Plant Ecology 138: 203–216, 1998.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.
203
The role of water depth and soil temperature in determining initial
composition of prairie wetland coenoclines
Eric W. Seabloom∗ , Arnold G. van der Valk & Kirk A. Moloney
Department of Botany, 353 Bessey Hall, Iowa State University, Ames, IA 50011–1020, USA; ∗ Present address:
National Center for Ecological Analysis and Synthesis, 735 State St., Suite 300, Santa Barbara, CA 93101–3351,
USA
Received 4 September 1997; accepted in revised form 26 May 1998
Key words: Environmental gradient, Germination, Regeneration niche, Seed bank, Zonation
Abstract
In this study, we examined the effects of water depth and temperature on seedling recruitment from a prairie wetland
seed bank. We collected seed-bank samples from natural and restored prairie pothole wetlands in northwestern
Iowa and combined them into a single sample. We examined seedling recruitment from this seed-bank sample in
an experimental study using a factorial design of 4 temperature treatments (5◦ night and 15◦ day to 20◦ night and
30◦ day) and 3 water-depth treatments (0, 2, and 7 cm).
Principal Components Analysis showed that both water depth and temperature had significant effects on the
composition of the seedling community as measured by changes in relative stem density and biomass. Water depth
had its strongest effects on stem density while temperature had its strongest effects on biomass.
For the 22 most common species, stem density varied with water depth for 95% of the species and with
temperature for 50% of the species. Most species with water depth responses had lower stem counts as water depth
increased, and for the majority of species with temperature responses stem density increased with temperature.
Total, annual, and perennial species richness was negatively correlated with water depth. Total and annual
species richness was positively correlated to temperature, while perennial species richness was unresponsive to
temperature. In addition, species found at low elevations as adults emerged at higher rates in the deep water
treatments while species that occurred at higher elevations as adults had their highest emergence rates in the low
water treatments.
Our results suggest that differences in environmental conditions along coenoclines can affect the initial distribution of species emerging from the soil seed bank. Water depth sorted seedlings according to their adult water-depth
tolerances, and temperature determined the proportion of annuals in the seedling community.
Introduction
Coenoclines, i.e., vegetation zones perpendicular to
environmental gradients, are formed when environmental conditions differentially affect rates of seed
production, propagule dispersal, seed germination,
seedling mortality, and/or adult mortality among
species (Grubb 1977; van der Valk & Welling 1988).
Environmental effects on dispersal, germination, and
seedling mortality define a species’ regeneration niche
(i.e., Grubb 1977), while adult mortality defines its
adult or habitat niche (i.e., Grubb 1977). The im-
portance of preemption in determining community
composition suggests that the recruitment of seedlings
may be a critical stage in the development of a coenocline (Grace 1987; Robinson & Dickerson 1987;
van der Valk & Welling 1988; Drake 1991; Weiher & Keddy 1995a, b). If the regeneration niche is
a critical factor in the development of a coenocline,
then the distribution of adult plants will be a function
of both the current environmental conditions and the
conditions predominating during recruitment events
(Gleason 1926; Sutherland 1974; Welling et al. 1988).
For this reason, an understanding of species’ regener-
204
ation niches may be fundamental to understanding the
of distribution adult plants.
In this study we examined the importance of the
regeneration niche in determining the initial composition of coenoclines in prairie wetlands. We did
this by subjecting a series of seed-bank samples to a
factorial combination of water-depth and temperature
treatments in growth chambers. While water depth and
temperature have been shown to affect the composition of the seedling community, we are not aware
of any studies that have examined the interactions
between these two environmental factors.
Wetlands are excellent systems in which to study
how regeneration niches affect vegetation patterns for
three reasons. First, the distribution of adult wetland plants is dominated by a single environmental
gradient, water depth (Spence 1982). Changes in water depth are associated with changes in a variety of
environmental factors (e.g., light, soil nutrients and
particle size, and gas exchange rates) that physiologically constrain species’ distributions (Keddy 1982;
Spence 1982). Despite the importance of water depth,
attempts to model the distribution of wetland plants
based solely on adult water depth tolerances have
generally been unsuccessful (de Swart et al. 1994).
This suggests that additional factors (e.g., species’ regeneration niches) are important in determining the
structure of the coenoclines that form along the elevational gradient in wetlands (Wilson & Keddy 1985;
de Swart et al. 1994).
Second, seedling recruitment plays an important
role in maintaining wetland communities. In prairie
wetlands, high-water induced mortality periodically
creates large openings in the emergent vegetation
(Harris & Marshall 1963). However, few emergent
species are able to germinate under standing water
(Kadlec 1962; van der Valk & Davis 1978), and these
openings are mostly revegetated by seedling recruitment from the seed bank during subsequent periods of
low water (van der Valk & Davis 1979; Welling 1987;
Poiani & Johnson 1989; Grillas et al. 1991). These
recruitment events are dominated by annual species
that are totally reliant on the creation of these open
mudflats (Salisbury 1971).
Third, species must differ in their regeneration
niches for recruitment events to change community
composition (Hutchinson 1961; Grubb 1977). In wetland systems, the regeneration niches of emergent
species are differentiated by both water depth (e.g.,
Keddy & Ellis 1985) and temperature gradients (e.g.,
Galinato & van der Valk 1986).
Water depth has been shown to affect seedling
community composition in observational studies
(Kadlec 1962; Harris & Marshall 1963; Smith &
Kadlec 1985) and experiments (Keddy & Ellis 1985;
Keddy & Constabel 1986; Weiher & Keddy 1995a).
Our expectation at the outset of this experiment was
that species with little tolerance for flooding would
have germination rates that were negatively correlated with water depth, because they would not be
able to survive in flooded soil. Flood-tolerant species
would be expected to have germination rates that were
positively correlated with water depth, because this
germination response would place their seedlings in
areas with fewer potential adult competitors (Keddy &
Ellis 1985). Most plants can grow in the mesic prairie
surrounding the wetlands, but only a few species are
able to tolerate flooded conditions deep in the wetland
basins (Seabloom 1997). It is important for these stress
tolerant species to avoid competition, because they often have a lower competitive ability (Wilson & Keddy
1986)
Germination rates of wetland plants also vary
along soil temperature gradients (Thompson & Grime
1983; Galinato & van der Valk 1986; Hogenbirk &
Wein 1992), and this response may cause shifts in
the composition of seedling communities (Welling
1987). Because annual species can only reproduce
in disturbance-created openings characterized by high
soil temperatures (Salisbury 1971), we expected that
the germination of many annual species would be triggered by high soil temperatures typical of exposed
mudflats, because these mudflats represent safe sites
for seedling establishment (i.e., Harper 1977).
We conducted a growth chamber experiment to determine the effects of water depth and temperature
on the seedling community recruited from a prairie
wetland seed bank.
