University of Groningen
SPECIES DYNAMICS AND NUTRIENT ACCUMULATION DURING EARLY PRIMARY
SUCCESSION IN COASTAL SAND DUNES
Olff, Han; VANTOOREN, BF; Tooren, B.F. van
Published in:
Journal of Ecology
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to
cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
1993
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
OLFF, H., VANTOOREN, B. F., & Tooren, B. F. V. (1993). SPECIES DYNAMICS AND NUTRIENT
ACCUMULATION DURING EARLY PRIMARY SUCCESSION IN COASTAL SAND DUNES. Journal of
Ecology, 81(4), 693-706.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the
author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the
number of authors shown on this cover page is limited to 10 maximum.
Download date: 15-06-2017
Species Dynamics and Nutrient Accumulation During Early Primary Succession in Coastal Sand
Dunes
Author(s): H. Olff, J. Huisman and B. F. Van Tooren
Source: Journal of Ecology, Vol. 81, No. 4 (Dec., 1993), pp. 693-706
Published by: British Ecological Society
Stable URL: http://www.jstor.org/stable/2261667 .
Accessed: 19/08/2013 10:11
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
British Ecological Society is collaborating with JSTOR to digitize, preserve and extend access to Journal of
Ecology.
http://www.jstor.org
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
Journalof
Ecology1993,
81, 693-706
Species dynamicsand nutrientaccumulationduring
earlyprimarysuccessionin coastal sand dunes
H. OLFF,*
J. HUISMAN
and B. F. VAN TOORENt
Department
ofPlantBiology,University
ofGroningen,
PO Box 14, 9750 AAHaren and tNatuurmonumenten,
Noordereinde60, 1243 JJ's-Graveland,theNetherlands
Summary
1 The presentstudyreportson a primarysuccessionserieswhichstartedon bare soil
on the Dutch island of Schiermonnikoogafter the building of a sand dike.
Vegetationalchanges were studiedfor 18 years by means of permanenttransects
along a topographic
gradientfroma moistplain to drydunes.Soil development
and
A fertilizer
werereconstructed
vegetationstructure
usinga chronosequence.
experimentwas setup in an intermediate
successionalstagein theplainand on thedune,in
whichsoil resourceslimitedproductivity.
orderto determine
2 Differences
in salinity,
floodingand moisturecontentwereimportant
determinants
in speciescompositionalongthetopographic
of thedifferences
gradient.In addition,
fluctuations
year-to-year
of thesefactorsseem to be responsiblefortheyear-to-year
in frequency
fluctuations
of occurrenceofmanyshort-lived
species.These factorsdid
not,however,showa consistent
long-term
trendovertime.
3 Fromsoil analysesand thenutrient
it is concludedthatnitroadditionexperiment,
biomassproduction.Over a periodof about 16 yearsthe
gen limitedabove-ground
totalamountofnitrogen
in theorganiclayerofthesoil increasedfrom7 to 50 g N m-2
in theplainsand from1 to 15 g N ml2on thedunes.
4 The accumulationof nitrogen
duringthesuccessionalseriesis accompaniedby an
increasedbiomass, a decreased lightpenetrationto the soil surface,a decreased
root/shoot
ratio,increasingdominanceof tall species,and a decreasingabundanceof
of lightcompetition
small,short-lived
species.These datasuggestthattheimportance
is increasingduringsuccession.
5 The importanceof plantheightversuslightreductionat the soil surfacein determiningtheoutcomeoflightcompetition
is discussed.
Keywords:fluctuations,
nitrogenaccumulation,sand dunes, succession,vegetation
structure
JournalofEcology 1993,81, 693-706
Introduction
693
for
of plantsuccessionfocusedon plantcompetition
limitingresources. This hypothesispredicts that
Although
succession
hasbeenintensively
studied
by
whennutrients
are shortin supplyduringearlysucplantecologists
sincethebeginning
ofthiscentury,
a
cession, competitionfor nutrientswill be more
singletheoretical
framework
has not yetemerged
important
thancompetition
forlight.This will favour
(Miles 1987). Severalattempts
at sucha synthesis
with
intoroots.With
small species
highinvestments
have been made (e.g. Clements1916; Connell&
theaccumulationof nutrients
in theecosystem,plant
Slatyer1977; Peet & Christensen
1980; Tilman
biomassis expectedto increase(at least at low levels
1985).Thesestudiesrevealedmanyimportant
eleof herbivory).Increased biomass will lead to
ments
ofsuccession;
e.g.resource
competition,
tolerincreasedlightinterception,
and therebyto reduced
anceto extreme
conditions,
plantdispersion,
species
light availabilityat the soil surface.As a conseonnutrient
effects
andvegetation
cycling
structure.
forlightmaybe moreimportant
quence,competition
Tilman's(1985, 1988)resource
ratiohypothesis
fornutrients
duringthelatersuccesthancompetition
will
favourtall species withhigh
sional stages.This
*Present address and correspondence: Centre for
speinvestments
intostemsand leaves. Accordingly,
AgrobiologicalResearch(CABO-DLO), PO Box 14, 6700
AA Wageningen,
theNetherlands.
cies renhIcementdiirinc succession can he exnlained
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
694
by changing availabilities of nutrientsand light
Primarysuccession (Tilman1985, 1988).
If, however,physicaland chemicalstressfactors
in coastal sand
such as high salinity,anaerobicconditionsor very
dunes
low pH prevail,otherplanttraitsmightbe important
for dealing with constraintson plant productivity.
Investmentsin physiological and morphological
traitswhichenable a plantto tolerate'extreme'conditionsmightcause stress-adapted
speciesto be weak
competitorsin situationswherethese stressfactors
are unimportant(Grime 1979; Taylor 1989).
it shouldbe expectedthata changein the
Therefore,
intensityof such stress factorswill also lead to
changesin speciescomposition.
Primarysuccessionscan be initiatedaftersandduneformation,
theretreat
of glaciers,volcaniceruptions,or in othersituationswherenew substrates
are
formed.A joint analysisof vegetationcomposition,
limibiomass distribution,
soil changesand nutrient
tationduringthecourseof primarysuccessionhas so
farbeen publishedforonly a few ecosystems,e.g.
Glacier Bay (Crocker & Major 1955; Crocker&
Dickson 1957; Lawrence et al. 1967; Bormann&
Sidle 1990), Lake Michigansanddunes(Olson 1957;
Robertson& Vitousek 1981; Robertson1982) and
china clay wastes (Robertset al. 1981; Marrset al.
1981).
In thispaper,we reporton the species dynamics,
biomass distribution,
soil changesand nutrient
limitationduringearlyprimarysuccession in a coastal
sand-dunearea. The data are discussedin thelightof
theaforementioned
conceptualmodels.
Studyarea
The studyarea (a 'Beach Plain') is located on the
Wadden Sea island of Schiermonnikoog,the
Netherlands
(53?29'N, 6?12'E). Up to thelate 1950s,
the area was a nearlybare sand flat,withscattered
young dunes (up to 2 m high). These dunes were
sparselyvegetatedwiththeperennialgrassesElymus
folfarctusandAmmophilaarenaria.(Nomenclature
lows Van der Meijden et al. 1983.) The lowerparts
were bare, or sparsely covered with the annuals
Salicornia stricta,Spergularia marina and Suaeda
maritima.In 1959, withthe construction
of a sand
dike,thearea was protectedfromthedirectinfluence
of the North Sea, and the vegetationsuccession
started(Fig. 1). However, heavy stormsin 1972
createda large opening in the dike which is still
present.In a 1-km2area aroundthisgap, vegetation
successionhas startedrepeatedly,
afterthevegetation
and top soil were removedby severewinterstorms.
This enabledus to samplesiteswheresuccessionhad
proceededforshorter
periods.In mostwinters,when
the water table is higherthan 1.8 m above NAP
(Dutch OrdnanceDatum),the entirearea is flooded
by sea watercomingthroughthe opening.This sea
water may remainfullysaturatingthe soil of the
Beach Plain forseveralmonthsbecause therelatively
low positionof thearea preventsit fromflowingout
again.The watertablein thelowerpartsusuallyfalls
below the soil surfaceby Aprilor May. It normally
reachesa depthof about30-50 cm at theend of the
summeralthougha depthof 1 m may be reachedin
in thelengthof
extremely
drysummers.Fluctuations
theinundation
periodand in theprecipitation/evaporationbalance resultin a strongly
fluctuating
salinity
oftheuppersoil layers(van Toorenetal. 1983).
At present,thesmalldunesand theirslopes in the
undisturbed
area (30 yearsof succession)are densely
coveredwithHippophaerhamnoides,a nitrogenfixing shrub,while the lower plain is dominatedby
Juncusgerardi,Scirpus maritimusand Phragmites
australis.
