fire severity affects vegetation and seed bank in a wetland

Fire severity affects vegetation and seed bank in awetland

Hideo Kimura and Shiro Tsuyuzaki

Keywords

Litter; Prescribed fire; Reed swamp; Seed

bank; Standing vegetation

Abbreviations

AIC = Akaike’s Information Criteria;

GLM = Generalized Linear Model;

LMM = Linear Mixed-Effects Model

Received 8 April 2010

Accepted 19 January 2011

Co-ordinating Editor: Juli Pausas

Tsuyuzaki, S. (corresponding author,

[email protected]): Graduate School of

Environmental Earth Science, Hokkaido

University, Sapporo 060-0810, Japan

Kimura, H. ([email protected]):

Graduate School of Environmental Science,

Hokkaido University, Sapporo 060-0810, Japan

.

Abstract

Questions: How does the severity of prescribed fires affect vegetation and seed

bank in a wetland?

Location: A fire-prone reed swamp in northern Japan (250 ha, 401490N,

1411220E, o10 m a.s.l.).

Methods: Vegetation, biomass and seed bank were monitored for the 2 yr after

annual prescribed fires were discontinued. Plant communities were placed into

three categories based on fire severity: high (H) – fire consumed litter

completely; moderate (M) – fire removed standing litter but left wet fallen

litter; and low (L) – fire incompletely removed standing litter and did not

remove fallen litter. Soil samples were collected in autumn 2007 and early

summer 2008, and germinable seed bank was investigated by greenhouse trials.

Results: High fire severity increased diversity in the next growing season by

the establishment of short herbs in the standing vegetation. The biomass of

forbs and grasses was greater in H where Phragmites australis biomass was

reduced. The density of seed bank was 4 30 000 seeds m�2 throughout all the

treatments. Perennial plants were dominant in the vegetation, while annuals,

biennials and rushes were dominant in the seed bank. Small seeds were more

abundant in the soil than in the litter. Qualitative and quantitative similarities

between seed bank and the vegetation were low, and tended to be higher in H.

Conclusions: Fire contributed to the development of diverse standing vegeta-

tion via the positive effects on seed bank dynamics, and can be considered a tool

to maintain species-rich marshes.

Introduction

Although prescribed fire has been used for increasing

productivity and biodiversity in various ecosystems (Iwa-

kuma 1996, Hiers et al. 2003), it threatens the sustain-

ability of the vegetation if the prescription is over-used

(Pausas & Keeley 2009). The effectiveness of fire differs

with severity, scale, frequency and patchiness (Hochkirch

& Adorf 2007; Cox & Allen 2008). For example, biomass

removal by fire decreases interspecific and intraspecific

competition. Moderate fire severity promotes seedling

recruitment by increases in light and nutrients (Parker &

Kelly 1989), while severe fire may decrease seed bank and

species diversity (Mamede & de Araujo 2008). Therefore,

fire severity affects post-fire vegetation development and

recovery.

Fire removes not only standing vegetation but also

litter that affects vegetation dynamics and the seed bank

(Owens et al. 2007; Allen et al. 2008; Egawa et al. 2009).

In burned forests, the severity fire affects the seed bank

more than the vegetation (Johnson 1992). In heath, some

fire regimes that promote overstory development are in

conflict with persistence of understory species because of

interactions between functional groups that are related to

growth, reproduction and competition (Keith & Brad-

stock 1994). Therefore, the removal of litter or biomass,

which is related to light environments, may be key for

predicting the effects of fire on ecosystem maintenance,

and clarifying relationships between seed bank and the

vegetation provides insight to the resilience of a commu-

nity against fire (Hopfensperger 2007).

In Hotoke Swamp, northern Japan, where Phragmites

australis is dominant, prescribed fire had been conducted

every early spring for about 30 yr. Prescribed fire was not

carried out in 2008, and this abeyance gave an opportu-

nity to monitor changes in vegetation and seed bank for

Applied Vegetation Science 14 (2011) 350–357

350Applied Vegetation Science

Doi: 10.1111/j.1654-109X.2011.01126.x r 2011 International Association for Vegetation Science

mailto:[email protected]







2 yr after the last prescribed fire. We expected that this

temporary release from prescribed fire would lead to

significant changes in seed bank and/or vegetation. The

hypothesis was that fire severity influences the seed bank

and the standing vegetation differently. To test this hy-

pothesis, we measured the effects of prescribed fire on

vegetation structure and species composition of the stand-

ing vegetation and seed bank.

Materials and Methods

Study site

Land reclamation was promoted for wetlands in Japan

soon after World War II, because of increases in rice

demand. Hotoke Swamp, located in northern Honshu

Island, Japan, was prepared for a change in land-use after

1964 by the construction of drains (250 ha, 401490 N,

1411220 E, o 10 m a.s.l.). Before the construction of

drains the swamp was a shallow pond with low plant

cover. After construction, groundwater was usually below

the ground surface except during times of snow-melt, and

a P. australis grassland subsequently developed. Such

reclamation was widely applied in Japan and led to the

dominance of P. australis (Iwakuma 1996). After the

1970s, the demand for rice declined and the government

switched the policy from reclamation to restoration (Su-

giura et al. 2003). Subsequently, the reclamation plan of

Hotoke Swamp was abandoned in early 1970s. However,

local farmers perform prescribed fire and remove water by

pumps every spring to keep reclaimed land from the

abandonment. The annual prescribed fires usually burn

the entire area. An unplanned wildfire occurred on 1 May

2007 before the prescribed fire was performed, and thus

prescribed fire was not conducted in 2007 and 2008. Soon

after the wildfire, we explored the swamp and established

monitoring plots.

The study region is in a warm–cool temperature zone.

In the city of Misawa, 15 km from the study site, annual

precipitation was 1165 mm in 2007 and 1000 mm in 2008

(ADMO 2009). Mean annual temperature in 2007 was

10.4 1C with a monthly minimum of 0.8 1C in Jan and

maximum of 23.7 1C in Augu, and was 10.0 1C in 2008

(minimum =� 2.1 1C, maximum = 20.6 1C in August).

Snow-free period is usually from Apr to Dec.

Vegetation, biomass and micro-environmental

characteristics

Three fire severities were recognized: high (hereafter, H) –

completely consumed standing and fallen litter; moderate

(M) – completely removed vegetation and standing litter

but did not burn fallen litter because the fallen litter was

wet as a result of high water level; and low (L) –

incompletely removed standing litter and did not remove

fallen litter. There were no unburned sites for control

plots, because the fire burned the whole area, with the

same patterns of ordinal prescribed fire. Three locations

for each severity were selected for monitoring the devel-

opment of vegetation (i.e. three replicates in the same

burn area). Vegetation was measured three times, mid-

Augt 2007, mid-Jul 2008 and mid-Aug 2008, in 20 1 m

� 1 m permanent plots at each location. More than 5 m

separated each plot. The percentage cover of each species

was visually estimated. As Pilea mongolica withered earlier

in 2008 than in 2007 and a few other species did not

emerge in mid-Jul 2008 but they emerged by Augu, plant

cover measured in 2008 was averaged across the two

surveys for the analysis in order to compare more accu-

rately with the data collected in 2007.

Biomass was harvested from 3 to 10 Sept 2007 and

from 5 to 12 September 2008 when P. australis ceased

growing. The harvested biomass was separated into eight

categories: P. australis leaf, P. australis stem, forb leaf, forb

stem, grass leaf, grass stem, sedge and fern. The biomass

was classified into photosynthetic parts (i.e. leaves, in-

cluding the whole of sedges and ferns) and non-photo-

synthetic parts (non-photosynthetic organs, i.e. stems).

Standing litter was also collected and weighed. The

weight was measured after drying up at 80 1C in an oven

for more than 3 d.

Light intensity and temperature were recorded at 1-h

intervals from Mar to Sep in 2008 using light/temperature

loggers (HOBO; Onset, Bourne, MA, USA) placed in each

of the three severities. Two loggers were set up on the

surface and subsurface of litter in M and L, respectively,

and a logger was only on the soil surface in H because of

the lack of litter. The sensors placed on subsurface were

located directly above the soil surface and were more than

8 cm below the surface of the litter.

Seed bank

Seed bank sampling was conducted in early Nov 2007

(autumn) and mid-Jun 2008 (summer). A 100-cm3 steel

tin (20 cm2 surface area and 5 cm depth) was used for

sampling soil. Seed bank samples were collected sepa-

rately from the litter and the soil, adjacent to the L and H

plots except that litter was not collected in H in autumn

2007 because of the very small amount. The total number

of samples was 20 in H and 40 in L in autumn 2007, and

40 in H and L in summer 2008. The soil collections were

made directly underneath the litter where the litter

covered the soil. To consider the vertical movements of

seeds, litter on L and soil on H were analysed as surface,

and soil on L was analysed as subsurface.

Seed germination experiments were conducted in a

greenhouse at Hokkaido University, Japan. The samples

Kimura, H. and Tsuyuzaki, S. Effects of fire in early spring on wetland

Applied Vegetation Science

Doi: 10.1111/j.1654-109X.2011.01126.x r 2011 International Association for Vegetation Science 351

were cold-stratified for 1 month at 3 1C in a dark refrig-

erator within plastic bags which kept the moisture in the

sample. The samples were then spread to less than 5 mm

thick on plastic trays filled with vermiculite. The trays

were covered with white sheer nets to prevent contam-

ination. Temperature was between 10 1C and 35 1C under

natural light. Water was automatically sprinkled seven to

12 times in a day, depending on air moisture. Seedlings

were marked by numbered flags everyday until no more

seedlings emerged for more than 1 wk. When seedlings

were not identified, the seedlings were transplanted to

large trays filled with culture soil and grown. Seed

germination was observed for 3 months in each green-

house experiment.

Data analysis

The alpha diversities of the vegetation and seed bank were

evaluated by three parameters; species richness, Shan-

non–Wiener diversity (H0) and evenness (J0). H0 and J0 are

expressed as: H0 =�P

(pi� log2pi), and J0 = H0/log2S,

where S is total number of species in the vegetation plot

or seed bank sample, and pi is relative dominance of

species i from 1 to S.

Vegetation analysis was performed using a linear mixed-

effects model (LMM) because of repeated measures, and

analysis of the seed bank and standing biomass used a

generalized linear model (GLM). The dependent variables

were species richness, diversity, evenness, plot cover, P.

australis cover, total seed number, and seed number of each

dominant species. Species richness and seed number were

assumed to follow a Poisson probability distribution, diver-

sity and cover a Gaussian, and evenness a negative bino-

mial. In GLM, the explanatory variables were fire severity,

year and the interaction term between them. The best

models of GLM were selected by stepwise procedures of

Akaike’s information criteria (AIC). Plot and year were

assigned as random effects in LMM. Litter biomass was

compared between severity, year and their interaction by

GLM with a Gaussian error distribution. For seed bank, two

independent variables, substrate (litter and soil) and layer

(surface and subsurface), were also examined. Mean

monthly temperature, temperature difference and light

illumination were examined between severities and be-

tween months by LMM with months used as random

effects. The significance level was adjusted to 0.01 to reduce

the chance of Type I errors associated with multiple tests.

Species in seed bank and the standing vegetation were

summarized by growth form, longevity, and seed-dispersal

type. The growth form was divided into grass, forb, rush and

sedge and longevity was into annual, biennial and peren-

nial. Seed size was obtained for each species from the

literature (mainly Ohwi & Kitagawa 1983; Ishikawa 1995).

In H and L, respectively, similarities were calculated on

a matrix made by (seed bank� vegetation� season). Two

similarity indices, qualitative S�rensen’s index and quan-

titative Goodall’s percentage similarity index, were ap-

plied to plant cover and number of seeds with both

transformed into relative percentages (Egawa et al.

2009). Ferns, for which dominance in seed bank were

not estimated, and unidentified species were not used for

the evaluations. The presence and absence of each domi-

nant species in each plot and in each soil sample were

investigated based on AIC after GLM assuming a binomial

distribution. In the analysis, explanatory variables were

growth form, longevity, seed dispersal, seed volume and

seed shape. Seed volume was calculated by length and

width, assuming an oval. As seed shape was expressed as

length divided by width, the lower values indicated more

circular shapes. All statistical analyses were made with

the package R 2.9.0 (R Development Core Team, Vienna,

Austria.

Results

Vegetation structures and environments

A total of 30 species was recorded from the standing

vegetation (see the Supporting Information, Appendix

S1). Phragmites australis was predominant throughout,

and was the tallest (220 to 300 cm). There were no shrubs

or trees. Except for lower species diversity in L than in H,

species richness, diversity and evenness did not differ

between fire severities, and did not change across the

2 yr (Table 1). These results imply that diversities evalu-

ated by plant cover were modified less by severe fire. Fire

severity did not affect the cover of all common species,

including P. australis (LMM, P4 0.01). Standing litter

biomass in 2007 was 224 g m�2� 67 (mean� standard

deviation) in H, 294 g m�2� 52 in M, and 695 g m�2� 103

in L, and was significantly higher in L (GLM, Po0.01)

because the litter was burnt out in H and M. In 2008, the

litter biomass did not differ between fire severities (543 g

m�2� 62 in H, 636 g m�2� 55 in M, and 833 g m�2� 124

in L; P4 0.01), showing that litter recovered promptly if

fire was absent.

Mean monthly temperature was nearly zero in the

three fire severities in Feb because of snow cover, and

then gradually increased to Jul. From Jul to Sep, the

temperature was higher than 15 1C. Depth below litter

showed small temperature differences within a day,

although mean temperatures did not differ between sur-

face and subsurface for any fire severity. Therefore, the

major effect of litter was to reduce temperature fluctua-

tions. After snowmelt, light illumination was not different

between months, and was lower in the subsurface

throughout the year. As plot cover did not differ between

Effects of fire in early spring on wetland Kimura, H. and Tsuyuzaki, S.



severities (Table 1), shading of the subsurface soil was

caused by litter.

The above-ground biomass of P. australis was highest in M,

and decreased from 2007 to 2008 (GLM, Po0.01). Non-

Phragmites photosynthetic biomass, including sedges and

ferns, was lower in L in 2007 than in 2008 (Po0.01), but

did not differ between H and M. The non-Phragmites non-

photosynthetic biomass did not differ between severities.

Litter thickness was less in H and M than in L (Po 0.01).

Photosynthetic non-Phragmites biomass was influenced

more by litter amount than by P. australis biomass, suggesting

that other species increased promptly after litter removal by

fire but then declined in the second year.

Seed bank composition

In total, 4903 and 4967 seeds germinated in autumn 2007

and summer 2008, respectively, equivalent to 39 469 m�2

and 31 044 m�2 (see the Supporting Information, Appen-

dix S2). A total of 39 species were recorded from the seed

bank (13 annuals, three biennials, 22 perennials, and one

unidentified). No shrubs and trees were detected in the

seed bank. Seed bank species richness was higher in H

than in L (Table 2), but did not differ between autumn

2007 and summer 2008: i.e. the richness in H was

6.7� 1.7 (mean� standard deviation) in 2007 and

8.0� 1.8 in 2008, and in L was 5.7� 1.5 and 6.6� 1.6.

Species diversity and evenness did not differ between

severities (diversity; H = 1.9� 0.5 in 2007, 2.1� 0.4 in

2008, L = 1.5� 0.5 in 2007, 2.0� 0.4 in 2008; evenness;

H = 0.7� 0.2 in 2007, 0.7� 0.1 in 2008, L = 0.6� 0.2 in

2007, 0.7� 0.1 in 2008). Seed bank species diversity was

higher in summer 2008 than in autumn 2007, while

evenness did not differ between the years. As diversity

was determined by the combination of species richness

and evenness, both of which increased but were statisti-

cally non-significant, diversity increased significantly

over the 2 yr.

Table 1. Mean species richness, diversity and evenness in vegetation (n = 20 in each severity) received different fire damages in 2007 and 2008. Each

numeral shows mean with standard deviation. Each numeral shows mean with standard deviation. The significance is investigated by linear mixed-

effects model. Explanatory variables that are not adopted by Akaike’s Information Criteria are not shown. �significant at Po 0.01. NS = not significant.

The difference was compared from H to L and M, and from 2007 to 2008. On the species diversity of vegetation, the interaction was observed between

year and severity. Abbreviations: Y = year, H = high fire severity, M = moderate, and L = low. – = not measured.

2007 2008 Significance

H M L H M L Intercept Severity Year Interaction

M L Y�M Y� L

Species

richness

6.1� 1.6 4.7� 1.3 5.0� 1.5 8.2� 2.5 6.0� 1.8 6.0� 1.8 11.80� � 0.25NS � 0.20NS 10.30NS � 0.10NS � 0.05NS

Species

diversity

1.9� 0.3 1.7� 0.3 1.5� 0.3 2.0� 0.4 1.7� 0.3 1.7� 0.3 11.88� � 0.22NS � 0.39� 10.11NS 10.08NS � 0.05NS

Evenness 0.7� 0.1 0.8� 0.1 0.7� 0.1 0.7� 0.1 0.7� 0.1 0.7� 0.1 11.05NS 10.20NS � 0.28NS � 0.32NS 10.30NS � 0.13NS

Total plant

cover

72� 8 80� 13 60� 14 85� 12 80� 11 74� 12 193.2NS 143.9NS � 51.6NS 180.3NS � 78.7NS � 15.9NS

Table 2. Coefficients of general linear model (GLM) that compares species richness, diversity, evenness, total number of seeds, and numbers seeds on

fire frequently germinated species in seed bank. Stepwise Akaike’s Information Criteria (AIC) was used to select the best models, and explanatory

variables not selected by AIC are not shown. Year was a comparison between 2007 and 2008, layer was between the surface and subsurface, and

substrate was between litter and soil. � = significant at Po 0.01. NS = not significant.

Severity Year Layer Substrate Interaction

High (intercept) Less Year�Depth Severity�Depth

Species richness 11.911� � 0.180� 10.164NS

Species diversity 11.531� 10.011NS 10.566� � 0.062NS 10.379NS � 0.265NS � 0.352NS

Evenness 10.919NS 10.323NS

Species

Pilea mongolica 14.285� � 0.205� � 1.299� � 0.927�

Cardamine flexuosa 13.465� � 0.262� � 0.686� � 1.081� � 0.447� 10.956� 10.632�

Stellaria alsine var. undulata 11.245� 10.329� � 0.126NS � 1.991� 11.964�

Juncus effusus var. decipiens 13.121� � 0.338NS 10.406� � 0.380NS 10.524� 11.186�

Juncus leschenaultia 12.209� � 0.566� � 0.976� 11.801� � 0.599�

Total number of seeds germinated 14.927� � 0.310� � 0.776� 10.252� � 0.870� 10.603� 10.997NS




Seed density decreased from autumn 2007 to summer

2008, litter to soil, H to L, and subsurface to surface (Table

2). Five seed bank species were recorded from all seve-

rities for 2 yr: Pilea mongolica, Cardamine flexuosa, Juncus

effusus var. decipiens, Stellaria alsine var. undulata, and

Juncus leschenaultii. Most species that accumulated seeds

in litter decreased seed densities in the second year. Juncus

leschenaultii and J. effusus var. decipiens had higher seed

densities in subsurface than in surface soil. The responses

of seed density to fire severity varied between the com-

mon species; the seeds of P. mongolica, C. flexuosa and J.

leschenaultia were fewer in L, S. alsine var. undulata more

abundant and J. effusus var. decipiens unchanged. Positive

interactions were detected between depth and year for

three species – C. flexuosa, S. alsine var. undulata and J.

effusus var. decipiens –, and between severity and depth for

C. flexuosa and J. effusus var. decipiens. These interactions

indicated that the seeds moved to deeper soil layers with

time. However, P. mongolica, which produced the largest

seeds of the five common seed bank species (see the

Supporting Information, Appendix S3), developed more

of a seed bank in the litter in H (Table 2), but decreased in

seed density with time. The two Juncus species produced

small seeds, and accumulated seeds more at greater depth.

Relationships between seed bank and standing

vegetation

The total number of identified seed plant species in the

standing vegetation and seed bank was 47 (Appendix S3).

In addition, three ferns and three unidentified taxa were

recorded. Perennial forbs were most common (i.e. 30

species, followed by 14 annuals and three biennials). Of

these annual and biennial plants, only five species ap-

peared in the vegetation. Most annuals and biennials

including P. mongolica produced flowers in 2007.

Of the seed plants identified, 33 species (70% of the

total) had gravity dispersal (Appendix S3). However, six

and three of the 33 gravity-dispersed species also use

another dispersal agent, wind or animal, respectively.

Therefore, seed dispersal distances of these species would

be short if secondary seed dispersal by water functioned

weakly. Species occurring in both standing vegetation and

seed bank produced larger seeds (GLM, Po0.01). There

were no other significant factors identified by AIC in the

relationship between standing vegetation and seed bank.

Qualitative similarities in seed banks between the 2 yr

were more than 60% in both L and H, and quantitative

similarities ranged from 81 to 87% (Fig. 1), indicating that

species composition and relative dominance did not differ

greatly between years, or between severities. Qualitative

similarities between seed bank and the standing vegeta-

tion ranged from 30 to 55%, while quantitative simila-

rities between them were extremely low (i.e. less than

2%), except for 14.2% between standing vegetation and

seed bank in 2007 and 7.5% between standing vegetation

in 2007 and seed bank in 2008. These showed that H

developed more similar species composition between seed

bank and standing vegetation than L. In addition, seed

bank in H in 2008 was influenced more by the standing

vegetation in the previous year (2007), indicating that

seed bank development is affected more by severe fire

that removed litter.

Discussion

Litter reduced light intensity and temperature fluctuation

without changing mean temperature, while the standing

Seedbank

Vegetation

High severity Less severity

2007

2007

2008

200886.363.8

1.853.8

7.541.7

14.243.2

1.243.9

95.985.7

2007 2008

2007 2008

1.546.2

0.234.3

1.130.3

80.773.9

0.540.9

95.985.7

Fig. 1. Similarities (%) between seed bank and the standing vegetation in each season and between seed bank in autumn and summer. Upper and lower

numerals show quantitative and qualitative similarities, respectively. Quantitative similarity is by Goodall’s index, and quantitative one is by S�rensen’s

index. Thickness of lines connected with two circles indicates the intensity of similarities. Thick line 4 80% in quantitative similarity; medium line: 2% to

15%, thin line: o 2%.




biomass did not. Therefore, the major environmental

changes between fire severities were light intensity and

temperature fluctuation when litter was removed.

The severe fire did not eliminate perennials in the

standing vegetation. Most perennials that do not develop

a seed bank should survive and recover by vegetative

reproduction after a fire. For example, P. australis can

resprout and dominate from rhizomes after fire (Thomp-

son & Shay 1985), but it does not develop a persistent seed

bank (Lenssen et al. 1999; Egawa et al. 2009). After

disturbances, perennial grasses often invest more in vege-

tative reproduction than seed reproduction (Gonzalez &

Ghermandi 2008). In general, the plant cover of a species

that can recover by vegetative reproduction is unlikely to

be influenced by the severity of fire. Furthermore, species

diversity in the standing vegetation was higher in H and

M than in L. The higher diversity is related to the greater

photosynthetic biomass of non-Phragmites species that

increased because of litter removal. These results indicate

that species under a P. australis canopy contribute to high

species diversity when litter on the ground surface was

removed by fire.

Pilea mongolica, which produced the largest seed of the

dominant seed bank species, developed its seed bank

more in litter layer and/or on the surface than at greater

depth. However, its seed density declined in the second

year, probably because of seedling emergence in the first

year and difficulties in vertical movements owing to large

seeds. Seed sizes often determine vertical distribution of

seed bank; for example small seeds are distributed more at

greater depth (Tsuyuzaki & Goto 2001; Cerabolini et al.

2003). In 2007 P. mongolica showed the highest cover on

M and established well in H. The significant difference

between the first and second years after fire was caused by

litter accumulation, suggesting that litter removal by fire

has an important role on the seedling emergence and seed

bank dynamics of this species.

Both annuals and biennials producing small seeds

made persistent seed banks, and recovered quickly after

fire (Thompson et al. 1997; Gonzalez & Ghermandi 2008).

However, about two-thirds of the species in the seed bank

did not establish well in the vegetation of Hotoke Swamp.

In particular, two biennials, C. flexuosa and S. alsine var.

undulata, both of which require light for seed germination

(Baskin & Baskin 1998), were frequent in the seed bank

of both litter and soil for the 2 years but had low cover in

the standing vegetation. These results suggest that the

biennial species maintained persistent seed banks even

where the severity of fire was greatest. Rushes (Juncus)

and annuals, most of which require light for seed germi-

nation, as well as the two biennials, are also dominant in

the seed bank of various grasslands (Allen et al. 2008). In

addition, Juncus did not establish well in the vegetation,

even though the seed bank was well developed (Milberg

1995). The seeds of Juncus are particularly small, and are

likely to move downwards easily. Numerous small seeds

should preserve the viability in the soil by the downward

movements. Juncus leschenaultii and J. papillosus develop

long-term persistent seed banks with the vertical move-

ments of seeds (Jensen 2004). In these ways, high seed

density of these species should be maintained.

Similarities between seed bank and standing vegetation

are low in Hotoke Swamp. In wetlands where seed-

dispersal distance is short the similarities are high (Jutila

2003). Short-distance gravity seed-dispersal was common

in Hotoke Swamp, however, only four of 13 annual

species in the seed bank were recorded from the standing

vegetation. Furthermore, species producing small seeds

such as Juncus did not appear in the vegetation but did in

seed bank. Low similarities between seed bank and the

vegetation are often derived from the predominance of

Juncus in seed banks that do not emerge in the vegetation

(Parker & Leck 1985). These inconsistencies should lead

to low similarities on Hotoke Swamp.

Disturbances increase the similarities between seed

bank and the standing vegetation, because the resultant

sparse vegetation and litter reduce competition for light

and promote seed germination (Osem et al. 2006), and/or

seedling emergence is promoted by disturbances (Grandin

2001). High severity showed the greatest similarity be-

tween seed bank and the standing vegetation in the first

year, while similarities between them became quite low

in low severity for the 2 yr. Fire that removes litter

completely in early spring promotes the link between

seed bank and the standing vegetation. The most severe

fire treatment, H, showed higher species diversity and

seedling emergence, indicating that the fire was not

catastrophic even when litter was completely removed.

For successful conservation and restoration, suitable ger-

mination and establishment conditions should be pro-

vided for seed banks (Leck & Sch +utz 2005). In conclusion,

high-severity prescribed fire is adequate for maintaining

diverse vegetation structure and high diversity via devel-

oping seed bank that contributes vegetation dynamics.

Acknowledgements

We thank all members in Plant Ecology Laboratory and S.

Suzuki for support, staff members of Misawa City Office

for permission and support, and F. Kobari in CAST for

technical help. Cordial thanks also J. Pausas, D. Keith,

J.H. Titus and two anonymous reviewers for their critical

reading of the manuscript. This work is partly supported

by grants from JSPS.




References

ADMO (Aomori District Meteorological Observatory) 2009.

Available at: http://www.data.jma.go.jp/obd/stats/etrn/

index.php?prec_no=31&prec_ch=%90%C2%90X%8C%

A7&block_no=0169&block_ch=%8EO%91%F2&year=&

month=&day=&view= (accessed 2 March 2011)

Allen, E.A., Chambers, J.C. & Nowak, R.S. 2008. Effects of a

spring prescribed burn on the soil seed bank in sagebrush

steppe exhibiting pinyon-juniper expansion. Western North

American Naturalist 68: 265–277.

Baskin, C.C. & Baskin, J.M. 1998. Seeds: ecology, biogeography,

and evolution of dormancy and germination. pp. 331–458.

Academic Press, San Diego, CA, US.

Cerabolini, B., Ceriani, R.M, Caccianiga, M., De Andreis, R. &

Raimondi, B. 2003. Seed size, shape and persistence in soil:

a test on Italian flora from Alps to Mediterranean coasts.

Seed Science Research 13: 75–85.

Cox, R.D. & Allen, E.B. 2008. Composition of soil seed banks in

southern California coastal sage scrub and adjacent exotic

grassland. Plant Ecology 198: 37–46.

Dobson, A.J.J. 1990. An introduction to generalized linear models.

Chapman and Hall, New York, NY, US.

Egawa, C., Koyama, A. & Tsuyuzaki, S. 2009. Relationships

between the developments of seedbank, standing vegetation

and litter in a post-mined peatland. Plant Ecology 203: 217–228.

Gonzalez, S. & Ghermandi, L. 2008. Postfire seed bank dynamics

in semi-arid grassland. Plant Ecology 199: 175–185.

Grandin, U. 2001. Short term and long-term variation in seed

bank/vegetation relations along an environmental and

successional gradient. Ecography 24: 731–741.

Hiers, J.K., Laine, S.C., Bachant, J.J., Furman, J.H., Greene, Jr.,

W.W. & Compton, V. 2003. Simple spatial modeling tool for

prioritizing prescribed burning activities at the landscape scale.

Conservation Biology 17: 1571–1578.

Hochkirch, A. & Adorf, F. 2007. Effects of prescribed burning

and wildfires on Orthoptera in Central European peat bogs.

Environmental Conservation 34: 225–235.

Hopfensperger, K.N. 2007. A review of similarity between seed

bank and standing vegetation across ecosystems. Oikos 115:

1438–1448.

Ishikawa, S. 1995. Seeds and fruits of Japan. 2nd ed. Shigeo

Ishikawa Picture Book Publication Council, Tokyo, JP. (in

Japanese).

Iwakuma, T. (ed.) 1996. Mires of Japan. Ecosystems and moni-

toring of Miyatoko, Akaiyachi and Kushiro mires. National

Institute for Environmental Studies, Tsukuba, JP.

