PRIMARY SUCCESSION BEACH PLAIN:
Species
and nuTrient
relation accumulation.
biomass
isman
dynamics
OIS" "' I) Lf')J
PRIMARY SUCCESSION ON A BEACH PLAIN:
SPECIES DYNAMICS IN RELATION TO BIOMASS AND NUTRIENT ACCUMULATION
Jef Huisman
Begeleiding: Han 01ff
RijksUniversiteit Groningen Doctoraal—onderwerp Plantenoecologie
Vakgroep Biologie van Planten 1990
Rijksunlversitelt Groflingefl BbllOtheek BjOIOOISCh Centrum
Kerklaafl 30 —
postbus 14 9750 AA HARENDe
resultaten vermeld in dit versiag worden niomenteel bewerkt voor publicatie.
Overname of gebruik van gegevens is
dan ook
niet toegestaan.CONTENTS
Summary 3
Introduction 4
Materials and methods
The study area 6
The permanent transects 7
Analyses on the data of species dynamics 7
Reconstruction of changes in biomass, light and soil 9
Results
Analyses on the data of species dynamics 12 Reconstruction of changes in biomass, light and soil 14
Discussion 17
References 22
Tables & figures 24
SUMMARY
The present study reports on a primary succession series which started on the bare beach of the Dutch island Schiermonnikoog after the building of a sand dike in 1959.
Vegetational changes were studied by means
of permanent transects. Soil development and vegetation structure during this successional series could be reconstructed, since several stages of this succession werepresent in the same part of the island.
Salinity, moisture content and flooding are considered to be major determinants of the spatial variation, ranging from saline, wet plains to dry, fresh dunes, and are likely to beresponsible for the year to year fluctuations of
short—living species.However, these factors did not show a trend in time. From soil analyses it is argued that nitrogen limits the total biomass. In about 16 years the total amount of nitrogen in the organic layer of the soil increased from 7 to 50 g N m in the plains and from 1 to 15 g N nf2 on the dunes. The accumulation of nitrogen during succession is accompanied by an increased biomass, a decreased
light penetration to
the soil—surface, a decreased proportion root, an increased proportion stem, an increased plant height at maturity, and a decreasing abundance of short—living species. Only during the first stage on the dry dunes sand—blowing is considered to prevent small species from establishing. With the exception of this early dune stage, our observations are in agreement with the resource ratio hypothesis of Tilman (1985). Causes of the high rates of nitrogen accumulation in the soil are discussed. The observed nitrogen accumulation, which isconsidered to be an important
determinant of the successional dynamics, is thought to be facilitated by the vegetation.INTRODUCTION
Primary succession starts on a bare soil, without plant propagules and organic matter. This can occur after for instance sand dune formation, the retreat of
glaciers, heavy erosion or volcanic eruptions. Except for nitrogen, all of the mineral elements required by plants occur in the parent material in which most soils form (Jenny, 1980). In general, the bare substrates on which primary successions start are therefore very low in nitrogen (Gerlach 1988, Crocker and Major, 1955). Therefore, nitrogen will often be the limiting soil resource during the early stages of primary successions.
The rate of accuu1ation of nitrogen in the system will be determined by for example the abundance of nitrogen—fixing organisms and the rate of atmospheric input. The total amount of the limiting nutrient in an ecosystem imposes an absolute limit on the maximal possible biomass of the system (DeAngelis et al, 1989)
Robertson and Vitousek (1981) and Robertson (1982) found that the mineralization potential of the sand dunes at Lake Michigan increased with their total nitrogen content. The same was found by Gerlach (1988) on the sand dunes of Spiekeroog. Willis (1963) observed that N—addition at the dunes of Braunton Burrows, England, was followed by an increased above—ground standing crop and a change in species composition. In his experiments P—addition did not increase the standing crop, but N+P—addition was followed by a larger
standing crop compared to
the N—addition, with other species reaching dominance. The addition of only P, trace—elements and K, Ca and Mg had no effects.An increased above—ground standing crop in a more productive habitat will lead to a decrease in the amount of light penetrating to the soil—surface.
This observation led Tilman (1985) to the formulation of the resource ratio hypothesis of succession (Tilman, 1985). This hypothesis is based on the fact that plants face a trade—off while foraging for light and below—ground resources (nutrients, water). This implicates that plants which are superior competitors for light will be less efficient in competing for nutrients and vice versa. Succession starting on a poor soil driven by the accumulation of nitrogen in the system will therefore implicate competitive replacement of nutrient competitors by light competitors, since the relative availabilities of nutrients and light changes through time. This hypothesis states that competition will be a major determinant of species dynamics during primary succession.
An increase of the total amount of a limiting nutrient, accompanied with an increase of the availability of this nutrient, and a constant loss rate, will lead to an increased total plant bioniass and a decreased light penetration to the soil—surface. In such a time series the resource ratio hypothesis predicts a succession characterized by a decrease in the allocation to root, an increase in the allocation to stem,
an increase in the plant height at
maturity and a decreasing in abundance of short—living species.The predictions of the resource ration hypothesis were not in correspondence with the data on two different secondary succession series. Tilman had to reject his hypothesis as an explanation for the first 60 years of secondary old field succession on a nitrogen poor Minnesota sand plain (Gleeson and Tilnian 1989). During this series the allocation to roots increased with
increasing nutrient availability. Berendse and Elberse (1989) and 01ff et al.
(1990), studying Dutch grassland—successions under hay—making, found that species from relatively poor successional stages had a higher shoot/root ratio than species from richer stages.
Other hypothesis have been proposed to
explain successional dynamics starting on a bare soil. Early species might allocate more to many small seeds which makes them good colonists. Later species might be poorer colonists but good nutrient competitors. This trade—off is likely to explain the old field succession at the Minnesota sand plain (Gleeson and Tilman, 1989) and comes close to a shift from 'Ruderals' to 'Competitors' sensu Grime (1979). A trade—off between maximal relative growth rate and competitive abilities isconsidered to
be important to explain succession patterns starting on nutrient—rich soils (Grime, 1979; Chapin, 1980; Tilman, 1988).Other trade—offs might be of importance when dealing with other physical constraints on plant growth. A change in these constraints might therefore also lead to a change in vegetation composition. Halofyts, for instance, invest a lot in physiological and morphological properties which enable them to withstand high salt concentrations. However, most halofyts seem to have their physiological growth optimum in freshwater conditions (Rozema, 1978).
A critical revision of the literature on halophytes led Barbour (1970) to conclude an absence of any obligate halophyts among the angiosperms. Yet a lot of reports have described successions where desalination was considered to be the main cause of species replacements (see for instance Feekes (1936) and Joenje (1974)). Apparently, under fresh conditions, the investments of halofyts only lead to growth reductions in competitive situations. Trade—offs beteen investments in avoidance and tolerance mechanisms against harmful effects of high salinity and investments in competitive ability for nutrients and light can be expected.
Although Tilman (1988) gives examples of primary successions at Glacier Bay, Alaska (Crocker and Major, 1955; Lawrence, 1967), and at the sand dunes of Lake Nichigan (Olson, 1957), which seem to agree with his resource—ratio hypothesis, no detailed comparison between the pattern of a primary succession and the predictions of the resource—ratio hypothesis has yet been published.
This paper will deal with succession on a former beach, which got more and more vegetated after the building of a sand dike.
The development of this vegetation is considered to be a good example of a priniary succession starting on a poor soil.
This successional series will be described in terms of species patterns, changes in biornass, soil development and allocation patterns, whereby trends will be separated from fluctuations.
The outcome of these analyses is compared with the predictions of conceptual models of plant succession.
