Adema et al 2002
Journal of Vegetation Science 13: 107-114, 2002
© IAVS; Opulus Press Uppsala. Printed in Sweden
- Alternative stable states in a wet calcareous dune slack - 107
Alternative stable states in a wet calcareous dune slack
in The Netherlands
Adema, Erwin B.1*; Grootjans, Ab P.1; Petersen, Jörg2 & Grijpstra, Johan3
1Laboratory of Plant Ecology, University of Groningen, P.O. Box 14, 9750 AA, Haren, The Netherlands;
2Institut für Geobotanik der Universität Hannover, Nienburger Str. 17, D-30167 Hannover, Germany;
3Kiwa Onderzoek en Advies, P.O.Box 1072, 3430 BB Nieuwegein, The Netherlands;
*Corresponding author: Fax +31503632273; E-mail e.b.adema@biol.rug.nl
Abstract. Evidence is presented for the occurrence of alterna- Calcareous dune slacks show large variation in the
tive stable states in a wet calcareous dune slack on the Frisian persistence of low-production early successional pio-
island of Texel, The Netherlands. An early pioneer stage (0.5 neer stages (Lammerts et al. 1995). Pioneer stages may
kg m–2 total standing crop) and a more productive later succes- be very short-lived in some slacks, while in others they
sional stage (2.9 kg m–2) occur side by side, with sharp bounda-
can persist for many decades. Lammerts et al. (1995)
ries between them. The pioneer vegetation has been recorded
at the site for more than 62 yr. These features indicate the and Sival (1996) related these differences in succession
occurrence of a positive-feedback mechanism that has led to rate to differences in the hydrological functioning of the
alternative stable states. slack. Alkaline groundwater at surface level, at least
Analyses of ground and surface water composition, and during wintertime, is necessary for the stable persistence
decalcification depths, indicated that hydrologically the study of pioneer vegetation. If the groundwater level is lower
site can be characterized as a flow-through slack, with exfiltration and never reaches the surface level, the topsoil will acidify
of calcareous groundwater on one side and infiltration of sur- and organic matter can start to accumulate. The nutrient
face water on the other side of the slack. These differences in availability will increase (Olff et al. 1993), which favours
hydrological conditions have led to distinct differences in envi-
later successional stages. This may explain some of the
ronmental conditions within the dune slack. The occurrence of
the two successional stages can, however, not be explained by variation between different dune slacks but it does not
differences in hydrological conditions since both stages occur explain the variation within one dune slack where compa-
side by side in the centre of the dune slack. It is, therefore, rable hydrological conditions are present. The occurrence
more likely that biotic interactions are the cause of the vegeta- of the two successional stages indicates the presence of
tion pattern. Three possible mechanisms for feedback proc- positive-feedback mechanisms that stabilize the pioneer
esses are discussed: (1) enhanced nitrogen loss; (2) sulfide stage (Wilson & Agnew 1992).
toxicity and (3) nutrient accumulation in internal cycle. A positive-feedback switch can occur if one or both
vegetation types contain species that modify their en-
Keywords: Hydrology; Microbial mat; Multiple stable state;
vironment to their own advantage (Wilson & Agnew
Pioneer species; Succession; Sulfide.
1992). These positive-feedback mechanisms may lead
Nomenclature: van der Meijden (1996) for phanerogams; to alternative stable states and this may have large
Schaminée et al. (1995) for syntaxa. consequences for nature management. It means that
successful restoration or development of early succes-
sional stages will need much more effort than simply
Introduction restoring the abiotic conditions (Scheffer 1998).
Alternative stable states may cause hysteresis and
In the past decades, theoretical and empirical work research in the last decades reveals that this phenom-
has been done on the occurrence of alternative stable enon is not uncommon in nature (May 1977; Rietkerk &
states in ecology, mainly focusing on the occurrence of van de Koppel 1997; van de Koppel et al. 1997; Wilson
alternative stable states caused by plant-animal interac- & Agnew 1992). The problem is how to recognize
tions (Noy-Meir 1975; May 1977). Plant-plant interac- alternative stable states in the field.
tions can also induce alternative stable states (Wilson & Since alternative stable states occur only if a posi-
Agnew 1992). We report a situation where there is prima tive-feedback mechanism is operating, we can recog-
facie evidence that alternative stable states exist in veg- nize possible alternative stable states in the field by
etation succession of wet calcareous dune slacks. those switches. A positive-feedback switch leads to
108 Adema, E.B. et al.
increasing differences between states, and consequently
to a bimodal distribution of the system states (Fig. 1).
Wilson & Agnew (1992) mention four possible out-
comes of positive-feedback switches, all resulting from
the bimodal distribution: (1) occurrence of a stable
vegetational mosaic in a previously uniform environ-
ment, or (2) intensification of a vegetational gradient
leading to sharp boundaries, (3) delay or (4) accelera-
tion of succession by displacement or sharpening of
temporal boundaries. A fifth outcome is the fact that
coexistence is not possible and therefore intermediate
states are lacking in the field (Scheffer 1998).
This study investigates whether alternative stable
states occur in dune slack succession and how these
may be related to the hydrological regime.
Study area
The ‘Buiten Muy’ on the Frisian island of Texel (53∞
07' N 5∞ 47' E; Fig. 2) originates from a beach plain that
was separated from the sea by a dune ridge between
1920 and 1925 (Westhoff & van Oosten 1991). The Fig. 1. Effects of a positive-feedback mechanism on the vegeta-
‘Buiten Muy’ is part of the ‘De Muy’ nature reserve, tion pattern in the field. The possible outcomes in the field are all
which was established in 1908. The study area in the based on the principle that a positive-feedback mechanism
creates a bimodal distribution in the system states.
‘Buiten Muy’ can be divided in three areas. 1. The
southwestern part (1; Fig. 2) that is yearly mown in
order to maintain rare species such as Anagallis tenella,
Epipactis palustris and Ophioglossum vulgatum. 2. The Methods
unmanaged central part (2; Fig. 2) in which a pioneer
stage (Samolo-Littorelletum) with a well-developed mi- Three transects, running from southeast to north-
crobial mat is still present, despite more than 70 yr of west, were established across the slack (T1, T2 and T3;
undisturbed succession. 3. The northeastern part (3; Fig. Fig. 2). On each transect 3 plots were situated: an inward
2) where the top soil was removed in 1994. Most of the plot (i), a seaward plot (s) and a plot in the middle of the
unmanaged area consists of tall Scirpo-Phragmitetum transect (m). Two extra plots, i’ and s’ were added to the
dominated by Phragmites australis or Carex riparia, first transect (T1) in the Samolo-Littorelletum stand.
which represents a late successional stage. The transects, T1, (low-productive state, Samolo-
Fig. 2. Overview of the ‘Buiten Muy’ with sur-
rounding area. 1. mown area; 2. unmanaged area;
3. sod-cut area. A = Samolo-Littorelletum with
microbial mat; B = Scirpo-Phragmitetum; C =
Caricetum ripariae; D = Caricetum trinervi-
nigrae dominated by Salix repens. T1, T2, and
T3 are the transects.
- Alternative stable states in a wet calcareous dune slack - 109
Littorelletum) and T2 (highly-productive state, Scirpo- 4 ∞C prior analysis. The cations were analysed by using
Phragmitetum) lie in the unmanaged part of the slack, AAS. For each plot saturation indices for Ca were
whereas T3 is situated in the sod-cut area. Height of the calculated after Stuyfzand (1989).
surface was measured on each plot at the transects using
an automatic level. Data analysis
Detrended correspondence analysis (DCA) was ap-
Vegetation plied to detect successional trends in the Samolo-
The vegetation of the ‘Buiten Muy’ was sampled in Littorelletum stand. All relevés of this site were analyzed
the summers of 1996 and 2000 and classified into veg- with log transformed species cover-abundance. Vegeta-
etation types according to Braun-Blanquet (1964). These tion samples of the Scirpo-Pragmitetum and Caricetum
data were supplemented by older data from this area ripariae stands were included to indicate local succes-
(Petersen 2000). sional stages.