Methods
Changes in species compositionwere studiedfrom
1972 to 1989 in permanenttransectsalong a topographicgradientfroma moistplain to drydunes.In
1990, changes in biomass composition,soil factors
werereconstructed
and canopystructure
by a spatial
comparisonof threesites which closely resembled
three differentstages of our primarysuccession
series.Nutrientlimitationwas investigated
by fertilizer additionat one of theselocations(a singlesuccessionalstage).
THE
PERMANENT
TRANSECTS
Set-upand recording
The courseof successionin theundisturbed
area was
recordedfrom1972 to 1989 in a gridconsistingof 12
each 20 m long,separatedby a
permanent
transects,
distanceof 1 m. Each year in August,the presence
(not abundance) of every species was recordedin
adjacent 1-m x 0.4-m plots along the 12 transects,
in data from12 x 20 = 240 plots.Since sucresulting
cessionstartedin 1960,thepermanent
transects
compriseda periodof 12 to 29 years of succession.In
some years,mainlybefore1979,partof thetransects
and/orsome species were not recorded,but in all
cases it was noted which parts and which species
were not recorded.Furtherdetails on the recording
and on species replacementduringthe first8 years
are givenin Van Toorenet al. (1983). The transects
compriseda flatlow part,a dune slope witha northem exposure,and a dry dune. Because the main
scope of our studywas to analysesuccessionunder
different
sets of physicaland chemicalconstraints,
the transect-areawas subdivided into five topographicpositionsaccordingto elevationand electrical conductivity
ofthesoil moisture(Table 1; Fig. 1).
The datafromthePlain,Slope and Dune plotswillbe
presentedhere.The data thuscontainedinformation
on both topographicvariation(Plain, Slope, Dune)
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
695
H. Olff;
J Huisman&
B.F. van Tooren
Table 1 Criteriaused for the division of the permanent
transects
area in fivetopographic
positions
Electrical
Topographic Numberof Heightabove conductivity
plots
NAP (cm)
position
(,uS/cm)
'Creek'
'Plain'
'Foot'
'Slope'
'Dune'
130 - 155
155- 185
155- 185
185 -230
230 -320
17
132
21
38
32
2200 -17000
750-6300
180- 1200
50 -200
12 - 200
and temporalvariation(from12 to 29 yearsof succession).
Analysisofthedata
The data fromthepermanent
transects
wereanalysed
followingthe responseanalysisapproachdeveloped
by Huisman et al. (1993). A successional trend
derivedfromthepresence/absence
data can be interpretedas theprobability
(p) to finda certainspecies
in a plot(0.4 m2) as a function
of time(t). Thuswhen
p(t) = 1 the species was foundin all plots at timet.
to as the probabilityof occurp(t) will be referred
rence of a species. For each species at each topographicposition - provided that the species was
foundat thattopographic
positionforat least4 years
- the data were fittedto the followinggeneral
responsemodel
p()=
p(t)
I
+ea+bt
+
I'T'
1+ec+dt
where a, b, c and d are the parametersto be estimated.A parameterwas only includedin themodel
when additionof thisparameterprovideda signifiNorth
seaSand dike
28
Beach
~~~~~20
1 km
researchlocations.
Fig. 1 The studyarea withthedifferent
The primarysuccessionwas initiatedafterthebuildingof a
sand dike in 1959. Permanent
transectswererecordedfrom
1972 until1989; a sketchof thethreetopographic
positions
is indicatedin the box ('Creek' and 'Foot' are omitted,
since theywere not used forthe succession analysis). A
reconstruction
of soil developmentand vegetationstructure
was based on locationswherethevegetationshowedclose
resemblanceto the vegetationof the permanenttransects
after12,20 and 28 yearsofsuccession.
cantlybetterfit.The parameterswere estimatedby
logistic regression,with the parameterselection
based on a %2-test
(Jongmanet al. 1987; McCullagh
& Nelder 1989; Huismanet al. 1993). As an indicationof thegoodnessof fitof theresponsemodel,R2
was calculatedon basis oftheobservedprobability
of
occurrencein each yearand theprobability
of occurrence 'predicted'by the model. See Huismanet al.
(1993) forfurther
detailson thefitting
and testingof
thesemodels,withseveralexamples.
Species mayexhibitconsiderablevariationaround
the successionaltrend(low R2). This may be due to
in the studsamplingerrorsor to some stochasticity
ied phenomena,but also to effectsof independent
variablesnotincludedin themodel.In orderto investigatethe possible causes of fluctuations
aroundthe
successionaltrend,the influenceof thetrendshould
firstbe eliminated.For thispurpose,thestandardized
residuals(SR) werecalculated(Huismanetal. 1993):
SR
residual
p(-p)R
of occurrenceprewherep represents
theprobability
dictedby the responsemodel. For each species at
each topographic
position,a multipleregressionwas
performedwith SR as the dependentvariable,and
and annual
rainfalldeficitovertheperiodApril-June
maximalheightof the sea level (relatedto depthof
flooding)as independentvariables. Meteorological
data,collected2 kmwestof thepermanent
transects,
of theFree
wereprovidedby theGeologicalInstitute
of Amsterdam.
Rainfalldeficitwas calcuUniversity
minusevaporationaccordingto
latedas precipitation
difPenman(1948). Thereweremarkedyear-to-year
in
but
ferencesof the rainfalldeficit spring, a longtermchange was not observed (linear regression:
R2 = 0.02, n = 18,P > 0. 1). Data on thesea level,collected 3 km to the south,were derived fromthe
The Hague. The
annual reportsof Rijkswaterstaat,
of
the
sea level fluctuated
annual maximal height
considerably,but no long-termchange could be
detected(linearregression:R2 = 0.01, n = 18, P >
0.1).
The responsemodelswerealso used forcomputationson theoccurrenceof groupsof specieswithdifferentlife-forms
and maximalheights.By usingthe
calculatedtrendsinsteadof the observedvalues for
werenotobfuscatedby annual
each year,thepatterns
of
fluctuations
and missingvalues. The classification
maxiforms
was
used
and
Raunkiaer(1934) of life
mal heightsofthespeciesweretakenfromtheDutch
botanical database (Anonymous 1986). Life-form
spectraperyearpertopographic
positionwerecalcuof
latedby weightingeach species by its probability
occurrence.The six specieswhichcould behaveboth
as hemicryptophyte
and geophytewere assignedto a
separategroup.The weightsofthethreeotherspecies
able to exhibitmorethanone life-form
weredivided
equallyamongtheirlife-form
groups.Mean maximal
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
696
Primarysuccession
in coastal sand
dunes
plantheightforeach topographicpositionand each
year was calculatedusing log-transformed
maximal
heightsweightedby theprobability
of occurrenceof
each species.
The importanceof plantheightwas investigated
further
by relatingthe occurrenceof several plant
species to the occurrenceof Hippophae rhamnoides
shrubsof severalheights.This is thefirstshrubwhich
entersthe succession on the Slope and Dune, and
role in the disapcould therefore
play an important
pearence of otherspecies throughshading.During
of
therecordingof theplots,no directmeasurements
the height of the Hippophae shrubs were made.
Therefore,the heightof this species was estimated
froma regressionbetweenplantage a (measuredby
countinggrowthrings)and plantheighth (in cm.).
This relationshipwas derivedfrom25 individuals,
and was h(a) = 278a/(7.83+ a) (nonlinearregression,
R2 = 0.81, n = 25). Since it was knownwhen the
shrubsinvadedthe plots,theirheightcould be estimatedforeveryplotin each year.
RECONSTRUCTION
Withinour researcharea, we searchedforlocations
where the vegetationshowed close resemblanceto
after12, 20
thevegetationin thepermanent
transects
and 28 yearsof succession,forthethreetopographic
positions(Fig. 1). Near the gap in the sand dike,
about2.5 kmfromthepermanent
we found
transects,
such locationswhere the soil had been cleared by
stormfloodsabout 12 and 20 yearsago, respectively,
as could be seen on older aerial photographs.For
each combinationof successional stage and topographicposition,we establishedthreeexperimental
plots at locationswherethe vegetationshowed the
closestpossibleresemblance(checkedby percentage
to thepermanent
transectsafter12 and
dissimilarity)
20 yearsof succession.The plotsrepresenting
thelast
successionalstage were locatednear the permanent
and showedclosestresemblanceto thepertransects,
after28 yearsof succession.By this
manenttransects
procedure,we situated3 replicateplots in each of
threesuccessional stages (12, 20 and 28 years) at
each of thethreetopographicpositions(Plain, Slope
and Dune), yieldinga totalof 3 x 3 x 3 = 27 experimentalplots.