Jensen, K. 2004. Dormancy patterns, germination ecology, and

seed bank types of twenty temperate fen grassland species.

Wetlands 24: 152–166.

Johnson, E.D. 1992. Fire and vegetation dynamics: studies from the

North American boreal forest. Cambridge University Press,

Cambridge, UK.

Jutila, H.M. 2003. Germination in Baltic coastal wetland

meadows: similarities and differences between vegetation

and seed bank. Plant Ecology 166: 275–293.

Keith, D.A. & Bradstock, R.A. 1994. Fire and competition in

Australian heath: a conceptual model and field investi-

gations. Journal of Vegetation Science 5: 347–354.

Leck, M.A. & Sch +utz, W. 2005. Regeneration of Cyperaceae,

with particular reference to seed ecology and seed banks.

Perspectives in Plant Ecology, Evolution and Systematics 7:

95–133.

Lenssen, J., Menting, F., van der Putten, W. & Blom, K. 1999.

Control of plant species richness and zonation of functional

groups along a freshwater flooding gradient. Oikos 86:

523–534.

Mamede, M.D. & de Araujo, F.S. 2008. Effects of slash and

burn practices on a soil seed bank of Caatinga vegetation

in Northestern Brazil. Journal of Arid Environments 72:

458–470.

Milberg, P. 1995. Soil seed bank after eighteen years of

succession from grassland to forest. Oikos 72: 3–13.

Ohwi, J. & Kitagawa, M. 1983. New flora of Japan. Shibundo,

Tokyo, JP.

Osem, Y., Avi, P. & Jaime, K. 2006. Similarity between seed

bank and vegetation in a semi-arid annual plant

community: the role of productivity and grazing. Journal

of Vegetation Science 17: 29–36.

Owens, A.B, Proffitt, C.E. & Grace, J.B. 2007. Prescribed fire

and cutting as tools for reducing woody plant succession in

a created salt marsh. Wetlands Ecology and Management 15:

405–416.

Parker, V.T. & Leck, M.A. 1985. Relationships of seed banks to

plant distribution patterns in a freshwater tidal wetland.

American Journal of Botany 72: 161–174.

Parker, V.T. & Kelly, V.R. 1989. Seed banks in California chaparral

and other Mediterranean climate shrublands. In: Leck, M.A.,

Paker, V.T. & Simpson, R.T. (eds.) Ecology of seed banks. pp.

232–255. Academic Press, San Diego, CA, US.

Pausas, J.G. & Keeley, J.E. 2009. A burning story: the role of

fire in the history of life. BioScience 59: 593–601.

Sugiura, T., Kobayashi, H., Tsutsumi, S. & Tsushina, Y. 2003.

The establishment of environmental education farm at

Hotokenuma reclaimed land in Misawa, Aomori. Journal

of the Agricultural Engineering Society of Japan 71: 825–828

(in Japanese).

Thompson, D.J. & Shay, J.M. 1985. The effects of fire on

Phragmites australis in the Delta Marsh, Manitoba.

Canadian Journal of Botany 63: 1864–1869.

Thompson, K., Bakker, J. & Bekker, R. 1997. The soil seed banks

of North West Europe: methodology, density, and longevity.

Cambridge University Press, Cambridge, UK.

Tsuyuzaki, S. & Goto, M. 2001. Persistence of seed bank under

thick volcanic deposits twenty years after eruptions of

Mount Usu, Hokkaido Island, Japan. American Journal of

Botany 88: 1813–1817.




http://www.data.jma.go.jp/obd/stats/etrn/index.php?prec_no=31&prec_ch=%90%C2%90X%8C%A7&block_no=0169&block_ch=%8EO%91%F2&year=&month=&day=&view=










Supporting Information

Additional Supporting Information may be found in the

online version of this article:

Appendix S1. Percentage cover on each taxa in the

summers of 2007 and 2008.

Appendix S2. Seed density (m�2) in soil and litter

substrates on heavily and less disturbed sites.

Appendix S3. Growth form, longevity, seed disper-

sal type and seed size on identified seed plants in vegeta-

tion and seed bank.

Please note: Wiley-Blackwell is not responsible for

the content or functionality of any supporting materials

supplied by the authors. Any queries (other than missing

material) should be directed to the corresponding author

for the article.




fire severity affects vegetation and seed bank in a wetland

Documents