(vanDiggelenetal.,2006).Thisisnowlegally acknowl-edgedthroughtheEU’sHabitatsDirective(Romão,1996).
Incomparisonwithmineralsoils,peatlandeutrophicationisa morecomplexprocessasnutrientenrichmentisnotnecessarily relatedtoanincreasedinputfromexternalsourcesalone(Bragazza etal.,2009).Aspeatsoilmainlyconsistsofreactiveorganicmatter, the slightest alterations in hydrological conditions can have disproportional effects on fen chemistry. In this respect, * Correspondingauthor.Tel.:+3232652268.
E-mailaddress:willem-jan.emsens@uantwerpen.be(W.-J.Emsens).
http://dx.doi.org/10.1016/j.ecoleng.2015.01.029
0925-8574/ã2015ElsevierB.V.Allrightsreserved.
ContentslistsavailableatScienceDirect
Ecological
Engineering
desiccationis consideredone of the majorthreats torich fens (vanDiggelenetal.,2006;Lamersetal.,2014).Desiccationcanbe triggered through direct local drainage (e.g., construction of drainageditches),butit canalso betheresultofalterations in regionalhydrology(e.g.,increasedratesofgroundwater abstrac-tioncanreduceregionalseepagefluxes(vanDiggelenetal.,2006)). Whenpeatsoildesiccates,intrusionofoxygenbecomesadriving forcefor increased rates of organicmatter decomposition and mineralization(Brounsetal.,2014), whicheventuallyresultsin nutrient release and eutrophication (Grootjans et al., 1986). Moreover, peat oxygenation triggers carbon loss (Laiho, 2006; Brouns et al., 2014), soil subsidence (Gambolati et al., 2006), regenerationofelectronacceptors(Fenneretal.,2011),acidi fica-tion(Beltmanetal., 2001;Cusell et al.,2013),leaching ofbase cations(Laihoetal.,1999),shiftsinvegetationassemblyandaloss of typicalbiodiversity (Malsonand Rydin, 2007; Malson et al., 2008). Long-term peat degradation can alter biogeochemical conditions to such an extent that successful fen restoration becomes notoriously difficult, even after restoration of the hydrological conditions (Zak et al., 2010, 2014; Brouns et al., 2014).Eventually,itislikelythatthesumofthesebiogeochemical alterationsinthetopsoilcantriggerashifttowardasystemthatis governed by a different set of positive feedback mechanisms, possiblyforcingthepeatlandtowardanalternativedegradedstate (Sudingetal.,2004).
Inthisstudy,wetestthehypothesisthatremovalofthetoplayer of(disturbed) peattorestorerichfen ecosystemscan trigger asystem “reset”(i.e.,aftertopsoilremoval,biogeochemicalconditionswill becomesimilartoinitialpristineconditions),andthatitimproves conditions for the establishment of target plant communities. Althoughtopsoilremovalisawell-establishedmeasureinnature conservationonmineralsoils(AllisonandAusden2004;Olssonand Ödman2014),theeffectoftopsoilremovalinrichfens(inwhichan underlyingpeatlayerisexposed)isonlydocumentedfragmentarily (Patzeltetal.,2001,2007;Klimkowskaetal.,2007,2015),oftenwith contrasting results. We believe this is due to the difficulty of predictingtheeffectsofexposinganunderlyinghigh-qualitypeat layer after topsoil removal, as rapid peat mineralization and concomitant re-eutrophication is not unlikely (Brouns et al., 2014).Toempiricallytesttheeffectsoftopsoilremovalonabiotic conditionsandrichfendevelopment,weconductedacomparative fieldstudyin whichweanalyzedthemid-term(3–18years)effects of pasttopsoilremovalinsixrichfens.
2.Methods 2.1.Studysites
Topsoil removal is an uncommon restoration measure for peatlands.Therefore,studysiteselectionwasdeterminedmainly bythe availabilityof suitable locations.Sites were included for samplingonlyiftheymetthefollowingcriteria:(1)peat-formation hadstarteddirectlyonmineralsoils,thusexcludingfloatingmires; (2)thepeatlandwasfedbybase-richgroundwater;(3)allpeatlands weredrainedinthepast,havingledtotopsoildegradation(Vonpost
humification topsoil>8); (4) topsoil removal has taken place>3yearsagoandtheareahasre-vegetated;(5)aftertopsoil removal, a new underlying peat layer was exposed (excluding locationswhereallpeatwasremoveddowntothemineralsubsoil); and(6)controlplots(plotsinwhichtopsoilremovalhasnottaken place)wereavailableinthevicinityofthetopsoilremovalplots,to allowameaningfulcomparison.Intotal,welocatedsixfensinthe NetherlandsandBelgiumthatmetallcriteria(Table1,Fig.1).Allsites haveahistoryofagriculturethatincludedhaymakingand(limited) fertilization, while two of the sites(DA and PE)had alsobeen rewettedbycanalandditchblocking.
2.2.Samplingdesign
InJune2013,withineachstudysiteweselectedeightplotsof 2m2m,resultinginatotalof48plots.Fourreplicateplotsper sitewereselectedrandomlyinthezonewheretopsoilremovalhad taken place, and four replicate control plots were selected on nearbyspotswherethedegradedtopsoilhadbeenleftuntouched. IneachplotwerecordedthecoverofallspeciesusingtheLondo scale (Londo, 1976). An Accupar LP-80 ceptometer (Decagon DevicesInc.,Pullman,WA,USA)wasusedtomeasurerelativelight intensity(RLI)belowthevegetationatsurfacelevel(=percentage ofincomingphotosyntheticallyactiveradiationasmeasuredwith areferencesensorabovethecanopy(KotowskiandvanDiggelen, 2004)), averaged over 4 measurements per plot. This is an important measure as a sufficiently high availability of light (indicatinglimitedherbproductivityandthuslesscompetition)is crucial for rich fen communities (Kotowski and van Diggelen, 2004;Kotowskietal.,2006).Toascertainthatahighrelativelight intensitycorrelates with a decreased productivity at our study sites,weharvestedtheabove-groundherbbiomassina randomly-placed sub-plotof0.4m0.4mwithineach ofthelargerplots. Next,foursoilsub-samplesperplotweretakenfromtheupper 10cmofthepeatsoil,andmixedintoonehomogeneoussample. Separate soil samplesweretaken for bulk densitycalculations. Elementpoolsizesinthesoilareanindicationoftotallong-term availability, but soil-bound elements are only partially plant-available.We,therefore,additionallycollectedporewatersamples fromtheupper10cmofthesoilineachplotusingmacro-rhizon samplers with a pore size of 0.15
m
m (Rhizosphere Research Products,theNetherlands).Elementsdissolvedintheporewater are directly plant-available, but concentrations are much more subjecttotemporalvariations.Porewatersampleswerestoredat 4untilfurthertreatment.2.3.Chemicalanalyses
WemeasuredpHandECofallporewatersamplesdirectlyin thefieldusingportablefieldequipment(WTWmulti340i).Total inorganiccarbon(TIC)wasanalyzedonaninfraredgasanalyzer (ABB Advance Optima): the resulting values were used to calculate HCO3 concentrations. Concentrations of NH4+ and NO3 were determined on an auto analyzer 3 system (Bran+ Luebbe) using ammonium molybdate, hydrazine sulphate and
Table1
Studysitelocationwithcoordinates,alongwithtime(yr)sincetopsoilremoval,andaveragedepth(cm)oftopsoilremoval.
Location Coordinates Timesincetopsoilremoval(yr) Averagedepthoftopsoilremoval(cm)
LeijerHooilanden(LH) 5238033.7300N;616043.2300E 10 15 Hellen(HE) 520033.9400N;534048.9300E 10 30 DrentscheAa(DA) 530049.6500N;637045.8800E 18 20 Holmers(HO) 5254012.4100N;637045.8300E 11 30 Peizermade(PE) 5310019.3300N;630012.1900E 3 30 Malendriesbeekvallei(MA) 5050056.4800N;452024.3900E 5 25
salicylate. Pore water sub-samples were acidified by adding 0.7mL 65% suprapure HNO3 per 100mL sample and were analyzed on ICP (IRIS Intrepid II) for the following elements: Ca,Mg,K,Na,Fe,Mn,P,S,andAl.
Soilsamplesforbulkdensity(101010cm)weredried(72h at 105C) and weighed;results were expressed in kgL1. Soil organicmattercontent(%)wasdeterminedbyloss-on-ignitionfor 4hat550C.KCl-extractionsandammonium-oxalateextractions (indarkness)onmoist soilallowed thedeterminationof pHKCl, NO3-N, NH4-N and P-oxalate (Pox), respectively. The latter is a measure for reactive soil-P bound to amorphous components includingFeandAl.Oven-drysoilsub-samples(48hat70C)were homogenizedandgroundinliquidN.CandNcontents(%)were determinedwithaCarloErbaNA1500elementalanalyser(Thermo FisherScientific). 200mg of soilwas digested with4mL HNO3 (65%) and 1mL H2O2 (30%) using a microwave labstation (Milestonesrl)tomeasuretotal Ca,Mg, K, P,S, Fe,Al,and Mn withICP.Valueswerecalculatedfordryweightsoil.Above-ground herbbiomasswasoven-dried(48hat70C)andweighed;results wereconvertedtotonsperhectare(tha1).
2.4.Dataanalyses
Assoil bulk density differs stronglybetween different fens, contentof soilchemical parameterswas expressed pervolume unit (mmolL1). Values of KCl-extractableNO3-N were largely below the detection limit and were, therefore, excluded from furtheranalyses.
Beforestatisticalanalysis,environmentaldatawerecheckedfor normalitybasedonvisualinspectionof histogramsand normal Q–Q plots. If needed, data were transformed using either logarithmic, square root, or inverse transformations to attain
approximate normal distributions. Species cover values from vegetationrelevéeswereconvertedintopercentages.
Totestforthemaineffectsoftopsoilremovalon biogeochemi-cal parameters, we ran a mixed-effect model using restricted maximumlikelihood(REML)estimationinwhichwetreatedthe factor“topsoilremoval”(no=0,yes=1)asafixedeffectand“study site”(LH,HE,DA,HO,PEorMA)asarandomeffect.Thelatterwasa deliberatechoiceasourstudysitescanbeconsideredacollection ofrandomsamplesdrawnfroma(theoretically)largepoolofrich fens to which we would like to extrapolate (Bennington and Thayne,1994).Thismodel,therefore,allowedtotestforthemain effects of the treatment “topsoilremoval” while correcting for inter-sitevariation,inwhichwearenotinterested.