With this experimentaldesign the problem of
cannotbe avoided,sincethefactor
pseudoreplication
successionalstagecannotbe trulyreplicated(a problem arising in most studies which use chronosequences). However,we have strongindicationsfrom
the vegetationdevelopmentin the permanentplots
and fromthe aerial photographsthatthe different
thedifferent
successionalstages.
locationsreflect
The combinationsof topographicposition and
successional age will be called Plain-12, Plain-20,
Plain-28 (succession on the Plain), Slope-12,
Slope-20, Slope-28 (succession on the Slope) and
Dune-12, Dune-20, Dune-28 (succession on the
Dune). Sampling of soil and biomass from the
experimental
plotswas done from1 to 15 July1990.
Biomassand lightprofiles
The above-ground
standingcrop,rootbiomass,vertical lightprofileand percentagecoverweremeasured
of succesforall 27 plots(n = 3 foreach combination
sional stage and topographicposition).The abovegroundstandingcrop,withoutthe shrubHippophae
rhamnoides,was sampled by clippingareas of 0.4
m x 0.4 m. These samplesweresortedto speciesand
litter,driedto constantmass at 70?C and weighed.
The above-groundbiomass of Hippophae was estimatedby a nondestructive
procedure,samplingover
a largerarea. Forthis,we measuredtheabove-ground
freshweight (B, in g), above-grounddry weight,
height(h, in cm) and diameterat the bottomof the
stem(d, in cm) of 13 Hippophaeshrubsin theBeach
Plain, rangingin heightfrom30 to 250 cm. The
above-ground
biomassofeach individualshrubcould
be estimatedby B = 2.815(d2h)09094
(R2 = 0.98,
n = 13). Dry weightof the shrubwas calculatedby
multiplying
the freshweightby 0.478 (R2 = 0.97,
n = 7). Average total above-groundbiomass (dryweight) of Hippophae rhamnoideswas estimated
fromthese relationshipsby takingan area around
each plotof at least 15 m2in whichwe estimatedthe
of each individualshrubby measuringh
dry-weight
andd.
Roots were collectedby takingtwo soil coresper
plot (20 cm deep, 38.5 cm2area). Only a fewroots,
mainly of Hippophae rhamnoidesand Ammophila
arenaria, were observed deeper than 20 cm. The
rootswererinsedfreeof soil undera finewaterspray.
Rhizomeswereseparatedfromthefineroots,and all
fractionswere driedat 70?C and weighed.For each
plot,on brightsunnydays,theverticallightprofilein
thevegetationwas measuredat 5-cmintervals,
using
a PAR collector(400-700 nm) witha measuringsurface of 1 m x 0.01 m. The lightintensityat each
heightwas expressedas a fractionof the ambient
lightintensity
above thevegetation.
Soil development
The thicknessof the organiclayerwas measuredat
10 randomlychosenspotsin each experimental
plot.
The darkbrownorganiclayershoweda sharpboundary with the underlyingyellow to greyishsand,
to as theminerallayer.At least
whichwillbe referred
fivecoresweretakenfromboththeorganiclayerand
fromthe mineral layer (depth 10-15 cm). These
cores were pooled by layerand mixeduntilat least
700 g per sample was collected,resultingin one
organiclayersample and one minerallayersample
per plot. These sampleswere analysedin duplicate
forNaCl (water solution),CaCO3 (titrationwith 1
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
697
H. Olff,
J.Huisman&
B.F. van Tooren
mol 1-1HCl), pH (KCl) (1 volume soil with5 volumes 1 mol 1-1KCI), organicmattercontent(loss on
ignitionat 550?C), totalcarbon(Carmhomatanalyser
withcorrectionforCaCO3), totalnitrogen(destruction withphenol-H2SO4+ Se, colorimetric
analysis
of NH3 using endophenolblue with salicylate)and
total phosphorus(destruction
with H2SO4 + HNO3,
colorimetric
analysisof P043- usingammoniummolybdate).The moisturecontentand bulkvolume(volume per weight)were measuredimmediatelyafter
sampling for each layer using 100-ml volumetric
rings.The bulk volume was used in all area-based
calculations.
1.0
(b) Plain(continued)
(a) Plain
17AS
o
<
a
~~~~~~~~~AP
(d) Slope (continued)
(c) Slope
0.8,
~~~~~~~~~~~~~~SN
~~~~~~~~~~~~~~CE
PA
FR
0-
L
PM
PH
0.4
C
The effectsof successionalage (12, 20 and 28
years) and topographicposition (Plain, Slope and
of the
Dune) on biomass and on soil characteristics
organic and of the minerallayer were statistically
tested using two-way analyses of variance with
amnong
means.All
Student-Newman-Keuls
contrasts
cell means were compared when the interaction
when only main effectswere
effectwas significant;
significant
thenthe means were comparedover the
factorlevels of each significantmain effect.The
priorto the
dependentvariablewas log-transformed
analysis if this improvedthe homogeneityof variances (as testedby Cochran'sC test).
~/
0.6
2L
0
1.0
0.8
0.6
(e) Dune
0.0
'
AS
(f)Dune(continued)
SN
/
16
p
CN
Cs
/A
____________
12
/
2LC
--HR
FR
?
'SA
04
0.2 /SE
2
jC
~~~AA
>N/
20
24
28
.-N
'
12
/
16
20
24
28
Age (years)
of occurrenceduringprimarysuccessionin a coastal sand dune area, forthreedifferent
Fig. 2 Fittedchangesin probability
transectsfrom1972 (age = 12) to 1989 (age
topographicpositions(Plain, Slope, Dune). Data were recordedin permanent
(withR2):
siteare shown.Species abbreviations
occurringspeciespertopographic
29). Onlythe 12 mostfrequently
pulchellum(0.40); GM, Glaux maritima
prostrata(0.19); AS, Agrostisstolonifera(-); CP, Centaurium
PLAIN: AP, Atriplex
(0.95); JG,Juncusgerardii(0.96); OV, Odontitesvernassp. serotina(0.59); PA, Potentillaanserina(0.81); PH, Phragmites
(0.93); SS, Spergulariasp.
australis(0.98); PM, Plantago maritima(0.96); SC, Salicornia sp. (0.45); SM, Scirpusmaritimus
(0.40).
SLOPE: AA, Ammophilaarenaria (0.74); AS, Agrostisstolonifera(0.89); CA, Cirsiumarvense(0.85); CE, Calamagrostis
epigejos (0.88); CL, Centauriumlittorale(0.58); FR, Festuca rubra (0.87); HR, Hippophae rhamnoides(0.90); LC, Linum
catharticum(0.44); LN, Leontodonnudicaulis(0.94); OV, Odontitesverna ssp. serotina(0.63); PA, Potentillaanserina
(0.70); SN, Sagina nodosa (0.98).
DUNE: AA, Ammophilaarenaria (0.73); AS, Agrostisstolonifera(0.74); CE, Calamagrostisepigejos (0.56); CF, Cerastium
(0.80); FR, Festuca rubra(0.27); HR,
(0.33); CS, Cerastiumsemidecandrum
fontanum(0.74); CN, Chamerionangustifolium
Hippophae rhamnoides(0.98); PP, Poa pratensis(0.90); SA, Sonchusarvensis(0.88); SE, Sedumacre (0.95); SN, Sagina
nodosa (0.97).
based on logisticregression(see text).
as testedbya X2-test
All presentedresponsemodelsweresignificant,
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
APPLICATION
698
FERTILIZER
Primarysuccession
To investigate
whichsoil resourcelimitedplantproin coastal sand
ductivity
duringearlysuccessionin thesesanddunes,
dunes
we added inorganicnitrogen,phosphate,potassium
and waterto thePlain-20and Dune-20 stagesof succession on 4-5 May 1991. Since we hypothesized
would
thatnitrogen
fromthe 1990 soil measurements
and
be limiting,
we used fourlevels forthisnutrient,
one level for P, K and water. The nutrientswere
added as solution(4 1) to each 2-m x 2-m plot,with
five replicateplots per site. Afteradditionof the
nutrient
solutions,another4 1 of waterwas added to
solutionfromthevegeeach plot,to rinsethenutrient
were:C (control,no addition),
tation.The treatments
r
N1 (2 g N m-2 N as
W (additionof water, 1 m-2),
NH4NO3solution),N2 (4 g N m-2)N3 (8 g N m-2),N4
(16 g N m-2), P (16 g P m2 as Na2HPO4solution),K
(16 g Km-2 as KCl) and N-P-K (16 g Nm2, 16 g P
m-2, 16 g K m2). These nine treatmentswere
arrangedin a randomizedblock design with five
occurredonce
blocks per site,whereeach treatment
in each block. This yieldeda totalof 9 x 5 x 2 = 90
plots.
The total above-groundvegetation (including
standingdead) was harvestedat 9 September1991 in
10-cm x 100-cmstripsin each plot,driedat 700C,
andweighed.