MATERIALS AND METHODS
The study area
The study area, the 'Beach Plain', extends from benchmark 7.4 to 10.4 on the eastern side of the Waddensea—island Schiermonnikoog (The Netherlands, 53°29'NL, 6°12'EL, Fig 1).
Several authors did earlier research in this area. The ecophysiology of some halophytes was related to the vegetation zonation of the Beach Plain (Rozema,
1978). A reconstruction of the vegetational development by aerial photographes and a few vegetation records was made by Van der Laan (1980). Wapenaar (1980) performed a vegetation mapping in 1979. Annual fluctuations of some
theophytes
were related to annual and seasonal fluctuations in the precipitation balance (Van Tooren, Schat and Ter Borg, 1982). The ecophysio—logy and population ecology of some therofytes and short—living perennials on the Beach Plain was studied by Schat (1982).
Originally the Beach Plain was a nearly bare sand flat, with scattered embryo—dunes up to 2 meters in height. These dunes were sparsely vegetated with Elymus farctus and Ammophila arenaria. The lower parts were bare, or sparsely covered with the annuals Salicornia stricta, Spergularia marina and Suaeda maritima (Van der Laan,l980). In 1959, with the construction of a sand ridge, the area was screaried off from the direct influence of the North Sea.
However, during heavy storms in 1972, a large opening was formed between benchmark 10 and 10.2. In most winters, when the water table is higher than at least 1.8 meter above NAP (Dutch Ordnance Level), the area is flooded by North Sea water coming through this opening. At higher tides, when the water, table is at least 2.1 meter above NAP, also water from the Wadden Sea can enter the area.
The seawater may stay in the Beach Plain for months. This occurs because the relatively low position of the flat parts of the Beach Plain prevents the inundation water from flowing out. In April or May, the water table in the lower parts usually falls below the soil surface, normally reaching a depth of about 40 — 70 cm. below soil surface in summer. In extremely dry summers a depth of 1 meter may be reached.
Fluctuations in the lenght of the inundation period and in the precipitation/—
evaporation balance result in a strongly fluctuating salinity of the upper soil layers of the lower parts (Van Tooren, Schat and Ter Borg 1983). Rozema (1978) found, during the summer, a 3 to 15 cm. higher groundwater table with a much lower salinity (50 mM NaC1) in the central parts of the small embryo—
dunes when compared to the groundwater of the flat area (295 mM), which indicates the existence of a fresh water lens in these small dunes. The capillary rise of the soil moisture was estimated at about 110 cm. (Rozema, 1978). This means that the higher parts of the dunes cannot be supplied with (fresh) capillary groundwater in contrast to the lower zones.
At present, the small dunes are densely covered with Hippophae rhamnoides, a thorny shrub able to fix nitrogen, and the lower parts are dominated by Juncus gerardii, Scirpus maritimus and Phragmites australis.
The permanent transects
The
wegetation development was recorded from 1972 onwards by Ter Borg, Van Toor en, Zonneveld, Keizer and a lot of undergraduate students, using a series of 1 Inetransects
lying near benchmark 8.2. This location comprises a flat low part and a dune slope with a northern exposure (Fig. 2), with the lowest point being 1.3 meter above NAP. Twelve permanent transects of 20 meters length eachwere
laid out perpendicular on the heightlines of the site, at a distance of 1 me -cer from each other. Every year in August, the presence of every species was recorded in 1 meter long and 40 cm. wide strips along these transects, resu1ting in data from 12 X 20 240 rectangles. In some years, mainly before 1979 , the recording was not complete: parts of the plot and/or some specieswere
not recorded. But in all cases it was known which parts and which specieswere
not recorded.Analysis of the data on species dynamics
T- nd
Becat15e the main scope of our study was to analyze vegetation dynamics under different sets of physical constraints, the transect—area was divided into five sub—areas, with the spatial variation within each sub—area reduced as far as possible (Table 1, Fig. 2). Criteria used for this division were mean height and soil—salinity (measured in summer by the electrical conductivity of the upper 10 cm. (unpublished data) of the rectangular strips.
Successional trend derived from presence/absence data can be seen as a
change in time (t)
of
the chance (p) to find a certain species in a given surface, in this case the surface of a strip (0.4 m2', in a given area. Trends were calculated for each combination of species and subarea, given that a species occurred in a subarea for at least four years. Five different models were used for determining the shape of p(t) as variations on the general modelp(t) f(t)/(1+f(t)) *
g(t)/(1+g(t))
(Eqn. 1)where the functions f(t) and g(t) were defined as f(t) exp(a), g(t)
(model I, Eqn. 2) f(t) exp(a+b*t), g(t)
(model II, Eqn. 3)
f(t) exp(a+b*t), g(t) exp(c) (model III, Eqn. 4)
f(t) exp(a+b*t+c*t2), g(t)
(model IV, Eqn. 5)
f(t) exp(a+b*t), g(t) exp(c+d*t) (model V,
Eqn. 7) where a, b, c and d are the coefficients to be estimated. p(t) will be called the expected frequency of a species. These models can be ranked according to an increasing number of parameters determining their shape, and are therefore of increasing complexity. Model I implies no trend (constant p(t), model II an increasing or decreasing trend, model III an increasing or decreasing trend with
an
upper limit, model IV implies an increase and a decrease at the samerate ("symmetrical around a maximum") and model V implies an increase and decrease with different rates ("skewed with a maximum"). Working with models
7
of increasing complexity allows to choose for the simplest possible model which sufficiently explains the observed pattern. The coefficients of model 1,11 and IV were estimated by logistic regression. The coefficients of model III and V were estimated using non—linear regression in SPSSPC
v.3.1
(1989).The Levenberg—Marquardt algorithm was used to solve the unconstrained models.
Logistic regression minimizes the statistic —2LL by iteration, defined as
—2LL SUM(_2*log(L0/Ll)) (Eqn 8)
where LO is the predicted probability of membership in the correct group in this model (for example, suppose the chance of finding a certain species in a strip is predicted to be 0.8 then, when this species is found LO —
0.8,
whenthis species is not found LO 0.2) and Li is the predicted probability in a perfect model, which is always one, because in a perfect model the prediction is always correct. This simplifies to
—2LL SUM(—2*log(L0)) (Eqn. 9)
Nonlinear regression minimizes the residual sum of squares by iteration, so for model III and V, —2LL was calculated after fitting the regression model.
Because —2LL has a Chi—Square distribution, with degrees of freedom equal to the number of cases minus the number of coefficients used in the model (SPSS v.3.1, 1989), the decrease of —2LL when putting in more coefficients has a Chi—Square distribution with degrees of freedom equal to the number of extra coefficients. Thus it can be tested whether this decrease is significant.
Starting with model I, only more complex models were accepted when the new model lead to a significant decrease in —2LL (p<.O5). Model III and IV have the same number of coefficients, therefore a choice between model III and model IV cannot be based on a significant improvement. Whenever a choice between model III and IV was necessary the one with the lowest —2LL was
chosen.
Missing data were accounted for by weighting each year by the number of strips recorded. The criteria used to stop the iterations where the default criteria of SPSS v.3.1.
The trends were also used for calculations on species—richness, on the rate of succession and on the occurrence of species with different life—forms and maximal heights. By using the trends in stead of the observed values for each year, the predicted patterns were not obfuscated by annual fluctuations.
The expected species—richness of a strip is just the summation over all occurring species of the expected chance to find them in a strip. Because the chance to find a species on a surface of a strip was estimated, also the chance to find a species on a surface of two strips could be estimated. So the expected species—richness could also be calculated for larger surfaces.