Replicate samples (n = 5) of 25 cm ¥ 50 cm were
harvested in T1 and T2 to determine above-ground
biomass, standing crop, separated in dead and alive. Results
Root biomass (n = 5) were taken with soil cores (diameter
= 7 cm; depth = 10 cm) after harvesting the litter. Vegetation
Soil parameters The DCA of the plots of the three vegetation types,
Calcium (Ca) content and pH (H2O) were measured (Fig. 3) shows a clear difference in the species composi-
in all transects at different depths up to 1 m. Total Ca tion. All the Samolo-Littorelletum plots lay on the left
content was measured after extraction with 1 M HCl by section of the graph, while the plots in the Caricetum
atomic adsorption spectrometry. pH (H2O) (glass elec- ripariae lay at the right side. The vegetation plots where
trode), bulk density, thickness of the organic layer, Phragmites dominates occurred in between these two
moisture content, and percentage organic matter (by on the first axis. This separation on the first axis of the
loss on ignition at 550 ºC) were determined in the topsoil DCA is very stable. The different years of the Samolo-
of all plots. All soil parameters were collected in the Littorelletum vary on the second axes of the DCA.
spring of 1997. The highly productive Scirpo-Phragmitetum stands
Oxygen saturation, redox potentials and sulfide (S2–) had higher values with respect to living and dead stand-
were measured in the soil profile using specific needle ing crop, as well as root and litter biomass, compared to
electrodes (van Gemerden 1993). The amount of free the low-production Samolo-Littorelletum stands (Table
sulfide was calculated from S2– and pH measuremens.
Profiles were measured up to 130 mm depth using a
micromanipulator for accurate depth positioning. The
measurements were carried out using step sizes of 1 mm
(0-15mm), 5 mm (15-100mm), or 10mm (100-130 mm).
These profile measurements were carried out in May
1998 in the inward and seaward parts of the slack – the
central part was inaccessible due to the high water level.
Groundwater parameters
Groundwater tubes were installed in each plot at
three different depths to measure the groundwater com-
position and water level. Water levels were measured
on 27-8-1997, 28-8-1997 (after a shower), and 10-10-
1997. Soil water samples were taken in October 1997,
and analysed for electric conductivity (EC25), and pH Fig. 3. The first two axes of the DCA showing the position of
(H2O) directly in the field. The concentrations of bicar- relevés of the three vegetation types distinguished; ▲ =
Scirpo-Phragmitetum; ▼ = Caricetum ripariae; all other sym-
bonate and carbon dioxide were measured by pH titra-
bols = Samolo-Littorelletum at different times. The Samolo-
tion within 24 hr after sampling. Chloride and sulfate
Littorelletum data shows a circular tendency in time. This is an
were measured by using an auto analyser (Skalar indication that this vegetation state is in equilibrium and
Analitical) directly after sampling. The samples for therefore stable in time. (Eigenvalues of DCA1 = 0.75 and
analyses of the cations: Na, K, Ca, Mg, and Fe were DCA2 = 0.26) The analysis is carried out with log transformed
adjusted to pH 2 using 4% HCl and stored in the dark at species cover-abundance.
110 Adema, E.B. et al.
Table 1. Plant biomass divided into live and dead standing Table 2. Soil parameters: pH, organic matter content (OM %)
crop, root biomass, and litter biomass. Values are mean and of the top 5 cm of the soil and the thickness of the organic
standard error in gm–2 from 5 samples. Asterisks mark plots matter layer. Asterisks mark plots that significantly differ
that significantly differ from T1-m in a non-parametric a from T1-m in a non-parametric a priori multiple comparison
priori multiple comparison (Zar 1984), * = P < 0.05 ** = P < (Zar 1984), * = P < 0.05 ** = P < 0.01.
0.01; n = 5. Plot pH (H2O) OM% OM
Plot Standing crop Standing crop Root Litter (mass) layer (cm)
live (g m–2) dead (g m–2) (g m–2) (g m–2) n =2 n =7 n =15
T1-s 235.4* 403.6** 1186 198 T1-s 7.92 1.9 0.9
T1-s’ 206.5* 53.4* 666 85** T1-s’ 7.83 3.1 * 0.7
T1-m 7.93 1.3 0.4
T1-m 92.9 0.0 382 0**
T1-i 7.61 ** 23.9 ** 8.2 **
T1-i 316.0** 515.1** 4140** 500
T2-s 6.23 ** 7.0 ** 12.3 **
T2-s 253.9* 713.5** 3895** 281** T2-m 7.55 ** 20.2 ** 7.3 **
T2-m 339.9** 85.4* 2195** 322** T2-i 7.53 ** 24.7 ** 10.0 **
T2-i 430.3** 383.9** 4181** 383** T3-s 8.22 ** 0.8 * 0.0 **
T3-s 147.2 74.4* 439 0** T3-m 8.03 * 0.8 0.1
T3-m 19.2 0.0 174 0 T3-i 8.38 ** 0.4 ** 0.0 **
T3-i 39.6 6.1 74 0
1). Litter and dead standing crop were not detectable in cut area, however, is more frequently inundated due to
the Samolo-Littorelletum stand. the lower surface level (Fig. 4). The pH of the topsoil
ranged from 6.23 to 7.93 in the unmanaged sites and
Soil parameters from 8.03 to 8.38 in the sod-cut area (Table 2). Large
differences existed between the sites in organic matter
The height difference of the two vegetational states content and thickness of the organic layer (Table 2).
within the unmanaged transects at soil surface level These two parameters were highest in the site with
were less than 6 cm, the sod-cut area is ca. 15 cm lower Phragmites australis. The higher organic matter con-
due to topsoil removal. We found no differences in tents were also reflected in lower bulk density values for
ground water level between the different sites. The sod- this site.
Fig. 4. Soil analyses of the three transects. Solid lines represent the soil surface and dashed lines the water table. The graphs at the
left show the calcium content. The grey area marks the CaCO3 containing soil, CA > 0.3 %. Numbers refer to the calcium content in
percentage of soil dry weight at the measurement locations. The graphs give the content of sulfate (middle) and iron (right) of the
groundwater (mg/l). The arrows mark the direction of the groundwater flow.
- Alternative stable states in a wet calcareous dune slack - 111
Groundwater parameters
Irrespective of the low calcium concentrations in the
soil at the seaward side Ca saturation indices (SIc) are all
between - 0.5 and + 0.5, indicating that the groundwater
is in equilibrium with Ca.
Depth distribution of sulfate and iron in the ground-
water differed between the three transects (Fig. 4). The
amount of sulfate in the groundwater was distinctly
higher at the inland side in all transects except for one
measurement in T3. The pattern for iron was somewhat
less distinct, but the highest iron concentrations were
also found at the inland side. The lowest concentrations
were found nearby the surface at the seaward sides for
T2 and T3 but closer to the middle for T1.
The surface water on the inland side of the slack
mainly consists of exfiltrated groundwater diluted by
some rainwater. The surface water on the inland side of
the slack differs from the surface water in the middle
and on the seaward side. In particular the concentrations
of Ca, iron, and bicarbonate are much higher at the
inward side. Chloride and sulfate concentrations in the
surface water did not differ (Table 3).
Fig. 5. Oxygen content, redox potential, and sulfide concen-
trations measured in soil profiles in exfiltration and infiltration Discussion
sites of the ‘Buiten Muy’. No free sulfide was measured in the
exfiltration site despite of the lower redox potential.
Hydrological system
The results from the water analyses indicate that the
Ca concentrations of the soil are markedly lower at
‘Buiten Muy’ is a dune slack containing a flow-through
the top of the profiles at the seaward side than in the rest
dune lake (Stuyfzand 1993; Stuyfzand & Moberts 1987)
of the dune slack. The highest value, more than 20% is
during a substantial part of the year (Fig. 6). The surface
found in the middle of T1 at the soil surface (0-2 mm),
water data in particular conform this idea. Furthermore,
due to calcium precipitation.
In contrast to the oxygen profile, the profiles of
redox potential and sulfide show significant differences
between the two sides of the dune slack (Fig. 5). Redox
potential was much higher on the seaward side than on
the inland side but sulfide is only found at the seaward
side.
Table 3. Composition of the surface water in the ’Buiten Muy’
on 26 May 1997.