Results
ANALYSIS
OF
PLANT
SPECIES
DYNAMICS
ThePlain
Glaux
Duringthe firstyears,the hemicryptophytes
maritimaand Agrostisstoloniferawere very abun(annudantin thePlain (Fig. 2A,B). The therophytes
als) Salicornia spp., Odontitesverna ssp. serotina
and Centaurium
pulchellumalso occurredfrequently
duringtheseearlystages of succession.Aftera few
monocotsJuncusgerardiiand
years,therhizomatous
Juncus maritimusand the stoloniferousPotentilla
anserina increasedgradually,whereasGlaux maritima, Limoniumvulgare,Plantago maritimaand the
decreased.At theend of
therophytes
aforementioned
the researchperiod, annual forbs,mainlyAtriplex
prostrataand Spergulariasp. could still be found.
Juncusgerardiihad passed itspeak abundance,while
Agrostisstoloniferawas stillverycommonand the
tall rhizomatousmonocots Scirpus maritimusand
Phragmitesaustraliswerestillincreasing(Fig. 2A).
In general,thetherophytes
had a lowerR2thanthe
perennialspecies (Fig. 2A,B), which indicatesthat
the therophytes
fluctuatedmore severely.For four
were
halophyticsummerannuals,these fluctuations
positivelycorrelatedwith the annual maximal sea
of Centaurium
pullevel (Table 2). The fluctuations
chellumand Centauriumlittoralewere negatively
correlatedwithrainfalldeficitin spring.Long-term
changes in life formscan be observed,therophytes
being replacedby geophytes,while hemicryptophytes remainedratherconstantin frequency
(Fig. 3A).
The weighted maximal plant height increased
withtime(Fig. 4).
butsignificantly,
slightly,
TheSlope
The perennial grasses Agrostis stolonifera,
Ammophilaarenaria and Festuca rubra initially
occurredwithhighfrequency
on theSlope, as did the
Centauriumlittorale,Linumcatharticum
therophytes
and Odontitesvernassp. serotina(Fig. 2C,D). Small
like Leontodonsaxatilis, Sagina
hemicryptophytes
nodosa, Armeria maritima,Trifoliumrepens and
as well as themonocotsJuncus
Trifoliumfragiferum,
alpino-articulatusand Carex distans increased
Potentillaanserina was
quickly.The stoloniferous
presentfromthebeginningand increasedduringthe
Table 2 Standardpartialregressioncoefficients
of a multipleregressionof standardizedresidualsagainstannualmaximal
are
heightof the sea level ('flooding'),and rainfalldeficitover the periodApril-June('rainfall'). Only those therophytes
shownwhichshowedat leastone significant
response
Plain
Slope
Dune
Species
Flooding Rainfall Flooding Rainfall Flooding Rainfall
Halophytes
Salicorniasp.
Spergulariasp.
Suaeda maritima
Atriplexprostrata
0.61**
0.68**
0.63*
0.54*
Glycophytes
Odontitesverna
NS
Centaurium
pulchellum NS
Centaurium
littorale
NS
Linumcatharticum
Arenariaserpyllifolia -
NS
NS
NS
NS
NS
-0.57*
-0.63*
-
NS
NS
-
-
NS
NS
NS
NS
NS
NS
NS
NS
-0.55*
NS
NS
-0.60*
-
-
NS
NS
X0.72** NS
NS
-0.77**
- = no trendcalculated,because speciesoccurredin less than4 years.
*P< 0.5; **P<.1.
NS=P>0.05;
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
699
H. OlffJ
J. Huisman &
B.F. van Tooren
(a) Plain
100
0
(b) Slope
20
12
0-
E
0
W,
16
24
20
28
12
16
24
20
Age (years)
28
(c) Dune
100
LIPhanerophytes
Geophytes
4020 -
12
Hemicryptophytes/Chamaephytes
/
~ ~
16
~
~
20
Age (years)
~ ~
24
//
/Hemicryptophytes
Chamaephytes
28
Therophytes
Fig. 3 Changesin the life-form
spectrain thePlain (A), on the Slope (B) and on the Dune (C) duringprimarysuccession.
of occurrence.
Each specieswas weightedforitsprobability
whole studyperiod.From 1981 onwardsall the earlier mentionedtherophytesand monocots started
decreasing.In contrast,the shrubHippophae rhamnoides,thetallforbsCirsiumarvenseand Chamerion
and thetallgrassesCalamagrostisepiangustifolium
gejos, Poa pratensis and Holcus lanatus strongly
increasedduringthe latterhalf of the studyperiod
were
of Linumcatharticum
(Fig. 2C,D). Fluctuations
negativelycorrelatedwith rainfalldeficitin spring
were thedominantlife
(Table 2). Hemicryptophytes
formon theSlope duringthewholestudyperiod.The
were graduallyreplacedduringsuccestherophytes
sion by both geophytes and phanerophytes.
reachedtheirmaximumfrequencyin
Chamaephytes
the intermediate
successional stage (Fig. 3B). The
weightedmaximal plant height increased concordantlyfrom1980 onwards(Fig. 4).
TheDune
Agrostis stolonifera,Festuca rubra, Ammophila
arenaria and Sonchus arvensis were the abundant
species in the earlysuccessionalstage on the Dune.
were almostabsentin this stage (Figs
Therophytes
2E,F & 3C). Sonchus arvensisand Agrostisstoloniferadecreasedduringthe whole studyperiod.The
therophytes
Euphrasiastricta,Cerastiumsemidecandrumand Airapraecox enteredtheDune aftera few
The smallchamaeyearsand increasedin frequency.
Sedum
Cerastium
acre and Sagina
fontanum,
phytes
their
nodosa reached
maximal frequencyaround
and small
1980. From 1981 onwards,all therophytes
disappeared,and theperennialgrasses
chamaephytes
Festuca rubraandAmmophilaarenariadecreasedas
well. At the same time,the shrubHippophae rham-
and the
noides,thetallforbChamerionangustifolium
tall grasses Poa pratensis,Calamagrostisepigejos
and Elymuspycnanthusincreased.Two glycophytic
annuals,Odontitesverna and Arenariaserpyllifolia,
were only foundon the Dune in years withmuch
rainfallin spring(Table 2). Fluctuationsof Linum
catharticumwere negativelycorrelatedwith the
annualmaximalsea level. Small species, especially
occurredmost frechamaephytesand therophytes,
successionalstages
intermediate
quentlyduringthe
(Figs 3C & 4). The averagemaximalplantheightwas
ratherhigh in 1972, thendecreasedtowards 1980,
and subsequentlyincreasedto the end of the study
period(Fig. 4).
The increasein occurrenceof Hippophae rhamnoideson theSlope and theDune seemedto be correlated with the decrease of several grass and forb
we
species (Fig. 2). To quantifythis effectfurther,
of
of
occurrence
probability
the
observed
plotted
these decreasingspecies against the probabilityof
occurrenceofHippophaewitha minimalheightcomparableto the maximalheightof each species (Fig.
5). This analysis revealed that the upper limit of
by the
occurrenceof these species was determined
the
maxishrubs
taller
than
of
Hippophae
occurrence
In
these
other
words,
of
these
species.
mal height
soon
were
as
as
over-topped
they
speciesdisappeared
bytheHippophaeshrubs.
RECONSTRUCTION
OF
BIOMASS
AND
LIGHT
PROFILES
withtimeat
The totalbiomassincreasedsignificantly
rates
each topographic
position,althoughat different
(Fig. 6A). Especiallyat theearlysuccessionalstages,
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
150
700
Primarysuccession
in coastal sand
dunes
0//
x ' 100vi3E
E
O
-
C
12
.
,
20
16
.
24
Age (years)
28
32
Fig.4 Average
maximal
plantheight
peryear,weighted
for
theprobability
ofoccurrence
ofeachspeciesin eachyear,
in the(0) Plain(R2 = 0.98,P < 0.001),(A) Slope(R2=
0.98,P<0.001) and(I) Dune(R2 = 0.96,P<0.001) primarysuccession.
a verylargefractionof thetotalbiomass was found
below ground.After28 yearsof succession,thetotal
biomass at each topographicpositionwas about the
same, butwitha quitedifferent
distribution
over the
different
plantparts.
The above-groundbiomass was dominatedby
only a few species (Fig. 6B). Juncusgerardii was
importantduring the later stages in the Plain.
Ammophilaarenaria dominatedthe earlystages on
the Dune. Duringthe laterstages on the Slope and
Dune most above-groundbiomass consistedof the
woody structures of Hippophae rhamnoides.
Chamerionangustifolium
was onlyimportant
during
the late stages on the Dune. The totalrootbiomass
increasedin the Plain and on the Dune, but did not
change on the Slope (Fig. 6C). In the Plain, this
increase was partlydue to the appearance of rhizomes ofJuncusgerardiiand Scirpusmaritimus,
and
on the Dune to the appearanceof woody roots of
Hippophaerhamnoides.
The profileof lightextinction
in thePlain,on the
Slope and in the laterDune stages showed similar
changes;thecanopyheightincreasedduringsuccession, withless lightpenetrating
to the soil (Fig. 7).
The vegetationof theDune-12 plotswas ratheropen,
butquitetall comparedwiththePlain-12and Slope12 plots.
RECONSTRUCTION
OF
SOIL
DEVELOPMENT
The Dune plotswere muchdrierthanthe Plain and
Slope. No major differencesbetweensuccessional
stages in soil moisturewere found,except a small
increaseon theDune (Fig. 8A,B). The NaCl content
of thesoil decreasedin thesequencePlain> Slope >
Dune, withagain no consistentdifferences
between
successionalstages(Fig. 8C,D). The CaCO3 concentrationof the organic layer also decreased in the
sequence Plain > Slope > Dune. The organiclayer
was very low in CaCO3 after28 years (Fig. 8E),
probablydue to thestrongincreasein organicmarter
concentration.The pH(KCl) of the organic layer
decreasedwithtimein thePlainand on theDune, but
remainedunchangedon theSlope (Fig. 8G). The pH
of the minerallayer remainedhigh, as could be
expectedfromthe high CaCO3 concentration
(Fig.
8F,H). In the Plain-20plots,we observedhighconcentrations
ofCaCO3 andNaCl in theorganiclayer.
Verylow N concentrations
werefoundin themineral layer (Fig. 9B), with a slightbut significant
increaseafter28 years. The P concentration
of the
minerallayerwas comparablewiththeconcentration
foundin the organiclayer,withthe highestlevels
foundin the Plain (Fig. 9D). Both theN concentration and the P concentration
of the organic layer
increasedsignificantly
withtimeat each topographic
position(Fig. 9A,C), withthe exceptionof P on the
Dune. The C/Nratioof theorganiclayerwas some-
1.0 (a)
' *ss
0.8
\ Sedumacre
Dune; 10cm
"
0.6
0
A
A
A
*
Trifolium
f.
Slope; 25 cm
* Trifolium
r.
a
0.44
Armeria
mar.
Slope; 30 cm
0.22
Slope;25 cm
N
0.0
1.0 (b)
0.8
o
*
A
o ?o\0
06
eQ
Cl)
0.4
Q.
0.2
*.
g
AA
A
\\
Juncusalp.
60cm
a
^Slope;
v.
0 Odontites
Slope; 80 cm
O
A0
*
.8
0.0 -
1.0
Cerastrium
f
Slope; 60 cm
.
'
(c)
0.8
\
0.6
*
*t
s
Ammophila
a.
Dune; 120 cm
0.4
0.2
0.c
0.0
0.2
0.4
0.6 0.8
P (Hippophae)
1.0
of occurrence
Fig. 5 Probability
of severalearlysuccessionalspeciesplotted
ofoccurrence
againsttheprobability
of Hippophaerhamnoides
of an estimated
minimal
plant
linesindicate
h. Thebroken
therelationship
height
P(species) = 1 - P(Hippophae
aboveheight
h). Eachspeciesis
plotted
whenitsmaximal
height
(see legend)was comparablewiththeminimal
height
h ofHippophae.
Heighth of
(A) 31 cm(1 yearold),(B) 57 cm(2 yearsold)
Hippophae:
and(C) 108cm(5 yearsold).
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
701
H. OlfJ
J.Huisman&
B.F. van Tooreun
Plain
4000 -(A)
b
E
>)cn2000
E
0
|
Slope
Dune
b
b
b
a
_
g
aa
a
Shoots
Hippophae
Shootsand Grasses
~~~~~~~~Forbs
the Plain and on the Slope thanon the Dune (Fig.
1OB). It shouldbe notedthatthetotalamountsof N
low, especiallyduringtheearliestsucare extremely
cessionalstage.
APPLICATION
FERTILIZER
applicaAnalysisofvarianceoftheeffectof fertilizer
vegetation
Dune-20
the
and
tion to the Plain-20
Roots
revealed that the block effectwas not significant.
0
was analysedas a two-way
theexperiment
Therefore,
a-. 1800 (B)
and two sites.The varitreatments
nine
with
ANOVA,
E
d
_ Others
werenotsignificantly
values
ances of untransformed
~~~~Juncus
c
c
1200
= 0.14,
C
(Cochran's
cells
differentbetween
gerardii
o
Ammophila
=
in
different
P
0.65). The sites were significantly
arenaria
Chamerion
=
a
<
P
17.56,
72
crop
totalabove-groundstanding
(Fl
c 600
angustifol.
The
11).
the
Dune
(Fig.
0.001) whichwas higheron
Hippophae
abab
Xn
0
"rhamnoides
11 ab ab
-c:
effectwas also significant
treatment
ab ab
(F8,72=6.29,P <
ab
was not
x
interaction
0.001) whilethesite treatment
=
=
differP
Therefore,
0.50).
significant
(F8,72 0.93,
3000 (C)
a-priori
using
tested
were
ences betweentreatments
I I Fineroots
C
c
R~~~~~~~hizomescontrasts(plannedcomparisons)while pooling over
2400
a.
Ammophila
N3
Onlythetreatments
bothsitesforeach treatment.
R__ hizomes
E 1800
X
=
Juncusg.
<
P
and
0.003)
(t= 3.42, P 0.001), N4 (t= 3.09,
0)
b
b
nhizomes
k~~ab
= 5.07, P < 0.001) were significantly
difN-P-K
(t
1200L
-~m ab 7
0
Scirpusm.
nhizomes
ILZF
0 61200
limthat
nitrogen
ferentfromthecontrol,indicating
ab
m 1000 -a
a
mHI
others
(a)
100 00-"--*;*
Age (years)
plots(Fig. 9E). Due to
whathigherin the28-year-old
in themineral
the verylow C and N concentrations
layer
was
veryvariable,
layer,the C/N ratioof this
occurring(Fig. 9F).
differences
withno significant
The C/Pratioof theorganiclayerincreasedstrongly
position(Fig. 9G). The
withtimeat each topographic
C/Pratiooftheminerallayerremainedlow (Fig. 9H)
whichwas due to thelow organicmattercontentand
ofthemineralsand.
thehighP concentration
The thicknessof theorganiclayerincreasedwith
successional age and varied between topographic
positions(Fig. lOA). Hence, the accumulationof N
and P in the organiclayerwas caused both by an
increasein thethicknessof thislayer(Fig. 1lOA)and
(Fig. 9A,C). Because
an increase in concentration
were foundin the
P
concentrations
high
relatively
minerallayer(Fig. 9D), the P accumulationin the
organiclayerwas verysmallcomparedwiththetotal
minerallayer.In contrast,
pool of P in theunderlying
the N accumulationin the organiclayer was very
large comparedwiththe total amountof N in the
minerallayer(Figs 9A,B & 1iOB).The accumulation
ratesof N in theorganiclayerwere muchhigherin
/1w+
801
80
of totalbiomass (A), above-ground
Fig. 6 Reconstruction
biomass(C) of different
biomass(B), and below-ground
stagesofpriplantspeciesinPlain,SlopeandDuneatthree
withineach
Totalswiththesameletter
marysuccession.
different.
werenotsignificantly
subfigure
/
60
,*
40 [I '
20f
100'
0 0)
0.
(b)
0
60,
a) 40Il,+
Ec,
20'
i
O'
+-'
,+
(c)
100
4-~~~~~~~~~20
20
80
.A- 8;-.'
'4
4
,
0gg
40
0
-
30
60
90
120
150
soil(cm)
from
Height
in (a) Plain,(b)
lightprofiles
Fig. 7 Changesin vertical
as reconsuccession,
primary
Slope and (c) Duneduring
agesof
Thesuccessional
usinga chronosequence.
structed
thesiteswere12years(0), 20 years(0) and28 years(+).
as a perwasexpressed
ateachheight
Thelightpenetration
abovethevegetation.
lightintensity
oftheambient
centage
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
702
Primarysuccessiot
in coastal sand
dunes
Organiclayer
Minerallayer
Plain Slope Dune
Plain Slope Dune
80 A 80
B
d d d
60c"D 60
ee
d
40.
dd
CD 0
d
C:
D
C 1.2 -
d
1.2
de~~~~~~~~
Z 0.4
ae 1 2
0.0
~i
bC
S
I
~aabab
bb0.0
4
4.
X0o 3
O
cu
*aa
caba[,
F
3-
2 -~
1-
b
2
b
~~~1
a~~~~
G
9g
~bc
8
~~~~b
b
bccd
a
9
H
8
cd
dcdda
b
b
P
7~~~~~~~
at
CL
122028
122028
7
122028
122028
122028
122028
Age (years)
Fig. 8 Changes of varioussoil chemicaland physicalfactors in the organicand minerallayer in Plain, Slope and
Dune duringprimarysuccession,as reconstructed
using a
chronosequence.Bars withthesame letterwithineach subdifferent.
figurewerenotsignificantly
Organiclayer
Minerallayer
Plain Slope Dune
Plain Slope Dune
4.0-
A
B
0I
Z .2' 2.0
_a
0.0
aab
0.4 -
-a ab
D
C
e
c:-30
0.3
2e
abab
a
i02oaia
aab
10)
0.c
d
a
a
d
a~~~~~~~~~~
a a~~~
a
0
1222
cig 9Cane
150D)-/ rai
122028
an dhmieadae
1202
1202
12202
in toalNcnet(,)toa
a208122
P/ aotn
8
88~~ba fabh
GI raiHGH
)ad122028
122028 122028
122028
ftesi
i li,SoeadDn
Age (years)
Discussion
SALINITY,
a
Ccb
0.4.c
ited shoot productionin this successional stage on
bothtopographicpositions.The pairwisedifferences
betweenthetreatments
N3, N4 and N-P-K were not
significant
(P > 0.05). There also seemed to be an
effectof wateron the Dune, but thiseffectwas not
significant.
raib
122028
Fig. 9 Changes in totalN content(A,B), total P content
(C,D), / ratio(IF)F andiC/P ratio(G,HoI f the organnic
duringprimarysuccession,as reconstructed
usinga chronosequence.Totals withthe same letterwithineach subfigure
werenotsignificantly
different.
SOIL
MOISTURE
AND
FLOODING
Species growingon thePlain have to cope withhigh
salinityand withfloodingin thewinterperiod.Many
species on the Plain are knownas halophytes(tolerant to saline conditions).Most of the species occurringin this area have no above-groundbiomass in
winter.The therophytes
occurringearlyin succession
are almostall summerannuals.Glaux maritimahas a
special kind of hibematingbuds (Rozema 1978).
Juncusgerardii,Scirpus maritimusand Phragmites
australishave extensiverhizomesforsurvivingthe
winterperiodwhen shootsdie off.The highabovegroundlosses of biomass in the Plain duringwinter
mightimplythatspecies which reducethese losses
by storing nutrientsand/or carbohydratesbelow
groundwill be bettercompetitors
fornutrients
and/or
light (Berendse & Jonasson 1992; De Kroon &
Schieving1990). Yet extremely
hightidesmayresult
in theformation
of largegaps in theperennialvegetation (personalobservation).Several annual species,
mainlyhalophytes,
can establishin thesegaps (Table
2). So far,we cannotoffera single explanationfor
the unexpectedhigh values of CaCO3 and NaCl in
the organiclayerof the Plain-20stage.High evaporatesin the summerperiodmay cause
transpiration
secondaryprecipitation
of thesemineralsin thetopsoil (Schlesinger1982, 1991), whichmightexplain
whythevalues in theorganiclayerwerehigherthan
in the underlyingminerallayer. But this does not
fromthePlain-12and Plain-28
explainthedifference
stages.Perhaps,thisis due to a spatiallydeviantsituationof thePlain-20plotswithrespectto thesevariables.
The Slope is muchless saline thanthe Plain,but
stillhas a ratherhighmoisturecontent.The species
are mainlyglycophytes
(nottolerantto saline conditions).The latersuccessionaldynamicson the Slope
are comparableto thoseon theDune (bothsitesshow
invasionof Hippophae),while at the earlystagethe
slope showsmoreresemblanceto thePlain.
Salinityand soil moisturecontentare bothlow on
theDune. Winterannualslike Cerastiumsemidecandrum,Arenariaserpyllifolia
andAirapraecox flower
in spring,die afterseed-setting
in May-Juneand survive duringthe drysummeras seeds (Rozijn 1984);
Sedum acre can reduce evaporationby closing the
stomataat daytime(CAM metabolism);Hippophae
rhamnoidesand Ammophilaarenaria can rootrather
deep; Ammophilaarenaria and Elymuspycnanthus
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
703
H. Olff,
J.Huisman&
B.F. van Tooren
10 -A
Plain
Slope
Dune
_8
a)
.0
C:
CZ
~~~~~~g
6-
0
CD
for a slightdecrease on the Slope. Similarly,van
Tooren,Schat& TerBorg(1983) did notobserveany
long-termtrendin the soil moistureconductivity
from1973 to 1980. Soil moisturecontentincreased
only on the Dune, accompanied by an increased
organicmattercontent(largerwaterholdingcapacity) and an increasedabove-groundstandingcrop
(moistmicroclimate);
so thismightbe rathera consequence than a cause of the observed succession.
Long-termtrendsin theheightof floodingor in the
in springdidnotoccur.
rainfall-deficit
NUTRIENT
d
/
.. d
E~~~~~~
aa
H~~~~
cd~~~~
bc
15
30
b5
..
Z
:.:.
1220 28
b c
1220 28
Age (years)
bc
1220 28
oftheorganic
layer(A) andtotalamount
Fig.10 Thickness
intheorganic
ofnitrogen
layer(B) ofthePlain,Slopeand
as reconstructed
usinga
primary
succession,
Duneduring
eachsubBarswiththesameletter
within
chronosequence.
different.
werenotsignificantly
figure
theirstomfoldtheirleaves whenitgetsdry,burying
ata deeplybetweenthe ribs.Fluctuationsof several
short-livedglycophyteswere negativelycorrelated
withrainfalldeficitin spring(Table 2). In thisstudy
area,a highrainfalldeficitcauses a low soil moisture
contentand, especially in the Plain and Slope, an
increasedsalinity(van Toorenet al. 1983). Bothlow
soil moisturecontentand highsalinityhave negative
of most
and establishment
effectson thegermination
oftheseglycophytes
(Schat 1982). Althoughwe have
notbeenable toprovethatwaterlimitstheproductivmight
experiments
ity,Fig. 11 suggeststhatfurther
reveal that,in additionto nitrogen,wateris also a
soil resourceon theDune.
limiting
Our resultssuggestthat differencesin salinity,
determiwateravailabilityand floodingare important
in speciescompositionalong
nantsof thedifferences
the topographicgradient.In addition,year-to-year
fluctuations
of these factorsseem to be responsible
of manyshort-lived
fortheyear-to-year
fluctuations
species. However,none of thesefactorscan be held
responsibleforthe observedsuccessionalsequence.
except
The NaCl contentdid notchangeconsistently,
DYNAMICS
The totalP contentof the minerallayerwas rather
high,althoughthisis probablymostlyCa-P whichis
not directlyavailable forplantuptake.Potassiumis
usuallydepositedin highamountsthroughsalt spray
fromthe sea. Nitrogenwas nearlyabsentfromthe
soil duringthe earlystagesof succession.The total
amountofnitrogen
in thetopsoilafter12 yearsseems
too littleto sustainthe totalbiomass observedafter
28 years.Additionof single nutrients
revealedthat
bioonlyN significantly
enhancedtheabove-ground
mass withrespectto thecontrol.The N contentofthe
soil was strictlycorrelatedwith its organicmatter
content,suggestingthatall N is organicN. A C/N
ratioof about 11, as foundin this study,was also
foundduringearlysuccessionat both the dunes of
Lake Michigan(Olson 1957) and the recedingglaciersin Alaska (Crocker& Dickson 1957). The conclusion that nitrogenwas an importantlimiting
nutrient
was also drawnby severalotherstudieson
in sand dunes (Willis et al. 1959;
limitation
nutrient
Willis 1963; Kachi & Hirose 1983; Doughertyet al.
1990).
For the Plain and Slope we estimatedan average
of2.8 g m-2year-1.This
increasein totalsoil nitrogen
E 800
EEO Plain
Dune
m
a) 600 -
T
*
+
o. 400-
0
C 200 CZ
/u
/
C
W
P
K Ni N2 N3 N4 NPK
Treatment
solutions
of application
of variousnutrient
Fig. 11 Effect
on the aboveground
biomassof (U) Plain-20and (Lj)
meanvalues? SD. (n =
Dune-20vegetation,
representing
in a 2-way
interaction
5). Since the site x treatment
to
werecompared
ANOVAwas notsignificant,
treatments
thecontrol
overbothsites,wherepairsnotedwithan asteriskweresignificantly
different
fromthecontrol
(P<O.05).
Thetreatments
were:C (control,
noaddition),
W (1 1 m-2),
Ni (2g N m2), N2(4g N m2) N3 (8g N m2), N4 (16g N
m-2),P(16 gPm-2),K(16g K i-2) andNPK (16g N m2,
16gPm-2,16gKm-2).
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
704
Primarysuccessioi
in coastal sand
dunes
is comparablewiththerateof nitrogenaccumulation
duringthefirst30 yearsof successionafterrecession
of glaciers in Alaska, which was 3-4 g m-2 year'
(Crocker& Major 1955; Crocker& Dickson 1957).