Methods
Seed-bank sample collection and preparation
We collected 43 seed-bank samples for this experiment in September of 1994 from 6 restored and 8 natural wetlands (mean number of samples per wetland =
3.1). These wetlands were located in the northwestern
Iowa counties of Clay, Dickinson, Emmet, and Palo
Alto in the United States. The wetlands ranged from
50 to 300 m in diameter, had closed drainages, and
had been free of human-caused hydrologic disturbance
205
for at least three years. Each of the collections was
composed of approximately 4 l of the top 5 cm of soil.
In order to make the comparison between species’
germination response to water depth and adult distribution along the water-depth gradients described
above, we needed to collect soil samples in a manner
that maximized the likelihood of including seeds from
species for which we had adult distribution information. We did this by collecting soil in each wetland
beneath stands of vegetation composed of species that
occurred in the adult distribution studies described
below. At the completion of the sampling, we had collected soil samples beneath stands of 34 species that
were in our adult field surveys.
We diluted these samples with water to make a
slurry and homogenized them into a single batch using
a concrete mixer. We removed undecomposed plant
material by hand picking and straining the soil slurry
through a coarse sieve (5 cm mesh).
We placed 500 ml of this seed-bank sample on top
of 1500 ml of sterile potting soil in each of 192 five
liter pots. Each pot had a layer of seed-bank sample
two cm thick and 333 cm2 in area. We randomly assigned the pots to the experimental treatments based
on the order they were taken from the concrete mixer
to control for the possible effects of seeds settling or
becoming more scarified in the bottom of the mixer.
After watering the soil to saturation, we covered
each pot with a plastic lid and stratified them in the
dark at between 0◦ and 5 ◦ C for 4 to 27 weeks (stratification period effects were accounted for by the blocking effect, Trial, described in Methods of Analysis).
The covered pots remained moist, and no germination
occurred during stratification. Pots were transferred
directly from the cooler to the growth chambers over
the course of the experiment.
Experimental design
The growth chamber experiment combined four replicate, 6-week trials, and used four 1.4 m2 growth
chambers (Conviron, Model E-15). The chambers
were lit by a combination of incandescent and fluorescent bulbs that produced a light intensity of 960
microeinsteins at 15 cm from the bulbs (light intensity
from model specifications) and were set on 16 hours
of light per day.
During each trial, pots were assigned to a factorial combination of four air temperature treatments (5◦
/15◦, 10◦ /20◦ , 15◦ /25◦ , and 20◦ /30◦ night/day) and
3 water-depth treatments (0, 2, and 7 cm). Tempera-
ture treatments were distributed among the chambers
and trials in a Latin square design, so that each of
the four temperatures was assigned once to each of
the four chambers. Each chamber contained 12 pots
placed in a randomized-block design with four pots
assigned to each of the three water-depth treatments.
These blocks accounted for environmental variability
within a growth chamber. The four pots in each block
were treated as subsamples, and their mean was used
in the data analysis. A growth chamber within a single
six-week trial was the experimental unit for the temperature treatment, and the four pots with the same
water depth within a chamber were the experimental
unit for the water-depth treatment.
We filled the pots in the growth chambers with water to one of three depths: 0, 2, or 7 cm. The pots in the
shallow treatment (0 cm) had drainage holes, but the
high levels of clay in the seed-bank samples and the
large amount of peat in the potting mix kept the seedbank samples near saturation. Over each 24-h period
between water additions, the water level dropped a
maximum of 1 cm in the high temperature treatments
and imperceptibly in the coldest treatments. Soil and
air temperature were recorded at 30-min intervals using a data logger. Soil temperatures were measured
using a thermocouple placed 1 cm into the soil in
one of the pots assigned to each of the three water
depths in each chamber. Maximum soil temperature
was typically five degrees higher than the air temperature, but the two measurements did not differ in their
minimums. Soil temperatures were similar among the
water-depth treatments.
Growth chamber data collection
At the end of every trial, we counted and removed all
identifiable seedlings from the pots. In the case of rhizomatous species, we counted genets. After making
the stem counts, we weighed the six most common
species in each trial (Table 1). These species were
readily identifiable at emergence, and we could collect
all of individuals present at the end of each trial. We
dried the biomass samples for at least 8 h at 70 ◦ C
and kept them sealed in a plastic box with silica gel
desiccant prior to weighing to the nearest 0.1 mg.
After making the biomass collections and initial
seedling counts, we moved pots with unidentified
plants to a greenhouse until they could be identified.
During this time, we removed any new seedlings,
so that only seedlings recruited in the growth chamber were included in the final analysis. Fifty species
206
Table 1. Adult traits and germination responses to water depth and temperature of the 22 species with significant overall regressions.
Species’ life span, adult elevational range, and biomass and stem density responses to water depth and temperature are summarized
by the linear component (slope) of the regression model for each term. Blank cells indicate regression coefficients that were not
significantly different from zero (p<0.05). Species included in the biomass analyses are indicated. Minimum, mean, and maximum of
adult elevational range are shown relative to maximum flooding level (0 cm). Twenty-eight additional speciesa occurred in the seed-bank
samples but did not have significant regressions. Mean stem-density and biomass values for all 50 species identified are presented in
Seabloom (1997).
Species
Alisma triviale Pursh.
Ambrosia artemisiifolia L.
Atriplex patula L.
Bidens cernua L.
Calamagrostis spp.f
Carex vulpinoidea Michx.
Cyperus diandrus Torr.
Cyperus odoratus L.
Echinochloa muricata (P. Beauv) Fern.
Epilobium ciliatum Raf.
Hypochoeris radicata L.
Leersia oryzoides (L.) Swartz.
Lycopus spp.g
Mentha arvensis L.
Mentha spicata L.
Mimulus ringens L.
Phalaris arundinaceae L.
Polygonum amphibium L.
Scirpus validus Vahl.
Solidago canadensis L.
Typha × glauca Godr.
Verbena hastata L.
Biomass Life Span
measured
Yes
Yes
Yes
Yes
Yes
Yes
Perennial
Annual
Annual
Annual
Perennial
Perennial
Annual
Annual
Annual
Perennial
Annual
Perennial
Perennial
Perennial
Perennial
Perennial
Perennial
Perennial
Perennial
Perennial
Perennial
Perennial
Adult Elevation
Min.
Mean Max.