Seaward Middle Inward
EC mS/cm 1035 1036 1245
pH (H2O) 9.9 9.6 7.8
CO2 mg/l 9.15
HCO3 - mg/l 64.3 76.1 273.8
CO3 2- mg/l 27.6 19.6
Cl- mg/l 230 274 232 Fig. 6. Schematic presentation of the hydrological system in
SO42- mg/l 12.8 13.5 9.7 the ‘Buiten Muy’ A = Scirpo-Phragmitetum; B = Samolo-
Na+ mg/l 148 144 125
K+ mg/l 5.1 5.4 9.7
Littorelletum; C = Caricetum ripariae. 1 = Incoming calcium
Ca2+ mg/l 27.3 29.5 88.8 and iron rich groundwater; 2 = Exfiltration of groundwater; 3 =
Mg2+ mg/l 15.7 15.5 17.6 Precipitation of iron and calcium; 4 = Infiltration of surface
Fe2++Fe3+ mg/l 1.2 0.9 7.1 water; 5 = Outgoing calcium- and iron-poor infiltration water.
112 Adema, E.B. et al.
the concentrations of Ca2+, HCO3-, and iron (Fe2+ + mechanism is present. 1. The vegetation pattern has been
Fe3+) in the groundwater and surface water were higher stable for more than 60 yr (Petersen 2000). 2. There is a
at the inland side of the slack, due to the discharge of Fe- sharp boundary between the two vegetational states of the
and Ca-rich groundwater here. Exfiltrated anoxic slack. 3. The normal succession, in which a late succes-
groundwater that originates from adjoining infiltration sional phase (dominated by Phragmites or Carex) will
areas contains reduced iron (Fe2+). At the surface, this replace the pioneer phase (Samolo-Littorelletum) within
soluble iron will be oxidized to Fe3+, which precipitates a few decades, is obviously delayed. 4. There are no
almost immediately at the soil surface, presumably as intermediate states in which both vegetation types coex-
iron oxides. ist. This clearly indicates that alternative stable states, or
The precipitation of Ca is caused by the release of at least positive-feedback switches, are operating in the
CO2 from the water at the surface. Algae in microbial ‘Buiten Muy’. There are also indications for the occur-
mats also consume much CO2 and therefore contribute rence of alternative stable states from other work on the
to the precipitation of CaCO3 (Chafetz 1994). The pre- Frisian Islands of Terschelling (Lammerts et al. 1995;
cipitation of iron and calcium carbonate can also be Petersen 2000) and Schiermonnikoog (Sival & Grootjans
clearly observed in the field. 1996). In some dune slacks the succession was extremely
As a result, when the surface water infiltrates at the slow (up to 80 yr), while in other slacks a vegetation shift
seaward side, most of the iron, calcium and bicarbonate occurred within 10 yr.
has been lost. This can explain why sulfide was present The theoretical consequences of alternative stable
in measurable quantities at the seaward side of the dune states are discussed by Saunders (1980) and Lockwood
slack. At this infiltration side, the soil contains less iron & Lockwood (1991) in their work on discontinuous
and is completely decalcified. When sulfate reduction phenomena in otherwise continuous systems (catastro-
occurs at the infiltration side, neither iron nor calcium phe theory). The theory predicts that small disturbances
is available in sufficient amounts to bind the sulfide that may induce rapid irreversible changes from a pioneer
is produced. stage with low nutrient accumulation to a highly pro-
In summary, we found hydrological and ecological ductive stage. Catastrophic events, such as prolonged
differences between the opposite sides of the slack, inundation in summer or human drainage activities can
which were clearly expressed in the vegetation, with trigger this irreversible shift.
Scirpo-Phragmitetum on the seaward side of the A practical consequence is that dune slacks that
Samolo-Littorelletum and Caricetum ripariae on the have entirely shifted towards a more productive state
inland side. However, the hydrological regime did not due to a disturbance will not simply return to a stable
differ clearly between pioneer, Samolo-Littorelletum, pioneer stage again when the disturbing agent is re-
and the late successional stage, Scirpo-Phragmitetum, moved. Extraction of drinking water and acid rainfall
in the middle of the unmanaged parts of the slack. The are examples of disturbances that can happen in dune
groundwater analyses do reflect differences in manage- slacks. To restore the desired early-successional vegeta-
ment (between unmanaged and sod-cut areas) but the tion, additional measures are necessary. For instance
hydrological system along the three transects is very sod cutting can be applied to make sure that the abiotic
similar. It is, therefore, likely that at the start, more than conditions are suitable for the pioneer vegetation, which
75 yr ago, the two successional stages in the centre of then suffers less from competition with the later species.
the slack had the same starting conditions. Nevertheless, first of all the restoration of a stable low-
productive dune slack vegetation requires a restored
Alternative stable states hydrological system.
A reconstruction of the past succession could be as Positive-feedback mechanisms
follows. After the beach plain became separated from
the influence of the sea, most of the vegetation of the Our field observations and earlier research give evi-
slack was in a pioneer stage. In the course of time, parts dence for three possible positive-feedback mechanisms
of the slack started to develop towards a late succes- that can occur in wet dune slacks. A first mechanism
sional stage dominated by Phragmites australis or Carex could be the adaptation of pioneer species to anoxic
riparia. The difference between the two vegetation types, and nutrient-poor conditions. Ernst (1991) showed that
the pioneer type and the subsequent one, increased with Schoenus nigricans has a very low nutrient demand.
time due to one or more positive-feedback mechanisms. The species is very efficient by withdrawing nutrients
This led to distinctly different stable vegetation states. from the top to the base of the shoots, from were the
The ‘Buiten Muy’ shows 4 of the 5 possible shoots grow. Several pioneer species show radial oxy-
vegetational results that can occur if a positive-feedback gen loss (ROL) from the roots (Armstrong 1982; Roelofs
- Alternative stable states in a wet calcareous dune slack - 113
Fig. 7. Three possible feedback mechanisms that can occur in a calcareous wet dune slack: (a) Enhanced nitrogen loss; (b) Sulfide
toxicity; and (c) Nutrient accumulation in internal cycle.
et al. 1984) which, in a calcareous environment, facili- different in calcareous dune slacks. Van Beckhoven
tates rapid decomposition of organic matter. Under such (1995) did not find significant differences in decompo-
conditions nitrification may occur on a very local scale sition rates in litter formed by Schoenus nigricans,
(Engelaar et al. 1991; Reddy et al. 1989; Vitousek & Calamagrostis epigejos and Molinia caerulea. Further-
Walker 1987). Further away from the roots this can lead more, litter of the pioneer species Littorella uniflora and
to denitrification and thus to nutrient losses from the Samolus valerandi decomposes very rapidly. Later suc-
system (Fig. 7a). If pioneer species are indeed capable cessional species are more efficient in retaining nutrients
of facilitating decomposition of their own litter or oth- (Ernst et al. 1996). They store nutrients in organic
erwise keep the nutrient accumulation at a low level by material and retrieve them the next growth season, so
stimulating nutrient losses from the system, they could the internal nutrient cycle increases (Fig. 7c).
efficiently stabilize the pioneer phase. If one of the first two mechanisms coexists with the
In the second possible mechanism a microbial mat is third mechanism we can explain the sharp boundaries.
involved. Microbial mats often cover the soil surface in However, sharp boundaries can also exist if only one of
open pioneer vegetation. Microbial mats are dominated the first two mechanisms occurs. Pioneer vegetation
by a few functional groups of microbes: cyanobacteria, cannot invade the later successional stages because pio-
colourless sulfur bacteria, purple sulfur bacteria, and neer species suffer from shortage of light under the taller
sulfate-reducing bacteria. Their combined metabolic productive species.
activities result in steep environmental micro-gradients, Further research will address those possible posi-
particularly of oxygen and sulfide (van Gemerden 1993). tive-feedback mechanisms. This research shows that it
Sulfide is toxic for most higher plants and inhibits is most likely that those mechanisms occur but more
the growth of most plant species. (Havill et al. 1985; knowledge about the precise mechanisms is necessary
Grootjans et al. 1997; Lamers et al. 1998). Therefore the for successful restoration and management of wet dune
vegetation remains open. However, some characteristic slack pioneer vegetation.
dune slack pioneer species can protect themselves against
the toxic sulfide by releasing oxygen from their root Acknowledgements. We wish to thank H. van Gemerden
system (Fig. 7b). Free sulfide will be detoxified by and J. van Andel for their valuable comments on earlier
colourless sulfur bacteria in the oxic rhizosphere. This versions of the manuscript.
results in open stable pioneer vegetation with a micro-
bial mat that cannot be invaded by later species not
adapted to anoxic soils containing free sulfide. References
A third possible feedback mechanism could be litter
Armstrong, W. 1982. Waterlogged soils. In: Etherington, J.R.
accumulation that increases the internal nutrient cycle.