VegetationdatawerestoredinTurboveg2.75(Hennekensand Schaminee,2001).Next,datawereexportedtotheJUICEsoftware (Tichy,2002)inordertolinkEllenberglight indicatorvaluesto eachofthespeciesinthedataset.Foreachplot,wethencalculated anaverage“species’lightrequirementindex(henceforth“SLRI”)”. TheseindiceswereobtainedbyaveragingthemeanEllenberglight indicatorvaluesforallindividualspeciesthatwerefoundwithin the same plot.To have a measure for restoration successafter topsoilremoval,wecountedthenumberoftargetspeciesperplot. Atargetspeciesmetatleastoneofthefollowingtwocriteria:(1) the speciesis listed onthe “red list” of either theNetherlands (vanderMeijdenetal.,2000)orFlanders(vanLanduytetal.,2006), or(2)thespeciescanbeconsideredtypicalforsmall-sedgeand brown-moss richfen vegetation inWestern Europe.Alistwith typicalspecieshadbeenconstructedinadvance(AppendixA).This listisbasedonabroadassessmentofrichfenrelevéesinwhichwe included all species witha frequency of >20% in the relevées, combined with rare low-frequency species that are considered highlycharacteristicforrichfens(Schamineeetal.,1995).
The relationship between plant communities and abiotic characteristics in plots with and without topsoil removal was analyzedthroughmultivariateanalysis.First,speciescovervalues wereLog10(x+1)transformed.Wethen rana detrended corre-spondenceanalysistodeterminethetotallengthofthegradient (length>5). Next, abiotic variables were selected by forward stepwiseselectioninacanonicalcorrespondenceanalysis(CCA) and a significance test based on permutations (p<0.05 at 499permutations).Toaccountforsiteeffects,permutationswere restrictedbyaddingtheblockingfactor“site”asacovariable.We used pore water variables rather than soil variables (pH and concentrationsofdissolvedHCO3,Al,Ca,Fe,K,Mg,Na,Mn,P,S, NH4+andNO3)intheordinationbecausethesearemostrelevant torootingplants. Additionally,weincluded groundwaterlevels (“waterlevel”)andrelativelightintensityatsurfacelevel(“RLI”). Extravariableswereonlyaddedtothemodeliftheyshowedno strongcorrelationwithanyoftheprecedingvariables(inflation factor<20).
UnivariatestatisticalanalyseswereperformedinSPSS20(SPSS Inc.).Formultivariateanalyses,weusedCANOCOforWindows4.5 (terBraakandŠmilauer,2002).
3.Results
3.1.Peatandporewaterchemistry
Largedifferencesinsoilpropertieswerefoundbetweenplots withandwithouttopsoilremoval(AppendixB).Overall,thetopsoil removalplots arecharacterized bya highersoilorganicmatter content (F1,41=36.94, p<0.001) and lower bulk density (F1,41=30.26, p<0.001) (Table 2), both of which are strongly correlated (Spearman’s rho=0.865, d.f.=10, p<0.001). This observeddecreaseinbulkdensitycorrelates witha decreasein totalpoolsizeofallminerals,withtheexceptionofstocksofCa,K andSwhichremainedunaltered.Interestingly,themostnotable effectswerefoundfortotalnutrientpoolsizes:poolsoftotal-Pand oxalate extractable-P weredrastically lower in topsoil removal plotsat all sites(totalpool size upto>6 times lower(P-total: F1,41=141.9,p<0.001,Pox:F1,41=118.4,p<0.001)),whilesoil C:N-ratiosincreased(F1,41=71.52,p<0.001)andKCl-extractableNH4+ -stocksdecreased(F1,41=13.82,p<0.001).
While mineralstocks in thesoilwereequal or lowerinthe topsoilremovalplotsthaninthecontrolplots,mineral concen-trationsintheporewaterfollowedanoppositetrend(AppendixC,
Table3).ConcentrationsofbasecationsCa(F1,41=20.77,p<0.001), Mg (F1,41=15.85, p<0.001), Na (F1,41=58.47, p<0.001) and K (F1,41=10.25, p<0.01) as well as concentrations of HCO3
(F1,41=34.58, p<0.001) and dissolved Fe (F1,41=4.98, p<0.05) weregenerallyhigherinthetopsoilremovalplots,correlatingwith increased groundwater levels (F1,41=116.85, p<0.001) and a higherpH(F1,41=15.79,p<0.001).Furthermore,inorganic nitro-gen concentrations (NO3 and NH4+) decreased (F1,41=27.71, p<0.001 and F1,41=16.75, p<0.001, respectively), but concen-trationsofdissolvedphosphorusremainedunaltered(F1,41=2.52, p>0.05).
3.2.Floristicresponsetotopsoilremoval
Intotal,weregistered116speciesofvascularplants(5were identifiedtogenuslevel)and16speciesofbryophytes(2togenus level)(AppendixD).37ofthesespecies(28%)wereclassifiedas typicalofrichfensand15(11%)wereredlistspecies.Asmostofthe redlistspecieswerealsoclassifiedastypicalfenspecies(e.g.,Carex diandra,Menyanthestrifoliata, ...),therewere38“targetspecies” inourdataset(29%oftotalspeciescount).Theremaining71%were considered general wetland or meadow species with a much broaderamplitude(e.g.,Juncuseffusus,Menthaaquatica, ...).
Comparedwiththecontrolplots,topsoilremovalplotshavea lower herb biomass at all study sites (F1,41=72.54, p<0.001, Appendix E, Table 4), which correlates with increased light intensity(RLI)at thesurface level(Pearson’s r=0.842, df=10, p<0.001,Fig.2).Consequently,wefoundanincreaseinbryophyte cover(F1,41=19.58,p<0.001)andSLRI(F1,41=28.43,p<0.001)in thetopsoilremovalplots.Bothbiodiversity(asdefinedbythetotal numberofspecies)aswellasthefractionoftargetspecieswas higherinallofthetopsoilremovalplots(F1,41=45.59,p<0.001and F1,41=37.63,p<0.001,respectively).
Canonicalcorrespondenceanalysisresultedinatotalofeight significant variables that partly explain variation in vegetation assemblagesbetween theplots(Fig.3):groundwaterlevels,RLI, Table2
Resultsofthemixed-effectmodelfortheeffectsoftopsoilremovalontopsoil chemistry(mmolL1), corrected for study site.NS, not significant; *
p<0.05;
**p<0.01;***p<0.001;and-,decreaseatallsites;--,decreaseatmostsites;0,no
change;+,increaseatmostsites;++,increaseatallsites.
Effect Dependentvariable d.f. F-value p-value Direction
Topsoilremoval pHKCl 1,41 18.55 *** + NH4-N 1,41 13.82 *** -P 1,41 141.90 *** - -Pox 1,41 118.40 *** - -C:N 1,41 71.52 *** ++ S 1,41 0.10 NS 0 K 1,41 2.16 NS 0 Ca 1,41 0.46 NS 0 Mg 1,41 8.15 ** -Fe 1,41 29.66 *** -Al 1,41 14.92 *** -Mn 1,41 47.18 *** - -Bulkdensity 1,41 30.26 *** -OM-content 1,41 36.94 *** ++ Table3
Resultsofthemixed-effectmodelfortheeffectsoftopsoilremovalonporewater chemistry (mmolL1), corrected for study site. NS,not significant; *
p<0.05;
**p<0.01;***p<0.001;and--,decreaseatallsites;-,decreaseatmostsites;0,no
change;+,increaseatmostsites;++,increaseatallsites.
Effect Dependentvariable d.f. F-value p-value Direction
Topsoilremoval pH 1,41 15.79 *** + HCO3 1,41 34.58 *** + NH4+ 1,41 16.75 *** -NO3 1,41 27.71 *** -P 1,41 2.52 NS 0/-S 1,41 29.54 *** - -Na 1,41 58.47 *** + K 1,41 10.25 ** + Ca 1,41 20.77 *** ++ Mg 1,41 15.85 *** ++ Fe 1,41 4.98 * + Al 1,41 5.52 * -Mn 1,41 0.59 NS 0 Groundwaterlevel 1,41 116.85 *** ++ Table4
Resultsofthemixed-effectmodelfortheeffectsoftopsoilremovalontotalherb biomass,totalnumberofspecies,mosscover,fractionoftargetspecies,andthe species’ lightrequirementindex(SLRI)per plot.NS,not significant;*
p<0.05;
**
p<0.01;***
p<0.001;and--,decreaseatallsites;-,decreaseatmostsites;0,no change;+,increaseatmostsites;++,increaseatallsites.
Effect Dependentvariable d.f. F-value p-value Direction
Topsoilremoval Herbbiomass(t/ha) 1,41 72.54 *** -
-No.ofspecies 1,41 45.59 *** ++
SLRI 1,41 28.43 *** ++
Bryophytecover(%) 1,41 19.58 *** +
andporewaterconcentrationsofNH4+,S,Ca,K,AlandMn.Axis1 and2combinedexplain42.1%ofthetotalspecies–environment relation.Species–environmentcorrelationswithaxis1 (eigenval-ue=0.49) and axis 2 (eigenvalue=0.35)equaled 0.95 and 0.93, respectively (total inertia=6.09). The plots with and without topsoilremovalwerelargelyseparatedalongthehorizontalaxis (axis1),whichcorrelatedbestwithgroundwaterlevels(r=0.88), RLI(r=0.50)and porewaterconcentrationsof NH4+(r=0.60), S(r=0.54),Ca(r=0.43),Al(r=0.35)andMn (r=0.26).Some separationcanalsobeobservedalongtheverticalaxis(axis 2), whichcorrelateswithporewaterconcentrationsofK(r=0.74, generallyhigherinthetopsoilremovalplots).
4.Discussion
Ourresultsshowthatremovalofadegradedandeutrophiedtop peatlayercanconsiderablyimprovegeochemicalconditionsforrich fendevelopment.Duringthefirsttwodecadesnutrientavailability (eitherN,Porboth)droppedsharplyineachofthesixstudysites, whereaslightavailability(RLI)andbasecation(Ca2+,Mg2+,Na+,K+) concentrationsintheporewaterincreased,correlatingwithhigher groundwater levels. In response, bryophytes expanded and the numberofrichfentargetspeciesincreased.
4.1.Peatandporewaterchemistry
In our study, hydro-geochemical conditions in the top soil differed markedly between plots with and without topsoil removal. One of the inherent effects of topsoil removal is the concomitantloweringofthesoilsurfacelevelandanincreasein relativegroundwaterlevels,whichweobservedinallsites.The resulting increased influence of base-rich groundwater leads to a higher baseavailability and alkalinity(Boeye et al.,1995; Lamersetal.,2014),whichcorrespondswithourobservationof higherporewaterconcentrationsofCa2+,Mg2+,Na+,K+andHCO
3 in the topsoil removal plots. Correspondingly, we observed significantly higher pHKCl-values in the topsoil removal plots (rangingfrom5.2to6.3).
Topsoil removal led to a drastic decline in nutrient concen-trations:concentrationsofnitrogenwerelowerbothinthepeatsoil (lowerNH4+-N,higherC:Nratios)andintheporewater(NH4+and NO3)ofthetopsoilremovalplots,andasimilardistinctpatternwas found for soil pools of total-P and oxalate-extractable P. Lower concentrations of nitrate after topsoil removal can be due to completeremovaloftheeutrophiedtoplayer,butitcanalsobea consequenceofwetter(moreanoxic)conditions,whichenhances denitrificationbymicro-organisms(WhitmireandHamilton,2005). Undersuchpermanentlywetconditions,however,reducednitri fi-cationratescanleadtoammoniumaccumulation(Paulissenetal., 2005). One would, therefore, expect that the observed higher groundwaterlevelsinthetopsoilremovalplotswouldcorrelatewith higherconcentrationsofNH4+,butweseetheopposite.Weassume that thehigherNH4+-concentrationsin thecontrol plotscanbe explainedby(past)peatmineralizationfollowedbythereleaseand adsorptionofammoniumtothecationexchangecomplexofthesoil. Inthetopsoilremovalplots,suchaccumulatedsoil-boundNH4+is removedtogetherwiththemineralizedtopsoil.Also,theincreased inputofbasecationsbygroundwateraftertopsoilremovalenhances thedesorptionofNH4+fromthesoiladsorptioncomplex,allowing NH4+tobewashedout(Lucassenetal.,2006).
Concentrationsoftotalphosphorusintheporewaterwerenot lowerinthetopsoilremovalplots,butwefinduptosixtimeslower soil pools of total-P and oxalate-P. Phosphorus is relatively immobile and tends to accumulate in aninorganic formin the topsoilofdegradedorfertilizedfens(Grahametal.,2005;Zaketal., 2010),particularlywhenironisabundant(Aggenbachetal.,2013). Such P-enriched layer is generally easily removed with topsoil removal,asiswellknownfromrestorationprojectsonmineralsoils (Allison and Ausden, 2004). The discrepancybetween lower P-concentrationsinthetopsoilbutunalteredP-concentrationsinthe porewaterof thetopsoilremoval plotsis possiblytheresultof lowered redox potentials thatfacilitate P-release inthe formof PO43-Ptotheporewater(Zaketal.,2010;vandeRietetal.,2013). 4.2.Floristicresponsetotopsoilremoval
We observed an increase in biodiversity as well as in the fractionoftargetspecies inalltopsoilremovalplots.Moreover, targetspeciesthatwerealreadypresentinthecontrolplotsare Fig.2.Relationship betweenaverage herbbiomass(tha1)andrelativelight
intensityatthesoilsurface(%),groupedforplotswithandwithouttopsoilremoval (y=31.1ln(x)+72.8;r2
=0.72).Dotsrepresentaveragespersiteandtreatment, barsrepresentstandarddeviations.
-1.0 1.0 1.0 NH4 K Ca S Water level Mn Al
RLI ENV. VARIABLES
SAMPLES
Control
Topsoil removal
-1.0
Fig. 3.CCA-biplot showing significant (p<0.05) explanatory environmental variables(vectors)inrelationtospeciescompositioninthe48studyplots(dots). Chemicalvariablesweremeasuredintheporewater,“Waterlevel”=groundwater leveland“RLI”=relativelightintensityatsurfacelevel.Plotsaregroupedinto topsoilremovalornotopsoilremoval(=“control”).Studysitewasincludedasa blockingfactorintheanalysis(siteblocksnotshowninfigure).
nearlyalwaysfoundinthetopsoilremovalplotsaswell,indicating activation of the seed bank after topsoil removal or rapid recolonizationfromnearbyareas.
Themultivariateanalysisindicatedthatvegetationcomposition inthetopsoilremovalplotsisstronglyrelatedtohighgroundwater levels,lowporewaterconcentrationsofNH4+,ahighavailabilityof light(RLI),andhighconcentrationsofbasecationsCa2+andK+.As discussed in Section 4.1, higher groundwater levels lead to rewetting,which correlates with higher concentrationsof base cationsandHCO3.AcontinuoussupplyofbothHCO3(ensuringa highalkalinityandabufferedpH)andbasecationsisessentialfor richfen species,which are vulnerabletoacidification andbase leaching(vanDiggelenetal.,1996;Grootjansetal.,2006;Cusell etal.,2013).Inthisrespect,aslightincreaseinconcentrationsof potassiummaybeofparticularimportanceasKleachesrelatively easilyfromdegradedpeatsoils,therebyhamperingfenrestoration (van Durenetal.,1997).Asmanyofthecontrolplotswerestill sufferingfrom(slight)desiccation,desiccation-relatedprocesses co-explainthelimitedoccurrenceoftargetspeciesinthecontrol plots.Furthermore,thelownitrogenlevels(especiallyNH4+)inthe topsoilremovalplotsareequallyimportantfortheestablishment of rich fen communities: many rich fen species are easily outcompetedby competitive helophytesin N-enrichedsystems (Verhoevenetal.,2011), whileexcessammoniumaccumulation canbephytotoxictotargetspecies(Paulissenetal.,2005).Finally, theCCAunravelsthesignificanteffectofrelativelight intensity (RLI) on target community establishment, which is inversely relatedtoproductivityandnutrientavailability.Siteaveragesfor thetopsoilremovalplotsalwaysexceeded 40%ofRLIatsurface level,whereasRLIinsomeofthecontrolplotsapproachedthe5% threshold of light compensation where respiration exceeds photosynthesisin herbs(Larcher,2003).Under suchconditions, light stress becomes a strong environmental filter for fen vegetation (Kotowski and van Diggelen, 2004).For typical rich fencommunitiesofsmallsedgesandbrownmosses,thresholdslie generallyaround40–60%ofRLI(KotowskiandvanDiggelen,2004; Kotowskietal.,2006).Consequently,weobservedahigherfraction
oflight-demanding targetspecies andanincrease inbryophyte coverinthetopsoilremovalplots.