14.5
89.0
NA
38.5
24.5
60.5
NA
NA
−5.0
43.0
NA
82.0
74.5
38.0
NA
−3.5
65.0
100.0
21.0
92.5
39.5
89.0
−21.2
19.1
NA
1.7
−12.1
13.8
NA
NA
−6.5
17.6
NA
9.4
7.7
−3.8
NA
−11.5
−3.9
−20.3
−24.8
28.8
−38.6
39.1
−59.5
−30.5
NA
−27.0
−51.5
−23.5
NA
NA
−8.0
1.0
NA
−22.0
−48.0
−33.5
NA
−19.5
−49.0
−92.5
−56.0
−14.5
−95.0
11.0
Stem Density
Depthb Temp.c
0.068
−0.058 −0.020
−0.071
−0.107
0.023
−0.072
−0.067
0.022
−0.051
0.027
−0.088
0.018
−0.037
0.039
−0.105
−0.016
−0.069
0.024
−0.086
0.002
−0.089
0.014
−0.072
−0.061
−0.057 −0.025
0.022
0.031
−0.090
0.018
−0.088
Biomass
Depthd Temp.e
−0.025 0.011
0.016
−0.008 0.011
−0.019
0.012 0.014
0.077 0.264
a Agrostis gigantea Roth., Amaranthus rudis Sauer., Bromus inermis Leysser., Cirsium arvense (L.) Scop., Carex atherodes Sprengel.,
Carex stricta Lam., Carex vesicaria L., Conyza canadensis (L.) Cronq., Eleochoaris palustris L., Galium boreale L., Gratiola neglecta
Torr., Juncus nodosus L., Juncus tenuis Willd., Lindernia dubia (L.) Pennell, Lobelia siphilitica L., Penthorum sedoides L., Phleum
pratense L., Poa pratensis L., Polygonum hydropiperoides Michx., Rorippa palustris (L.) Besser., Rumex crispus L., Scirpus fluviatilis
(Torr.) A.Gray, Scirpus validus Vahl., Sium suave Walter, Sparganium eurycarpum Englem., Spartina pectinata Link., Taraxacum
officinale Weber., and Trifolium spp.
−1/2 ·cm−1
b stems1/2 ·stems
max
c stems1/2 ·stems
−1/2 ·◦ C
max
d mg1/2 ·mg
−1/2 ·cm−1
max
e mg1/2 ·mg
−1/2 ·◦ C
max
f Includes C. canadensis (Michx.) P. Beauv and C. stricta (Timm) Koeler
g Includes L. americanus Muhl. and L. asper Greene
were identified in the samples (Table 1). Nomenclature
follows Gleason & Cronquist (1991).
Adult distribution data collection
One of our hypotheses was that a species’ tolerance for
flooding as an adult would be related to its germination
response to water depth. In making this comparison
we used data collected during the 1993 and 1994 field
seasons on adult distributions along the water depth
gradients in 27 prairie wetlands. During these years,
we conducted topographic and vegetative surveys in
10 natural and 17 restored prairie pothole wetlands.
The seed-bank samples were collected in a subset of
these wetlands. We established a grid, oriented on a
random bearing, in each wetland and adjusted the size
of the grid to provide 4 to 5 points in each wetland
vegetation zone (i.e., Stewart & Kantrud 1971). The
number of samples per wetland ranged from 11 to 38
207
with a mean of 24. Grid size ranged from 15 to 30 m
between nodes.
The elevation of each point on the grid, relative to
the highest attainable water level, was surveyed to the
nearest cm. By definition, an elevation of 0 cm was
the highest point that standing water could reach. In
restored wetlands, artificial water-control structures,
such as a standpipes, determined the maximum potential flooding elevation and was assigned an elevation
of zero. In the natural wetlands, we used the highwater mark during the spring of 1993 as the zero
elevation, a year of exceptional flooding throughout
the midwestern United States.
Our vegetation surveys followed a 2–3-year period of normal to high water levels. Because seedling
recruitment occurs during low-water periods and the
wetlands had been flooded longer than the time it takes
for flood-induced mortality to occur, adult plants were
probably not present at water depths greater than those
in which they could survive over the long term. Active
growth and sexual reproduction cease rapidly under
flooded conditions (McKee et al. 1989; Squires and
van der Valk 1992; van der Valk et al. 1994), and mortality usually occurs in less than three years (Millar
1973; van der Valk and Davis 1980; van der Valk et al.
1994).
We recorded the percentage cover of each emergent species within 1 m2 around each grid node.
Further detail on the vegetation surveys is provided by
Seabloom (1997).
Data analysis
In our analyses, we measured the response of the
seedling community and the constituent species to the
water-depth and temperature treatments. In all analyses, we used a regression model composed of two
treatment parameters (Water Depth and Temperature),
interactions among the treatment parameters, and
three experimental design terms (Chamber, Trial, and
their interaction) (Table 2). Chamber represented the
four growth chambers and Trial represented the four 6week trials. The interaction term Chamber × Trial was
nested within Chamber, Trial and Temperature. Tests
of the significance of temperature treatment used the
Chamber × Trial as an error term, because the temperature treatment was applied to specific Chamber and
Trial combinations.
We analyzed the following univariate measures
of community response: stem density, biomass, total species richness, richness of annual species, and
richness of perennial species. We also used a combination of Principal Components Analysis (PCA) and
regression analysis to determine the effects of water
depth and temperature on the overall species composition of the seedling community. First we transformed
the original biomass and stem density for all species
using PCA. The goal in conducting these PCA’s was
to produce several independent axes (PC axes) that
described the majority of the variation in species composition in the seedling community. In the second step,
we used regression analysis to determine how much
of the seedling community variability described by
each PC axis was accounted for by changes in water depth and temperature. Finally, we calculated the
amount of the variation in total community composition accounted for by each independent variable. This
allowed us to compare the regression results among all
of the PC axes.
All six species for which we had biomass data were
included in the biomass PCA (Table 1). We had stem
density data on 50 species, however only 22 were included in the stem density PCA (Table 1). These 22
species were common enough or consistent enough
in their occurrence to produce a significant regression models as individual species. Seabloom (1997)
presents stem density treatment means for all species
identified in the experiment.
We square-root transformed the stem density and
biomass data to stabilize their variance. The PCA’s
used correlation matrices, and we included PC axes in
our analysis if they had eigenvalues greater than one
(Manly 1986).
In PCA, each sample in the original data set is
given a set of scores that corresponds to its location
along the PC axes. We regressed these scores onto
the complete experimental model to determine the
experimental factors that accounted for the greatest
amount of change in the seedling community. One of
the benefits of using PC scores in regressions is that
the PC axes are orthogonal to each other. This is not
necessarily true of the original variables (species).
The independence of the PC axes allowed us to
make comparisons among regressions by converting
the results to the comparable units of total community
variance. We did this in three steps. First, we calculated the proportion of the total variance in the original
data accounted for by each PC axis, the eigenvalue in a
PCA using a correlation matrix. A PCA that includes
n species will produce n PC axes. We will designate
the ith of these n axes as PCi and the proportion of the
original variance accounted for by PCi as Vi .