(ed.) Environment and plant ecology, pp. 290- 330. Wiley,
Later successional species grow on sites with higher Chichester, UK.
nutrient pools (Olff et al. 1993; Lammerts et al. 1999). Berendse, F. 1993. Ecosystem stability, competition and nu-
Berendse (1993) found that in wet heathlands the spe- trient cycling. In: Schulze, E.D. & Mooney, H.A. (eds.)
cies of relatively nutrient-rich habitats had a higher Biodiversity and ecosystem function, pp. 409- 431. Springer-
nutrient turnover and produced litter that is more rap- Verlag, Berlin, DE.
idly decomposable than species of poor habitats Berendse, F. & Elberse, W.T. 1990. Competition and nutrient
(Berendse & Elberse 1990). The situation appears to be availability in heathland and grassland ecosystems. In: Grace,
114 Adema, E.B. et al.
J.B. & Tilman, D. (eds.) Perspectives on plant competi- Reddy, K.R., Patrick, W.H.J. & Lindau, C.W. 1989. Nitrifica-
tion, pp. 94-116. Academic Press, New York. NY. tion-denitrification at the plant root-sediment interface in
Braun-Blanquet, J. 1964. Pflanzensoziologie. Grundzüge der wetlands. Limnol. Oceanogr. 34:1004-1013.
Vegetationskunde. Springer, Berlin, DE. Rietkerk, M. & van de Koppel, J. 1997. Alternate stable states
Chafetz, H.S. 1994. Bacterially induced precipitates of cal- and threshold effects in semi-arid grazing systems. Oikos
cium carbonate and lithification of microbial mats. In: 79: 69-76.
Krumbein, W.E., Paterson, D.M. & Stal, L.J. (eds.) Roelofs, J.G.M., Schuurkes, J.A.A.R. & Smits, A.J.M. 1984.
Biostabilization of sediments, pp. 149-164. BIS Verlag, Impact of acidification and eutrophication and macrophyte
Oldenburg, DE. communities in soft waters. II. Experimental studies. Aquat.
Engelaar, W.M.H.G., Bodelier, P.L.E., Laanbroek, H.J. & Bot. 18: 389-411.
Blom, C.W.P.M. 1991. Nitrification in the rhizosphere of Saunders, P.T. 1980. An introduction to catastrophe theory.
a flooding-resistant and a flooding-non-resistant Rumex Cambridge University Press, Cambridge, UK.
species under drained and waterlogged conditions. FEMS Schaminee, J.H.J., Weeda, E.J. & Westhoff, V. 1995 De vege-
Microbiol. Ecol. 86: 33-42. tatie van Nederland 2. Plantengemeenschappen van wateren,
Ernst, W.H.O. 1991. Ökophysiologie von Pflanzen in Küsten- moerassen en natte heiden. Opulus Press, Uppsala, SE.
dünen Europas in einem Gradienten von Nordsee zum Scheffer, M. 1998. The ecology of shallow lakes. Chapman &
Mittelmeer. Ber. Reinhold Tüxen-Ges. 3: 157-172. Hall, London, UK.
Ernst, W.H.O., Slings, Q.L. & Nelissen, H.J.M. 1996. Pedo- Sival, F.P. 1996. Mesotrophic basiphilous communities af-
genesis in coastal wet dune slacks after sod-cutting in fected by changes in soil properties in two dune slack
relation to revegetation. Plant Soil 180: 219-230. chronosequences. Act. Bot. Neerl. 45: 95-106.
Grootjans, A.P., van der Ende, F.P. & Walsweer, A.F. 1997. Sival, F.P. & Grootjans, A.P. 1996. Dynamics of seasonal
The role of microbial mats during primary succession in bicarbonate supply in a dune slack: Effects on organic
calcareous dune slacks: an experimental approach. J. matter, nitrogen pool and vegetation succession. Vegetatio
Coastal Conserv. 3: 95-102. 126: 39-50.
Havill, D.C., Ingold, A. & Pearson, J. 1985. Sulphide toler- Stuyfzand, P.J. 1989. An accurate relatively simple calcula-
ance in coastal halophytes. Vegetatio 62: 279-285. tion of the saturation index of calcite for fresh to salt water.
Lamers, L.P.M., Tomassen, H.B.M. & Roelofs, J.G.M. 1998. J. Hydrol. 105: 95-107.
Sulfate-induced eutrophication and phytotoxity in fresh Stuyfzand, P.J. 1993. Hydrochemistry and hydrology of the
water wetlands. Environ. Sci. Technol. 32: 199-205. coastal dune area of the western Netherlands. Ph.D. The-
Lammerts, E.J., Grootjans, A.P., Stuyfzand, P.J. & Sival, F.P. sis, Vrije Universteit, Amsterdam, NL.
1995. Endangered dune slack plants; gastronomers in need Stuyfzand, P.J. & Moberts, F.M.L. 1987 De bijzondere
of mineral water. In: Salman, A.H.P.M., Berends, H. & hydrologie van kwelplassen in duinen met kunstmatige
Bonazountas, M. (eds.) Coastal management and habitat infiltratie. H2O 20: 52-57. (In Dutch, with summary.)
conservation, pp. 355-369. EUCC, Leiden, NL. van Beckhoven, K. 1995. Rewetting of coastal dune slacks:
Lammerts, E.J., Pegtel, D.M., Grootjans, A.P. & van der effects on plant gowth and soil processes. Ph.D. Thesis,
Veen, A. 1999. Nutrient limitation and vegetation changes Vrije Universiteit, Amsterdam, NL.
in a coastal dune slack. J. Veg. Sci. 10: 111-122. van de Koppel, J., Rietkerk, M. & Weissing, F.J. 1997. Cata-
Lockwood, J.A. & Lockwood, D.R. 1991. Rangeland grass- strophic vegetation shifts and soil degradation in terres-
hopper (Orthoptera: Acrididae) population dynamics: trial grazing systems. Trends Ecol. Evol. 12: 352-356.
Insights from catastrophe theory. Environ. Entomol. 20: van der Meijden, R. 1996. Heukels Flora van Nederland.
970-980. Wolters-Noordhof, Groningen, NL.
May, R.M. 1977. Thresholds and breakpoints in ecosystems van Gemerden, H. 1993. Microbial mats: A joint venture.
with a multiplicity of stable states. Nature 269: 471-477. Mar. Geol. 113: 3-25.
Noy-Meir, I. 1975. Stability of grazing systems: an applica- Vitousek, P.M. & Walker, L.R. 1987. Colonization,succession
tion of predator prey graphs. J. Ecol. 63: 459-481. and resource availability ecosystem-level interactions.
Olff, H., Huisman, J. & van Tooren, B.F. 1993. Species In:Gray A.J., Crawley M.J. & Edwards P.J. (eds.) Coloni-
dynamics and nutrient accumulation during early primary zation, succession and stability, pp. 207- 223. Blackwell,
succession in coastal sand dunes. J. Ecol. 81: 693-706. Oxford, UK.
Petersen, J. 2000. Die Dünentalvegetation der Wattenmeer- Westhoff, V. & van Oosten, M. F. 1991. De plantengroei van
Inseln in der südlichen Nordsee. Eine pflanzensoziologische de waddeneilanden. KNNV, Utrecht, NL.
und ökologische Vergleichsuntersuchung unter Berück- Wilson, J.B. & Agnew, A.D.Q. 1992. Positive-feedback switches
sichtigung von Nutzung und Naturschutz. Husum Druck- in plant communities. Adv. Ecol. Res. 23: 263-337.
und Verlagsgesellschaft, Husum, DE. Zar, J.H. 1984. Biostatistical analyses. Prentice-Hall,
Englewood Cliffs, NJ.
Received 7 June 2001;
Revision received 16 November 2001;
Accepted 30 November 2001.