The rate of succession was computed as the percentage dissimilarity (PD) for each year with the previous year with 1973 and with 1989 (01ff & Bakker, in press), with
PD(y) — 100 * (1 —
(2c/(a+b))
(Eqn. 10)where a is the sum over all species of their expected chances in year y, b is the sum over all species of their expected chances in the year with which year y is compared, c is the sum over all species of the minimum of these two expected chances. The classification of Raunkiaer (1934) of life forms was used and maximal heights of the species were taken from the Dutch botanical database (CBS, 1987). Life—form spectra per year per subarea were calculated by weighting each species by it's expected frequency. The 6 species which
couTJd behave both as hemicryptofyt and geofyt were assigned to a separate gro'i The weights of the 3 other species able to exhibit more than one life—
form
were equally divided among their life—form groups. Mean maximal plant heLlt for each subarea and each year was calculated as the mean of the lOlog trasformed maximal heights weighted by the expected chance of each species.F],uc tuations
The standardized residual (SR) is defined as the residual (observed relative
frequency minus expected chance) divided by an estimate of its standard
deviatiOfl, in this case:residual(t)
SR(t)
— (Eqn. 11)SQRT(p(t)*(l—p(t))
Mu1tJ-le regressions were computed using SR(t) as the dependent variable and precIPitation deficit over the period april—june and maximal height of the flooding seawater in the period september—april as independent variables.
MetOr0l0giCal data, collected 2 km west of the permanent plot, were supplied by the Geological Institute of the Free University of Amsterdam. Precipitation deficit is defined as actual precipitation minus evaporation, according to Penma1. Data on the waterlevel of the WaddenSea, collected 3 km to the south, were derived from the annual reports of Rijkswaterstaat, The Hague.
CorilatiOflS between species
It might be expected that decreasing species are negatively correlated with increasing species. But whenever the disappearance of a species is directly caused by competitive replacement due to the appearance of another species,
it might be expected that the maximum chance of finding the decreasing species in a very small, homogeneous area is 1 minus the chance to find the increasing species. It is the maximum chance, because other factors (e.g. dispersal, competition with other species) might cause the chance to be lower. Since Hjppphae rhamnoides is considered to be a major determinant of the structure of the community, the decrease of other species was related to the increase of this species.
The age of the shrub Hippophae rhamnoides in a certain year in a certain strip was estimated by counting the number of years Hippophae had already been present in this strip. The first year the age was taken zero (seedlings);
whenever in a certain year in a certain strip Hippophae was absent, the counting started anew. The ages derived in this way were in agreement with the present ages of the shrubs of the permanent transects.
Reconstruction of changes in biomass, light and soil
The data of the line—transects only yielded information on the frequency of occurrence, not on the abundance of the different species. Also changes in biomaS5 or in the distribution of this biomass over the diffent functional units of a plant are unknown and no data on soil—development, amount of nutrients or light—profile were gathered. Therefore, to get some insight in the mechanisms of succession in the Beach Plain a reconstruction of the changes in the above—mentioned factors had to be made.
The vegetation of Plain, Slope and Dune during the intervals 1972/73, 1980/81 and 1988/89 were characterized by means of the trends. In and around the Beach Plain, we searched for sites were the vegetation showed close resemblance to these 9 characteristics. In these vegetations 3 plots per characteristic were situated. These plots were considered to be representative for the situation which occurred at the line—transects at the place and time of their characteristics. The subsites will be called Plain'72, Plain'80, and so on.
The location of the plots is shown in fig.1. The Plain'72, Slope'72 and Dune'72 plots were placed in front of the gap at benchmark 10.2. The plots of
'80 were located east of this gap in a dune area which was bare
sand, according to aerial photographs, until about ten years ago. The '88 plots were laid down in the neighborhood of the permanent line transects.Line—transects were made in the beginning of September in the same way as described earlier but with 3 cords of 5 meters length resulting in 15 strips per plot. This yielded to the possibility to investigate if the plots were a good representation for their successional stage in their spatial context. All other samplings at the plots were done during 1—15 july 1990.
Biomass and light
The above—ground standing crop, root biomass, light profile and percentage cover were measured for each plot.
Above—ground standing crop, without the shrub Hippophae, was sampled by clipping an area of 0.4 x 0.4 m. per plot. These samples were sorted to living species and dead organic material, dried to constant mass at 70°C and weighted. The above—ground biomass of Hippophae was estimated by a non—
destructive procedure, sampling over a larger area.
For this, we measured the above—ground fresh weight (B, in grams) and 70°C dried biomass, height (h, in cm), diameter at the bottom of the stem (d, in cm) and age of several Hiptovhae shrubs in the Beach Plain, ranging in height from 30 to 250 cm. The age of each shrub was determined as the number of yearrings counted at the bottom of the stem. The above—ground biomass of each individual shrub could be estimated by B 2.8l5*(d2*h)°9094 (r 0.99, N =
13).
Dry weight of the shrub was calculated by multiplying the fresh weight by 0.478 (r 0.98, N 7). Avarage total above—ground biomass (dry—weight) of 1-Iippophae rhamnoides was estimated by taking an area around each plot of at least 15 m2 in which we estimated the dry—weight of each shrub by measuring h and d.Roots were collected by taking two 20 cm. deep, 7 cm. diameter soil cores per plot. Nearly all roots were within these 20 cm. The cores were rinsed of soil and litter under a fine water spray to obtain the roots, which were dried and weighed. The roots could not be separated into living and dead. At each plot the vertical light profile in the vegetation was measured using 5 centimeter intervals, using a PAR collector (400—700 rim) with a measuring surface of 100 x 1 cm. The light extinction at each height was expressed as a fraction of the
ambiant light intensity above the vegetation.
Allocation patterns
Proportion root was calculated as root biomass divided by total biomass. For
a lot of Cramineae, Juncaceae and Cyperaceae
it was quit arbitrary to
discriminate between leaves and stems. Therefore, asa measure for the
"proportion stem", we divided the vegetation height (height at which 5% of the light was intercepted) by total biomass. This measure was called height per biomass (HB). Similarly, the cover per biomass (CB, percentage cover divided by total biomass) was used as measure for "proportion leaves".
10
:L measurements
The thickness of the organic layer was measured at ten randomly chosen sites in .ach plot. The dark brown organic layer showed a sharp boundary with the
jrLying yellow to greyish sand, which will be
referred to as the mineral layr Samples were taken from the organic layer and from the mineral layer at depth between 10 and 15 cm, at least at from five samples taken near each These cores were pooled and mixed until at least 700 gram per sample wasgat1eth This
resulted in one organic—layer—sample and one mineral—layer—sanipJe per plot. Soil samples taken near each plot were analyzed for moisture contt, specific weight of the soil, NaC1, CaCO3, pl1(l-120), pH(KC1), organic
matt content,
total carbon, total nitrogen and total phosphorus, using staridard procedures.The moisture content and specific dry—weight
were mea1ed by taking 2 samples per plot from the upper 5 centimeters of the soil and2
samples from the mineral layer at a depth between 10 and 15 cm. using 100 l. pF—rings. The samples of the upper—soil were cut of at the depth of the rganic layer and for these samples only this organic layer was used. The samples were freshly weighted, dried at 70°C and weighed again.At .1ain'72, Dune'72 and Dune'80, the organic layer was too poorly developed for .sampling it separately; in these cases the organic—layer—samples were
sampl of the upper 2 cm. of the soil.
The total amounts of organic matter, C, N and F in the organic layer (A) were estirmat for all plots, except the Plain'72, Dune'72 and Dune'80 plots, by th[AJ x sw[AJ x ct{A], where ththickness, sw=specific weight and ctcontent.