208
Table 2. Regression models used for the analysis of the growth chamber seedling recruitment data. Stem density,
biomass, and species richness were analyzed using the full model. Principal components were analyzed using the
reduced model. The main-effects model was used to calculate linear temperature and depth regression coefficients
for main effects that were significant in the full model.
Source
Degrees of freedom
Full model
Reduced
model
Description
Main-effects
model
Chamber
Trial
Temperaturea
Temperature2a
Temperature3a
Chamber × Trial b
Depth
Depth2
Temperature × Depth
Temperature2 × Depth
Temperature3 × Depth
Temperature × Depth2
Temperature2 × Depth2
Temperature3 × Depth2
Error
3
3
1
1
1
6
1
1
1
1
1
1
1
1
24
3
3
3
3
6
2
2
24
42
Total
47
47
47
Among chamber effects
Among trial effects
Linear temperature effects
Quadratic temperature effects
Cubic temperature effects
Temperature effects error term
Linear depth effects
Quadratic depth effects
6
a Temperature treatments were allocated as a Latin square within the four growth chambers and four trials.
b Chamber × Trial is nested within Chamber, Trial, and Temperature.
Secondly, we regressed PCi onto the reduced
model in Table 2 and used the resulting ANOVA table
to calculate the proportion of Vi that was attributable
to each independent variable in the regression. The
proportion of the variation in PCi accounted for by
each independent variable, pij, can then be calculated
from an ANOVA table as
!
Sij2
pij =
,
(1)
Si2
where j designates one of the independent variables in
the regression of PCi scores onto the regression model
(Table 2), Sij2 is the sum of squares for variable j , and
Si2 is the total sum of squares in the regression (cf.,
Snedecor & Cochran 1989).
Finally, by multiplying pij by Vi we calculated the
amount of the total variation in abundance (stem density or biomass) of all the species in all of the samples
that was accounted for by a given experimental factor
(e.g., water depth and temperature).
In addition to the community level analysis, we
also examined the response of individual species to the
water-depth and temperature treatments by regressing
the stem density and biomass of each species onto
the full regression model in Table 2. The original
data exhibited nonstationarity of variance in relation
to predicted values and regressions were run on three
datasets: non-transformed, log10 transformed, and
square root transformed. The square root transform
was the most effective in achieving both normality and
stationarity of residual variance.
We tested the relationship between the seed germination responses to water depth and temperature
and adult niche characters of mean water depth and
life span (annual or perennial) by using the linear
water-depth and temperature coefficients from regression models with no interaction or polynomial terms
(Table 2). This slope provided a simple summary of
the overall response of the species to the two experimental factors (water depth and temperature) with the
other factor held constant (e.g., species germination
was either increased, decreased, or unchanged by the
water-depth treatments independent of the temperature).
The linear terms we included in these models were
all significant (p<0.05) in the regression using the
full model (Table 2). In these linear regressions, the
stem densities (stems•pot−1 ) of species were trans-
209
formed by dividing them by the mean number of
stems in the water-depth and temperature combination with the highest stem density (stemsmax •pot−1 ).
This transformation removed the effects of seed abundance in the sample, because the highest treatment
mean was set equal to one for all species and we made
among species comparisons using the relative density
(stems•stems−1
max ).
Results
Community level analyses
The PCA’s of stem density and biomass showed the
strong effects of water depth and temperature on overall composition of the seedling community. There
were six significant PC axes (eigenvalues greater than
1) included in the PCA of the stem density data (Table 3). These PC axes accounted for 70.4% of the total
species variance for the original 22 species. The regressions of PCA scores had few significant quadratic
or cubic water-depth or temperature terms. We present
only the results of regressions using linear terms and
linear interactions in the interest of clarity (Table 3).
The first five PC axes had significant (p < 0.05)
overall regression models. The first PC axis was most
highly correlated with water depth and the second
PC axis with temperature (as calculated in equation
1). The remaining PC axes were dominated by the
experimental design terms Chamber, Trial, and Plot.
There were two significant PC axes (eigenvalues
greater than 1) in the PCA of the biomass measurements (Table 3). These PC axes accounted for 73.5%
of the total species variance from the original six
species. The first axis was most highly correlated with
temperature, and the second axis with water depth
(Table 3).
Total species richness was negatively correlated
with water depth and positively correlated with temperature (Figure 1c). Over the range of our experimental treatments, the water-depth response was greater
than that of the temperature. As with total species
richness, both annual and perennial species richness
declined rapidly with increasing water depth. However, annuals and perennials had different responses
to the temperature treatments. Annual species richness was positively correlated with temperature in the
shallow treatments, while perennial species richness
showed little response to temperature (Figures 1a, 1b,
Table 4).
Figure 1. Effect of water depth and temperature on mean (n = 4)
annual, perennial, and total species richness in soil seed bank samples from a 6-week growth chamber experiment. Temperature was a
main plot effect and depth was a split plot effect. Daytime soil temperatures were approximately 5 ◦ C higher than the air temperature,
and nightime air and soil temperatures were identical. Separate standard errors are given for making comparisons among depth means
(Depth SEM) and among temperature means (Temp SEM).
Total stem density was positively correlated with
temperature and negatively correlated with water
depth (Figure 2, Table 4). Water depth and temperature only affected biomass in the in the two highest
temperature treatments (15◦ night/25◦ day and 20◦
night/30◦ day). In these warm treatments, biomass was
positively correlated with temperature and negatively
worrelated with water depth (Figure 3).
There were significant Trial effects in both the
Stem Density PCA (Table 3) and the species richness, total stem density, and biomass regressions
(Table 4). Total, annual, and perennial species richness declined over the course of the experiment due
to increased stratification period (Figure 4). Although
biomass and stem density varied among trials, there
were no perceptible trends over the course of the entire
experiment.
210
Table 3. Results of the stem density and biomass Principal Components Analyses (PCA) and the regression of the stem density and biomass
principal components (PC) scores onto temperature and water depth experimental terms. Regression results are presented as the proportion
of the total species variance accounted for by each term in the design. Empty cells denote terms that were not significantly different from
zero at the 0.05 level. Total sums of squares and overall model significance are also shown. Twenty-two species were included in the stem
density PCA, and six species were included in the biomass PCA.
Eigenvalue
Stem density PCA results
proportion of species variance
PC1
6.739
0.306
PC2
2.700
0.123
Biomass PCA results
PC3
2.074
0.094
PC4
1.597
0.073
PC5
1.355
0.062
PC6
1.057
0.046
Stem Density Regression Results
Water Depth
Temperaturea
Temperature × Water Depth
Growth Chamber
Trial
Chamber × Trial interaction b
Overall Model r2
Total Sums of Squares.