Coordinating Editor: J.B. Wilson.
© IAVS; Opulus Press Uppsala. Printed in Sweden
- Alternative stable states in a wet calcareous dune slack - 107
Alternative stable states in a wet calcareous dune slack
in The Netherlands
Adema, Erwin B.1*; Grootjans, Ab P.1; Petersen, Jörg2 & Grijpstra, Johan3
1Laboratory of Plant Ecology, University of Groningen, P.O. Box 14, 9750 AA, Haren, The Netherlands;
2Institut für Geobotanik der Universität Hannover, Nienburger Str. 17, D-30167 Hannover, Germany;
3Kiwa Onderzoek en Advies, P.O.Box 1072, 3430 BB Nieuwegein, The Netherlands;
*Corresponding author: Fax +31503632273; E-mail e.b.adema@biol.rug.nl
Abstract. Evidence is presented for the occurrence of alterna- Calcareous dune slacks show large variation in the
tive stable states in a wet calcareous dune slack on the Frisian persistence of low-production early successional pio-
island of Texel, The Netherlands. An early pioneer stage (0.5 neer stages (Lammerts et al. 1995). Pioneer stages may
kg m–2 total standing crop) and a more productive later succes- be very short-lived in some slacks, while in others they
sional stage (2.9 kg m–2) occur side by side, with sharp bounda-
can persist for many decades. Lammerts et al. (1995)
ries between them. The pioneer vegetation has been recorded
at the site for more than 62 yr. These features indicate the and Sival (1996) related these differences in succession
occurrence of a positive-feedback mechanism that has led to rate to differences in the hydrological functioning of the
alternative stable states. slack. Alkaline groundwater at surface level, at least
Analyses of ground and surface water composition, and during wintertime, is necessary for the stable persistence
decalcification depths, indicated that hydrologically the study of pioneer vegetation. If the groundwater level is lower
site can be characterized as a flow-through slack, with exfiltration and never reaches the surface level, the topsoil will acidify
of calcareous groundwater on one side and infiltration of sur- and organic matter can start to accumulate. The nutrient
face water on the other side of the slack. These differences in availability will increase (Olff et al. 1993), which favours
hydrological conditions have led to distinct differences in envi-
later successional stages. This may explain some of the
ronmental conditions within the dune slack. The occurrence of
the two successional stages can, however, not be explained by variation between different dune slacks but it does not
differences in hydrological conditions since both stages occur explain the variation within one dune slack where compa-
side by side in the centre of the dune slack. It is, therefore, rable hydrological conditions are present. The occurrence
more likely that biotic interactions are the cause of the vegeta- of the two successional stages indicates the presence of
tion pattern. Three possible mechanisms for feedback proc- positive-feedback mechanisms that stabilize the pioneer
esses are discussed: (1) enhanced nitrogen loss; (2) sulfide stage (Wilson & Agnew 1992).
toxicity and (3) nutrient accumulation in internal cycle. A positive-feedback switch can occur if one or both
vegetation types contain species that modify their en-
Keywords: Hydrology; Microbial mat; Multiple stable state;
vironment to their own advantage (Wilson & Agnew
Pioneer species; Succession; Sulfide.
1992). These positive-feedback mechanisms may lead
Nomenclature: van der Meijden (1996) for phanerogams; to alternative stable states and this may have large
Schaminée et al. (1995) for syntaxa. consequences for nature management. It means that
successful restoration or development of early succes-
sional stages will need much more effort than simply
Introduction restoring the abiotic conditions (Scheffer 1998).
Alternative stable states may cause hysteresis and
In the past decades, theoretical and empirical work research in the last decades reveals that this phenom-
has been done on the occurrence of alternative stable enon is not uncommon in nature (May 1977; Rietkerk &
states in ecology, mainly focusing on the occurrence of van de Koppel 1997; van de Koppel et al. 1997; Wilson
alternative stable states caused by plant-animal interac- & Agnew 1992). The problem is how to recognize
tions (Noy-Meir 1975; May 1977). Plant-plant interac- alternative stable states in the field.
tions can also induce alternative stable states (Wilson & Since alternative stable states occur only if a posi-
Agnew 1992). We report a situation where there is prima tive-feedback mechanism is operating, we can recog-
facie evidence that alternative stable states exist in veg- nize possible alternative stable states in the field by
etation succession of wet calcareous dune slacks. those switches. A positive-feedback switch leads to
108 Adema, E.B. et al.
increasing differences between states, and consequently
to a bimodal distribution of the system states (Fig. 1).
Wilson & Agnew (1992) mention four possible out-
comes of positive-feedback switches, all resulting from
the bimodal distribution: (1) occurrence of a stable
vegetational mosaic in a previously uniform environ-
ment, or (2) intensification of a vegetational gradient
leading to sharp boundaries, (3) delay or (4) accelera-
tion of succession by displacement or sharpening of
temporal boundaries. A fifth outcome is the fact that
coexistence is not possible and therefore intermediate
states are lacking in the field (Scheffer 1998).
This study investigates whether alternative stable
states occur in dune slack succession and how these
may be related to the hydrological regime.
Study area
The ‘Buiten Muy’ on the Frisian island of Texel (53∞
07' N 5∞ 47' E; Fig. 2) originates from a beach plain that
was separated from the sea by a dune ridge between
1920 and 1925 (Westhoff & van Oosten 1991). The Fig. 1. Effects of a positive-feedback mechanism on the vegeta-
‘Buiten Muy’ is part of the ‘De Muy’ nature reserve, tion pattern in the field. The possible outcomes in the field are all
which was established in 1908. The study area in the based on the principle that a positive-feedback mechanism
creates a bimodal distribution in the system states.
‘Buiten Muy’ can be divided in three areas. 1. The
southwestern part (1; Fig. 2) that is yearly mown in
order to maintain rare species such as Anagallis tenella,
Epipactis palustris and Ophioglossum vulgatum. 2. The Methods
unmanaged central part (2; Fig. 2) in which a pioneer
stage (Samolo-Littorelletum) with a well-developed mi- Three transects, running from southeast to north-
crobial mat is still present, despite more than 70 yr of west, were established across the slack (T1, T2 and T3;
undisturbed succession. 3. The northeastern part (3; Fig. Fig. 2). On each transect 3 plots were situated: an inward
2) where the top soil was removed in 1994. Most of the plot (i), a seaward plot (s) and a plot in the middle of the
unmanaged area consists of tall Scirpo-Phragmitetum transect (m). Two extra plots, i’ and s’ were added to the
dominated by Phragmites australis or Carex riparia, first transect (T1) in the Samolo-Littorelletum stand.
which represents a late successional stage. The transects, T1, (low-productive state, Samolo-
Fig. 2. Overview of the ‘Buiten Muy’ with sur-
rounding area. 1. mown area; 2. unmanaged area;
3. sod-cut area. A = Samolo-Littorelletum with
microbial mat; B = Scirpo-Phragmitetum; C =
Caricetum ripariae; D = Caricetum trinervi-
nigrae dominated by Salix repens. T1, T2, and
T3 are the transects.
- Alternative stable states in a wet calcareous dune slack - 109
Littorelletum) and T2 (highly-productive state, Scirpo- 4 ∞C prior analysis. The cations were analysed by using
Phragmitetum) lie in the unmanaged part of the slack, AAS. For each plot saturation indices for Ca were
whereas T3 is situated in the sod-cut area. Height of the calculated after Stuyfzand (1989).
surface was measured on each plot at the transects using
an automatic level. Data analysis
Detrended correspondence analysis (DCA) was ap-
Vegetation plied to detect successional trends in the Samolo-
The vegetation of the ‘Buiten Muy’ was sampled in Littorelletum stand. All relevés of this site were analyzed
the summers of 1996 and 2000 and classified into veg- with log transformed species cover-abundance. Vegeta-
etation types according to Braun-Blanquet (1964). These tion samples of the Scirpo-Pragmitetum and Caricetum
data were supplemented by older data from this area ripariae stands were included to indicate local succes-
(Petersen 2000). sional stages.