The anmounts in the organic layer at Plain'72, Dune'72 and Dune'80 were estixit by 2 x sw[AJ x ct[AJ — (2 —
th[AJ)
x sw[BJ x ct[B) where Bmineral layei. The total amount of organic matter in the upper 20 cm. of the soil wascalcL1'at by th[A] x sw[AJ x ct[AJ
+ (20 —th[A))
x sw[BJ x ct[BJ with th[AJ=2 for Plain'72, Dune'72 and Dune'80.analyses
The effects of time interval ('72,'80 and '88) and place (Plain, Slope and Dune) on biomass and on soil characteristics of the organic and of the mineral layer were determined using two—way analyses of variance with Student—Newman—
KeulS contrasts among treatment means.
11
RESULTS
Analysis of the data on species dynamics Trends
The Plain
During the first years, hemicryptofyts like Glaux inaritima and Agrostis stolonifera are very abundant in the Plain (Fig. 3 and Fig. 6A). Liinonium vulgare and Plantago inaritima, both halofyts, can be found too. Also a lot of therofytic species like the halofyts Spergularia spec, and Salicornia spec.
and the glycofyts Odontites verna ssp. serotina, Centauriuln pulchelluin and Centauriurn littorale (a biannual) occur. The rhizomatous species Juncus gerardii, Scirpus maritimus, Juncus maritimus and Phragmites australis and the hemicryptofyt Potentilla anserina gradually increase, whereas Glaux maritima, Limoniuni vulgare, Plantago maritima and the afore mentioned
therofyts decrease.
At the end of the research
period therofyts can still be found, mainly Atriplex prostrata and Spergularia spec.. Juncus gerardii is over it's peak abundance, while Agrostis stolonifera is still very common and the tall Scirpus maritimus and Phragmites australis are increasing.The mean maximal plant height slightly increases in time (Fig. 7A). The expected species richness is rather constant until about 1980 and then gradually declines (Fig. 8).
The Slope
The Slope can initially be characterized by a very high frequency of the grasses Agrostis stolonifera and Festuca rubra and by the occurence of the therofyts Centaurium littorale, Linuni catharticuni and Odontites verna ssp.
serotina (Fig. 4 and Fig. 6B). Also Ammophila arenaria and Sonchus arvensis (geo/hemicryptofyts) are common. Festuca rubra increases in frequency until it is present in every strip, whereas Agrostis stolonifera decreases. Small hemicryptofyts like Leontodon nudicaulis, Sagina nodosa (which can behave as a chainaefyt as well as a hemicryptofyt), Armeria maritima, Trifolium repi and Trifoliusn fragiferuni even as the monocots Juncus alpino—articulalatUS, Carex distans and Carex flacca increase after they have colonized the area.
Furthermore, flippophae rhamnoides and Potentilla anserina, both present from the beginning, increase.
From 1981 onwards all the afore mentioned therofyts and monocots disappear.
Arnmophila arenaria, Sonchus arvensis and Festuca rubra decrease. In contrast, Hippohae rhamnoides, Potentilla anserina, Cirsiuni arvense, Chamaenerion
angustifoliuni and the tall grasses Calamagrostis epigejos, Poa pratensis and Holcus lanatus occur more frequently.
Mean maximal plant height increases (Fig. 7B). Species richness increases until about 1981, when on average 12 species were counted per strip. In total, on an area of about 15 in2 40 species were found. But after 1981 this very high species richness decreases very rapidly and is about halved in 1989 (Fig. 8).
The Dune
Like on the Slope in the early succession stages Agrostis stolonifera,
Festuc
rubra, Amniophila arenaria and Sonchus arvensis are the abundant species, although the last two occur in higher frequency compared to the slope. But in contrast, therofyts are almost absent in this stage (Fig. 5 and Fig. 6C).
Sonchus arvensis and Agrostis stolonifera decrease from the beginning. The
sumirer
annuals
Lir.um catharticum, Euphrasia stricta, and the winter annualsCerastium semidecadruin and Aira praecox enter the succession area and
increase in frequency. The small charnaefyts Cerastium fontanum, Sedum acre and Sagina nodosa reach their maximal frequency around 1980. From 1981 onwards, all therofyts and small chamaefyts disappear and Festuca rubra and Ammophila arenaria decrease, while Flippophae rhamnoides, Chamaenerion angustifolium and the Large grasses Poa pratensis, Calamagrostis epigejos, Elymus pycnanthus and Holcus lanatus increase. Mean maximal plant height is rather high in 1972, reaches a minimum in 1980 and then increases (Fig. 7C). The trend in species richness is the same as on the Slope, although the number of species on the Dune is always lower (Fig. 8).Rate of succession
The rate of succession measured by changes in percentage dissimilarity (PD), is iore or less constant in time in each area (fig. 9A). In the Plain, the rate of succession seems somewh lower than on the Slope and the Dune, In course of time the vegetation co position of Plain, Slope and Dune differed more and more from 1973 (Fig. 9B). The vegetation of the Plain corresponded more and more to the latest year 1989, whereas Slope and Dune showed an
increasing correspondence with 1989 from only 1981 onwards (Fig. 9C).
Fluctuations
Although the highest rainfall—deficit over the months april, may, june and the highest flooding occurred in the same year (Fig. 10), there was no significant correlation between rainfall deficit and height of highest flooding (r —
0.19,
N 18, p >0.1).
Also no trends in time of these two variables could be found (rainfall—deficit: r 0.12, N 18, p > 0.1; height of highest flooding: r003, N =
18, p > 0.1). All bi-J1est floodings of the periods september—april occurred between november ai march, with the exception of 1973 (3 april) and 1979 (12 september 1978). In almost all winters the area was flooded by seawater from both the Waddensea and the North Sea.The standardized residuals of the trends in 40 % of the therofytic species were significantly explained by rainfall—deficit in 'spring' and/or height of the highest flooding, whereas 13% of the perennial species—trends showed a significant effect (table 2).
The standardized residual of the trend had a positive correlation with the height of the highest flooding for five halofytic surnnier—annuals, which species mainly occurred in the lower subareas. It had a negative correlation with the rainfall—deficit over the period april—june for five non—halofytic therofyts and for the perennials Plantago coronopus and Cerastium fontanum, all mainly occurring on Slope and Dune.
In the Plain the difference between observed species richness/strip and expected species richness/strip increased with height of if g -st flooding but not significantly (r 0.57, N 11, P 0.066).
Correlations between species
The increase in height (h, in cm,) with increasing age (x, in years) of Hippophae rhamnoides could be described by the equation h(x) 278*x/(7.83+x)
(r
0.90, p<.OOl, N —25).
The decrease of several species was very closely correlated with the increase of Hippophae rhamnoides. The relative frequencies of these species were at maximum 1 minus the relative frequencies of Hippophae rhamnoides of a given minimal age (Fig. 11); whereby this mininal age of Hippophae, thus it's minimal height, had to be taken higher to reach thismaximum, whenever the maximal heights of these species were higher.
Reconstruction of changes in biomass, light and soil The plots
The percentage dissimilarity of the vegetations
of the plots with the
characterized vegetations of the successional stages these plots represented was always lower than with the other characterized vegetations, with the exception of the somewhat higher PD of the Slope '72 plots (table 3). This implicated that the plots were a good representation of the chronosequence.Biomass and light
In all three subareas, the living above—ground standing crop significantly increased, with Dune finally reaching the highest standing crop and Plain the lowest (Fig. 12).
Plain'72 is dominated by Glaux maritija, Agrostis stolonifera and Scirpus maritimus (Fig. 13A). Plain'80 and '88 are dominated by Juncus gerardii. In Plain'88 Glaux maritima is absent and Agrostis stolonifera takes only a small part in the total standing crop, although Agrostis stolonifera occurred in almost all strips of the Plain'88 plots. The proportion of Scirpus maritimus
increases from Plain'80 to Plain'88.