Probability > F for Model
PC1
0.164
0.013
PC2
0.005
0.052
PC3
0.018
PC1
2.932
0.489
PC2
1.475
0.246
Biomass Regression Results
PC4
PC5
PC6
PC1
PC2
0.127
0.411
0.006
0.009
0.053
0.816
316.735
<0.001
0.02
0.032
0.015
0.746
126.881
<0.001
0.680
97.498
0.001
0.583
75.067
0.015
0.015
0.669
63.683
0.001
0.024
0.016
49.677
0.475
0.889
137.811
<0.001
0.794
69.306
<0.001
a Significance test uses Chamber × Trial as the error term.
b Chamber × Trial is nested within Chamber, Trial, and Temperature.
Table 4. The effects of water depth and soil temperature on recruitment of seedlings from the soil seedbank. Results of the regression of stem density
and total, annual, and perennial species richness in growth chamber trials. Degrees of freedom (D.F.), sums of squares (S.S.), and the probability of a
Type I error (α) are presented for each of the four regressions.
Dependent variable
Model r 2
Source
D.F.
Chamber
3
Trial
3
Temperaturea
1
Temperature2a
1
Temperature3a
1
Chamber × Trialb
6
Depth
1
Depth2
1
Temperature × depth
1
Temperature2 × depth
1
Temperature3 × depth
1
Temperature × depth2
1
Temperature2 × depth2 1
Temperature3 × depth2 1
Error
24
Total
47
Stem
Density
0.95
S.S.
21 994.75
26 420.42
70 795.35
6 960.08
2 982.15
7 968.50
109 363.04
6 440.08
9 235.33
396.19
450.50
579.40
108.35
108.28
13 816.83
277 619.25
Biomass
0.96
S.S.
α
<0.001
<0.001
<0.001
0.062
0.185
0.067
<0.001
0.003
0.001
0.415
0.385
0.326
0.668
0.668
0.0001
α
5 597 763
0.077
8 558 188
0.020
350 204 757
0.000
9 079 929
0.015
6 698 672
0.027
4 720 012
0.397
4 652 873
0.018
33 793 304 <0.001
22 445 460 <0.001
86 990
0.732
2 852 115
0.058
209 607
0.595
3 221 593
0.045
3 347 437
0.042
17 328 504
472 797 204
Total
Richness
0.88
S.S.
α
Annual
Richness
0.79
S.S.
α
Perennial
Richness
0.83
S.S.
α
30.92
0.242
328.75 <0.001
20.42
0.026
16.33
0.039
0.00
1.000
14.17
0.908
605.78 <0.001
172.51 <0.001
14.89
0.155
0.78
0.740
1.30
0.668
3.42
0.489
0.01
0.966
0.98
0.710
165.67
1 375.92
0.0001
7.73
0.498
68.73
0.001
30.10
0.002
1.02
0.368
0.10
0.766
6.46
0.908
117.41 <0.001
30.47
0.005
9.56
0.095
1.06
0.568
1.06
0.568
2.75
0.360
0.23
0.789
0.30
0.762
75.83
352.81
0.0009
8.06
0.490
105.40 <0.001
0.94
0.423
9.19
0.036
0.10
0.784
7.63
0.876
189.81 <0.001
57.98 <0.001
0.59
0.674
0.02
0.936
0.01
0.952
0.04
0.917
0.35
0.744
0.20
0.808
77.67
457.98
0.0001
a Temperature treatments were allocated as a Latin square within the four growth chambers and four trials. The temperature error term is Chamber ×
Trial.
b Chamber × Trial is nested within Chamber, Trial, and Temperature.
211
Species level analyses
Figure 2. Effect of water depth and temperature on mean (n = 4)
stem density (stems·pot−1 ) in soil seed bank samples from a six
week growth chamber experiment. Temperature was a main plot
effect and depth was a split plot effect. Separate standard errors are
given for making comparisons among depth means (Depth SEM)
and among temperature means (Temp SEM).
Figure 3. Effect of water depth and temperature on mean (n = 4)
biomass (g·pot−1 ) in soil seed bank samples from a six week growth
chamber experiment. Temperature was a main plot effect and depth
was a split plot effect. Separate standard errors are given for making comparisons among depth means (Depth SEM) and among
temperature means (Temp SEM).
Prior to analyzing the germination responses of individual species, we had to establish that the species
were responding to the experimental treatments independently of one another. If there was little competitive effect on plant growth, then we could assume that
the response to the experimental treatments by individual species was largely independent of coexisting
species. We tested for potential competitive effects by
using a regression of biomass on stem density.
We regressed biomass onto the stem density of
the six species included in the biomass measurement
(Table 1). The six species included in the biomass
measurements were very common and accounted for
82% of the total stems in the experiment. In addition,
the remaining 18% of the stems were significantly correlated with the stem density of the six species for
which we had biomass measurements (r 2 = 0.463,
p < 0.001), and using the stem density of only the
six species with biomass measurements accounted for
90.3% of the total variation in stem density.
We used a log10(x) transform of the stem density
and the biomass data to achieve stationarity of variance. We examined the residuals of the transformed
data using Normal Probability plots, plots of residuals
vs. predicted values, and the Shapiro–Wilk test and
found that, after the transformation, the residuals were
not significantly different from a normal distribution
and the variance was stationary.
The regression of biomass on stem density had
a significant positive linear trend (Figure 5). The
quadratic term in this model was not significant (p =
0.246), indicating there was no substantial decrease
in the response of biomass to increasing stem density.
Biomass (mg·pot−1), b, can be expressed in terms of
stem density (stems·pot−1 ), s, using the expression:
log10 b = 1.112 + 1.096 log10 s .
(4)
Because these data were log transformed, we could
detect the presence of competitive effects on yield by
testing if the slope of the regression was significantly
different from 1.0. We were not able to reject the null
hypothesis that the slope in Equation (4) equals 1.0 (p
= 0.405), and Equation (4) reduces to
Figure 4. Effect of stratification length on mean (n = 4) total, annual, and perennial species richness in soil seed bank samples from
a six week growth chamber experiment.
b = 12.9s
(5)
showing a proportional increase in biomass with increasing stem density and indicating the absence of detectable competitive effects on plant growth, a strong
212
Figure 5. The relationship between stem density and biomass
production over a range of water depth and temperature conditions in four growth chamber experiments (r 2 = 0.43,
p < 0.0012). The regression model predicts biomass (mg·pot−1 ),
b, based on stem density (stems·pot−1 ), s, using the relationship:
log10 b = 1.112 + 1.096log10 s.
indication that there were no density effects on yield
(Begon & Mortimer 1986).
Stem density means of all 50 species in the experiment are presented by Seabloom (1997). Of the
22 species that were common or consistent enough in
their occurrence to have a significant regression model
in the individual species analyses, most were affected
by the water-depth and temperature treatments (Table 1). Stem densities were significantly correlated to
water depth in 95% of the species, and 86% of these
species had lower stem densities as water depth increased. Fifty percent of the species with significant
regressions responded to temperature. Of the species
that responded to temperature, 82% had higher stem
densities in the warmer treatments.
All species, except Echinochloa muricata, had
similar biomass and stem density responses to water
depth (i.e., the signs of the slopes are the same in Table 1). Echinochloa muricata did not have a significant
water-depth slope in the biomass regressions, while
its stem density was positively correlated with water
depth. Unlike the stem density response, no species
had higher biomass in cooler temperatures (a negative
temperature slope in Table 1).
The hypothesis that annual species should have
higher germination rates at high temperatures was
tested by using a one-tailed t-test to determine if
the mean slope of temperature response of seedlings
was higher for annual species than perennial species.
In other words, does the germination rate of annual
species increase more rapidly with increasing temperature than the germination rate of perennial species?
The average response of the seven annual species
with significant regressions was to have higher
stem counts with increasing temperature (0.012
stems1/2 •stemsmax −1/2 •◦ C−1 ) (Table 1). This slope
was greater than the mean slope of the 15 perennial
−1/2
species (0.005 stems1/2 ·stemsmax ·◦ C−1 ) in a onetailed t-test for samples with unequal variance (p =
0.005).
The hypothesis that the distribution of adults along
an elevational gradient was correlated with the slope of
the water-depth response of the seedlings was tested
using a one-tailed t-test to determine if there was a
negative correlation between the mean elevation at
which a species was found as an adult and the slope
of the water-depth response. In other words, do fewer
seeds of species found at high elevations in a wetland
germinate as water depth increases, and do more seeds
of species that are found at low elevations germinate as
water depth increases? We had both adult distribution
and seedling water-depth response data for 17 species
(Table 1).
There was a significant correlation between a
species’ response to water depth as a seedling and the
mean water depth of its distribution as an adult. The
slope of the response in a species’ germination rate to
−1/2
water depth (stems1/2 •stemsmax •cm), β, can be predicted by the mean of its distribution as an adult using
the following relationship (r 2 = 0.498, p < 0.001):
β = −15.3 − 289.2z ,
(6)
where z is the mean elevation (cm) at which adults of
the same species are found in the field vegetation surveys (Figure 6). Equation (6) shows that species found,
on average, at high elevations will have lower germination rates as water depth increases, while species
found at low elevations will have higher germination
rates with increasing water depth.
Discussion
We found that water depth and temperature had strong
effects on the composition of wetland seedling communities independent of competitive effects. Water
depth and temperature also changed total stem density
and biomass and had the potential to alter the intensity of competition among seedlings. We also found a
correlation among regeneration and adult niches.
213
Figure 6. Correlation between mean adult elevation and species’
germination response to water depth (r 2 = 0.498, p = 0.001).
Mean adult elevation is shown relative to the maximum attainable
water depth (0 cm). Germination response to water depth, β, is
calculated as the change in the square-root transformed relative stem
density (stems•stems−1
max ) per cm change in water depth. Stemsmax
is defined as the greatest number of stems in any experimental treatment. Beta is predicted for each species from the mean elevation
of that species’ distribution as an adult, z, using the relationship:
β = −0.052 − 1.722z. Species names are abbreviated as follows:
Ali = Alisma triviale, Amb = Ambrosia artemisiifolia, Bid = Bidens
cernua, Cal = Calamagrostis canadensis, Car = Carex vulpinoidea,
Ech = Echinochloa muricata, Epi = Epilobium ciliatum, Lee =
Leersia oryzoides, Lyc = Lycopus spp. , Men = Mentha arvensis, Mim = Mimulus ringens, Pha = Phalaris arundinaceae, Pol
= Polygonum amphibium, Sci = Scirpus validus, Sol = Solidago
canadensis, Typ = Typha glauca, and Ver = Verbena hastata.
Increased levels of flooding decreased stem density, biomass, and species richness and sorted species
according to their adult tolerance for flooding. As the
water depth in which a seed germinated declined, the
density and total biomass of its neighbors increased.
The magnitude of the water-depth effect on biomass
increased with increasing temperature.
While our trials ended without any detectable
effects of competition, the plants in high-density
shallow-water treatments would have experienced a
more competitive environment as they increased in
size. The presence of neighboring plants can cause
a significant reduction in plant growth and survival
(Putwain & Harper 1970; Pinder 1975; Allen & Forman 1976; Fowler 1981; Silander & Antonovics 1982;
Mithen et al. 1984; Silander & Pacala 1985; Goldberg 1987), because autotrophic plants have similar
resource requirements (Aarssen 1983; Goldberg &
Werner 1983; Ågren & Fagerström 1984; Shmida &
Ellner 1984; Hubbell & Foster 1986; Goldberg 1987).
Biomass and plant size can also account for differ-
ences in the magnitude of neighbor effects among
species (e.g., Gaudet & Keddy 1988).
Species richness declined with increasing water
depth leaving a reduced pool of species in deeper sites.
Reduced species richness in submersed treatments has
been found in other studies (van der Valk & Davis
1979; Pederson 1983). The species found as seedlings
in flooded conditions were not a random subset of the
total species pool. The species level analysis showed
that, while most species had the highest seedling densities in shallow treatments, there was a guild of
species that had consistently higher germination rates
in the flooded treatments. This guild was composed of
the same stress-tolerant species that grow in flooded
conditions as adults, such as Scirpus validus, Typha ×
glauca, and Alisma triviale.
The correspondence between the effects of water
depth on adult and regeneration niches confirms the
qualitative results in a similar study by Keddy & Ellis (1985) that had four species in common with our
study. For two of these species, the germination responses along a water-depth gradient were concordant
between these two studies. Bidens cernua, an annual,
germinated at a lower rate with increasing water depth
in both studies. Alisma triviale, an emergent perennial,
showed increased germination rates with increasing
water depth in both studies.
Keddy & Ellis (1985) found Typha angustifolia
and Scirpus validus to be unresponsive to a range
of water depths from 5 cm below the soil surface to
10 cm above the soil surface, while our study showed
that Typha × glauca and Scirpus validus increased
germination rates with increasing water depth. Harris & Marshall (1963) and Smith & Kadlec (1962)
also found that Typha sp. and Scirpus sp. germinated at higher rates in submersed conditions. Van
der Valk & Davis (1978) found that Typha angustifolia and Scirpus validus germinated at lower rates in
flooded conditions, however they used flooding treatments (10 cm) that were higher than those in our study
(7 cm).
The relationship between adult distributions and
germination characteristics may create a positive feedback in which species germinate at their highest rates
in areas where they are most common as adults. This
feedback may speed the formation of coenoclines, by
creating an incipient coenocline at the time of seedling
emergence (van der Valk & Welling 1988). This positive feedback could be further accentuated in cases
where seed abundance in the seed bank was correlated
with adult distributions, because species would have
214
higher germination rates and higher densities of seeds
at water depths where they most frequently occur as
adults (van der Valk & Welling 1988).
The effect of water depth on germination is an
inherent character of individual species and will be
maintained among sites. In contrast, the effect of seed
abundance gradients on coenocline formation is likely
to be site-specific. There are examples of wetlands that
have patterns of seed density along water-depth gradients that correspond to adult community composition
(e.g., Hogenbirk & Wein 1992) or to the distribution
of specific species (e.g., Pederson 1983). At other
sites, secondary dispersal by water creates a wellmixed seed bank (e.g., van der Valk & Davis 1976)
or driftlines of seeds along the shore (e.g., Pederson
& van der Valk 1984), patterns that are independent of
adult distributions.
We found that the primary effect of temperature
was to increase total stem density and biomass and
to increase the proportion of annual species in the
seedling community. Annual species richness declined
with decreasing temperature, while perennial richness
showed a small, though detectable, increase. This result is consistent with our hypothesis that the high soil
temperatures typical of exposed mudflats should trigger the germination of annual species, because these
mudflats represent safe sites for the establishment of
ruderal species.
In wetlands, climatic cycles and topography cause
water depth and soil temperature to vary spatially
along an elevational gradient and temporally on a
seasonal and annual basis. Soil temperature is also
affected by the presence of litter and emergent vegetation. Because wetland species lack a shared germination response, this variability in environmental
conditions will affect the distribution of species at the
time of recruitment from the seed bank due to their
different regeneration niches. Changes in water depth
will sort species according to their adult water-depth
tolerances, and temperature fluctuations will change
the proportion of annuals appearing in the seedling
community. The sensitivity of the seedling community
to environmental conditions may explain the difficulty
of relating the composition of the seed bank directly
to seedling or adult community composition (Leck &
Simpson 1995) as well as the inability to predict the
distributions of adults plants along elevation gradients based solely on current environmental conditions
(Wilson & Keddy 1985; de Swart et al. 1994).
Given the importance of seedling recruitment in
wetland systems (Salisbury 1971; van der Valk &
Davis 1978; Keddy & Reznicek 1985; Poiani & Johnson 1989; Grillas et al. 1991), this variability in the
initial pool of seedlings will likely be reflected in the
composition of the adult coenocline. This will happen because the environment at the time of seedling
recruitment will act as a filter that removes species
from the pool available to colonize an area as adults.
Due to the importance of the regeneration niche, the
distribution of adult plants will reflect both the current
environmental conditions as well as the environment
during historical recruitment events.
Acknowledgements
We wish to thank Deven Nice for his work throughout
this research and Kenneth Koehler from the Iowa State
University Department of Statistics for his statistical
advice. Funding was provided by the U.S. Environmental Protection Agency as well as the Department
of Botany, the Ecology and Evolutionary Biology
Interdepartmental Program, and the Geographic Information Systems Support and Research Facility at Iowa
State University.
References
Aarssen, L. W. 1983. Ecological combining ability and competitive
combining ability in plants: toward a general evolutionary theory
of coexistence in systems of competition. Am. Nat. 122: 707–
731.
Ågren, G. I. & Fagerström, T. 1984. Limiting dissimilarity in
plants: randomness prevents exclusion of species with similar
competitive abilities. Oikos 43: 369–375.
Allen, E. B. & Forman, R. T. T. 1976. Plant species removals and
old-field community structure and stability. Ecology 57: 1233–
1243.
Begon, M. & Mortimer, M. 1986. Population Ecology. Sinauer
Associates, Inc., Sunderland, Massachusetts, U.S.A.
de Swart, E. O. A. M., van der Valk, A. G. , Koehler, K. J. &
Barendregt, A. 1994. Experimental evaluation of realized niche
models for predicting responses of plant species to a change in
environmental conditions. J. Veg. Sci. 5: 541–552.
Drake, J. A. 1991. Community-assembly mechanics and the structure of an experimental species ensemble. Am. Nat. 137: 1–26.
Fowler, N. 1981. Competition and coexistence in a North Carolina
grassland. II. the effects of the experimental removal of species.
J. Ecol. 69: 843–854.
Galinato, M. I. & van der Valk, A. G. 1986. Seed germination
traits of annuals and emergents recruited during drawdowns in
the Delta Marsh, Manitoba, Canada. Aquatic Bot. 26: 89–102.
Gaudet, C. L. & Keddy, P. A. 1988. Predicting competitive ability
from plant traits: a comparative approach. Nature 334: 242–243.
Gleason, H. A. 1926. The individualistic concept of the plant
association. Bull. Torrey Bot. Club 53: 7–26.
215
Gleason, H. A. & Cronquist, A. 1991. Manual of vascular plants
of Northeastern United States and adjacent Canada, 2nd edition.
New York Botanical Garden, Bronx, New York.
Goldberg, D. E. 1987. Neighborhood competition in an old-field
plant community. Ecology 68: 1211–1223.
Goldberg, D. E. & Werner, P. A. 1983. Equivalence of competitors
in plant communities: a null hypothesis and a field experimental
approach. Am. J. Bot. 70: 1098–1104.
Grace, J. B. 1987. The impact of preemption on the zonation of two
Typha species along lakeshores. Ecol. Monog. 57: 283–303.
Grillas, P., Van Wijk, C. & Bonis, A. 1991. Life history traits: a
possible cause for higher frequency of occurrence of Zannichellia pedunculata than of Zannichellia obtusifolia in temporary
marshes. Aquatic Bot. 42: 1–13.
Grubb, P. J. 1977. The maintenance of species-richness in plant
communities: the importance of the regeneration niche. Biol.
Rev. 52: 107–145.
Harper, J. L. 1977. Population biology of plants. Academic Press,
London.
Harris, S. W. & Marshall, W. H. 1963. Ecology of water-level
manipulations on a northern marsh. Ecology 44: 331–343.
Hogenbirk, J. C. & Wein, R. W. 1992. Temperature effects on
seedling emergence from boreal wetland soils: implications for
climate change. Aquatic Bot. 42: 361–373.
Hubbell, S. P. & Foster, R. B. 1986. Biology, chance, and history and
the structure of tropical rain forest tree communities. pp. 314–
329. In: Diamond, J. & Case, T. J. (eds), Community Ecology.
Harper and Row, New York.
Hutchinson, G. E. 1961. The paradox of the plankton. Am. Naturalist 95: 137–145.
Kadlec, J. A. 1962. Effects of a drawdown on a waterfowl impoundment. Ecology 43: 267–281.