Replicate samples (n = 5) of 25 cm ¥ 50 cm were
harvested in T1 and T2 to determine above-ground
biomass, standing crop, separated in dead and alive. Results
Root biomass (n = 5) were taken with soil cores (diameter
= 7 cm; depth = 10 cm) after harvesting the litter. Vegetation
Soil parameters The DCA of the plots of the three vegetation types,
Calcium (Ca) content and pH (H2O) were measured (Fig. 3) shows a clear difference in the species composi-
in all transects at different depths up to 1 m. Total Ca tion. All the Samolo-Littorelletum plots lay on the left
content was measured after extraction with 1 M HCl by section of the graph, while the plots in the Caricetum
atomic adsorption spectrometry. pH (H2O) (glass elec- ripariae lay at the right side. The vegetation plots where
trode), bulk density, thickness of the organic layer, Phragmites dominates occurred in between these two
moisture content, and percentage organic matter (by on the first axis. This separation on the first axis of the
loss on ignition at 550 ºC) were determined in the topsoil DCA is very stable. The different years of the Samolo-
of all plots. All soil parameters were collected in the Littorelletum vary on the second axes of the DCA.
spring of 1997. The highly productive Scirpo-Phragmitetum stands
Oxygen saturation, redox potentials and sulfide (S2–) had higher values with respect to living and dead stand-
were measured in the soil profile using specific needle ing crop, as well as root and litter biomass, compared to
electrodes (van Gemerden 1993). The amount of free the low-production Samolo-Littorelletum stands (Table
sulfide was calculated from S2– and pH measuremens.
Profiles were measured up to 130 mm depth using a
micromanipulator for accurate depth positioning. The
measurements were carried out using step sizes of 1 mm
(0-15mm), 5 mm (15-100mm), or 10mm (100-130 mm).
These profile measurements were carried out in May
1998 in the inward and seaward parts of the slack – the
central part was inaccessible due to the high water level.
Groundwater parameters
Groundwater tubes were installed in each plot at
three different depths to measure the groundwater com-
position and water level. Water levels were measured
on 27-8-1997, 28-8-1997 (after a shower), and 10-10-
1997. Soil water samples were taken in October 1997,
and analysed for electric conductivity (EC25), and pH Fig. 3. The first two axes of the DCA showing the position of
(H2O) directly in the field. The concentrations of bicar- relevés of the three vegetation types distinguished; ▲ =
Scirpo-Phragmitetum; ▼ = Caricetum ripariae; all other sym-
bonate and carbon dioxide were measured by pH titra-
bols = Samolo-Littorelletum at different times. The Samolo-
tion within 24 hr after sampling. Chloride and sulfate
Littorelletum data shows a circular tendency in time. This is an
were measured by using an auto analyser (Skalar indication that this vegetation state is in equilibrium and
Analitical) directly after sampling. The samples for therefore stable in time. (Eigenvalues of DCA1 = 0.75 and
analyses of the cations: Na, K, Ca, Mg, and Fe were DCA2 = 0.26) The analysis is carried out with log transformed
adjusted to pH 2 using 4% HCl and stored in the dark at species cover-abundance.
110 Adema, E.B. et al.
Table 1. Plant biomass divided into live and dead standing Table 2. Soil parameters: pH, organic matter content (OM %)
crop, root biomass, and litter biomass. Values are mean and of the top 5 cm of the soil and the thickness of the organic
standard error in gm–2 from 5 samples. Asterisks mark plots matter layer. Asterisks mark plots that significantly differ
that significantly differ from T1-m in a non-parametric a from T1-m in a non-parametric a priori multiple comparison
priori multiple comparison (Zar 1984), * = P < 0.05 ** = P < (Zar 1984), * = P < 0.05 ** = P < 0.01.
0.01; n = 5. Plot pH (H2O) OM% OM
Plot Standing crop Standing crop Root Litter (mass) layer (cm)
live (g m–2) dead (g m–2) (g m–2) (g m–2) n =2 n =7 n =15
T1-s 235.4* 403.6** 1186 198 T1-s 7.92 1.9 0.9
T1-s’ 206.5* 53.4* 666 85** T1-s’ 7.83 3.1 * 0.7
T1-m 7.93 1.3 0.4
T1-m 92.9 0.0 382 0**
T1-i 7.61 ** 23.9 ** 8.2 **
T1-i 316.0** 515.1** 4140** 500
T2-s 6.23 ** 7.0 ** 12.3 **
T2-s 253.9* 713.5** 3895** 281** T2-m 7.55 ** 20.2 ** 7.3 **
T2-m 339.9** 85.4* 2195** 322** T2-i 7.53 ** 24.7 ** 10.0 **
T2-i 430.3** 383.9** 4181** 383** T3-s 8.22 ** 0.8 * 0.0 **
T3-s 147.2 74.4* 439 0** T3-m 8.03 * 0.8 0.1
T3-m 19.2 0.0 174 0 T3-i 8.38 ** 0.4 ** 0.0 **
T3-i 39.6 6.1 74 0
1). Litter and dead standing crop were not detectable in cut area, however, is more frequently inundated due to
the Samolo-Littorelletum stand. the lower surface level (Fig. 4). The pH of the topsoil
ranged from 6.23 to 7.93 in the unmanaged sites and
Soil parameters from 8.03 to 8.38 in the sod-cut area (Table 2). Large
differences existed between the sites in organic matter
The height difference of the two vegetational states content and thickness of the organic layer (Table 2).
within the unmanaged transects at soil surface level These two parameters were highest in the site with
were less than 6 cm, the sod-cut area is ca. 15 cm lower Phragmites australis. The higher organic matter con-
due to topsoil removal. We found no differences in tents were also reflected in lower bulk density values for
ground water level between the different sites. The sod- this site.
Fig. 4. Soil analyses of the three transects. Solid lines represent the soil surface and dashed lines the water table. The graphs at the
left show the calcium content. The grey area marks the CaCO3 containing soil, CA > 0.3 %. Numbers refer to the calcium content in
percentage of soil dry weight at the measurement locations. The graphs give the content of sulfate (middle) and iron (right) of the
groundwater (mg/l). The arrows mark the direction of the groundwater flow.
- Alternative stable states in a wet calcareous dune slack - 111
Groundwater parameters
Irrespective of the low calcium concentrations in the
soil at the seaward side Ca saturation indices (SIc) are all
between - 0.5 and + 0.5, indicating that the groundwater
is in equilibrium with Ca.
Depth distribution of sulfate and iron in the ground-
water differed between the three transects (Fig. 4). The
amount of sulfate in the groundwater was distinctly
higher at the inland side in all transects except for one
measurement in T3. The pattern for iron was somewhat
less distinct, but the highest iron concentrations were
also found at the inland side. The lowest concentrations
were found nearby the surface at the seaward sides for
T2 and T3 but closer to the middle for T1.
The surface water on the inland side of the slack
mainly consists of exfiltrated groundwater diluted by
some rainwater. The surface water on the inland side of
the slack differs from the surface water in the middle
and on the seaward side. In particular the concentrations
of Ca, iron, and bicarbonate are much higher at the
inward side. Chloride and sulfate concentrations in the
surface water did not differ (Table 3).
Fig. 5. Oxygen content, redox potential, and sulfide concen-
trations measured in soil profiles in exfiltration and infiltration Discussion
sites of the ‘Buiten Muy’. No free sulfide was measured in the
exfiltration site despite of the lower redox potential.
Hydrological system
The results from the water analyses indicate that the
Ca concentrations of the soil are markedly lower at
‘Buiten Muy’ is a dune slack containing a flow-through
the top of the profiles at the seaward side than in the rest
dune lake (Stuyfzand 1993; Stuyfzand & Moberts 1987)
of the dune slack. The highest value, more than 20% is
during a substantial part of the year (Fig. 6). The surface
found in the middle of T1 at the soil surface (0-2 mm),
water data in particular conform this idea. Furthermore,
due to calcium precipitation.
In contrast to the oxygen profile, the profiles of
redox potential and sulfide show significant differences
between the two sides of the dune slack (Fig. 5). Redox
potential was much higher on the seaward side than on
the inland side but sulfide is only found at the seaward
side.
Table 3. Composition of the surface water in the ’Buiten Muy’
on 26 May 1997.
Seaward Middle Inward
EC mS/cm 1035 1036 1245
pH (H2O) 9.9 9.6 7.8
CO2 mg/l 9.15
HCO3 - mg/l 64.3 76.1 273.8
CO3 2- mg/l 27.6 19.6
Cl- mg/l 230 274 232 Fig. 6. Schematic presentation of the hydrological system in
SO42- mg/l 12.8 13.5 9.7 the ‘Buiten Muy’ A = Scirpo-Phragmitetum; B = Samolo-
Na+ mg/l 148 144 125
K+ mg/l 5.1 5.4 9.7
Littorelletum; C = Caricetum ripariae. 1 = Incoming calcium
Ca2+ mg/l 27.3 29.5 88.8 and iron rich groundwater; 2 = Exfiltration of groundwater; 3 =
Mg2+ mg/l 15.7 15.5 17.6 Precipitation of iron and calcium; 4 = Infiltration of surface
Fe2++Fe3+ mg/l 1.2 0.9 7.1 water; 5 = Outgoing calcium- and iron-poor infiltration water.
112 Adema, E.B. et al.
the concentrations of Ca2+, HCO3-, and iron (Fe2+ + mechanism is present. 1. The vegetation pattern has been
Fe3+) in the groundwater and surface water were higher stable for more than 60 yr (Petersen 2000). 2. There is a
at the inland side of the slack, due to the discharge of Fe- sharp boundary between the two vegetational states of the
and Ca-rich groundwater here. Exfiltrated anoxic slack. 3. The normal succession, in which a late succes-
groundwater that originates from adjoining infiltration sional phase (dominated by Phragmites or Carex) will
areas contains reduced iron (Fe2+). At the surface, this replace the pioneer phase (Samolo-Littorelletum) within
soluble iron will be oxidized to Fe3+, which precipitates a few decades, is obviously delayed. 4. There are no
almost immediately at the soil surface, presumably as intermediate states in which both vegetation types coex-
iron oxides. ist. This clearly indicates that alternative stable states, or
The precipitation of Ca is caused by the release of at least positive-feedback switches, are operating in the
CO2 from the water at the surface. Algae in microbial ‘Buiten Muy’. There are also indications for the occur-
mats also consume much CO2 and therefore contribute rence of alternative stable states from other work on the
to the precipitation of CaCO3 (Chafetz 1994). The pre- Frisian Islands of Terschelling (Lammerts et al. 1995;
cipitation of iron and calcium carbonate can also be Petersen 2000) and Schiermonnikoog (Sival & Grootjans
clearly observed in the field. 1996). In some dune slacks the succession was extremely
As a result, when the surface water infiltrates at the slow (up to 80 yr), while in other slacks a vegetation shift
seaward side, most of the iron, calcium and bicarbonate occurred within 10 yr.
has been lost. This can explain why sulfide was present The theoretical consequences of alternative stable
in measurable quantities at the seaward side of the dune states are discussed by Saunders (1980) and Lockwood
slack. At this infiltration side, the soil contains less iron & Lockwood (1991) in their work on discontinuous
and is completely decalcified. When sulfate reduction phenomena in otherwise continuous systems (catastro-
occurs at the infiltration side, neither iron nor calcium phe theory). The theory predicts that small disturbances
is available in sufficient amounts to bind the sulfide that may induce rapid irreversible changes from a pioneer
is produced. stage with low nutrient accumulation to a highly pro-
In summary, we found hydrological and ecological ductive stage. Catastrophic events, such as prolonged
differences between the opposite sides of the slack, inundation in summer or human drainage activities can
which were clearly expressed in the vegetation, with trigger this irreversible shift.
Scirpo-Phragmitetum on the seaward side of the A practical consequence is that dune slacks that
Samolo-Littorelletum and Caricetum ripariae on the have entirely shifted towards a more productive state
inland side. However, the hydrological regime did not due to a disturbance will not simply return to a stable
differ clearly between pioneer, Samolo-Littorelletum, pioneer stage again when the disturbing agent is re-
and the late successional stage, Scirpo-Phragmitetum, moved. Extraction of drinking water and acid rainfall
in the middle of the unmanaged parts of the slack. The are examples of disturbances that can happen in dune
groundwater analyses do reflect differences in manage- slacks. To restore the desired early-successional vegeta-
ment (between unmanaged and sod-cut areas) but the tion, additional measures are necessary. For instance
hydrological system along the three transects is very sod cutting can be applied to make sure that the abiotic
similar. It is, therefore, likely that at the start, more than conditions are suitable for the pioneer vegetation, which
75 yr ago, the two successional stages in the centre of then suffers less from competition with the later species.
the slack had the same starting conditions. Nevertheless, first of all the restoration of a stable low-
productive dune slack vegetation requires a restored
Alternative stable states hydrological system.
A reconstruction of the past succession could be as Positive-feedback mechanisms
follows. After the beach plain became separated from
the influence of the sea, most of the vegetation of the Our field observations and earlier research give evi-
slack was in a pioneer stage. In the course of time, parts dence for three possible positive-feedback mechanisms
of the slack started to develop towards a late succes- that can occur in wet dune slacks. A first mechanism
sional stage dominated by Phragmites australis or Carex could be the adaptation of pioneer species to anoxic
riparia. The difference between the two vegetation types, and nutrient-poor conditions. Ernst (1991) showed that
the pioneer type and the subsequent one, increased with Schoenus nigricans has a very low nutrient demand.
time due to one or more positive-feedback mechanisms. The species is very efficient by withdrawing nutrients
This led to distinctly different stable vegetation states. from the top to the base of the shoots, from were the
The ‘Buiten Muy’ shows 4 of the 5 possible shoots grow. Several pioneer species show radial oxy-
vegetational results that can occur if a positive-feedback gen loss (ROL) from the roots (Armstrong 1982; Roelofs
- Alternative stable states in a wet calcareous dune slack - 113
Fig. 7. Three possible feedback mechanisms that can occur in a calcareous wet dune slack: (a) Enhanced nitrogen loss; (b) Sulfide
toxicity; and (c) Nutrient accumulation in internal cycle.
et al. 1984) which, in a calcareous environment, facili- different in calcareous dune slacks. Van Beckhoven
tates rapid decomposition of organic matter. Under such (1995) did not find significant differences in decompo-
conditions nitrification may occur on a very local scale sition rates in litter formed by Schoenus nigricans,
(Engelaar et al. 1991; Reddy et al. 1989; Vitousek & Calamagrostis epigejos and Molinia caerulea. Further-
Walker 1987). Further away from the roots this can lead more, litter of the pioneer species Littorella uniflora and
to denitrification and thus to nutrient losses from the Samolus valerandi decomposes very rapidly. Later suc-
system (Fig. 7a). If pioneer species are indeed capable cessional species are more efficient in retaining nutrients
of facilitating decomposition of their own litter or oth- (Ernst et al. 1996). They store nutrients in organic
erwise keep the nutrient accumulation at a low level by material and retrieve them the next growth season, so
stimulating nutrient losses from the system, they could the internal nutrient cycle increases (Fig. 7c).
efficiently stabilize the pioneer phase. If one of the first two mechanisms coexists with the
In the second possible mechanism a microbial mat is third mechanism we can explain the sharp boundaries.
involved. Microbial mats often cover the soil surface in However, sharp boundaries can also exist if only one of
open pioneer vegetation. Microbial mats are dominated the first two mechanisms occurs. Pioneer vegetation
by a few functional groups of microbes: cyanobacteria, cannot invade the later successional stages because pio-
colourless sulfur bacteria, purple sulfur bacteria, and neer species suffer from shortage of light under the taller
sulfate-reducing bacteria. Their combined metabolic productive species.
activities result in steep environmental micro-gradients, Further research will address those possible posi-
particularly of oxygen and sulfide (van Gemerden 1993). tive-feedback mechanisms. This research shows that it
Sulfide is toxic for most higher plants and inhibits is most likely that those mechanisms occur but more
the growth of most plant species. (Havill et al. 1985; knowledge about the precise mechanisms is necessary
Grootjans et al. 1997; Lamers et al. 1998). Therefore the for successful restoration and management of wet dune
vegetation remains open. However, some characteristic slack pioneer vegetation.
dune slack pioneer species can protect themselves against
the toxic sulfide by releasing oxygen from their root Acknowledgements. We wish to thank H. van Gemerden
system (Fig. 7b). Free sulfide will be detoxified by and J. van Andel for their valuable comments on earlier
colourless sulfur bacteria in the oxic rhizosphere. This versions of the manuscript.
results in open stable pioneer vegetation with a micro-
bial mat that cannot be invaded by later species not
adapted to anoxic soils containing free sulfide. References
A third possible feedback mechanism could be litter
Armstrong, W. 1982. Waterlogged soils. In: Etherington, J.R.
accumulation that increases the internal nutrient cycle.
(ed.) Environment and plant ecology, pp. 290- 330. Wiley,
Later successional species grow on sites with higher Chichester, UK.
nutrient pools (Olff et al. 1993; Lammerts et al. 1999). Berendse, F. 1993. Ecosystem stability, competition and nu-
Berendse (1993) found that in wet heathlands the spe- trient cycling. In: Schulze, E.D. & Mooney, H.A. (eds.)
cies of relatively nutrient-rich habitats had a higher Biodiversity and ecosystem function, pp. 409- 431. Springer-
nutrient turnover and produced litter that is more rap- Verlag, Berlin, DE.
idly decomposable than species of poor habitats Berendse, F. & Elberse, W.T. 1990. Competition and nutrient
(Berendse & Elberse 1990). The situation appears to be availability in heathland and grassland ecosystems. In: Grace,
114 Adema, E.B. et al.
J.B. & Tilman, D. (eds.) Perspectives on plant competi- Reddy, K.R., Patrick, W.H.J. & Lindau, C.W. 1989. Nitrifica-
tion, pp. 94-116. Academic Press, New York. NY. tion-denitrification at the plant root-sediment interface in
Braun-Blanquet, J. 1964. Pflanzensoziologie. Grundzüge der wetlands. Limnol. Oceanogr. 34:1004-1013.
Vegetationskunde. Springer, Berlin, DE. Rietkerk, M. & van de Koppel, J. 1997. Alternate stable states
Chafetz, H.S. 1994. Bacterially induced precipitates of cal- and threshold effects in semi-arid grazing systems. Oikos
cium carbonate and lithification of microbial mats. In: 79: 69-76.
Krumbein, W.E., Paterson, D.M. & Stal, L.J. (eds.) Roelofs, J.G.M., Schuurkes, J.A.A.R. & Smits, A.J.M. 1984.
Biostabilization of sediments, pp. 149-164. BIS Verlag, Impact of acidification and eutrophication and macrophyte
Oldenburg, DE. communities in soft waters. II. Experimental studies. Aquat.
Engelaar, W.M.H.G., Bodelier, P.L.E., Laanbroek, H.J. & Bot. 18: 389-411.
Blom, C.W.P.M. 1991. Nitrification in the rhizosphere of Saunders, P.T. 1980. An introduction to catastrophe theory.
a flooding-resistant and a flooding-non-resistant Rumex Cambridge University Press, Cambridge, UK.
species under drained and waterlogged conditions. FEMS Schaminee, J.H.J., Weeda, E.J. & Westhoff, V. 1995 De vege-
Microbiol. Ecol. 86: 33-42. tatie van Nederland 2. Plantengemeenschappen van wateren,
Ernst, W.H.O. 1991. Ökophysiologie von Pflanzen in Küsten- moerassen en natte heiden. Opulus Press, Uppsala, SE.
dünen Europas in einem Gradienten von Nordsee zum Scheffer, M. 1998. The ecology of shallow lakes. Chapman &
Mittelmeer. Ber. Reinhold Tüxen-Ges. 3: 157-172. Hall, London, UK.
Ernst, W.H.O., Slings, Q.L. & Nelissen, H.J.M. 1996. Pedo- Sival, F.P. 1996. Mesotrophic basiphilous communities af-
genesis in coastal wet dune slacks after sod-cutting in fected by changes in soil properties in two dune slack
relation to revegetation. Plant Soil 180: 219-230. chronosequences. Act. Bot. Neerl. 45: 95-106.
Grootjans, A.P., van der Ende, F.P. & Walsweer, A.F. 1997. Sival, F.P. & Grootjans, A.P. 1996. Dynamics of seasonal
The role of microbial mats during primary succession in bicarbonate supply in a dune slack: Effects on organic
calcareous dune slacks: an experimental approach. J. matter, nitrogen pool and vegetation succession. Vegetatio
Coastal Conserv. 3: 95-102. 126: 39-50.
Havill, D.C., Ingold, A. & Pearson, J. 1985. Sulphide toler- Stuyfzand, P.J. 1989. An accurate relatively simple calcula-
ance in coastal halophytes. Vegetatio 62: 279-285. tion of the saturation index of calcite for fresh to salt water.
Lamers, L.P.M., Tomassen, H.B.M. & Roelofs, J.G.M. 1998. J. Hydrol. 105: 95-107.
Sulfate-induced eutrophication and phytotoxity in fresh Stuyfzand, P.J. 1993. Hydrochemistry and hydrology of the
water wetlands. Environ. Sci. Technol. 32: 199-205. coastal dune area of the western Netherlands. Ph.D. The-
Lammerts, E.J., Grootjans, A.P., Stuyfzand, P.J. & Sival, F.P. sis, Vrije Universteit, Amsterdam, NL.
1995. Endangered dune slack plants; gastronomers in need Stuyfzand, P.J. & Moberts, F.M.L. 1987 De bijzondere
of mineral water. In: Salman, A.H.P.M., Berends, H. & hydrologie van kwelplassen in duinen met kunstmatige
Bonazountas, M. (eds.) Coastal management and habitat infiltratie. H2O 20: 52-57. (In Dutch, with summary.)
conservation, pp. 355-369. EUCC, Leiden, NL. van Beckhoven, K. 1995. Rewetting of coastal dune slacks:
Lammerts, E.J., Pegtel, D.M., Grootjans, A.P. & van der effects on plant gowth and soil processes. Ph.D. Thesis,
Veen, A. 1999. Nutrient limitation and vegetation changes Vrije Universiteit, Amsterdam, NL.
in a coastal dune slack. J. Veg. Sci. 10: 111-122. van de Koppel, J., Rietkerk, M. & Weissing, F.J. 1997. Cata-
Lockwood, J.A. & Lockwood, D.R. 1991. Rangeland grass- strophic vegetation shifts and soil degradation in terres-
hopper (Orthoptera: Acrididae) population dynamics: trial grazing systems. Trends Ecol. Evol. 12: 352-356.
Insights from catastrophe theory. Environ. Entomol. 20: van der Meijden, R. 1996. Heukels Flora van Nederland.
970-980. Wolters-Noordhof, Groningen, NL.
May, R.M. 1977. Thresholds and breakpoints in ecosystems van Gemerden, H. 1993. Microbial mats: A joint venture.
with a multiplicity of stable states. Nature 269: 471-477. Mar. Geol. 113: 3-25.
Noy-Meir, I. 1975. Stability of grazing systems: an applica- Vitousek, P.M. & Walker, L.R. 1987. Colonization,succession
tion of predator prey graphs. J. Ecol. 63: 459-481. and resource availability ecosystem-level interactions.
Olff, H., Huisman, J. & van Tooren, B.F. 1993. Species In:Gray A.J., Crawley M.J. & Edwards P.J. (eds.) Coloni-
dynamics and nutrient accumulation during early primary zation, succession and stability, pp. 207- 223. Blackwell,
succession in coastal sand dunes. J. Ecol. 81: 693-706. Oxford, UK.
Petersen, J. 2000. Die Dünentalvegetation der Wattenmeer- Westhoff, V. & van Oosten, M. F. 1991. De plantengroei van
Inseln in der südlichen Nordsee. Eine pflanzensoziologische de waddeneilanden. KNNV, Utrecht, NL.
und ökologische Vergleichsuntersuchung unter Berück- Wilson, J.B. & Agnew, A.D.Q. 1992. Positive-feedback switches
sichtigung von Nutzung und Naturschutz. Husum Druck- in plant communities. Adv. Ecol. Res. 23: 263-337.
und Verlagsgesellschaft, Husum, DE. Zar, J.H. 1984. Biostatistical analyses. Prentice-Hall,
Englewood Cliffs, NJ.
Received 7 June 2001;
Revision received 16 November 2001;
Accepted 30 November 2001.
Coordinating Editor: J.B. Wilson.