Slope'72 is dominated by Agrostis stolonifera, Festuca rubra and Centaurium littorale (Fig. l3B). In Slope'EO Agrostis stolonifera has a much smaller portion than in Slope'72, the portion of Centaurium littorale is about the same, the portion of Festuca ru1 has doubled. The standing crop of Slope'88 consisted for 70% of Hippcphae rhaninoides, Centaurium littorale has disappeared.
Dune'72 is dominated by Ammophila arenaria; also Sonchus arvensis and Festuca rubra are important contributors to the total standing crop (Fig.
13C). The portion of Festuca rubra has doubled in Dune'BO, Ammophila arenaria and Sonchus arvensis contribute less. In '88 Hippophae rhamnoides shares again about 70% of the standing crop. Chamaenerion angustifolium takes about half of the 30% left. In all the plots the therofyts only take a very small part (18.5% at maximum in Slope'72, most of a biannual species) in the total standing crop. In all plots only a few species (5 species at maximum in Slope'80) account for more than 90% of the standing crop.
The amount of above—ground dead organic material is low in Plain'72 and Slope'72 in contrast with Dune'72, where a lot of dead standing leaves of Ammophila arenaria were found (Fig. 14).
Above—ground dead organic material is less than 10% of the total above—
ground organic matter in Plain'72, Slope'88 and Dune'88; but even more than 50% in Slope'80, Dune'72 and Dune'80. In Plain and Dune root biomass increases significantly in time, on the Slope root biomass is constant (Fig. 15).
Highest root biomass is achieved by Plain'80 and Plain'88; in these plots rhizomes are very abundant and account for about 25% of the root biomass.
Total biomass increases significantly with time (Fig. 16). In '88 the total biomass of all three subareas is the same, but with a very different distribution over roots and shoots. In '88 the above—ground standing crop of forbs and grasses is also the same in the three subareas.
The '72
plots
were only scarcely vegetated, with a lot of bare sand; between 40 and 90% of the PAR reached the soil—surface. In the '88 plots this is at maximum 10% (Fig. 17). The vegetation is much higher in the '88 plots than in14
the '72 plots. The vegetation—height of Plain'72 and Slope'72, about 10 cm., is in sharp contrast with the height of Dune'72, being about 50 cm. Standing crops of more than 300 g/rn2
resulted
in low light penetrations at the soil—surface (Fig. 18).
Allocation iatterns
In the Plain, the proportion roots decreased from 95% in '72 to 85% in '88 (FLg. 19). From '72 to '80 there is an increase in cover per unit biomass (GB), while hight per unit bioinass (HB) is constant. From '80 to '88 it is the other way round: GB is constant and FIB increases. On the Slope, the proportion root declines from 95% to 50%. CR is maximal in '80 and FIB increases. From '72 to '80 on the Dune, FIB falls down very rapidly, GB also declines, but proportion root increases from 65 to 85%. From '80 to '88 the allocation
pattern on the Dune is the same as on the Slope.
SoL1 measurements
The specific weight of the organic layer was strongly decreasing with time in all three subareas (Fig. 20). In the mineral layer there was only a slightly decrease of the specific weight on the Dune. Both in the organic and in the
mineral layer the moisture content of the Dune was much lower than the
moisture content of Plain and Slope (Fig. 21). In almost all cases the organic layer had a higher moisture content than the mineral layer. No trends in time were observed, except a slight increase on the Dune.The NaCl content is much higher in the Plain than on Slope and Dune (Fig.
22) . Whereas on the Slope some NaC1 can be found, the Dune was already completely. desalinated during the early stages of succession. In the Plain and on the Slope the organic layer had a higher NaC1 content than the mineral layer. No major trends in time were observed.
Although in all three subareas a decrease in the CaCO3 content of the mineral layer was found, this decrease was only significant on the Dune (Fig.
23) . The GaCO3 content of the organic layer showed a strong decline from the '80 plots to the '88 plots. The CaCO3 content of the organic layers of the '88 plots was less than 0.3%.
The pH(H20) and pH(KC1) of the mineral layers were between 7.8 and 8.2 (Fig.
24 and Fig. 25). On the Slope, a significant increase in pH(H20) was found, while on the Dune the pI-1(KG1) significantly decreased. The organic layer of both Plain and Dune showed a significant decrease of pH(H20) and pH(KC1) in time. The organic layer of the Slope has no significant change in pH(H20) or pH(KC1). In all three subareas there is a significant increase of the thickness of the organic layer in time (Fig. 26). In the Dune'72 plots and in two of the three Plain'72 plots no well defined organic layer could be found.
Slope has a much thicker organic layer than Plain and Dune both in '72 and in '88. In all three subareas there is a significant increase of the organic matter content of the organic layer in time (Fig. 27). The content is lowest on the Dune for every year. In the mineral layer almost no organic matter could be found, although there is a significant increase in time on the Dune.
In a lot of plots the total organic matter content of the upper 20 cm. of the soil consisted almost completely of roots (Fig. 28). One Plain'88 plot had a thicker organic layer with a higher organic matter content than the other two Plain'88 plots, whereas root biornass was about the same. In the Slope'88 plots there is also considerable accumulation of organic matter other than roots.
The C—content (C), N—content (N) and organic matter content of the samples
were all closely correlated, according to C — —0.048 +
l0.8*N
(fig. 29, r 0.98, N 54), C — —0.046 + 0.332*OM (Fig. 30A, r —0.98,
N — 54) and N 0.0026 + 0.0295*OM (Fig. 30B, r —0.96,
N —54).
No effects of time, place and/or layer on these correlations were found and the intercepts were not significantly different from zero. Stated otherwise: in all plots the C—content was 33.2% of the organic matter content, the N—content was 2.95% of the organic matter content and the C/N—ratio was 10.8. Thus C— and N—content show the same pattern as the organic matter content: in all three subareas the C— and N—content of the organic layer increase significantly in time, with lowest contents on the Dune; and the C— and N—content in the mineral layer is low, although the increase in time is significant (Fig. 31 and Fig. 32).
P—content (P) was not that closely correlated with organic matter content as C— and N—content; also the intercept is not that low compared to the slope that it could be neglected: P 0.0108 + 0.0021*OM (Fig. 30C, r —
0.86,
N54). The P—content of the mineral layer is not that low when compared to the organic layer (Fig. 33). Plain had a significantly higher content than Slope and Dune. The P—content significantly increased with time in the organic layer of Plain and Slope from '72 to '80, but not from '80 to '88. In Dune no significant change was found, although the direction of change is comparable.
The P content of Dune'80 and '88 is significantly lower than the P content of Plain'80, Plain'88 and Slope'80. The C/P—ratio in the organic layer significantly increased with time in all subareas, reaching a ratio of about 100 in '88 (Fig. 34). The total amount of organic matter, N and P in the organic layer increased significantly in time, with Dune having the lowest accuiu1ation (Fig. 35 and Fig. 36). The average rates of N accumulation in the soil were 3 g N m2 year' for Plain and Slope and 1 g N nf2 year' for Dune.