Keddy, P. A. 1982. Shoreline vegetation in Axe Lake, Ontario:
effects of exposure on zonation patterns. Ecology 64: 331–344.
Keddy, P. A. & Constabel, P. 1986. Germination of ten shoreline
plants in relation to seed size, soil particle size and water level:
an experimental study. J. Ecol. 74: 133–141.
Keddy, P. A. & Ellis, T. H. 1985. Seedling recruitment of 11 wetland plant species along a water level gradient: shared or distinct
responses? Can. J. Bot. 63: 1876–1879.
Keddy, P. A. & Reznicek, A. A. 1985. Vegetation dynamics, buried
seeds, and water level fluctuations on the shorelines of the great
lakes. pp. 33–58. In: Prince, H. H. & D’Itri, M. (eds), Coastal
Wetlands. Lewis Publishers, Inc.
Leck, M. A. & Simpson, R. L. 1995. Ten-year seed bank and vegetation dynamics of a tidal freshwater marsh. Am. J. Bot. 82:
1547–1557.
Manly, B. F. J. 1986. Multivariate statistical methods: a primer.
Chapman and Hall, London.
McKee, K. L., Mendelssohn, I. A., & Burdick, D. M. 1989. Effects of long-term flooding on root metabolic response in five
freshwater marsh plant species. Can. J. Bot. 67: 3446–3452.
Millar, J. B. 1973. Vegetation changes in shallow marsh wetlands
under improving moisture regime. Can. J. Bot. 51: 1443–1457.
Mithen, R., Harper J. L. & Weiner, J. 1984. Growth and mortality of
individual plants as a function of ‘available area’. Oecologia 62:
57–60.
Pederson, R. L. 1983. Abundance, distribution, and diversity of
buried seed populations in the Delta Marsh, Manitoba, Canada.
Ph.D. Dissertation, Iowa State University, Ames, Iowa.
Pederson, R. L. & van der Valk, A. G. 1984. Vegetation change and
seed banks in marshes: ecological and management implications.
Trans. North Am. Wildlife Nat. Res. Conf. 49: 271–280.
Pinder III, J. E. 1975 Effects of species removal on an old-field plant
community. Ecology 56: 747–751.
Poiani, K. A. & Johnson, W. C. 1989. Effect of hydroperiod on
seed-bank composition in semipermanent prairie wetlands. Can.
J. Bot. 67: 856–864.
Putwain, P. D. & Harper, J. L. 1970. Studies in the dynamics of plant
populations III. The influence of associated species on populations of Rumex acetosa and R. acetosella in grassland. J. Ecol.
58: 251–264.
Robinson, J. V. & Dickerson Jr., J. E. 1987. Does invasion sequence
affect community structure? Ecology 68: 587–595.
Salisbury, E. 1971. The pioneer vegetation of exposed muds and its
biological features. Phil. Trans. Roy. Soc. London B 259: 207–
255.
Seabloom, E. W. 1997. Vegetation dynamics in prairie wetlands.
Ph.D. Dissertation, Iowa State University, Ames, Iowa.
Shmida, A. & Ellner, S. . 1984. Coexistence of plant species with
similar niches. Vegetatio 58: 29–55.
Silander, J. A. & Antonovics, J. 1982. Analysis of interspecific interactions in a coastal plant community – a perturbation approach.
Nature 298: 557–560.
Silander Jr., J. A. & Pacala, S. W. 1985. Neighborhood predictors of
plant performance. Oecologia 66: 256–263.
Smith, L. M. & Kadlec, J. A. 1985. The effects of disturbance on
marsh seed banks. Can. J. Bot. 63: 2133–2137.
Snedecor, G. W. & Cochran, W. G. 1989. Statistical Methods – 8th
edition. Iowa State University Press, Ames, Iowa
Spence, D. H. N. 1982. The zonation of plants in freshwater lakes.
pp. 37–125. In: Macfadyen, A. & Ford, E. D. (eds), Advances in
ecological research. Academic Press, New York.
Squires, L. & van der Valk, A. G. 1992. Water-depth tolerances
of the dominant emergent macrophytes of the Delta Marsh,
Manitoba. Can. J. of Bot. 70: 1860–1867.
Stewart, R. E. & Kantrud, H. A. 1971. Classification of natural
ponds in the glaciated prairie region. U.S. Fish and Wildlife
Service Resource Publication No. 92.
Sutherland, J. P. 1974. Multiple stable points in natural communities. Am. Natur. 108: 859–873.
Thompson, K. & Grime, J. P. 1983. A comparative study of germination responses to diurnally-fluctuating temperatures. J. Appl.
Ecol. 20: 141–156.
van der Valk, A. G. & Davis, C. B. 1976. The seed banks of prairie
glacial marshes. Can. J. Bot. 54: 1832–1838.
van der Valk, A. G. & Davis, C. B. 1978. The role of seed banks in
the vegetation dynamics of prairie glacial marshes. Ecology 59:
322–335.
van der Valk, A. G. & Davis, C. B. 1979. A reconstruction of the
recent vegetational history of a prairie marsh, Eagle Lake, Iowa,
from its seed bank. Aquatic Bot. 6: 29–51.
van der Valk, A. G. & Davis, C. B. 1980. The impact of a natural
drawdown on the growth of four emergent species in a prairie
glacial marsh. Aquatic Bot. 9: 301–322.
van der Valk, A. G., Squires, L., & Welling, C. H. 1994. Assessing
the impacts of an increase in water level on wetland vegetation.
Ecol. Appl. 4: 525–534.
van der Valk, A. G. & Welling, C. H. 1988. The development
of zonation in freshwater wetlands: an experimental approach.
pp. 145–158. In: Werger, H. J., During, M. J. A. & Willems, J.
H. (eds), Diversity and pattern in plant communities. Academic
Publishing. The Hague, The Netherlands.
Weiher, E. & Keddy, P. A. 1995a. The assembly of experimental
wetlands. Oikos 73: 323–335.
216
Weiher, E. & Keddy, P. A. 1995b. Assembly rules, null models,
and trait dispersion: new questions from old patterns. Oikos 74:
159–164.
Welling, C. H. 1987. Reestablishment of perennial emergent macrophytes during a drawdown in a lacustrine marsh. M.S. Thesis,
Iowa State University, Ames, Iowa.
Welling, C. H., Pederson, R. L. & van der Valk, A. G. 1988.
Recruitment from the seed bank and the development of zonation
of emergent vegetation during a drawdown in a prairie wetland.
J. Ecol. 76: 483–496.
Wilson, S. D. & Keddy, P. A. 1985. Plant zonation on a shoreline
gradient: physiological response curves of component species. J.
Ecol. 73: 851–860.
Wilson, S. D. & Keddy, P. A. 1986. Species competitive ability and
position along a natural stress/disturbance gradient. Ecology 67:
1236–1242.