4.3.Siteeffects
Wedidnotanalyzedifferent sitesseparately,but itappears thatthestrength of theeffect of topsoilremovalis somewhat sitedependent. Moreover, depth of topsoilremoval as well as timesince topsoilremovalvaried between sites,and thismay affecttheoutcomeaswell(Klimkowskaet al.,2007).Wehave toofew study sitesto statistically disentangleall thedifferent siteeffectsthataffectthemagnitudeofrestorationsuccess,but visual inspection of ourdata combinedwith the CCA-analysis suggests that higher groundwater levels and lower nutrient stocks play a major role, with most successful results on (previously)desiccatedlocationswithaheavily-mineralizedtop soil (locationLH, HE, HO andMA).The two locationsthathad beenrewettedinthepast(DA,PE)showedalessdistinct,albeit positive,responsetotopsoilremoval.
4.4.Topsoilremovalasarestorationstrategyfororganicsoils Topsoilremovalonpeatsoils(therebyexposinganunderlying peatlayer)isanuncommonpracticeinnaturerestoration(butsee
Patzeltetal.,2001;Klimkowskaetal.,2009,2015),butourresults show that it can significantly improve conditions for rich fen development.It should be notedthat we cannotascertain that topsoilremovaltriggersacomplete“ecosystemreset”topristine conditions,asthisrequirescompleteknowledgeoftheconditions priortodegradation.However,itisclearthathydro-geochemical conditionsinthedegradedfensshifttowardconditionsthatare,at least,moretypicalforpristinerichfens(seeAggenbachetal.,2013), and that targetspecies respond positively. Atthe sametime, a completeremovalofadegradedpeatlayerisirreversibleandnot withoutrisk.Therefore,weprovideasimplifieddecisionflow-chart withcriteriathatshouldbemetbeforetopsoilremovalisconsidered (Fig.4).First,itisevidentthattopsoilremovalisnotfeasibleifthe
Groundwater at surfacelevel?
Rewettingpossible? No topsoil removal yes no yes
Anyrisk of creatingopen water? Anyrisk of
negativeeffects on neighboring areas?
Topsoilremoval
no
High nutrient(NP) pools in
top soil? no
yes
yes
no
Target vegetationpresent? yes
no
- Definetarget vegetation - Evaluatecurrentvegetationtypes - Evaluateoccurenceof target species - Evaluatepresenceof propagules
- Block ditches - Restoreinitialhydrology - Maintainyearroundexfiltrationof
mineral-richgroundwater
- Monitor groundwaterlevels (yearround)
- Canexcesswater easilyrun off? - Is therea risk of drainingneighboring
areas?
- Measure pools of N andP in peatsoil (andporewater ) at different depths - Does light stress playa role (RLI at
surface< 30 %)?
- Determineexact depthof topsoil removal
Actions andquestions Start
target rich fen vegetation is already present. Second, if other restorationstrategiesdeservepriority(forrichfensthisincludes rewettingwithminerotrophicwater(vanDiggelenetal.,2006)), thentopsoilremovalshouldbeconsideredonlyif(1)rewettingisnot possibleor(2)nutrient(NorP)poolsaresohighthattypicalrichfen speciesareoutcompetedbynon-targetspecies(Lamersetal.,2014). Finally, if topsoil removal is expected to negatively affect any neighboringarea ofhighecologicalvalue (e.g.,throughdrainage effects),then potentialgainsintherestorationareamustbebalanced withpotentiallossesintheneighboringarea.
5.Conclusions
Ourstudyhasshownthattopsoilremovalondegradedpeat soilscan significantly improveconditions for rich fen develop-ment.Wesuggestthatthebestresultsaretobeexpectedinareas where raising groundwater levels to the surface level is not possible,sothat topsoilremovalleadstoimmediate rewetting. Moreover,removal of thedegradedtop layer reducessoil bulk densityandnutrientpools,therebyexposinganunderlyingpeat layerofbetterphysio-chemicalquality.Generally,targetspecies respondrelativelyfast.Weproposethattopsoilremovalshouldbe morefrequently consideredin degraded groundwater-fed peat-lands.AsmostpeatlandsinEuropearealreadyinastageofsevere degradation,suchdrasticmeasuresmaybecrucialtoimprovethe conservationprospectsoftheseendangeredhabitats.
Acknowledgements
Wethank Staatsbosbeheer, Landschap Overijssel, Natuurmo-numentenandtheAgentschapvoorNatuurenBosforproviding background information on and access tothe study sites. We gratefullyacknowledgefieldsupportbyGuyEmsens.Thisstudy wasfinancedbytheDutchO+BNResearchProgramandtheFund forScientificResearchinFlanders(toW.J.Emsens).
Appendix.ASupplementarydata
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. ecoleng.2015.01.029.
References
Aggenbach,C.J.S.,Backx,H.,Emsens,W.J.,Grootjans,A.P.,Lamers,L.P.M., Smolders,A.J.P.,Stuyfzand,P.J.,Wolejko,L.,VanDiggelen,R.,2013.Dohigh ironconcentrationsinrewettedrichfenshamperrestoration?Preslia85, 405–420.
Allison,M.,Ausden,M.,2004.Successfuluseoftopsoilremovalandsoil ameliorationtocreateheathlandvegetation.Biol.Conserv.120,221–228.
Beltman,B.,VandenBroek,T.,Barendregt,A.,Bootsma,M.C.,Grootjans,A.P.,2001. RehabilitationofacidifiedandeutrophiedfensinTheNetherlands:effectsof hydrologicmanipulationandliming.Ecol.Eng.17,21–31.
Bennington,C.C.,Thayne,W.V.,1994.Useandmisuseofmixed-modelanalysisof varianceinecological-studies.Ecology75,717–722.
Boeye,D.,Vanstraaten,D.,Verheyen,R.F.,1995.Arecenttransformationfrompoor torichfencausedbyartificialgroundwaterrecharge.J.Hydrol.169,111–129.
Bragazza,L.,Buttler,A.,Siegenthaler,A.,Mitchell,E.A.,2009.Plantlitter decompositionandnutrientreleaseinpeatlands.In:Baird,A.J.,Belyea,L.R., Comas,X.,Reeve,A.S.,Slater,L.D.(Eds.),CarboncyclinginNorthernpeatlands. Geophys.Monogr.Series,184,pp.99–110.
Brouns,K.,Verhoeven,J.T.A.,Hefting,M.M.,2014.Shortperiodofoxygenation releaseslatchonpeatdecomposition.Sci.TotalEnviron.481,61–68.
Cusell,C.,Lamers,L.P.M.,vanWirdum,G.,Kooijman,A.,2013.Impactsofwaterlevel fluctuationonmesotrophicrichfens:acidificationvs.eutrophication.J.Appl. Ecol.50,998–1009.
Fenner,N.,Williams,R.,Toberman,H.,Hughes,S.,Reynolds,B.,Freeman,C.,2011. Decomposition‘hotspots'inarewettedpeatland:implicationsforwaterquality andcarboncycling.Hydrobiologia674,51–66.
Gambolati,G.,Putti,M.,Teatini,P.,Stori,G.G.,2006.Subsidenceduetopeat oxidationandimpactondrainageinfrastructuresinafarmlandcatchment southoftheVeniceLagoon.Environ.Geol.49,814–820.
Graham,S.A.,Craft,C.B.,McCormick,P.V.,Aldous,A.,2005.Formsandaccumulation ofsoilPinnaturalandrecentlyrestoredpeatlands–UpperKlamathLake, Oregon,USA.Wetlands25,594–606.
Grootjans,A.P.,Schipper,P.C.,Vanderwindt,H.J.,1986.Influenceofdrainageon N-mineralizationandvegetationresponseinwetmeadows.2. Cirsio-molinietumstands.ActaOecol.-Oecol.Plant.7,3–14.
Grootjans,A.P.,Adema,E.B.,Bleuten,W.,Joosten,H.,Madaras,M.,Janakova,M., 2006.Hydrologicallandscapesettingsofbase-richfenmiresandfenmeadows: anoverview.Appl.Veg.Sci.9,175–184.
Hautier,Y.,Niklaus,P.A.,Hector,A.,2009.Competitionforlightcausesplant biodiversitylossaftereutrophication.Science324,636–638.
Hennekens,S.M.,Schaminee,J.H.J.,2001.TURBOVEG,acomprehensivedatabase managementsystemforvegetationdata.J.Veg.Sci.12,589–591.
Klimkowska,A.,VanDiggelen,R.,Bakker,J.P.,Grootjans,A.P.,2007.Wetmeadow restorationinWesternEurope:aquantitativeassessmentoftheeffectivenessof severaltechniques.Biol.Conserv.140,318–328.
Klimkowska,A.,Kotowski,W.,vanDiggelen,R.,Grootjans,A.P.,Dzierza,P., Brzezinska,K.,2009.Vegetationre-develomentafterfenmeadowrestoration bytopsoilremovalandhaytransfer.Restor.Ecol.18,924–933.
Klimkowska,A.,Elst,D.J.,Grootjans,A.P.,2015.Understandinglong-termeffectsof topsoilremovalinpeatlands:overcomingthresholdsforfenmeadows restoration.Appl.Veg.Sci.18,110–120.
Kooijman,A.M.,1992.ThedecreaseofrichfenbryophytesintheNetherlands.Biol. Conserv.59,139–143.
Kotowski,W.,Thorig,W.,vanDiggelen,R.,Wassen,M.J.,2006.Competitionasa factorstructuringspecieszonationinriparianfens–atransplantation experiment.Appl.Veg.Sci.9,231–240.
Kotowski,W.,vanDiggelen,R.,2004.Lightasanenvironmentalfilterinfen vegetation.J.Veg.Sci.15,583–594.
Laiho,R.,2006.Decompositionin peatlands:reconcilingseemingly contrastingresults ontheimpactsofloweredwaterlevels.SoilBiol.Biochem.38,2011–2024.
Laiho,R.,Sallantaus,T.,Laine,J.,1999.Theeffectofforestrydrainageonvertical distributionsofmajorplantnutrientsinpeatsoils.PlantSoil207,169–181.
Lamers,L.P.M.,Vile,M.A.,Grootjans,A.P.,Acreman,M.,vanDiggelen,R.,Evans,M.G., Richardson,C.J.,Rochefort,L.,Kooijman,A.M.,Roelofs,J.G.M.,Smolders,A.J.P., 2014.EcologicalrestorationofrichfensinEuropeandNorthAmerica:fromtrial anderrortoanevidence-basedapproach.Biol.Rev.doi:http://dx.doi.org/ 10.1111/brv.12102.
Larcher,W.,2003.PhysiologicalPlantEcology:EcophysiologyandStressPhysiology ofFunctionalGroups.Springer.
Londo,G.,1976.Decimalscaleforrelevesofpermanentquadrats.Vegetation33, 61–64.
Lucassen,E.C.H.E.T.,Smolders,A.J.P.,Boedeltje,G.,vandenMunckhof,P.J.J.,Roelofs,J. G.M.,2006.Groundwaterinputaffectingplantdistributionbycontrolling ammoniumandironavailability.J.Veg.Sci.17,425–434.
Malson,K.,Backeus,I.,Rydin,H.,2008.Long-termeffectsofdrainageandinitialeffects ofhydrologicalrestorationonrichfenvegetation.Appl.Veg.Sci.11,99–106.
Malson,K.,Rydin,H.,2007.Theregenerationcapabilitiesofbryophytesforrichfen restoration.Biol.Conserv.135,435–442.
Matson,P.A.,Parton,W.J.,Power,A.G.,Swift,M.J.,1997.Agriculturalintensification andecosystemproperties.Science277,504–509.
Olsson,P.A.,Ödman,A.M.,2014.Naturalestablishmentofspecialistplantspecies aftertopsoilremovalandsoilperturbationindegradedcalcareoussandy grassland.Restor.Ecol.22,49–56.
Patzelt,A.,Wild,U.,Pfadenhauer,J.,2001.Restorationofwetfenmeadowsbytopsoil removal:vegetationdevelopmentandgerminationbiologyoffenspecies. Restor.Ecol.9,127–136.
Paulissen,M.P.C.P.,Besalu,L.E.,DeBruijn,H.,VanderVen,P.J.M.,Bobbink,R.,2005. Contrastingeffectsofammoniumenrichmentonfenbryophytes.J.Bryol.27, 109–117.
Phoenix,G.K.,Hicks,W.K.,Cinderby,S.,Kuylenstierna,J.C.I.,Stock,W.D.,Dentener,F. J.,Giller,K.E.,Austin,A.T.,Lefroy,R.D.B.,Gimeno,B.S.,Ashmore,M.R.,Ineson,P., 2006.Atmosphericnitrogendepositioninworldbiodiversityhotspots:theneed foragreaterglobalperspectiveinassessingNdepositionimpacts.Global ChangeBiol.12,470–476.
Romão,C.,1996.InterpretationManualofEuropeanUnionHabitats.European Commission.
Schaminee,J.H.J.,Weeda,E.J.,Westhoff,V.,1995.DevegetatievanNederland2: Plantengemeenschappenvanwateren,moerassenennatteheiden.Opulus Press,Uppsala/Leiden,pp.358.
Suding,K.N.,Gross,K.L.,Houseman,G.R.,2004.Alternativestatesandpositive feedbacksinrestorationecology.TrendsEcol.Evol.19,46–53.
terBraak,C.,Šmilauer,P.,2002.CANOCOreferencemanualandCanoDrawfor Windowsuser’sguide:softwareforcanonicalcommunityordination(version 4.5).SectiononPermutationMethods.MicrocomputerPower,Ithaca,NewYork.
Tichy,L.,2002.JUICE,softwareforvegetationclassification.J.Veg.Sci.13, 451–453.
vandeRiet,B.P.,Hefting,M.M.,Verhoeven,J.T.A.,2013.Rewettingdrainedpeat meadows:risksandbenefitsintermsofnutrientreleaseandgreenhousegas exchange.WaterAirSoilPolutt.224,1440–1452.
vanderMeijden,R.,Odé,B.,Groen,C.L.G.,Witte,F.J.,Bal,D.,2000.Bedreigdeen kwetsbarevaatplanteninNederland.BasisrapportmetvoorstelvoordeRode Lijst.Gorteria26,85–208.
vanDiggelen,R.,Middleton,B.,Bakker,J.,Grootjans,A.,Wassen,M.,2006.Fensand floodplainsofthetemperatezone:Presentstatus,threats,conservationand restoration.Appl.Veg.Sci.9,157–162.
vanDiggelen,R.,Molenaar,W.J.,Kooijman,A.M.,1996.Vegetationsuccessionina floatingmireinrelationtomanagementandhydrology.J.Veg.Sci.7,809–820.
vanDuren,I.C.,Boeye,D.,Grootjans,A.P.,1997.Nutrientlimitationsinanextantand drainedpoorfen:implicationsforrestoration.PlantEcol.133,91–100.
vanLanduyt,W.,Vanhecke,L.,Hoste,I.,etal.,2006.Rodelijstvandevaatplantenvan vlaanderenenhetBrusselsHoofdstedelijkGewest.In:vanLanduyt,W.(Ed.), AtlasvandeFloravanVlaanderenenhetBrusselsGewest.INBOendenationale plantentuinvanBelgië,Brussel.
Verhoeven,J.T.A.,Beltman,B.,Dorland,E.,Robat,S.A.,Bobbink,R.,2011.Differential effectsofammoniumandnitratedepositiononfenphanerogamsand bryophytes.Appl.Veg.Sci.14,149–157.
Wassen,M.J.,Venterink,H.O.,Lapshina,E.D.,Tanneberger,F.,2005.Endangered plantspersistunderphosphoruslimitation.Nature437,547–550.
Whitmire,S.L.,Hamilton,S.K.,2005.Rapidremovalofnitrateandsulfatein freshwaterwetlandsediments.J.Environ.Qual.34,2062–2071.
Zak,D.,Gelbrecht,J.,Zerbe,S.,Shatwell,T.,Barth,M.,Cabezas,A.,Steffenhagen,P., 2014.Howhelophytesinfluencethephosphoruscycleindegradedinundated peatsoils–implicationsforfenrestoration.Ecol.Eng.66,82–90.
Zak,D.,Wagner,C.,Payer,B.,Augustin,J.,Gelbrecht,J.,2010.Phosphorus mobilizationinrewettedfens:theeffectofalteredpeatpropertiesand implicationsfortheirrestoration.Ecol.Appl.20,1336–1349.