16
DISCUSSION
The vegetation composition at the plots, on which the reconstruction is based, were very similar to the vegetation at the permanent transects area, both according to their percentage dissimilarity (table 3) and to their structure as could be seen from photographs of the permanent transects area. The PD of Slope'72 with it's own characteristic is somewhat too high, this is a consequence of the absence of Linum catharticum and a high frequency of Glaux niarit.ma. Maybe these plots were laid down at a place slightly wetter or more saline than the Slope of the permanent transects. Apart from these two species the vegetation of the Slope'72 plots closely resembled to the successional stage it represented. The succession series at the permanent transects is probably a good representation for what happened in most aprts of the Beach Plain. Rozema (1978), Van der Laan (1980), Wapenaar (1980) and Schat (1982) describe vegetation zonations from different years and from different places of the Beach Plain. These descriptions are in agreement with the observed patterns at the permanent transects area. Rozema (1978) determined the biomass
per species along a transect from plain to
dune. These biomasses were somewhere intermediate between the biomasses found at the '72 plots and those at the '80 plots. At present all the dunes from the most Eastern point to benchmark 9 are densely covered with Hippophae. All the plain areas from benchmark 7.8 to 9 are covered with vegetations of Juncus gerardii, Scirpus maritimus, Phragmites australis, etc.The succession is still going on. Phragmites australis, the tallest species of the Plain, is increasing. In older dune slacks on the island which are also flooded by seawater, Phragmites australis is the dominant species. On some dunes in the Beach Plain young Sambucus nigra, shrubs taller at maturity than 1-lippophae rhamnoides, are growing up. On older dunes near the Beach Plain, Sambucus nigra has replaced Hippophae rhamnoides. The tree Acer pseudoplatanus is a possible candidate for following up Sambucus nigra.
Species growing on the Plain have to cope with high salinity and with flooding in the winterperiod. Waterlogging of the soil will probably also in summer impose anaerobic root conditions for long periods. A lot of species in the Plain are known as halofyts. Most of the species occurring in this area have no above—ground biomass in winter. The therofyts are almost all summer annuals. The shoots of Glaux maritima, Juncus gerardii, Scirpus maritimus and Phragmites australis die off in autunin. Glaux maritima has a special kind of hibernating buds (Rozenia, 1978); the geofyts can survive the winterperiod in underground rhizomes. Also the leaves of Potentilla anserina die off in autumn. In spring and summer Agrostis stolonifera and Potentilla anserina can rapidly spread by their stolons. Fluctuations of annuals were positively correlated with the height of flooding, probably because high floodings create large gaps in the vegetation, where annuals can establish.
The Slope is much less saline than the Plain, but still has a rather high moisture content. The species are mainly glycofyts. Fluctuations of therofyts
and some perennials were negatively correlated with rainfall—deficit in spring. A high rainfall— deficit will cause a low soil moisture content.
Lowering of the soil moisture content will increase the salinity (Van Tooren, Schat and Ter Borg, 1982).
Both a low soil moisture content and a high
salinity will have a negative effect on the germination and establishment of the glycofytic therofyts, Odontites verna ssp. serotina, Centaurium pulchellum, Centaurium littorale and Linum catharticuni all germinate in spring; Plantago coronopus mainly germinates in spring, but can also germinatein the autumn (Schat, 1982).
Salinity and soil moisture content are both low on the Dune. Winterannuals like Cerastium semidecandruin, Arenaria serpyllifolia and Aira praecox flower
in early spring, die after seed—setting in may/june and survive during the dry summer as seeds (Rozijn, 1984). Odontites verna sst. serotina, a summer annual, appears on the Dune only after wet springs. Sedum acre can reduce evaporation, by closing the stomata at daytime (CAM—metabolism). Hippophae rhamnoides and Ammophila arenaria can root rather deep. Arrunophila arenaria and Elyrnus pycnanthus fold their leaves when it gets dry, burrying their stomata deeply between the ribs. Thus most species of the Dune are adapted to long periods of drought.
It can be concluded that differences in salinity, water availability and flooding are the physical constraints determining the differences in species composition of Plain, Slope and Dune. Year to year fluctuations of the same
factors seem to be responsible for fluctuations in the occurrence of a lot of short—living species.
However, non of these factors can be held responsible for the succession patterns observed. NaCl—content did not change with time, except a slight
decrease on the Slope, which can also be an effect of the site where the Slope'72 plots were laid. Van Tooren, Schat and Ter Borg (1982), measured the conductivity of the soil moisture of the Plain in august from 1973 to 1980 and did not observe any desalination. Moisture content only increased on the Dune, accompanied by an increased organic matter content (larger water holding capacity) and an increased above—ground standing crop (moist microclimate).
So this might be rather a consequence than a cause of the observed succession.
No trends in the height of flooding or in the rainfall—deficit in spring did occur. Also no changes in the access of flooding seawater to the area were observed.
Loss—rates imposed by herbivores (rabbits, hares and geese) are thought to be not high enough to be of great influence on the observed successional patterns at the Beach Plain. As discussed above, loss—rates in the Plain are higher than on Slope and Dune, because of loss of above—ground biomass in autumn and of losses due to high floodings.
The rate at which P is made available in soils with high pH and high CaCO3—
content is low, But this low rate does not necessarily mean that in principle not all P can be made available. The total amount of P in the upper soil is enough to sustain a much higher biomass than observed in '88. Assuming that plants have a P—content of 0.2%, P—content in the mineral soil is 0.1 g/kg and
specific weight of the soil is 1.4 kg/dm3, then the total amount of P in a 10 cm. thick layer can sustain a plant biomass of 7 kg/rn2. Plants will use nutrients from layers thicker than 10 cm. The K—content of the mineral layers on the Beach Plain ranges from 0.2 to 1 g/kg dry soil, with a considerable water—soluble part (Rozema, 1978). This K—content is also enough to sustain a much higher biomass than observed, Assuming that plants have a K—content of 1% and K—content in the mineral soil is 0.2 g/kg, then the total amount of K in a 10 cm. thick layer can sustain a plant biornass of 2.8 kg/rn2. The N—
content is strictly correlated with the organic matter content, suggesting that all N is organic N. A C/N—ratio of about 11, as on the Beach Plain, was
also found at the first stages of succession at both the dunes of Lake
Michigan (Olson, 1957) and the recessing glaciers in Alaska (Crocker and Dickson, 1957). This C/N—ratio is in the same order as the C/N—ratio of most living organisms. Almost all organic matter on the Beach Plain consisted of roots, although the fraction dead roots is not known. The total amount of N at the '72 plots seems too less to sustain the biomasses observed in the '88 plots. Assuming that plants have a N—content of 1% (which is three times lower than the regression—coefficient of N—content with organic matter content) and N—content in the mineral soil is 0.06 g/kg, then the total amount of N in a 10 cm. thick mineral layer can sustain a plant biomass of only 0.8 kg/rn2. The amount of N in the organic layer increases rapidly, but is in '72 still very18
low. This leads to the conclusion that the total amount of N limits the total bionass at the Beach Plain. This conclusion is in agreement with the findings of Villis (1963) at the dunes of Braunton Burrows. The soils at Braunton Burrows are comparable with the Beach Plain: pH is between 8 and 9, P—content is about 0.1 g/kg and N—content ranges from 0.11 to 2.8 g/kg. Addition of only
P, only K or P+K didn't lead to an increased plant biomass, but after N—
addition biomass increased (Willis, 1963). In his thesis Rozema (1978) mentions an unpublished N—addition experiment at the Beach Plain, where bioniass increased 10—20 fold compared to control plots.
On Slope and Dune the increasing amount of N is accompanied by an increase in frequency and height of Hippophae rhamnoides. The decrease of a lot of species directly follows the increase of 1-Iippophae, with taller species decreasing when Hippophae gets taller. This suggests that Hippophae takes away the light, thereby outcompeting these species. Indeed a lot of light can be taken away in a Hippophae stand; at Dune'88 only 10% of the incident PAR reaches a height of 50 cm. above the soil—surface. Some other species on Slope and Dune seem to disappear for other reasons; they decrease before they can be shaded by Hivophae.
On the Plain an increased amount of N is also accompanied by an increasing height of the vegetation, an increased above—ground standing crop and a decreased light penetration to the soil. The decrease and/or disappearance of small species, for instance Glaux rnaritima, Spergularia spec, and Salicornia spec., might also be caused, at least partly, by shading by taller species.