On theDune, on average0.9 g N m-2year-'accumulatedin the soil, whichis comparablewiththeaccumulationrateduringtheearlysuccessionalstagesof
the drydunes of bothLake Michigan(Olson 1957)
and the island of Spiekeroog,Germany(Gerlachet
al. 1988). An accumulationof 1.0 g N m-2 year-'was
also foundat the higherpart of the salt marshat
Schiermonnikoog
(Olff 1992b). The totalincreasein
nitrogenin theecosystemmusthave been larger,but
the N contentof the
we have not (yet) investigated
vegetation.
In our studyarea, thetotalrateof nitrogeninput
intothe ecosystemis expectedto be determined
by
therateof atmospheric
deposition,theabundanceof
nitrogen-fixing
organismsand the input of N by
floodingof sea water.On the East Frisianislands,
Germany,the atmosphericinputfromdry and wet
depositionis estimatedat about 1.5 g N m -2 year-l
(Gerlach et al. 1989). On the Dutch mainlandthis
input is estimatedat about 4 g N m-2 year-'.
Probably,in ourstudyarea theatmospheric
inputwill
be somewherebetweenthesetwo estimates.Stewart
(1965) estimatesthatnitrogenfixationby free-living
bacteriain dune slacks is about 6 g N m-2year-'.
Hippophaeshrubscan fixbetween1.5 g N m2 year'
(Akkermans1971) and 17.9 g N m-2 year' (Becking
1970). In the period 1973-89, the NorthSea water
near our studyarea had a mean (? SD) totalN contentof 1.33 ? 0.95 mg 1-l(n = 95, Rijkswaterstaat,
Thusevaporationof20 cm
personalcommunication).
of inundatedNorthSea water(as happensin spring)
will add only about 0.26 g N m2 to the soil. The
actual accumulationof N will be determined
by the
differences
betweentheseinputs,and all losses that
occur. The magnitudeof N losses fromthe soil
organic matterpool (mineralization)and fromthe
ecosystem(denitrification,
leaching,or litterexport
duringfloodings)has notyetbeen measuredforsites
similarto ourstudyarea.
The vegetationitselfis likelyto be one of the
maindeterminants
of theobservednitrogen
accumulation.Withoutstorageof nitrogenin plantsand soil
organicmatter,nitrogeneasily leaches out of sandy
soils (Schlesinger 1991). Moreover, as discussed
above, N-fixation
by plantsand theirassociatesmay
be one of the mostimportant
sourcesof N inputto
theecosystem.This can resultin a positivefeedback
whena specieswhichis a superiorcompetitor
at high
ratesofN supplyproducesa typeof litterwhichrapidlydecomposesand therefore
resultsin higherrates
of N mineralization(Vitousek & Walker 1987;
Berendse1990). Whenthesespecieshave established
highratesofN supply,theymayin tumn
be out-competedforlightby specieswhichare taller.Whilediscussing the primarysuccession at Glacier Bay,
Lawrenrce
et al. (1967, p. 812) state: 'it is only the
latter[thenitrogen
fixingAlnus]thatraisesthenitrogen supplyto a level enablingthe spruceforestto
achieve dominance'. These types of interactions
betweenplantgrowthand nutrient
supplycome close
to thefacilitation
hypothesis
of Clements(1916) and
Connell& Slatyer(1977).
SUCCESSION
AND
SYNTHESIS
The spatial variationin stressfactorslike salinity,
and floodinghas led to verydifferent
drought
successional sequences, in termsof species composition.
However,the generalpattemsof these successional
sequences in termsof soil development,nutrient
accumulation,
plantheightand biomasscompartmentationwereverysimilar.Nitrogen,
whichlimitedproductivity,
was nearlyabsentfromthesoil duringthe
earlystagesof succession.The accumulationof soil
nitrogenwas accompaniedby an increasedbiomass
and,thereby,
by an increasedlightinterception
bythe
ratiodecreasedand small,
vegetation.The root/shoot
short-livedspecies were replaced by tallergrasses,
forbsand shrubs.Only theearliestDune stagedeviated fromthisgeneralpattem.In thisearlystage,the
vegetationwas composedof tall species, withrelativelymuch above-groundbiomass. Small species
were absent althoughmuch lightpenetratedto the
soil surface.This deviatingstagemightbe explained
as theresultof drifting
sand,whichcan lead to depositionof manycentimetres
of sand on thevegetation
withina fewdays (personalobservation).Small species wouldprobablybe buriedunderthesand in this
earlydunestage.
The general successional pattemsin our study
area, withthe exceptionof theearlydune stage,are
in agreement
withthepattemspredictedby Tilman's
resourceratiohypothesisof primarysuccession(e.g.
Tilman 1988, p. 132). This hypothesispredictsa
gradual change from nutrientcompetitionduring
successionalstagestowardslight
early,nutrient-poor
at later,nutrient-rich
competition
stages.However,a
comparisonbetweenobservedand predictedpattems
can provideonlyweak supportforsucha mechanistic
hypothesis.The resourceratiohypothesisis derived
froma graphicalisoclineapproach,in whichsuperior
nutrient
reducelimitingnutrients
competitors
to low
levels, while superiorlightcompetitors
reducelight
near the soil surfaceto low levels. It cannotyetbe
decidedwhethernutrient
or colonization
competition
determinants
of the
dynamics,or both,are important
communitystructureduringthe early successional
stages in our studyarea. It may be that,although
nutrient
availabilityis low, theplantsare not really
able to competeeffectively
witheach otherso that
the colonizationcharacteristics
of the plant species
aremoreimportant
thantheircompetitive
ability(e.g.
Gleeson & Tilman 1990). This possibilityneeds furtherexperimental
investigation.
The accumulationof
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
705
H. Olff;
J.Huisman&
B.F. van Tooren
nitrogenled to an increasedabove-groundbiomass
to a reducedlightavailabilityat thesoil
and,thereby,
smallspecieswerereplacedby
surface.Furthermore,
tallerspecies. In thecase of Hippophaerhamnoides,
the firstshrub enteringsuccession on Slope and
Dune, it was shownthatmanyof the earlysuccessional grass and forbspecies disappearedfromthe
transectsas soon as theywereovergrown
permanent
by the shrubs.This suggeststhatthe importanceof
is increasingduringthissuccession.
lightcompetition
However, it also indicatesthat plant heightis an
of theoutcomeof competition
determinant
important
for light (e.g. Grime 1979; Givnish 1982; Olff
will of coursebe mediated
1992a). Lightcompetition
byreducingthelightavailabilityforotherplants,but,
in additionto lightreductionabilities(i.e. 'lightnear
soil surface'),the positionat whichlightis reduced
(i.e. 'plant height') has to be taken into account.
Informationon plant height is not included in
Tilman's graphicalisocline approach.For this reason,Tilman's (1985, 1988) resourceratiohypothesis
of primarysuccession is consideredinadequateas
determisoon as plantheightbecomes an important
forlight.But our
nantof theoutcomeof competition
data cannotrejectthe more generalhypothesisthat
species replacementin our studyarea was mainly
and
govemedby changingavailabilitiesof nutrients
light.
In conclusionlet us evaluateto whichextentour
frameto a moregeneraltheoretical
resultscontribute
work forprimarysuccession.It seems that,in our
and flooding
drought,
case, stressfactorslikesalinity,
settingin whicha
have providedthe environmental
particularsuccession has taken place. Given that
changes,
thesestressfactorsdid not show long-term
the observedspecies replacementmay be attributed
of lightcompetimainlyto an increasingimportance
in
the
system.This
tion when nutrients
accumulate
in
tum, is probably
nutrientaccumulationrate,
closely linkedto the vegetationdevelopmentitself.
Withineach successionalstage however,the spatial
variationin species compositionalong the topographicgradientseemed closely relatedto spatial
variationin salinity,moisturecontentand flooding.
This indicatesthatthesestressfactorsmayhave considerableimpacton species compositionand on the
wheneverstress
pathwayof succession.Accordingly,
factorsexhibitlong-term
changeswithtime,theanalin abilityto cope with
differences
ysisof interspecific
additional
such stressfactorsshouldbe an important
succession.
of studieson primary
component
Acknowledgements
to SinyterBorg,Peter-Jan
Keizerand
We are grateful
TitiaZonneveldforfieldassistance.J.van Andel,J.P.
theanonymous
Bakker,F. Berendse,W. H. 0. Emnst,
forPlant
refereesand themembersof theLaboratory
Ecologymade valuablecommentson an earlierdraft.
We thankAlbrechtGerlachand David Tilmanfordiscussionson thesubject.Figure1 was drawnby Evert
by the
weresupported
Leeuwinga.The investigations
and by the
fundCooperationGroningen-Oldenburg
FoundationforBiologicalResearch(BION) whichis
subsidized by the NetherlandsOrganizationfor
Research(NWO).
