reducing the germinable weed seedbank with soil disturbance and cover crops

DOI: 10.1111/j.1365-3180.2010.00792.x

Reducing the germinable weed seedbank with soildisturbance and cover crops

S B MIRSKY*, E R GALLANDT�, D A MORTENSEN�, W S CURRAN� &D L SHUMWAY§*Sustainable Agricultural Systems Laboratory, USDA, Agricultural Research Service, Beltsville, MD, USA, �Department of Plant, Soil and

Environmental Sciences, University of Maine, Orono, ME, USA, �Department of Crop and Soil Sciences, The Pennsylvania State University,

University Park, PA, USA, and §Department of Statistics, The Pennsylvania State University, University Park, PA, USA

Received 26 May 2009

Revised version accepted 24 February 2010

Summary

Ecologically based weed management relies heavily on a

greater integration of cultural and mechanical control

tactics than conventional weed management. As such,

management outcomes are more dependent on biotic

interactions among the crop, weed and methods of

management. In this study, we assessed the influence of

soil disturbance and cover cropping on the germinable

seedbank of Chenopodium album, Abutilon theophrasti

and Setaria spp. across a range of initial weed seedbank

densities. Cover crop system treatments ranged from

greater reliance on tillage at one extreme, to a greater

reliance on cover cropping to accomplish weed suppres-

sion at the other. In general, seedbank losses to

germination were highest when soil disturbance was

associated with cover cropping. Of the five cover crop

systems (1 year duration), the summer fallow and the

yellow mustard ⁄buckwheat ⁄winter rape mixture consis-

tently decreased the weed seedbank. The magnitude of

decline in these high disturbance systems reached

complete depletion of the Setaria spp. germinable

seedbank and an 85% and 80% reduction for C. album

and A. theophrasti respectively. Both systems included

tilling the soil three to four times throughout the

growing season. In contrast, treatments which involved

less soil disturbance had weed escapes that set seed in

one or more site years. Cover crop systems that

stimulated weed seed germination and where weeds

were either suppressed by the cover crop or where

subsequent tillage provided control, resulted in the

greatest weed seedbank declines.

Keywords: cover crop, seedbank, sustainable agricul-

ture, population dynamics, mechanical weed control,

Chenopodium album, Abutilon theophrasti, Setaria.

MIRSKY SB, GALLANDT ER, MORTENSEN DA, CURRAN WS & SHUMWAY DL (2010). Reducing the germinable weed

seedbank with soil disturbance and cover crops. Weed Research.

Introduction

Weed management outcomes are linked to infestation

size, as larger resident weed seedbanks often require

more weed control to adequately manage the population

(Hartzler & Roth, 1993). Ecologically based weed

management (EBWM) strategies rely on an integrated

suite of control tactics to achieve an acceptable level of

weed suppression (Liebman & Gallandt, 1997). Such

EBWM strategies are used by growers interested in

reducing reliance on herbicides and are central to

organic production. Cropping systems that rely less on

herbicides, have a greater emphasis on cultural (crop

rotation diversification and cover cropping) and

mechanical weed control practices (e.g., cultivation).

However, the efficacy of these practices is more variable

than chemical weed management (Ryan, 2007). For

example, EBWM that relies on timely tillage operations

can be compromised when fields are too wet to till

(Ryan, 2007) with a resulting increase in the weed

Correspondence: Steven B Mirsky, Sustainable Agricultural Systems Laboratory, USDA, Agricultural Research Service, 10300 Baltimore Avenue,

Bldg. 001, Rm. 117, BARC-W, Beltsville, MD 20705, USA. Tel: (+01) 301 5045324; Fax: (+01) 301 5048370; E-mail: [email protected]

� Published 2010. This article is a US Government work and is in the public domain in the USA

Journal Compilation � 2010 European Weed Research Society Weed Research

seedbank (Forcella & Lindstrom, 1988). It is therefore

critical to design and assess new tactics that reduce this

variation in performance.

The size of the weed seedbank strongly influences the

success of anyweedmanagement practice and is critical to

the success of EBWM (Forcella et al., 1993). Jordan

(1996) suggests that EBWM �must be buttressed by efforts

to reduce weed seedbanks and maintain them at low

levels�. Ecologically based strategies for manipulating

weed population and community dynamics include:

(i) reducing the seedbank pool size by stimulating seed

death, (ii) stimulating germination, (iii) limitingweed seed

production of emerged plants or (iv) completely removing

above-ground weed biomass (Swanton & Booth, 2004).

The germinable fraction of the seedbank has the most

substantial impact on weed seedbank flux, with the

potential to account for more than half of seedbank

losses annually (Gallandt et al., 2004). Therefore, EBWM

that targets the germinable weed seedbank and prevents

seed production will likely have the greatest impact on

seedbank size. Gallandt (2006) suggests an effective way

to achieve these reductions is through tillage-induced

germination and subsequent control.

Germination of many species is stimulated by soil

disturbance (Cavers & Benoit, 1989); in this way the

expressed germinable fraction can be increased. Poten-

tial mechanisms for this germination response include:

exposure to light for light-sensitive species (Gallagher &

Cardina, 1998); temperature fluctuations (Thompson &

Grime, 1983), and movement of seed to favourable

germination sites (Milberg et al., 1996). Tillage also

increases microbial biomass through enhanced soil

aeration and by exposing occluded labile soil organic

matter fractions (Plante & McGill, 2002). The increase

in microbial activity can also increase biosynthesis of

ethylene (Hilhorst & Darssen, 2000), which is a known

germination stimulant (Schonbeck & Egley, 1980).

Synchrony of management tactics and germination

periodicity can be an effective way of �expressing� andthen controlling emerged plants. Growers using EBWM

have relied on weed expressive tactics, such as the stale-

seedbed approach before planting where germination

stimulation of weeds are used to lower weed seedbanks

(Lonsbary et al., 2003). However, because the germina-

tion period is species-specific, the timing of EBWM can

result in weed community changes (Booth & Swanton,

2002). Identifying weed management practices that

consistently reduce weed seedbanks across a range of

weed life history strategies (i.e., broadleaved vs. grasses;

annuals vs. perennials; early vs. late emerging weeds)

will further the adoption of EBWM. Additionally,

coupling cover crop use with �expressive tactics� couldoffset the environmental downsides of excessive tillage as

cover crop residues reduce weed emergence and early

growth (Mohler, 2001; Reberg-Horton et al., 2005) and

can help build soil carbon content (Mirsky et al., 2008).

This study assessed whether the weed seedbank could

be depleted by a range of cover crop and tillage treatments

arranged in a system representative of vegetable cropping

practices in north-eastern North America. We hypothes-

ised that: (i) the magnitude of decline in the germinable

weed seedbank will increase with increasing frequency of

soil disturbance and cover cropping, (ii) the frequency of

soil disturbance and cover cropping must increase as the

density of the weed seedbank increases in order to

preserve a constant level of weed suppression and (iii)

consistency in weed decline across geographic regions and

weed life history traits will increase with increasing soil

disturbance and cover crop system intensity.

Methods

Field experiments were conducted from 2003 to 2006 at

the Rogers Farm in Old Town, Maine (ME) (44�55¢N,

68�41¢W) and the Russell E. Larson Agricultural

Research Center in Rock Springs, Pennsylvania (PA)

(40�44¢N, 77�57¢W) to evaluate the influence of soil

disturbance and cover cropping practices on the ger-

minable weed seedbank for three summer annual weeds.

The experiment was initiated in the fall of 2003 and

repeated in 2004 in PA and winter of 2004 and 2005 in

ME, in adjacent fields at the two locations. Site years

will be referenced by the year that the cover cropping

systems received management that included soil distur-

bance (2004 and 2005). The study was conducted on a

Hagerstown silt loam soil (Fine, mixed, mesic, Typic

Hapludalfs; 12% sand, 70% silt, 18% clay) in PA and a

Buxton silt loam soil (Fine, illitic, frigid Aquic Dystric

Eutrudepts; 11% sand, 47% silt, 42% clay) in ME.

The experiment was arranged as a split-plot rando-

mised complete block with four replicates. Main plot

treatments were five cover crop systems representing

different crops, levels of soil disturbance, duration of

cover and quantity and quality of cover crop biomass

(Fig. 1). Subplots were supplemented with a mixture of

weed seeds at a range of weed seedbank densities per

species (four factor levels). An experimental replicate

consisted of five main plots (12.2 · 12.2 m), and five

weed seedbank density sub-plots (2.7 · 1.5 m).

Cover crop systems

Cover crop treatments were established in the spring of

2004 and 2005 in adjacent fields. In PA, the previous

crops were field maize and soyabean and in ME, fallow

and maize silage for 2004 and 2005, respectively. The

timing and number of field operations for each cover

crop system are provided in Fig. 1. The five cover crop

2 S B Mirsky et al.



treatments included: (i) summer fallow where no crop

was grown (SF); (ii) an oat (Avena sativa L. �Ogle�) cashcrop followed by rye (Secale cereale L. �Aroos-

took�) ⁄hairy vetch (Vicia villosa Roth.) in PA (O-R ⁄HV) or a green bean (Phaseolus vulgaris L. �Provider�)cash crop followed by rye ⁄hairy vetch in ME (GB-

R ⁄HV); (iii) oat ⁄pea (Pisum sativum L. �Maxum�)followed by rye ⁄hairy vetch (O ⁄P-R ⁄HV); (iv) oats ⁄ redclover (Trifolium pratense L. �Mammoth�) (O ⁄RC); and

(v) yellow mustard (Brassica juncea L. �Idagold�) fol-

lowed by buckwheat (Fagopyrum esculentum Moench.)

followed by winter rape (Brassica napus L. �Dwarf

Essex�) (M-BW-C).

In PA, fields were disked and culti-mulched twice to

ensure a smooth seedbed prior to establishing the

seedbank density sub-plot. In ME, primary tillage con-

sisted of chisel ploughing followed by two passes with a

soil conditioner (Perfecta Harrow, Unverferth Manufac-

turing, 18107 U.S. 224 West P.O. Box 357, Kalida, OH

45853). Prior to establishing the cover crop treatments in

the spring, primary tillage was completed in ME with a

Perfecta II s-tine cultivator (2·), while in PA, fields were

tilled with a cultimulcher (2·). All legume cover crops

were inoculated with appropriate strains of Rhizobium

spp. prior to planting. Crops were seeded in 19 cm spaced

rows with either a Great Plains drill (1006NT) in PA or a

Massy Harris drill in ME, or for buckwheat and mustard

species with a Brillion (�SS-Solid Stand�; 1.5 m width)

seeder (both locations) in 10 cm rows. The green bean

cash crop was planted in 76 cm rows.

The oat cash crop was treated with 0.56 kg a.e ha)1

MCPA amine, post-emergence about 6 weeks after

seeding for control of annual broad-leaved weeds, while

the green bean cash crop was managed without herbi-

cides relying on cultivation for weed control. The oats

and green beans were machine harvested for grain or

fresh market produce, respectively in early August. The

oat cover crop within the O ⁄RC system was terminated

by mowing the oats to a height of 15 cm (c. 1 cm above

the red clover canopy). Cover crops were flail or rotary

mowed prior to soil incorporation in the O-R ⁄HV,

O ⁄P-R ⁄HV, O ⁄RC and M-BW-C systems. For treat-

ments with more than one crop grown in a season, the

preceding cover crop was flail or rotary mowed or

the cash crop harvested (oats and green bean) and the

residue incorporated and the soil was then prepared for

the subsequent crop using a rotary cultivator. In PA,

plots were additionally cultimulched (2·) prior to

seeding. The direction of travel during each tillage

operation was alternated to minimise the movement of

weed seed out of the density plots. Above-ground cover

crop and weed biomass were measured prior to each

termination event by clipping all above-ground plant

material from two 0.25 m2 quadrats in each plot. The

biomass was dried at 50�C for 1 week prior to weighing.

Biomass was reported for the spring and summer sown

cover crops.

Supplemented weed seedbank

Weed seedbanks were supplemented in the late fall (PA),

or early spring (Maine). Prior to planting the cover crops,

weed seeds (collected in nearby fields in the autumn prior

to sowing) were hand applied at PA and then lightly

incorporated using a leaf rake or tine weeder (<3 cm

deep), while in Maine weed seeds were frost-seeded in

early spring. The following commonly occurring weeds

were selected for the study:Setaria faberiL. (giant foxtail)

in PA or Setaria glauca (L.) Beauv. (yellow foxtail) in

Maine, and Chenopodium album L. (fat-hen or common

lambsquarters) and Abutilon theophrasti Medik. (velvet-

leaf) at both sites. Seed additions of 0 (control), 60 (low),

450 (medium), and 2100 (high) seeds m)2 of each species

were applied for a combined seedbank supplementation

of 0, 180, 1350, 6300 seeds m)2. To assess the efficiency of

weed seed recovery, 2100 ceramic beads m)2 (Macrolite

0.06–0.02 sphere diameter, ML1430; Freeman Manufac-

turing and Supply, 1101 Moore Road, Avon, OH 44011)

were added to the high weed seedbank density plot

(Mohler et al., 2006).

Quantifying changes in the weed seedbank

The initial and final weed seedbank densities were

enumerated at both locations in the spring of the estab-

lishment year andagain in the springof the following year,

using glasshouse germination and direct extraction meth-

ods (Gross, 1990; Gallandt et al., 1998). Glasshouse

germination was used to estimate the readily germinable

(non-dormant to partially dormant) fraction of the

Primary tillage Mowing Herbicide

Secondary tillage Living cover/cash crop

J F M AA M J J A S O N D

Month of the year

O/RC

SF

O-R/HV

GB-R/HV

M-BW-C

O/P-R/HV

Fig. 1 Aschematic depicting the disturbance events andduration of

living cover in the cropping systems: SF, summer fallow;O-R ⁄HV,oats

green bean-rye ⁄hairy vetch;M-BW-C, yellowmustard-buckwheat-

rape;O ⁄P-R ⁄HV, oats ⁄pea-rye ⁄hairy vetch;O ⁄RC, oats ⁄ red clover.

Soil disturbance, cover crops and seedbanks 3



https://www.researchgate.net/publication/232683232_How_can_we_target_the_weed_seedbank?el=1_x_8&enrichId=rgreq-84bfca8c-02c1-465e-b7bf-515f01f78793&enrichSource=Y292ZXJQYWdlOzIyOTk4MzEyNDtBUzoxMDMwMzkyNTU5NzM4ODlAMTQwMTU3Nzg4MTExNw==

seedbank from the density sub-plots (Gallandt et al.,

1998).Ten (8.3 cmdiameterby10 cmdeep) soil coreswere

collected from control, low and medium density subplots

and 15 cores from the high density sub-plots. Soil cores

were randomly collected from half (2.7 · 0.75 m) of each

density sub-plot, while recruitment data were taken from

the other half of the sub-plot. Direct extraction of Setaria

spp., A. theophrasti and the ceramic bead seed surrogates

was accomplished using a modified hydropneumatic root

elutriator (Gillison�s Variety Fabrication) (Gross, 1990).

This procedurewas performed on soil taken from the high

density subplots, to check against the results of the

glasshouse assay. Weed seed and organic residues were

separated fromthe soil byfirst running the sample through

the elutriator, then seeds were sorted under a dissecting

microscope. Seed viability was estimated by applying

pressure to seeds with forceps; firm seeds were assumed to

be viable (Andersson &Milberg, 1998).

Recruitment was measured in the field prior to each

field disturbance between May and August. Seedling

emergence was quantified by placing a 0.5 m2 quadrat

within each density sub-plot and counting the target

weed species, including other dominant weed species

present in the field. Although fecundity of mature weeds

in the different cover crop systems was not quantified,

mature weed populations capable of introducing seeds

were documented.

Statistical methods

The relationship between the two seedbank enumeration

methods was determined using correlation analysis.

Analysis of variance (ANOVA) was conducted using a

Proc Mixed procedure in SAS v. 9.1 (SAS Institute Inc.,

2004) to test for variations in the initial germinable

seedbank at both locations and site years. Preliminary

analyses showed site year to be significant at both

locations; therefore, the change in seedbank, hereafter

referred to as D germinable seedbank, was evaluated

across cover crop systems by location and site year using

analysis of covariance (ANCOVA), with initial seedbank

density as a covariate. Annual seedbank change can be

examined using proportional change (final ⁄ initial) or bycomputing the difference. The latter method was

selected for the population dynamics analysis, because

of zero initial weed seedbank counts in some sub-plots.

ANCOVA was performed using Proc Mixed in SAS

version 9.1. The ANCOVA output included a linear

expression of the relationship between initial germinable

seedbank and D germinable seedbank (slope and inter-

cept). ANOVA was also used to test the effects of site year,

location and cover crop system on cumulative weed

seedling recruitment. All mean comparisons were per-

formed using the Tukey–Kramer method (P < 0.05).

The influence of the cover crop treatment on weed

community composition was examined using redun-

dancy analysis (RDA) with CANOCO 4.5 (ter braak

CJFAS, 2002; Reberg-Horton et al., 2006). A con-

strained form of principle component analysis, RDA is

similar to principle component analysis, in that RDA

identifies optimal linear models describing community

structure and predicts species response. Regression

coefficients in RDA were constrained by factors in the

experimental design, as well as environmental gradients,

which allowed for hypothesis testing by partitioning

treatment effects into individual factors (Reberg-Horton

et al., 2006). RDA was performed on the proportional

change (final ⁄ initial seedbank) of the three target weed

species by constraining the weed community with the

following explanatory variables: location, cover crop

systems, site year and the disturbance gradient. This

analysis was only performed on high density plots,

because of zero seedbank counts in low density treat-

ments. A Monte Carlo permutation test (499 permuta-

tions) was used to test for a significant community

gradient. The significance of the community gradient

was determined by calculating the proportion of

F-statistics generated from the permutated environmen-

tal matrices that were greater than the F-statistic

calculated from the empirical species-environmental

matrix. The environmental variables that best predict

the species response was then determined by using a

forward selection procedure on the environmental vari-

ables. Biplots, developed in CanoDraw (graphical

package bundled with CANOCO 4.5) were then used

to visualise the degree of association of the target weed

species and explanatory variables. Those visualisations

were supported by a complimentary permutational

multivariate analysis of variance, PERMANOVA

(Anderson, 2001), to account for treatment effects and

interactions. PERMANOVA, a permutation method, simul-

taneously tests for responses of multiple variables (i.e.,

weed species) on multiple treatment factors (i.e., cover

crop systems, site year and location) based on distance

measures, retaining the structure of the experimental

design (Anderson, 2001). A single distance measure is

computed from community signatures, distance mea-

sures of community composition are then compared

between samples, and an F-statistic was calculated as

mentioned above.

Results

Cover and cash crop biomass

The cover crop treatments (with the exception of SF)

had living cover present for much of the experimental

period. Cover crop systems typically produced biomass

4 S B Mirsky et al.



levels in excess of 3000 kg ha)1 during the 12 month

experimental period. The following cover crop biomass

levels are reported by site and cropping system and

pooled over site year. The O ⁄P treatment produced the

greatest cover crop biomass, averaging 5200 and

8000 kg ha)1 in Maine and PA, respectively. The

O ⁄RC and M-BW-C treatments produced 2700 and

3800 kg ha)1 of cover crop biomass in Maine and PA,

respectively. It is likely that the cooler, shorter growing

season in Maine (Fig. 2) accounts for the lower biomass

production. Setaria glauca reached reproductive matu-

rity in 2004 in Maine in the O ⁄P-R ⁄HV and O ⁄RC

systems, while in PA, S. faberi set seed in both site years

in O ⁄RC and in 2004 in O-R ⁄HV. Some C. album

produced seed in both site years in Maine�s O ⁄P-R ⁄HV

system and the 2005 field site in PA�s O ⁄RC. Abutilon

theophrasti did not reach maturity in any of the cover

crop systems at either location.

Seedbank quantification

Ceramic bead recovery efficiency was consistent across

cover crop treatments at both study sites (data not

shown). A significant positive linear relationship was

observed for Setaria spp. and A. theophrasti seedbanks

between the glasshouse germination assay and direct

extraction P < 0.001; r = 0.692). The glasshouse assay

was selected as the primary means for quantifying Dgerminable seedbank and is the basis for the results

reported in this paper, because it showed a level of

precision similar to direct extraction and was conducted

on the full range of weed seedbank density treatments.

Population dynamics

The slope for the relationship between D germinable

seedbank and initial seedbank was significant for most

cover cropping systems. Here, slope reflects the weed

suppressive effects of a cover cropping treatment

(Tables 1 and 2) and represents the change in the decline

of the weed seedbank as a function of the initial weed

seedbank. For example, SF, GB-R ⁄HV and B ⁄BW ⁄Bconsistently resulted in reduced Setaria spp. seedbanks.

Due to a significant result in the ANCOVA test for equal

slopes, comparisons of fitted values of the D germinable

weed seedbank at three points over the range of initial

germinable seedbank were conducted for S. faberi,

C. album and A. theophrasti (Tables 3 and 4). The rate

of seedbank decline (slope) was unaffected by initial

weed seedbank density.

The magnitude of weed seedbank change varied

across cover crop treatments and weed species (Tables 1

and 2). Weeds emerging in less suppressive cover crop

treatments often produced seed resulting in net seedbank

increases. There was complete depletion of the germin-

able Setaria spp. seedbank in the M-BW-C and SF

treatments at both site years in Maine and the 2004 field

site in PA. The C. album seedbank declined by 85% over

the same period in the three high disturbance systems

(SF, GB-R ⁄HV or O-R ⁄HV and M-BW-C) across all

site years and locations. The A. theophrasti seedbank

consistently declined in all systems, but the SF and

M-BW-C system had the highest rates of decline

(c. 80%). The O ⁄P-R ⁄HV, O ⁄RC (the two low distur-

bance treatments) and the cash crop systems still

resulted in high rates of decline in the germinable

seedbank, averaging about a 73% reduction.

Setaria spp. increased in O ⁄RC in the 2004 field site

in Maine, in both site years in PA, and also increased in

O ⁄P-R ⁄HV in Maine in 2004 (Tables 3 and 4). Che-

nopodium album increased in the O ⁄P-R ⁄HV in both site

years in Maine and in O ⁄RC in PA (2005). In Maine, it

is likely that increases in C. album and S. glauca weed

seedbanks in O ⁄P-R ⁄HV were the result of a poor cover

0

100

200

300

Prec

ipita

tion

(mm

)

–15

0

15

Sept

embe

r

Nov

embe

r

Janu

ary

Mar

ch

May

July

Sept

embe

r

Nov

embe

r

Janu

ary

Mar

ch

May

July

Sept

embe

r

Nov

embe

r

Janu

ary

Mar

ch

Tem

pera

ture

(°C

)

Fig. 2 Mean monthly precipitation and air

temperature from 2003 to 2006 for

Pennsylvania and Maine.




crop stand during this exceptionally wet field season

(Fig. 2). In PA, Setaria spp. populations were high when

the oat cash crop establishment was less than ideal,

resulting in stand gaps that allowed S. faberi establish-

ment and reproductive success (Table 4).

Proportional decline in Setaria spp. seedbanks was

highest in high disturbance systems at medium and high

initial seedbank densities (Tables 3 and 4). Contrary to

the cash crop system in Maine (green beans), the PA

cash crop system had lower rates of S. faberi seedbank

decline or actual increases to the seedbank as was the

case in 2004 (Table 4). The C. album seedbank was

reduced by all cropping systems, except in O ⁄P-R ⁄HV

in Maine (both site years) (Table 3) and O ⁄RC in PA

(2004) (Table 4). In PA, SF typically had the greatest

seedbank decline (2004). The O ⁄RC system generally

reduced the C. album seedbank population less at the

medium to high weed densities compared to the other

Table 1 Coefficients for the linear regression of Setaria spp., Chenopodium album and Abutilon theophrasti D germinable seedbank as a

function of initial germinable seedbanks across the cropping systems in Maine

Weed species

Cropping

system

2004 2005

ß0 ß1 P > T R2 ß0 ß1 P > T R2

Setaria spp. SF 2.21 (22.90) )1.02 (0.24) <0.0001 0.99 2.28 (4.70) )0.98 (0.02) <0.0001 0.99

GB-R ⁄ HV )7.34 (22.37) )0.47 (0.29) 0.1092 0.20 5.84 (4.75) )1.00 (0.02) <0.0001 0.98

M-BW-C 7.14 (23.97) )1.02 (0.39) 0.0093 0.85 )0.03 (4.53) )1.00 (0.01) <0.0001 0.90

O ⁄ P-R ⁄ HV 54.14 (22.91) 0.01 (0.30) 0.9686 0.03 6.55 (4.59) )0.96 (0.02) <0.0001 0.92

O ⁄ RC 38.18 (22.86) 0.31 (0.32) 0.3342 0.07 21.06 (4.44) )0.86 (0.01) <0.0001 0.96

C. album SF 1328.75 (1404.76) )0.83 (0.31) 0.0080 0.56 393.73 (751.68) )0.91 (0.16) <0.0001 0.89

GB-R ⁄ HV 892.45 (1051.79) )0.71 (0.23) 0.0024 0.68 )55.15 (644.55) )0.80 (0.15) <0.0001 0.96

M-BW-C 750.43 (1068.95) )0.73 (0.22) 0.0012 0.83 )174.64 (630.29) )0.78 (0.15) <0.0001 0.96

O ⁄ P-R ⁄ HV )148.64 (1275.88) 1.99 (0.26) <0.0001 0.37 5155.80 (588.36) )0.94 (0.11) <0.0001 0.28

O ⁄ RC 1906.96 (1250.49) )0.57 (0.19) 0.0025 0.24 1411.69 (576.19) )0.91 (0.13) <0.0001 0.83

A. theophrasti SF 2.05 (6.72) )0.87 (0.13) <0.0001 0.89 16.95 (10.46) )0.95 (0.04) <0.0001 0.92

GB-R ⁄ HV 2.17 (6.66) )0.74 (0.15) <0.0001 0.45 8.15 (10.86) )0.81 (0.05) <0.0001 0.90

M-BW-C 4.99 (6.83) )0.56 (0.17) 0.0015 0.32 7.23 (9.87) )0.94 (0.03) <0.0001 0.99

O ⁄ P-R ⁄ HV 1.51 (6.53) )0.60 (0.14) <0.0001 0.73 26.88 (10.56) )0.97 (0.04) <0.0001 0.90

O ⁄ RC 13.10 (6.59) )0.71 (0.12) <0.0001 0.27 19.70 (10.36) )0.78 (0.03) <0.0001 0.94

Standard errors are presented next to coefficients within parentheses.

SF, summer fallow; GB-R ⁄HV, green bean-rye ⁄ hairy vetch; M-BW-C, yellow mustard-buckwheat-rape; O ⁄P-R ⁄HV, oat ⁄ pea-rye ⁄ hairyvetch; O ⁄RC, oat ⁄ red clover.

Table 2 Coefficients for the linear regression of Setaria spp., Chenopodium album and Abutilon theophrasti D germinable seedbank as a

function of initial germinable seedbanks across the cropping systems in Pennsylvania

Weed species Cropping system

2004 2005

ß0 ß1 P > T R2 ß0 ß1 P > T R2

Setaria spp. SF 302.27 (198.24) )1.01 (0.09) <0.0001 0.88 21.82 (59.47) )0.73 (0.11) <0.0001 0.91

O-R ⁄ HV 1122.50 (216.23) 0.01 (0.09) 0.9680 0.03 )10.43 (59.47) )0.37 (0.12) 0.0017 0.19

M-BW-C 224.71 (186.25) )1.00 (0.10) <0.0001 0.92 )9.31 (61.23) )0.61 (0.12) <0.0001 0.80

O ⁄ P-R ⁄ HV 81.51 (175.94) )0.80 (0.07) <0.0001 0.95 24.75 (61.16) )0.57 (0.14) <0.0001 0.74

O ⁄ RC 296.39 (214.76) 0.08 (0.11) 0.4663 0.05 206.78 (64.04) 0.75 (0.13) <0.0001 0.23

C. album SF 359.23 (187.58) )1.05 (0.09) <0.0001 0.94 29.88 (219.34) )0.94 (0.30) 0.0019 0.97

O-R ⁄ HV 419.37 (219.84) )0.90 (0.13) <0.0001 0.62 107.02 (206.73) )0.93 (0.18) <0.0001 0.94

M-BW-C 702.50 (219.57) )0.83 (0.07) <0.0001 0.61 43.04 (214.48) )0.88 (0.26) 0.0010 0.89

O ⁄ P-R ⁄ HV 444.41 (184.52) )0.85 (0.07) <0.0001 0.88 63.45 (204.09) )0.89 (0.25) 0.0004 0.97

O ⁄ RC 706.57 (187.34) )0.89 (0.05) <0.0001 0.80 814.68 (217.76) 1.35 (0.36) 0.0002 0.05

A. theophrasti SF 14.56 (14.72) )0.77 (0.05) <0.0001 0.89 10.69 (12.13) )0.79 (0.05) <0.0001 0.94

O-R ⁄ HV 31.13 (14.52) )0.76 (0.05) <0.0001 0.83 20.11 (11.71) )0.72 (0.04) <0.0001 0.88

M-BW-C 20.95 (15.52) )0.78 (0.07) <0.0001 0.90 15.82 (11.85) )0.79 (0.05) <0.0001 0.88

O ⁄ P-R ⁄ HV 31.27 (14.59) )0.72 (0.05) <0.0001 0.85 4.81 (12.56) )0.52 (0.06) <0.0001 0.62

O ⁄ RC 26.745 (14.32) )0.70 (0.05) <0.0001 0.84 15.37 (11.91) )0.71 (0.05) <0.0001 0.90

Standard errors are presented next to coefficients within parentheses.

SF, summer fallow; O-R ⁄HV, oat-rye ⁄ hairy vetch; M-BW-C, yellow mustard-buckwheat-rape; O ⁄ P-R ⁄HV, oat ⁄ pea-rye ⁄ hairy vetch;

O ⁄RC, oat ⁄ red clover.

6 S B Mirsky et al.



Table 3 Comparisons for fitted values of D germinable weed seedbanks (seeds m)2) at three points over the range of initial germinable

seedbank of the Setaria spp., Chenopodium album and Abutilon theophrasti using ANCOVA with initial germinable weed seedbank as the

covariate (Maine)

Species

Site

year

Initial

seedbank

D germinable seedbanks

SF GB-R ⁄ HV B ⁄ BW ⁄ B O ⁄ P-R ⁄ HV O ⁄ RC

Setaria spp. 2004 25 )23a )23a )18a 54b 46b

60 )59a )59a )54a 55b 57b

120 )121a )121a )115a 56b 76b

2005 50 )47a )44a )50a )41a )22b

400 )390ab )395ab )398a )378b )322c

750 )734a )745a )746a )714a )622b

C. album 2004 2200 )495a )666a )860a 4220b 647a

4500 )2402a )2296a )2544a 8786c )669b

6500 )4060ab )3713ab )4009a 12757c )1815b

2005 1200 )701ab )1018a )1116a 4030c 316b

3000 )2342a )2461a )2528a 2342b )1328a

4800 )3984a )3905ab )3939ab 653c )2973b

A. theophrasti 2004 10 )7a )5a )1a )4a 6a

55 )46a )39ab )26b )31ab )26b

100 )85a )72a )51a )58a )58a

2005 50 )30a )32a )40a )22a )19a

300 )267ab )235bc )275a )265ab )213c

500 )456a )398b )463a )460a )369b

Mean values within a row for a given location and site year followed by the same letter are not significantly different according to the Tukey–

Kramer test for mean separation (P < 0.05).

SF, summer fallow; GB-R ⁄HV, green bean-rye ⁄ hairy vetch; M-BW-C, yellow mustard-buckwheat-rape; O ⁄ P-R ⁄HV, oat ⁄ pea-rye ⁄ hairyvetch; O ⁄RC, oat ⁄ red clover.

Table 4 Comparisons for fitted values of D germinable weed seedbanks (seeds m)2) at three points over the range of initial germinable

seedbank of the Setaria faberi, Chenopodium album and Abutilon theophrasti using ANCOVA with initial germinable weed seedbank as the

covariate (Pennsylvania)

Species Site year Initial seedbank

D germinable seedbanks

SF O-R ⁄ HV B ⁄ BW ⁄ B O ⁄ P-R ⁄ HV O ⁄ RC

S. faberi 2004 1000 )711a 1126c )779a )718a 373b

2000 )1723a 1130c )1784a )1518a 450b

3000 )2736a 1134d )2788a )2317b 527c

2005 50 )15a )29a )40a )4a 244b

400 )270a )159b )253ab )203ab 507c

700 )489a )271b )435a )374ab 733c

C. album 2004 750 )429a )256ab )81b )194ab 41b

1800 )1532a )1202b )790c )1088b )892bc

3800 )3633a )3003ab )2448b )2791b )2668b

2005 100 )65a 14a )45a )26a 950b

800 )726a )638a )663a )650a 1894b

1600 )1481a )1383a )1368a )1364a 2972b

A. theophrasti 2004 50 )24a )6a )18a )5a )8a

350 )254a )232a )251a )222a )218a

550 )408a )382a )406a )367a )358a

2005 50 )29a )16a )24a )21a )20a

350 )264a )233a )260a )178b )234a

500 )382a )341a )378a )256b )341a

Mean values within a row for a given location and site year followed by the same letter are not significantly different according to the Tukey–

Kramer test for mean separation (P < 0.05).

SF, summer fallow; O-R ⁄HV, oat-rye ⁄ hairy vetch; M-BW-C, yellow mustard-buckwheat-rape; O ⁄P-R ⁄HV, oat ⁄ pea-rye ⁄ hairy vetch;

O ⁄RC, oat ⁄ red clover.




systems. The A. theophrasti germinable seedbank was

consistently reduced by all cropping systems at both

locations and field sites (Table 4). In Maine in 2005, the

decrease in the A. theophrasti seedbank was lower at the

medium to high initial seedbank densities for the GB-

R ⁄HV and O ⁄RC compared with the other systems

(Table 3). The D germinable seedbank for A. theophrasti

was similar across cropping systems in PA; only

O ⁄P-R ⁄HV in 2005 resulted in a lower rate of seedbank

decline at medium to high population densities

(Table 4).

Cumulative recruitment

Cumulative weed seedling recruitment was used to

document seedbank losses to germination. Cropping

system effects were observed when density plots were

pooled (data not shown); however, the large variation in

the background weed population limited our ability to

resolve density treatment effects. Although site year was

a significant source of variance for A. theophrasti and

C. album in Maine (Table 5), site year was pooled

because preliminary analysis showed a similar relation-

ship for both species. Cumulative recruitment differed

by cropping system and density, but there were no

interactions between these treatments for any of the

target weeds at either location (Table 5).

Summed over cover crop treatments, cumulative

emergence of Setaria spp., C. album and A. theophrasti

was 39, 211 and 69 plants m)2 in Maine and 68, 77 and

44 plants m)2 in PA. Seedling densities ranged from 0 to

242 and 0–448 for Setaria spp., 37–582 and 0–388

for C. album and 0–386 and 0–250 plants m)2 for

A. theophrasti, in Maine and PA, respectively.

Significant differences were not observed for C. album

cumulative recruitment by system at either location

(Table 6). We attribute this lack of response to high

initial background weed populations in both locations

and site years. In Maine, the greatest S. glauca recruit-

ment occurred in O ⁄P–R ⁄HV and O ⁄P-R ⁄HV and

surprisingly these were the low disturbance systems,

while recruitment was greatest for A. theophrasti in

O ⁄RC. In PA, cumulative recruitment of S. faberi and

A. theophrasti was highest in the M-BW-C and SF

systems where disturbance was highest (Table 6). Gen-

erally, seedbank supplementation resulted in higher

Setaria spp. and A. theophrasti recruitment, a pattern

that was particularly pronounced at higher supplemen-

tation rates in PA (Table 7). Chenopodium album seed-

bank supplementation only resulted in higher

recruitment under the highest supplementation rate

and only in PA.

Weed community analysis

Abutilon theophrasti, C. album and Setaria spp.

responded strongly to cover crop treatments and field

sites (Fig. 3). Using RDA, the primary axis in the

Table 5 Analysis of variance table of cumulative seedling recruitment for Setaria spp., Chenopodium album and Abutilon theophrasti

populations in Maine (ME) and Pennsylvania (PA) for the 2004 and 2005 recruitment year

Location Effects

Setaria spp. C. album A. theophrasti

DF F-value P DF F-value P DF F-value P

ME Cropping system 4 7.10 0.0001 4 0.69 0.6147 4 4.81 0.0107

Density 2 158.17 0.0001 2 2.74 0.0701 2 273.99 0.0001

Cropping system · density 8 0.82 0.5901 8 0.90 0.5171 8 0.62 0.7591

Site year 0.2827 0.0001 0.0001

PA Cropping system 4 4.99 0.0093 4 2.77 0.0767 4 8.47 0.0009

Density 2 8.88 0.0003 2 20.16 0.0001 2 91.85 0.0001

Cropping system · density 8 0.33 0.9517 8 0.36 0.9375 8 1.15 0.3370

Site year 0.2420 0.2404 0.2584

Table 6 Cumulative seedling recruitment for Setaria spp.,

Chenopodium album and Abutilon theophrasti (plants m)2)

populations across the cropping systems in Maine (ME) and

Pennsylvania (PA) (density and year pooled). Means comparison

for C. album and A. theophrasti (ME) by year (only treatments with

year effect) did not vary between year and was therefore pooled

Location System Setaria spp. C. album A. theophrasti

ME SF 30b 240a 54b

GB-R ⁄ HV 21b 172a 46b

B ⁄ BW ⁄ B 34ab 206a 56ab

O ⁄ P-R ⁄ HV 65a 209a 101a

O ⁄ RC 41ab 230a 90a

PA SF 138a 99a 68a

O-R ⁄ HV 38b 71a 50ab

B ⁄ BW ⁄ B 90a 89a 50a

O ⁄ P-R ⁄ HV 27b 43a 22b

O ⁄ RC 43b 104a 33b

Mean values within a given location and year followed by the same

letter are not significantly different according to Tukey�s tests

(P = 0.1).

SF, summer fallow; GB-R ⁄HV, green bean-rye ⁄ hairy vetch,

O-R ⁄HV, oat-rye ⁄ hairy vetch; M-BW-C, yellow mustard-

buckwheat-rape; O ⁄P-R ⁄HV, oat ⁄ pea-rye ⁄ hairy vetch; O ⁄RC,

oat ⁄ red clover.

8 S B Mirsky et al.



bi-plot, which accounts for the greatest amount of

variation in the weed communities, represents a soil

disturbance gradient (F = 11.85, P = 0.002) (Fig. 3).

The cover crop treatments were strongly associated with

the disturbance gradient; as a result, SF, M-BW-C, GB-

R ⁄HV and O-R ⁄HV treatments were on one extreme

and O ⁄RC and O ⁄P-R ⁄HV on the other. The species

vectors were negatively correlated with disturbance.

Setaria spp. was strongly associated with O ⁄RC,

particularly in PA. Chenopodium album increased in

Maine and was strongly positively associated with

O ⁄P-R ⁄HV. Abutilon theophrasti displayed little associ-

ation with either disturbance gradient or site year.

Species vectors tended to associate with locations where

the greatest increases in the populations were observed:

S. faberi and C. album in PA in 2004 and C. album for

both years in Maine (Table 8). The M-BW-C and SF

systems were both negatively correlated with A. theo-

phrasti, C. album and Setaria spp. (Fig. 3). The cash

crops systems were negatively correlated with C. album

and not correlated to A. theophrasti. The orthogonal

secondary axis represented a growing condition gradient

(site years) (F = 3.11, P = 0.002). Location had a

minor influence on the community gradient, as indicated

by proximity to the centroid in the biplot, and based on

the forward selection analysis not being significant

(P = 0.940).

Weed community was significantly influenced by site

year and cover crop systems, and interactions between

site year and location, and site year and cover crop

systems (Fig. 3). These results confirm the significant

community gradient observed in the ordination biplots,

where species vectors and several cover crop systems

strongly associated with a given site year. Location was

not a significant source of variance, but there was an

interaction with site year. Interactions between location

and site year may have resulted from variation in

background seedbank populations between locations,

causing greater changes in the seedbank, PA having a

higher A. theophrasti population in 2004 and Maine in

2005 (Table 8). As both cover crop systems and the

species vectors were distributed along the disturbance

gradient, the results indicate the SF and M-BW-C were

the only cover crop systems that consistently reduced all

weed populations. These results confirmed our hypoth-

esis that consistent depletion of the germinable seedbank

increases with greater frequency of soil disturbance and

cover cropping.

Discussion

These results indicate the combination of tillage and

cover crops resulted in very high seedbank depletion

rates, rates considerably higher than under more con-

ventional management. Teasdale et al. (2004) reported

>50% reductions in the Amaranthus hybridus L.

germinable seedbank in years of good weed management

in a soyabean cash crop. In another study that included

eight locations across the US Corn Belt, C. album,

S. faberi and A. theophrasti seedbanks declined 40–47%,

Table 7 Cumulative seedling recruitment for Setaria spp.,

Chenopodium album and Abutilon theophrasti (plants m)2)

populations in the density plots for the recruitment year (years

pooled) across the supplement seedbank densities in Maine (ME)

and Pennsylvania (PA) (cropping system and year pooled). Means

comparison for C. album and A. theophrasti (ME) by year (only

treatments with year effect) did not vary between year and was

therefore pooled

Location Density Setaria spp. C. album A. theophrasti

ME Low 4c 194a 6c

Medium 28b 206a 35b

High 83a 234a 167a

PA Low 54c 46b 6c

Medium 68b 55b 25b

High 121a 101a 102a

Mean values within a given location and site year followed by the

same letter are not significantly different according to Tukey�s tests(P = 0.05) for the high density plots.

1.0

1.0

SET. spp.

PA

2004

2005

M-BW-C

–1.0

–1.0

CHEAL

ABUTH

SF ME

O-R/HV &

GB-R/HV

O/P-R/HV

O/RC

Tillage

Treatment P(perm)Field site year 0.002

Location 0.182System 0.018

Field site year x location 0.002Field site year x system 0.002

Fig. 3 Biplot ordinations from a redundancy analysis displaying

the associative relationships between the proportional change

(final ⁄ initial) in the weed community assembled from the three

target weed species (CHEAL – Chenopodium album; SETFA –

Setaria spp.; ABUTH – Abutilon theophrasti) and the explanatory

variables: site years ( ) (2004; 2005), location (m) (Maine-ME;

Pennsylvania-PA), cropping system (j) (O – Oats; RC – red clover;

HV – hairy vetch; P – pea; GB – green bean; M – Mustard; BW –

buckwheat; C – rape), and tillage (- - -). Multivariate analysis of

variance table using permutational multivariate analysis of

variance of site year, location, and system treatment factors.




78% and 42%, respectively over a 1 year period (Davis

et al., 2005).

The stimulating effect of disturbance on weed

seedling recruitment in PA is consistent with other

reports in the literature. Mulugeta and Stoltenberg

(1997) observed up to a sixfold increase in S. faberi

and C. album emergence in tilled vs. untilled soils.

Additionally, these authors reported 16-fold reductions

in C. album and sixfold reductions in S. faberi seedbanks

in a single season. Therefore, it is likely that specific

treatments, through disturbance and microclimate,

favoured germination losses to the seedbank more than

others, and that this germination is the primary mech-

anism for drawing down the seedbank (Forcella, 2003).

While cumulative weed emergence strongly corre-

sponded to initial seedbank densities, it is unclear why

emergence was not similarly influenced at the Maine

study site. It is possible that cooler soils at the time of

tillage, or the presence of higher concentration of cover

crop-derived allelochemicals may have played a role

(Mohler, 2001).

The species-specific response to cover crop treat-

ments, resulted in a restructuring of the weed commu-

nity. While we chose to study three summer annual weed

species, they reflect differences in seed size, longevity and

responsiveness to germination stimulants (disturbance

and cover crop induced germination). Therefore, we

infer that these systems would act in a similar way on a

broader spectrum of summer annual weed species. In

their framework for long-term management of weed

seedbanks, Swanton and Booth (2004) identified four

management tactics that target critical life history stages:

increase seed mortality, manipulate seed germination

and emergence, reduce seed production, or complete

removal of the above-ground biomass. The success of

our community-based weed management approach was

a result of the direct effects of cover crops and tillage

targeting all of these critical life history stages. A distinct

feature of the SF and M-BW-C systems is the distribu-

tion of soil disturbance evenly throughout the growing

season, thereby stimulating germination and then

controlling the germinated plants. In effect, the SF and

M-BW-C serve as a full season form of stale-seedbed

management. The success of this stale-seedbed approach

is only realised, if resulting weed cohorts are managed

and not allowed to reproduce (Bond & Grundy, 2001;

Gallandt, 2006). Given the M-BW-C system has a lower

disturbance frequency than the cash crops systems,

which varied in its weed suppressive performance, the

distribution of the disturbance events are likely to be as

important as frequency.

The SF system served as the control system and was

expected to provide consistent suppression of weed

populations, because it directly targeted weeds before

maturity. However Cyperus esculentus L., one of the

naturally occurring perennial weeds at the study site,

increased threefold in the SF compared with the

M-BW-C system in PA (data not shown). Most of the

C. esculentus emergence in the glasshouse assays

appeared to arise from vegetative structures, rather than

from seed. Clearly tillage alone (SF) was not enough to

manage C. esculentus. In contrast, the M-BW-C system

provided significant suppression of C. esculentus. In

addition to the disturbance and cover crop biomass

produced by this system, additional weed suppression

may have been provided by the Brassica cover crop

which is known to produce glucosinolates (Bhowmik &

Inderjit, 2003), a weed suppressive allelochemical. This

finding is significant, in that it identifies a cover crop

treatment that is broadly suppressive of both annual and

perennial weeds common to such cropping systems.

Our results also suggest that the timing of soil

disturbance is as important as frequency in reducing

weed seedbank populations. For example, managing for

high cover crop biomass in the O ⁄P-R ⁄HV system in

Maine, resulted in large increases in the C. album weed

seedbank. Similarly, poor crop stand establishment and

delays in management in the oat cash crop in PA

resulted in S. faberi surviving to reproductive maturity,

with corresponding increases in the Setaria spp. seed-

bank. Variations in timing of disturbance can cause

species shifts. If the disturbance coincides with the

emergence period of a species, that species will be

effectively controlled (Crawley, 2004).

Table 8 Mean, standard deviation and range of initial germinable seedbank for Setaria spp., Chenopodium album and Abutilon theophrasti

(seeds m)2) populations at Maine (ME) and Pennsylvania (PA) for 2004 and 2005

Location Site year

Setaria* spp. C. album A. theophrasti

Mean SD Range Mean SD Range Mean SD Range

ME 2004 41 58 0–301 4270 2152 844–12 357 26 41 0–181

2005 152 233 0–1266 2938 2107 121–11 212 131 197 0–1236

PA 2004 1734 1229 91–6030 1768 1182 364–6333 166 212 0–939

2005 309 276 0–1364 447 639 0–4061 113 176 0–788

*Setaria faberi in Pennsylvania and Setaria glauca in Maine.

10 S B Mirsky et al.



Conclusion

Field experiments, conducted in PA and Maine, demon-

strated that soil disturbance associated with cover crop-

ping encouragedweed seed germination, thereby reducing

the density of germinable seed in the weed seedbank. Of

the five cover cropping systems, one, M-BW-C, consis-

tently depleted the weed seedbank through weed expres-

sion and subsequent control of emerged plants. That same

system also provided the greatest suppression of the

troublesome perennial C. esculentus. The frequency and

even temporal distribution of soil disturbance in the

M-BW-C system stimulated weed seed germination,

controlled emerged plants and pre-empted weed seed

rain. We also found that in slower growing perennial

cover crops, weedy species like Setaria spp. were able

grow and set seed in the growing cover crop. When this

happened, the �debits� resulting fromweed seed emergence

were overwhelmed by the resulting seed rain �credits�(Forcella, 2003). The M-BW-C system provided an

effective and consistent community-based approach to

weed management (Swanton & Booth, 2004). The lower

disturbance and cash crop systems may still provide a

useful population-based approach to managing weeds.

However, their weaknesses were revealed during the

course of this study.

A first step in the development of weed management

systems that are durable enough to withstand the

inherent variability of weather and management is the

reduction of the weed seed bank. Our work indicates

that the seedbanks of weedy fields can be dramatically

reduced in a single field season using cover crops and

tillage. These results may be particularly relevant to

farms in transition to organic production or small

holder vegetable production where crops are less com-

petitive and where herbicides are less likely to be used.

This practice primarily served as a rescue practice when

weed management in the previous year is compromised.

The highly dormant fraction of the seedbank will still

persist and will require continuous management. Future

research addressing the persistent fraction of the

seedbank could identify practices that would further

reduce seedbanks, a step that would make EBWM more

robust.

References

ANDERSON MJ (2001) A new method for non-parametric

multivariate analysis of variance. Austral Ecology 26, 32–46.

ANDERSSON TN & MILBERG P (1998) Weed flora and the

relative importance of site, crop, crop rotation, and nitrogen.

Weed Science 46, 30–38.

BHOWMIK PC & INDERJIT (2003) Challenges and opportunities

in implementing allelopathy for natural weed management.

Crop Protection 22, 661–671.

BOND W & GRUNDY AC (2001) Non-chemical weed manage-

ment in organic farming systems. Weed Research 41,

383–405.

BOOTH BD & SWANTON CJ (2002) Assembly theory applied

to weed communities. Weed Science 50, 2–13.

CAVERS PB & BENOIT DL (1989) Seed banks in arable land.

In: Ecology of Soils Seed Banks (eds MA LECK, BT PARKER &

R SIMPSON), 309–328. Academic Press, San Diego, CA,

USA.

TER BRAAK CJFAS P (2002) CANOCO Reference Manual and

CanoDraw for Windows User�s Guide: Software for CanonicalCommunity Ordination (Version 4.5). Micro Computer

Power, Ithaca, NY, USA.

CRAWLEY MJ (2004) Timing of disturbance and coexistence

in a species-rich ruderal plant community. Ecology 85,

3277–3288.

DAVIS AS, CARDINA J, FORCELLA F et al. (2005) Environmental

factors affecting seed persistence of annual weeds across

the US corn belt. Weed Science 53, 860–868.

FORCELLA F (2003) Debiting the seedbank: priorities and

predictions. Aspects of Applied Biology 69, 151–162.

FORCELLA F & LINDSTROM MJ (1988) Movement and

germination of weed seeds in ridge-till crop production

systems. Weed Science 36, 56–59.

FORCELLA F, ERADATOSKOUI K & WAGNER SW (1993)

Application of weed seedbank ecology to low-input crop

management. Ecological Applications 3, 74–83.

GALLAGHER RS & CARDINA J (1998) The effect of light

environment during tillage on the recruitment of various

summer annuals. Weed Science 46, 214–216.

GALLANDT ER (2006) How can we target the weed seedbank?

Weed Science 54, 588–596.

GALLANDT ER, LIEBMAN M, CORSON S, PORTER GA &

ULLRICH SD (1998) Effects of pest and soil management

systems on weed dynamics in potato. Weed Science 46,

238–248.

GALLANDT ER, FUERST EP & KENNEDY AC (2004) Effect of

tillage, fungicide seed treatment, and soil fumigation on seed

bank dynamics of wild oat (Avena fatua). Weed Science

52, 597–604.

GROSS KL (1990) A comparison of methods for estimating seed

numbers in the soil. Journal of Ecology 78, 1079–1093.

HARTZLER RG & ROTH GW (1993) Effect of prior year�s weedcontrol on herbicide effectiveness in corn (Zea mays). Weed

Technology 7, 611–614.

HILHORST HWM & DARSSEN CM (2000) Effect of chemical

environment on seed germination. In: The Ecology of

Regeneration in Plant Communities (ed M FENNER), 293–309.

CABI Publishing, Wallingford, UK.

JORDAN N (1996) Weed prevention: priority research for

alternative weed management. Journal of Production

Agriculture 9, 485–490.

LIEBMAN M & GALLANDT ER (1997) Many little hammers:

ecological management of crop-weed interactions. In:

Ecology in Agriculture (ed LE JACKSON), 291–343.

Academic Press, San Diego.

LONSBARY SK, O�SULLIVAN J & SWANTON CJ (2003)

Stale-seedbed as a weed management alternative for

machine-harvested cucumbers (Cucumis sativus). Weed

Technology 17, 724–730.




MILBERG P, ANDERSSON L, ELFVERSON C & REGNER S (1996)

Germination characteristics of seeds differing in mass.

Seed Science Research 6, 191–197.

MIRSKY SB, LANYON LE & NEEDELMAN BA (2008) Evaluating

soil management using particulate and chemically labile soil

organic matter fractions. Soil Science Society of America

Journal 72, 180–185.

MOHLER CL (2001) Enhancing the competitive ability of crops.

In: Ecological Management of Agricultural Weeds

(eds M LIEBMAN, CL MOHLER & CP STAVER), 269–321.

Cambridge University Press, Cambridge, UK.

MOHLER CL, FRISCH JC & MCCULLOCH CE (2006) Vertical

movement of weed seed surrogates by tillage implements and

natural processes. Soil & Tillage Research 86, 110–122.

MULUGETA D & STOLTENBERG DE (1997) Weed and seedbank

management with integrated methods as influenced by

tillage. Weed Science 45, 706–715.

PLANTE AF & MCGILL WB (2002) Soil aggregate dynamics and

the retention of organic matter in laboratory-incubated soil

with differing simulated tillage frequencies. Soil & Tillage

Research 66, 79–92.

REBERG-HORTON SC, BURTON JD, DANEHOWER DA et al.

(2005) Changes over time in the allelochemical content

of ten cultivars of rye (Secale cereale L.). Journal of Chemical

Ecology 31, 179–193.

REBERG-HORTON C, GALLANDT ER & MOLLOY T (2006)

Measuring community shifts in a weed seedbank study

with the use of distance-based redundancy analysis. Weed

Science 54, 861–866.

RYAN MR (2007) Stability of Weed Management Efficacy and

Weed-Crop Competition in a Long-Term Cropping Systems

Trial. MS thesis, The Pennsylvania State University,

University Park.

SCHONBECK MW & EGLEY GH (1980) Effects of temperature,

water potential, and light on germination responses of

redroot pigweed seeds to ethylene. Plant Physiology 65,

1149–1154.

SWANTON CJ & BOOTH BD (2004) Management of weed

seedbanks in the context of populations and communities.

Weed Technology 18, 1496–1502.

TEASDALE JR, MANGUM RW, RADHAKRISHNAN J & CAVIGELLI

MA (2004) Weed seedbank dynamics in three organic

farming crop rotations. Agronomy Journal 96, 1429–1435.

THOMPSON K & GRIME JP (1983) A comparative study of

germination responses to diurnally-fluctuating temperatures.

Journal of Applied Ecology 20, 141–156.

12 S B Mirsky et al.



reducing the germinable weed seedbank with soil disturbance and cover crops

Documents