On Slope and Dune species richness reaches an optimum around 1980. Species richness on the Plain is lower and fluctuates more. Species richness will partly depend on the number of species which can tolerate extreme values of physical factors as salinity, flooding and drought. The number of species from the Dutch flora which can survive moist and fresh conditions, as on the Slope, is niuch larger compared to the number of species which survive saline, anaerobic and often flooded conditions, as on the Plain. Species richness will also partly depend on competitive interactions and the heterogeneity of resources and constraints in space and time. This heterogeneity can be expected to be high on the Slope, since it is the small transition zone between a wet, saline subarea and a dry, fresh subarea, yielding possibly a
checkerboard of conditions of the two subareas, Furthermore, the heterogeneity a species experiences will depend on the ratio between the scale of environmental heterogeneity and the size of the species. The decrease of
species richness on Slope and Dune might be a consequence of the increase of taller species. Disturbances can cause a loss of competitively gained dominance. In gaps caused by floodings a lot of species can enter, thereby
increasing species richness.
The competitive abilities of plant species will be greatly influenced by their allocation patterns. The height which a plant, given it's biomass, can reach,
will depend on
it's allocation to stern. Suppose an increase in height(h) is a simple function of stem biomass(B):h
.=c*Bs'
1=>
hB5As*B
Jwith
B total biomass, A — allocation to stem, c and x are constants. Whenstem—biomass
is not only used for an increase in height, but also for increase in diameter, then x < 1. An increasing diameter is necessary for bearing a higher weight. What does happen with HB; height divided by total biornass:FIB h/B = c*AsX*Bl
When the allocation pattern is fixed (As is constant) and x < 1, 1-lB will
decrease with increasing biomass. Thus, when x < 1, HB can only increase with increasing biomass, when A8 increases. Therefore HB is a good measure for the
"proportion stem".
On Plain and Slope not only mean plant height increases, but also HB,
suggesting an increased allocation to stem. In contrast, there is a decrease in the mean plant height and HR until about 1980 at the Dune. In 1972 small species were almost absent at the Dune, which was dominated by Ainmophila arenaria and Sonchus arvensis, both being able to reach a height of more than 1 meter. These species are characteristic for places with much sand—blowing.Sand—blowing explains the absence of small species, they will drown in the sand. The biomass of roots deeper than 20 cm. were not estimated. On Plain and Slope no roots going deeper than 20 cm. were observed. On the Dune some roots of Hipophae rhamnoides and Anmiophila arenaria went deeper. Rut because there were not much roots going deeper, the vast portion of root biomass is thought to be sampled. Proportion root decreases at Plain and Slope and from '80 to '88 also on the Dune.
Although on the early Dune stages sand—blowing will have been an important constraint, leading to somewhat other successional patterns than predicted from a hypothesis about resource interactions, all other observations, also at the later Dune stages, are in agreement with the predictions of the resource—ratio—hypothesis (Tilman, 1985): An increase of the total amount of N is accompanied by an increased total plant biomass, a decreased light penetration to the soil—surface, a decreased proportion root, an increased proportion stem, an increased plant height at maturity and a decreasing abundance of short—living species.
This does not imply that differences in colonization abilities or RGRmax did not play a role in this successional dynamics. These traits will for example be of importance for species entering the gaps created by high floodings. Rut in general soil—driven dynamics (sensu Tilman, 1988) seem to be of overruling importance for this succession series.
On the Reach Plain the total amount of N will have increased by an atmosferic input, nitrogen—fixing organisms and the input of N by flooding seawater. On the East Frisian islands, Germany, the atmosferic input from dry and wet deposition is estimated at about 1.5 g N m2 yr' (Gerlach, 1989). On the Dutch main land this input is estimated at 4 g N m2 yr'. Hippophae shrubs can fix between 1.5 g N m2 yr' (Akkermans, 1971) and 17.9 g N rn2 yr' (Becking, 1970). Stewart (1965) estimates the nitrogen fixation by free—living bacteria in dune—slacks at 6 g N m2 yr'. In the period 1973—1989 North Sea water, 10 km. West from the Reach Plain, had a total N—content of 1.33 0.64 mg/l (N 95; Rijkswaterstaat, personal communication); when 20 cm. of flooded North Sea water damps in, it looses about 0.26 g N m2 to the soil. Leaching will depend on the nitrogen uptake by the vegetation. Other N—losses might be suffered from denitrification, mainly on the wet Plain and Slope, and probably from allocation of biomass by floodings. The relative importance of these losses has not yet been estimated for sites comparable with the Beach Plain.
We estimated an increase of 45 g N m2 in 16 years for the organic layer of Plain and Slope. This is comparable with the rate of N—accunulation in the soil at the first 20 years of succession after recession of glaciers in Alaska, which was about 4 g N m2 year' (Crocker and Major, 1955; Crocker and Dickson, 1957). On the Dune, about 15 g N m2 accumulated in the soil in 16 years, this is comparable with the accumulation at the early successional stages of the dunes of both Lake Michigan (Olson, 1957) and Spiekeroog (Gerlach, 1989). The total increase of N on the Reach Plain and at the other
20
successions must have been higher, because the increasing above—ground biomass will also have increased above—ground N. At the glaciers and the dunes of Lake Michigan the rate of N—accumulation decreased in time.
The increase of the P—content of the organic layer might be due to inputs from outside, but also from reallocation of P in the soil by the vegetation.
By 'influencing fixation,
denitrification and leaching and by uptake of
nitrogen, the vegetation is an important determinant of the rate of nitrogen accumulation. It is not likely that without a vegetation, the total amounts of nitrogen at the Beach Plain and in other primary successions had ever become as much as they are now, because without storage, mainly in plants, nitrogen easily leaches out from these sandy soils. When discussing the importance of nitrogen—fixation at Glacier Bay, Lawrence et al.(l967, p.812) argue: "it
is only the latter [the nitrogen—fixing Alnus] that raises the nitrogen supply to a level enabling the spruce forest to achieve dominance".Thus, although a lot of primary successions might be mainly "soil—driven" with the outcome of competition determined by the interactions between organisms and resources, soil—development in terms of N—accumulation might be mainly
"plant—driven", which makes the usefulness of the first term doubtfull.
Examples of theoretical and emperical results about positive feedbacks between the growth of autotrophs and the total amount of limiting nutrients in open systems (i.e. systems with inputs and outputs of nutrients) are reviewed by DeAngelis et al (1989), Such positive feedbacks come close to the facilitation hypothesis of Clements (1916), although it is facilitation in an indirect way, distrusted by so many plant ecologists (Drury and Nisbet, 1973; Colinvaux, 1973; Connell and Slatyer, 1977; Peet and Christensen, 1980; Miles, 1979). On the Beach Plain nitrogen accumulation, probably facilitated by plants, seems to be the main cause of the observed successional patterns.
REFERENCES
Akkermans ADL (1971) Nitrogen fixation and nodulation of Alnus and Hippophae under natural conditions, PhD Thesis, Leiden.
Barboür MG (1970) Is any angiosperm an obligate halophyte ? Am. Midl. Nat.
84: 105—120.
Becking JH (1970) Plant—endophyte symbiosis in non—leguniinous plants. Plant
& Soil 32: 611—654.
Berendse F, Elberse WTh
(1989)
Competition and nutrient losses from the plant. In: Lambers H, Cambridge ML,Konings H and Pons TL
(eds) Variation in Growth Rate and Productivity, Academic Publishers, The Hague SPB.Chapin FS (1980) The mineral nutrition of wild plants. Annual Review of Ecology and Systematics 11: 233—260.
Clements FE (1916) Plant succession. Carnegie Institute Washington Publica- tion 242.
Colinvaux PA (1973) Introduction to Ecology. Wiley, New York.