Scientific
References
Akkermans,A.D.L. (1971) Nitrogenfixationand nodulationofAlnusand Hippophaeundernaturalconditions.
ofLeiden,theNetherlands.
Thesis,University
Anonymous(1986) Botanisch Basis Register. Centraal
Heerlen.
Bureauvoorde Statistiek,
Becking, J.H. (1970) Plant-endophytesymbiosis in
plants.Plant & Soil, 32, 611-654.
non-leguminous
Berendse,F. (1990) Organicmatteraccumulationand nitrogen mineralizationduring secondary succession in
heathland ecosystems. Journal of Ecology, 78,
413-427.
use and nutriBerendse,F. & Jonasson,S. (1992) Nutrient
ecosystems.ArcticEcosystems
entcyclingin Northern
in a ChangingClimate - an Ecological Perspective
& R.L. Reynolds),pp.
(eds F.S. Chapin,R.L. Jefferies
337-358, AcademicPress,New York.
Bormann,B.T., & Sidle,R.C. (1990) Changesin productivin a chronosequenceat
of nutrients
ityand distribution
GlacierBay,Alaska.JournalofEcology,78, 561-578.
Clements,F.E. (1916) Plant Succession.CarnegieInstitute
Publication242.
Washington
Connell,J.H. & Slatyer,R.O. (1977) Mechanismsof succession in naturalcommunitiesand theirrole in comAmericanNaturalist,
munitystabilityand organization.
111, 1119-1144.
on
Crocker,R.L. & Dickson,B.A. (1957) Soil development
therecessionalmorainesoftheHerbertand Mendenhall
Alaska.JournalofEcology,45,
glaciersof southeastern
169-185.
Crocker,R.L. & Major,J.(1955) Soil developmentin relation to vegetationand surface age at Glacier Bay.
JournalofEcology,43, 427-448.
De Kroon, H & Schieving,F. (1990) Clonal growthin
plants, regulationand function.Clonal Growth in
Plants: Regulation and Function (eds J. van
Groenendael & H. de Kroon), pp. 113-130. SPB
The Hague.
AcademicPublishers,
F.J.
K.M., Mendelsohn,I.A. & Monteferrante,
Dougherthy,
(1990) Effectof nitrogen,phosphorusand potassium
contentof a
additionson plantbiomassand soil nutrient
in Louisiana.Annalsof
swale barrierstrandcommunity
Botany,66, 265-271.
Gerlach,A., Albers, E. & Broedlin,W. (1988) Nitrogen
contentand turnoverof coastal dunes on Spiekeroog
(East-Frisian Islands, German Bight). Bulletin
Ecological SocietyofAmerica,69, 143.
Gerlach,A., Albers,E. & Broedlin,W. (1989) Nitrogen
contentand turnoverin coastal dune succession on
Spiekeroog,East FrisianIslands.CongressBook EUDC
Congress,Seville,Spain.
Givnish,T.J. (1982) On the adaptivesignificanceof leaf
height in forest herbs. American Naturalist, 120,
353-381.
Gleeson, S. & Tilman,D. (1990) Allocationand the transientdynamicsof successionon poorsoils. Ecology71,
1144-1155.
Grime, J.P. (1979) Plant Strategies and Vegetation
Processes.JohnWileyand Sons,Chichester.
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions
Huisman,J.,Olff,H. & Fresco,L.F.M. (1993) A hierarchi706
cal set of modelsforspecies responseanalysis.Journal
Primarysuccession
of Vegetation
Science,4, 37-46
in coastal sand
Jongman,
J.G.H.,terBraak,C.J.F.& van Tongeren,A.F.R.
dunes
(1987) Data Analysis in Community
and Landscape
Robertson,G.P. & Vitousek, P.M. (1981) Nitrification
potentials in primary and secondary succession.
Ecology,62, 376-386.
Rozema,J.(1978) On theecologyofsome halophytes
from
a beach plain in the Netherlands. Thesis, Free
Ecology.Pudoc,Wageningen.
University
ofAmsterdam.
Kachi, N. & Hirose,T. (1983) Limitingnutrients
forplant
Rozijn,N.A.M.G. (1984) AdaptiveStrategiesofsomeDune
growthin coastal sand dune soils. Journalof Ecology,
Annuals.Thesis,FreeUniversity
of Amsterdam.
71, 937-944.
Schlesinger,W.H. (1982) Carbonstoragein the caliche of
Lawrence,D.B., Schoenike,R.E., Quispel,A. & Bond, G.
aridsoils: a case studyfromArizona.Soil Science,133,
(1967) The role of Dryas drummondiin vegetation
247-255.
developmentfollowingice recessionat Glacier Bay,
Schlesinger,W.H. (1991) Biogeochemistry:
an Analysisof
Alaska,withspecial reference
to itsnitrogen
fixationby
Global Change.AcademicPress,San Diego.
rootnodules.JournalofEcology,55, 793-813.
Schat,H. (1982) On theecologyofsome Dutch duneslack
Marrs,R.H., Roberts,R.D., Skeffington,
R.A. & Bradshaw,
plants.Thesis,FreeUniversity
ofAmsterdam.
A.D. (1981) Ecosystemdevelopment
on naturally-colo- Stewart,W.P.D. (1965) Nitrogenturnoverin marineand
nized chinaclay wastes.II. Nutrient
compartmentation.
brackishhabitats.AnnalsofBotany,29, 229-239.
JournalofEcology69, 163-169.
Taylor, G.J. (1989) Maximumpotentialgrowthrate and
McCullagh,P. & Nelder,J.A. (1989) GeneralizedLinear
allocationof respiratory
energyas relatedto stresstolModels,2ndedn.Chapman& Hall, London.
erance in plants.Plant Physiologyand Biochemistry,
Miles, J. (1987) Vegetationsuccession: past and present
27, 605-611.
perceptions.Colonization,Successionand Stability(eds
Tilman,D. (1985) The resource-ratio
hypothesisof plant
A.J. Gray,M.J. Crawley& P.J. Edwards),pp. 1-27.
succession.AmericanNaturalist,125, 827-852.
BlackwellScientific
Publications,Oxford.
Tilman,D. (1988) Plant Strategiesand theDynamicsand
Olff,H. (1992a) Effectsof lightand nutrient
availabilityon
Structureof Plant Communities.
PrincetonUniversity
drymatterand N allocationin six successionalgrassPress,Princeton.
land species: testing for resource ratio effects.
Van der Meijden,R., Weeda, E.J.,Adema,F.A.C.B. & de
Oecologia,89, 412-421.
JoncheereG.J. (1983) Flora van Nederland.Wolters
Olff,H. (1992b) On themechanismsof vegetationsuccesGroningen.
Noordhoff,
sion.Thesis,University
ofGroningen.
Van Tooren, B.F., Schat, H. & Ter Borg, S.J. (1983)
Olson, J.S. (1957) Rates of successionand soil changeson
Successionand fluctuation
in thevegetationof a Dutch
southernLake Michigansanddunes.Botanical Gazette,
Beach Plain. Vegetatio,
53, 139-151.
119, 125-169.
Vitousek,P.M. & Walker,L.R. (1987) Colonization,sucPeet,R.K. & Christensen,
N.L. (1980) Succession,a popucessionand resourceavailability:ecosystem-level
interlationprocess.Vegetatio,
43, 131-140.
actions.Colonization,Successionand Stability(eds A.
Penman,H.L. (1948) Naturalevaporationfromopen water,
J.Gray,M. J.Crawleyand P. J.Edwards),pp. 207-224.
bare soil and grass. Proceedingsof theRoyal Society
BlackwellScientific
Publications,
Oxford.
London,A193, 120-145.
Willis, A.J.,Folkes, B.F., Hope-Simpson,J.F. & Yemm,
Raunkiaer, C. (1934) The Life Forms of Plants and
E.W. (1959) BrauntonBurrows:The dune systemand
StatisticalPlant Geography.OxfordPress,London.
itsvegetation.
JournalofEcology,47, 1-24.
Roberts,R.D., Marrs,R.H., Skeffington,
R.A. & Bradshaw,
Willis, A.J. (1963) BrauntonBurrows:the effectson the
A.D. (1981) Ecosystemdevelopment
on naturally-colovegetationof the additionof mineralnutrients
to the
nized china clay wastes. I. Vegetationchanges and
dunesoils.JournalofEcology,51, 353-374.
overall accumulationof organicmatterand nutrients.
JournalofEcology,69, 153-161.
Robertson,G.P. (1982) Factors regulatingnitrification
in
primary and secondary succession. Ecology, 63,
Received23 July1992
1561-1573.
Revisedversionaccepted21 April1993
This content downloaded from 129.125.63.113 on Mon, 19 Aug 2013 10:11:29 AM
All use subject to JSTOR Terms and Conditions