Connell JH, Slatyer RO (1977) Mechanisms of succession in natural communities
and their role
in community stability and organization. American Naturalist ill: 1119—1144.Crocker RL, Major J (1955) Soil development in relation to vegetation and surface age at Glacier Bay. Journal of Ecology 43: 427—448.
Crocker RL, Dickson BA (1957) Soil development on the recessional moraines of the Herbert and Mendenhall glaciers of southeastern Alaska. Journal of Ecology 45: 169—185.
DeAngelis DL, Mulholland PJ, Palumbo AV, Steinxnan AD, Huston MA, Elwood JW (1989) Nutrient dynamics and food—web stability. Annual Review of Ecololgy and Systematics 20: 71—95.
Drury WH, Nisbet ICT (1973) Succession. J. Arnold Arboretum 54: 331—368.
Feekes W (1936) De ontwikkeling van de natuurlijke vegetatie in de Wieringer—
meerpolder, de eerste grote droogmakerij van de Zuiderzee. PhD Thesis, Wageningen.
Cerlach A, Albers E, Broedlin W (1988) Nitrogen content and turnover of coastal dunes on Spiekeroog (EastFrisian Islands, German Bight). Bull.
Ecol. Soc. Amer. 69: 143.
Cerlach A, Albers E, Broedlin W (1989) Nitrogen content and turnover in coastal dune succession on Spiekeroog, East Frisian Islands. EUDC—
Congress Seville, March 1989, Congress book.
Cleeson SK, Tilman D (1989) Allocation and the transient dynamics of succession on poor soils. Ecology 71.
Gririie, JP (1979) Plant strategies and vegetation processes. John Wiley &Sons, Chichester.
Jenny, H (1980) Soil genesis with ecological perspectives. Ecological studies, vol.37. Springer—Verlag, New York.
Joenje, W (1978) Plant colonization and succession on embanked sand flats.
PhD Thesis Groningen.
Lawrence, DB Schoenike, RE Quispel, A Bond, C (1967) The role of Dryas drummondi in vegetation development following icerecession at Glacier Bay, Alaska, with special reference to itsnitrogen fixation by root nodules. Journal of Ecology 55: 793—813.
Miles, J (1979) Vegetation Dynamics. London Chapman and Hall.
01ff, H Van Andel, J Bakker, JP (1990) Biomass and shoot/root allocation of five species from a grasslandsuccession series at different combina- tions of light and nutrient supply. Functional Ecology 4: 193—200.
22
Olson, JS (1957) Rates of succession and soil changes on southern Lake Michigan sanddunes. Botanical Gazette 119: 125—169.
Peet, RK Christensen, NL (1980) Succession: a population process. Vegetatio 43: 131—140.
Robertson, GP (1982) Factors regulating nitrification in primary and secondary succession. Ecology 63: 1561—1573.
Robertson, GP Vitousek, PM (1981) Nitrification potentials in primary and secondary succession. Ecology 62: 376—386.
Rozema, J (1978) On the ecology of some halophytes from a beach plain in the Netherlands. PhD Thesis, Free University of Amsterdam,
Rozijn, NAMG (1984) Adaptive strategies of some dune annuals. PhD Thesis, Free University of Amsterdam.
Schat, H (1982) On the ecology of some Dutch dune slack plants. PhD Thesis, Free University of Amsterdam.
Stewart, WDP (1965) Nitrogen turnover in marine and brackish habitats. Ann.
Bot: 229—239.
Tilman, D (1985) The resource—ratio hypothesis of plant succession. American Naturalist 125: 827—852.
Tilman, D (1988) Dynamics and Structure of Plant Communities. Princeton University Press, Princeton.
Van Tooren, BF Schat, H Ter Borg, SJ (1983) Succession and fluctuation in the vegetation of a Dutch Beach Plain. Vegetatio 53: 139—151.
Van der Laan, F (1980) Synthese van het vegetatiekundig onderzoek in de strandvlakte op Schiermonnikoog van 1952 t/m 1979. MSc Thesis, Free University of Amsterdam.
Wapenaar, P (1980) Vegetatiebeschrijving en —kartering van de Strandvlakte van Schiermonnikoog in 1979. MSc Thesis, Free University of Amsterdam.
Willis, AJ Folkes, BF Hope—Simpson, JF Yemm, EW (1959) Braunton Burrows: The dune system and its vegetation. Journal of Ecology 47: 1—24.
Willis, AJ (1963) Braunton Burrows: The effects on the vegetation of the addition ofmineral nutrients to the dune soils, Journal of Ecology 51:
353—374.
23
TABLES & FIGURES
Fig.l: (A) Schiermonnikoog, an island in the Wadden Sea. The study area, the Beach Plain, is shown within the rectangle. Arrow indicates permanent tran- sects area. (B) The Beach Plain, with the locations of the '72, '80 and '88 plots. The box, close to the '88 plots, indicates the permanent transects area.
Fig.2: The permanent transects area, (A) a three—dimensional view and (B) mean height of each strip above the lowest point and the division in five
subareas.
Fig.3: Trends in the changes in frequency (based on 132 strips per year) of the most frequent species of the Plain.
(A)
AS = Agrostis
stolonifera, CP Centaurium puichellum,GM =
Glaux maritima, JG = Juncus gerardii, OV Odontites verna ssp. serotina, PA =Potentilla
anserina, PH =Phragmites
australis, SM =Scirpus
maritimus.(B) AP =
Atriplex
prostrata, CL = Centaurium littorale, JB = Juncus bufonius, JM = Juncus maritimus, PM Plantago maritima, SC =Salicornia
spec. , SS =Spergularia
spec..(C) CD = Carex distans, CR = Chenopodium rubrum, ES =
Eleocharis
spec, LV = Limonium vulgare, PS Puccinellia spec., SU = Suaeda maritima.Fig.4: Trends in the changes in frequency (based on 38 strips per year) of the most frequent species of the Slope.
(A) AA Ammophila arenaria, AS =
Agrostis
stolonifera, CL =Centaurium
littorale, FR =
Festuca
rubra, HR Hippophae rhamnoides, JA = Juncus alpino—articulatus, PA Potentilla anserina, PP = Poa pratensis.
(B) CA =
Cirsium
arvense, CE =Calamarostis
epigeios, CF Cerastium fon—tanum, LC = Linum catharticum, LN = Leontodon nudicaulis, OV =
Odontites
verna ssp. serotina, SA = Sonchus arvensis, SN Sagina nodosa.
(C) AM =
Armeria
maritima, CD Carex distans, CH = Chamaenerion angusti—folium, CX Carex flacca, GM = Glaux maritima, HL =
Holcus
lanatus, TF =Trifolium
fragiferum, TR Trifolium repens.Fig.5: Trends in the changes in frequency (based on 32 strips per year) of the most frequent species of the Dune.
(A) AA = Amrnophila arenaria, AS Agrostis stolonifera, CA = Chamaenerion angustifolium, FR =
Festuca
rubra, HR Hippophae rhamnoides,PP =
pratensis,
SA = Sonchus arvensis, SM = Sedum acre.(B) AP =
Aira
praecox, CE =Calamagrostis
epigejos, CF =Cerastium
fontanum,CS =
Cerastium
semidecandrum, EP = Elymus pycnanthus, LN Leontodon nudi—caulis, SN = Sagina nodosa, SJ = Senecio iacobaea.
(C) CI =
Cirsium
arvense, ES =Euphrasia
stricta, HL = Holcus lanatus, LC = Linum catharticum, PC =Plantago
coronopus.Fig.6: Trends in changes in the life—form spectra. Each species was weighted for its frequency as estimated by the trends. (A) Plain, (B) Slope, (C) Dune.
Fig.7:
