Content uploaded by Hugh Mcallister
Author content
All content in this area was uploaded by Hugh Mcallister on Mar 22, 2016
Content may be subject to copyright.
Content uploaded by Hugh Mcallister
Author content
All content in this area was uploaded by Hugh Mcallister on Mar 22, 2016
Content may be subject to copyright.
The effectiveness of old and new strategies for the
long-term control of Pteridium aquilinum, an 8-year
test
GMILLIGAN*,ESCOX*,JGALDAY*,VMSANTANA*,HAMCALLISTER*,
RJPAKEMAN†,MGLEDUC*&RHMARRS*
*School of Environmental Sciences, University of Liverpool, Liverpool, UK, and †The James Hutton Institute, Aberdeen, UK
Received 16 February 2015
Revised version accepted 8 January 2015
Subject Editor: David Clements, Trinity Western University, Canada
Summary
There is a need for management strategies to control
dominant perennial weeds and restore seminatural
communities. We compared the effects of five weed
control treatments on dense Pteridium aquilinum
relative to an untreated experimental control over an
8-year period with the aim of restoring acid grass-
land. The weed control treatments tested were as
follows: cutting and bruising, both twice and thrice
annually, and herbicide treatment (asulam in year 1
followed by annual spot retreatment of all emergent
fronds). Pteridium aquilinum performance and plant
species composition were monitored. Data were anal-
ysed using Bayesian mixed-effect models and multi-
variate techniques. Cutting twice and thrice yearly
and the asulam treatment all reduced frond density
to zero; both bruising treatments were ineffective.
The plant communities in the cut and asulam-treated
plots showed differences from the untreated and
bruised plots; the asulam-treated plots contained
more ruderal species and the cut plots were more
typical of acid grassland. Acid grassland recovery
was fastest in the asulam-treated plots, but the cut
plots caught up after approximately 5 years. There
were two important conclusions. First, an intractable
weed like P. aquilinum can be eradicated and a vege-
tation more suited for grazing can be achieved by the
continuous application of some treatments over many
years. Here, success was achieved by cutting twice/
thrice annually, or by a single asulam application fol-
lowed by annual spot spraying of all emergent fronds
for 8 years. Second, bruising, a treatment favoured
by some conservation organisations, did not work
and cannot be recommended. The use of long-term,
continuously applied treatments might be considered
for all perennial weeds with large underground root/
rhizome systems.
Keywords: Bayesian mixed-effects models, canonical
correspondence analysis, bracken, asulam, cutting,
bruising.
M
ILLIGAN G, COX ES, ALDAY JG, SANTANA VM, MCALLISTER HA, PAKEMAN RJ, LE DUC MG & MARRS RH
(2016). The effectiveness of old and new strategies for the long-term control of Pteridium aquilinum, an 8-year test.
Weed Research.
Introduction
Perennial weeds with an extensive underground root/
rhizome system cause major problems for agriculture,
especially in extensively grazed systems, and plant
community conservation worldwide (Naylor, 2002).
Once established, they can come to dominate the
ecosystem, becoming an effective monoculture with
Correspondence: Prof. R H Marrs, School of Environmental Sciences, University of Liverpool, Liverpool L69 3GP, UK. Tel: (+44) 151 795
5172; Fax: (+44) 151 795 4644; E-mail: calluna@liv.ac.uk
© 2016 European Weed Research Society
DOI: 10.1111/wre.12203
few other species present. Attempts to restore such
invaded ecosystems must involve at least two inter-
linked processes. First, the weed species must be either
eradicated or be reduced below the level at which it
loses its competitive advantage. Second, an appropriate
and stable, native community must be established
(Alday et al., 2013). Where such weeds have a large
store of underground reserves, control can be very dif-
ficult; there are many examples where after good initial
control, there is a rapid return of the weed species
(Lowday & Marrs, 1992). Clearly, the only way to test
appropriate management for both weed control and
restoration of appropriate plant communities is
through experiments where management interventions
are followed over many years. Here, we assess weed
control and concurrent restoration over 8 years using
Pteridium aquilinum (L.) Kuhn as the test weed species.
Pteridium spp. are a worldwide perennial weed
problem (Marrs & Watt, 2006; Miatto et al., 2011),
capable of producing a dense canopy and often domi-
nating the invaded communities, reducing economic
and conservation value, and the aesthetic appeal of
natural areas (Pakeman & Marrs, 1992; Marrs &
Watt, 2006). In Britain, the current ecological success
of P. aquilinum has been suggested to result from
changes in cultural and environmental practices that
have increased the amount of suitable habitat, and
this, combined with improved climatic conditions, has
provided opportunities for range expansion (Pakeman
et al., 2000). Pteridium aquilinum possesses several
biological and ecological attributes that allow it to
capitalise on these opportunities for range expansion:
(i) an extensive rhizome system possessing consider-
able stores of carbohydrate, nutrients and dormant
buds; (ii) high productivity, capable producing a
dense canopy that casts deep shade over understory
vegetation; (iii) deep litter accumulation which sup-
presses colonisation of other species; and (iv) possibly
production of allelopathic substances (Marrs & Watt,
2006).
The attributes which contribute to the ecological
success of P. aquilinum also provide it with a resili-
ence to control treatment, which is often deemed diffi-
cult (Lowday & Marrs, 1992; Chapman et al., 2009).
Over the last 40 years, control has usually been imple-
mented either by cutting or by herbicide use (usually
asulam). Cutting aims to continuously deplete rhi-
zome carbohydrate and nutrient stores (Williams &
Foley, 1976; Marrs & Watt, 2006), resulting in a slow
reduction in frond cover to a very low level (Marrs &
Lowday, 1992; Marrs et al., 1998); in the longest
experiment carried out to date, a few fronds remained
even after 18 years of twice-yearly cutting (Marrs
et al., 1998). Cutting also breaks up the litter layer,
accelerating vegetation recolonisation (Marrs et al.,
2007).
Asulam is a pteridophyte- and Rumex-selective
post-emergence, systemic carbamate herbicide (Low-
day & Marrs, 1992), considered to be of relatively
low ecological risk (USEPA, 1995) and it has been
licensed for aerial application in the UK. In 2012,
asulam was withdrawn from use within the European
Union (EU) because of a lack of information with
respect to metabolites resulting from its breakdown in
crop plants (European Union Regulation (EC) No.
1045/2011 (Anon 2011). However, its use has been
continued as a result of derogated powers in four EU
member states, including the UK, and asulam is cur-
rently going through the EU re-registration/approval
process.
Recently, however, two alternative approaches have
been suggested for P. aquilinum control within the
UK. The first is asulam application to provide an ini-
tial reduction in P. aquilinum cover, followed by spot
spraying all emergent fronds in subsequent years,
without respite. Indeed, this has been suggested as an
effective way of eradicating P. aquilinum (Robinson,
2000). The second is the re-introduction of bruising
(variously termed breaking, crushing and rolling) as
an alternative to cutting. Bruising aims to knock the
fronds down and produce breaks/nicks along the
frond rachis, which is damaged, but not severed
(Braid, 1959). This approach was commonly used in
Britain up to the Second World War, before the
advent of adequate cutting machines and effective her-
bicides (Braid, 1959). Bruising is currently undergoing
a revival of interest for a variety of reasons; it can be
incorporated into organic farming regimes where it is
preferred over herbicide use, and can be applied much
faster than cutting especially on rocky, steep or
uneven ground (Lewis et al., 1997; Bacon et al.,
2001). Neither bruising nor continuous spot spraying
has been assessed experimentally in randomised con-
trol trials.
Here, therefore, we compared the effectiveness of
bruising (twice and thrice yearly) and annual spot
spraying of asulam against two cutting treatments
(twice and thrice yearly) and an untreated control. Our
hypotheses were as follows:
(1) Intensive treatment can eradicate P. aquilinum
through either (a) multiple annual cuts/bruisings
or (b) an initial asulam application plus annual
follow-up spot spraying.
(2) Bruising was an effective means of P. aquilinum
control.
(3) Differing control treatments produce different
trajectories of change in the underlying plant
community.
© 2016 European Weed Research Society
2 G Milligan et al.
Methods
Experimental design
The experiment was set up at Bamford Edge in the
Peak District National Park, Derbyshire, UK (1°41
0
W
53°41
0
N; GB National Grid Reference SK213 841).
The site is a steep escarpment, most of which has been
covered with P. aquilinum for over a century (N Tay-
lor, pers. comm.). The site was sprayed with asulam in
1990 (Marrs et al., 1992), but by 2005, there had been
substantial frond recovery. The soil is a podzol with a
pronounced organic layer and the site is grazed by
sheep at ca. 0.5 sheep ha
!1
, in keeping with agri-envir-
onment scheme prescriptions (Alday et al., 2013).
In November 2004, three replicate blocks, each con-
taining six 20 9 20 m plots separated by 2-m strips, were
marked out; this plot size exceeds that needed to remove
violations of independence stemming from between-plot
interference from underground rhizome effects (Le Duc
et al., 2003). Plots were selected randomly for
application of one of six treatments: (i) P. aquilinum
cut twice yearly (Cut29); (ii) P. aquilinum cut thrice
yearly (Cut39); (iii) P. aquilinum bruised twice yearly
(Bruise29); (iv) P. aquilinum bruised thrice yearly
(Bruise39); (v) overall asulam application in 2005
followed by annual spot spraying with asulam of all
emerging fronds (Asulam); and (vi) the untreated con-
trol (Untr). The experimental layout is illustrated in
Supporting Information, Figure S1. The strips around
and within each block were bruised thrice per year for
access.
Treatments
Cutting and bruising were both applied in late June
(92, 93), late July (92, 93) and late August (93)
from 2005 to 2012 inclusive. Peak frond biomass
usually occurs in late July (Williams & Foley, 1976)
and the timing was recommended by Braid (1959)
for cutting, that is to produce three flushes of fronds
to be treated each season. In the cut treatments, the
fronds were cut using a petrol-driven strimmer; bruis-
ing was carried out using a ‘Bracken Bruiser’ (sup-
plied by Peter Gotham, Bracken Bruisers, Sidmouth,
UK), trailed by a 4WD ATV (Figure S2). Over the
8 years of the study, cutting/bruising was applied 16
times to the 92 plots and 24 times to the 93 plots.
The herbicide treatment followed methodology pro-
posed by Robinson (2000) for P. aquilinum clearance
and started with an initial application of asulam
(commercial product, Asulox
!
, produced by Bayer
CropScience Ltd and United Phosphorus, Ltd) by
knapsack sprayer at a rate of 4.4 kg a.i. ha
!1
(11 L Asulox ha
!1
) in 400 L of water in early
September 2005. Thereafter, follow-up spot spraying
of every emergent frond was carried out annually
from 2006 to 2012 (7 spot applications) using a
knapsack sprayer at a dose of approximately 2 ml
per squirt at a ratio of 6% vol:vol Asulox to water
(~0.05 g a.i. per frond). From 2009 to 2012, the
number of spot-spray squirts was counted in each
plot to provide a whole-plot estimate of success.
Monitoring
Each treatment plot was divided into a grid of 1 m
2
squares. Every June, from 2005 (pre-treatment) to
2013 inclusive, a 1 m
2
quadrat was placed at five dif-
ferent randomly selected co-ordinates each year within
this grid. The cover (%) of all plant species present in
each quadrat was estimated visually. A 0.25 m
2
sub-
quadrat was then placed centrally, and all P. aquilinum
fronds were cut at soil level, counted to provide an
estimate of density (corrected to number m
!2
) and
length measured to provide a measure of P. aquilinum
productivity (mean length per quadrat was calculated).
The results for the 2013 sampling reflect the response
to the last treatment application in 2012.
By 2013, it was apparent that the asulam and both
cutting treatments reduced the P. aquilinum infestation
to a very low level. Accordingly, in September 2013,
an additional whole-plot sampling of these treatments
was performed; the replicate plots were gridded into
100 2 9 2 m cells and the fronds in each counted.
Data analysis
All data analyses were performed using R version
2.13.2 (R Development Core Team, 2011).
Five response variables, P. aquilinum cover, frond
density (frond m
!2
), mean frond length (cm), species
richness and Shannon–Weiner species diversity (H),
were analysed to assess treatment effects through time
on P. aquilinum performance and plant community
diversity. When assessing P. aquilinum performance in
control experiments, it is important to consider both
frond length and density, because P. aquilinum often
exhibits a trade-off between them especially in cutting
treatments (Lowday & Marrs, 1992); cover was also
included because this is the easiest one to assess under
field conditions by managers. Given the randomised
block design of this experiment, an error term was
needed to account for any block and/or plot-level vari-
ation. Moreover, initial analyses of the temporal effects
for each treatment identified heterogeneous variance
structures. Therefore, it was deemed more suitable
to analyse these data using generalised linear mixed
© 2016 European Weed Research Society
Bracken control strategies 3
models (GLMMs) in a Bayesian framework (as
opposed to the somewhat simpler ANOVA/ANCOVA
approach), as this offered a more robust method for
estimation of the parameters and their associated con-
fidence intervals, whilst also circumventing problems
associated with data transformation (Pinheiro & Bates,
2000; Bolker et al., 2009). Hence, here, all data were
analysed untransformed. All univariate analyses were
implemented using the ‘MCMCglmm’ function in the
‘MCMCglmm v.2.16’ package (Hadfield, 2010); models
incorporated treatment and a treatment 9 time inter-
action as fixed terms and block and plot identity as
random covariates; polynomial contrasts were included
(Gurevitch & Chester, 1986), but only the first-order
ones are discussed here. A Markov chain Monte Carlo
(MCMC) routine was used to estimate the posterior
distributions for the mean effects and their correspond-
ing 95% confidence intervals and associated Bayesian
P-values (P
MCMC
). The models incorporated parame-
ter-expanded priors. All models were run for a 1 9 10
5
iteration burn-in with sampling every 500th iteration
for a further 2 9 10
6
iterations, resulting in an effec-
tive sample for each parameter estimate of 4 9 10
3
from the approximated posterior distribution. Trace
plots of all parameters were checked for convergence.
In addition, frond distribution maps for the 2013
whole-plot assessment were created using the ‘interp’,
‘image’ and ‘contour’ functions within the ‘akima’
package v.0.5-11 (Crawley, 2013).
Species composition data (Hellinger transformed,
function ‘decostand’) with respect to each treatment
through time were analysed using canonical correspon-
dence analysis (CCA, function ‘cca’) in the ‘vegan’
package v.2.0-2 (Oksanen et al., 2011). Hellinger trans-
formation does not weight rare species differentially
and has been shown to be appropriate for testing the
significance of relationships between community com-
position and a set of explanatory variables, particularly
when there are many zero values within the data
(Legendre & Gallagher, 2001). Significance of the over-
all model, the first canonical axis and treatment effects
were tested using Monte Carlo tests with 999 permuta-
tions.
Results
Treatment effects on P. aquilinum and species
diversity
All first-order GLMM-estimated results for the four
response variables between 2005 and 2013 are pre-
sented in Table 1, along with assessment of significant
treatment effects relative to the untreated response. At
the start of the experiment in 2005, there were no
significant differences ( P < 0.05) between treatments in
terms of frond length (overall experimental
mean " SE, n = 18; 35.9 " 3.6 cm), frond density
(30.8 " 3.6 fronds m
!2
), cover (23.0 " 2.8%), rhizome
mass (2.11 " 0.75 kg m
!2
), P. aquilinum litter cover
(49.4 " 4.6%), cover of understorey species
(5.08 " 4.3%) or Shannon–Weiner Index (1.38 " 0.06;
all mean " SE, n = 18). Thereafter, P. aquilinum cover
showed significant responses through time for all treat-
ments (P < 0.001); in three treatments, there was an
increase between 2005 and 2013. The untreated con-
trols showed the largest increase (53%), closely fol-
lowed by thrice- and twice-yearly bruising with a 50%
and 35% increase respectively (Fig. 1A). The remain-
ing three treatments showed a reduction in cover: 16%
in cutting thrice yearly, 17% in the asulam treatment
and 19% in cutting twice yearly (Table 1, Fig. 1A).
No significant effect through time was found for
frond density in the untreated plots (P > 0.05, Table 1.
Fig. 1B), but there was an increased frond density of
+15 fronds m
!2
(P < 0.05) and +35 fronds m
!2
(P < 0.001) over the starting density of 31 fronds m
!2
in the twice- and thrice-bruised plots (Fig. 1B) respec-
tively. The bruising treatments showed a dip in frond
density between 2007 and 2011, but then, there was a
strong increase (Fig. 1B). The cutting and asulam
treatments reduced frond density over the starting den-
sity: !15 fronds m
!2
in the asulam treatment
(P < 0.05), !19 fronds m
!2
in the thrice-yearly cut
(P < !0.05) and !24 fronds m
!2
in the twice-yearly
cut (P < 0.001; Fig. 1B). At the whole-plot scale, the
number of fronds squirted each year during spot
spraying declined, reflecting the continued reduction in
frond density (Fig. 2).
The only significant increase in mean frond length
was found in the untreated plots (increase of 32 cm,
P < 0.001). A reduction of 27.1 cm was found in the
asulam treatment and 16.1 cm and 22.0 cm in the twice-
and thrice-yearly cut respectively (all P < 0.001). No sig-
nificant effects through time were found for either of the
two bruising regimes on frond height (Fig. 1C).
No significant treatment effects were found for
species richness, but significant increases in Shannon–
Weiner index (H) were found for two treatments, cut-
ting twice and thrice yearly, where the index increased
to 0.34 and 0.31 respectively (both (P < 0.001,
Fig. 1D).
The spatial distribution of all three replicate blocks
for the three treatments that reduced the P. aquilinum
infestation successfully (Fig. 3) showed that frond den-
sity in the middle of each plot had been reduced to
effectively zero. Almost all fronds were located around
the periphery and the densities around the periphery
were greatest in the cut treatments.
© 2016 European Weed Research Society
4 G Milligan et al.
Treatment effects on plant community composition
The CCA explained approximately 11.0% of the spe-
cies-treatment variance with eigenvalues for the first
two axes of 0.16 and 0.08 respectively. Both the overall
model and first canonical axis were significant
(P = 0.005). With the exception of twice-yearly bruis-
ing (P = 0.10), all other P. aquilinum control treat-
ments had significant overall effects on community
composition (P < 0.01), and significant interactions
through time (P < 0.01). The species biplot showed
most of the dominant species, typical of acid grassland
close to the centroid (P. aquilinum, Agrostis capillaris,
A. vinealis, Deschampsia flexuosa) with the less com-
mon species around the edges (Fig. 4A). The untreated
plots showed a trajectory increasing in the direction of
P. aquilinum and away from grassland species
(Fig. 4B). Both bruising treatments showed a less well-
pronounced trajectory in approximately the same
direction. The asulam spot-sprayed plots showed an
opposite response moving away from P. aquilinum and
towards a community dominated by ruderal species
(Aira praecox L., Epilobium angustifolium L, Poa annua
L), whereas both cutting treatments showed an almost
orthogonal response to the untreated/bruising-asulam
trajectories with a positive correlation with acid grass-
land species (Cerastium fontanum Baumg., Festuca
ovina L. agg., Galium saxatile L., Rumex acetosella L.
and Stellaria media (L.) Vill.), and a negative one with
a group of species in the lower quadrant of Fig 4;
these were a Carex L. spp., a Fraxinus excelsior L.
seedling, Luzula multiflora (Ehrh.) Lej., Veronica offici-
nalis L., and the bryophytes Hypnum cupressiforme
Hedw., Lophocolea bidentata (L.) Dumort and Ptilid-
ium ciliare (L.) Hampe.
Discussion
Intensive treatment can eradicate P. aquilinum
(Hypothesis 1)
To test this hypothesis, it is necessary to consider the
results obtained here with those of a multisite study
testing some of these treatments in different regions of
Great Britain (Alday et al., 2013). The Bamford Edge
site would be expected to develop into an acid grass-
land and Alday et al. (2013) suggested that such sites
would produce alternative stable acid grassland states
for at least 10 years with either annual cutting once
or twice yearly, or with a single asulam application.
Table 1 Summary of GLMM posterior mean estimates of the first-order (linear) response of three bracken response variable and Shan-
non–Weiner index to treatment through time with the lower (L 95%) and upper (U 95%) 95% confidence intervals and associated Baye-
sian P-values (P
MCMC
). Included are contrasts for each mean estimate with the control obtained using t-tests performed on the posterior
distributions
Response Treatment
Effect through time Contrast with control
Mean L95% U95% P
MCMC
Difference " SE tP
MCMC
P. aquilinum Cover (%) Untreated 53.38 43.18 63.77 <0.001*** –––
Asulam !17.11 !27.73 !7.32 0.002** !70.48 " 0.14 !957.86 <0.001***
Bruise29 34.47 24.26 44.80 <0.001*** !18.90 " 0.14 !256.09 <0.001***
Bruise39 50.02 39.85 60.14 <0.001*** !3.35 " 0.14 !45.48 <0.001***
Cut29 !19.30 !28.16 !9.62 <0.001*** !72.68 " 0.14 !1029.79 <0.001***
Cut39 !16.05 !25.11 !7.34 <0.001*** !69.42 " 0.14 !1005.91 <0.001***
P. aquilinum frond
density (No m
!2
)
Untreated 6.30 !6.60 19.58 0.340 –––
Asulam !14.88 !28.19 !1.82 0.029* !21.18 " 0.19 !224.23 <0.001***
Bruise29 15.24 2.38 28.53 0.025* 8.93 " 0.18 95.04 <0.001***
Bruise39 34.70 21.60 48.08 <0.001*** 28.40 " 0.19 299.85 <0.001***
Cut29 !23.52 !35.37 !10.59 <0.001*** !29.83 " 0.18 !326.46 <0.001***
Cut39 !19.26 !30.59 !7.70 0.022* !25.56 " 0.17 !288.57 <0.001***
P. aquilinum frond length (cm) Untreated 32.16 !23.95 40.38 <0.001*** –––
Asulam !27.11 !34.80 !18.70 <0.001*** !59.27 " 0.12 !1014.34 <0.001***
Bruise29 !1.97 !10.17 5.81 0.621 !
34.13 " 0.11 !585.61 <0.001***
Bruise39 3.77 !4.31 11.87 0.338 !28.40 " 0.11 !488.15 <0.001***
Cut29 !16.07 !22.76 !8.94 <0.001*** !48.23 " 0.11 !884.87 <0.001***
Cut39 !21.98 !28.42 !15.41 <0.001*** !54.15 " 0.11 !1019.07 <0.001***
Shannon–Weiner Index Untreated !0.16 !0.42 0.10 0.230 –––
Asulam 0.20 !0.07 0.46 0.131 0.36 " 0.00 188.69 <0.001***
Bruise29 0.12 !0.15 0.38 0.365 0.27 " 0.00 144.67 <0.001***
Bruise39 !0.12 !0.39 0.14 0.345 0.03 " 0.00 17.23 <0.001***
Cut29 0.34 0.10 0.57 0.009* 0.49 " 0.00 275.09 <0.001***
Cut39 0.31 0.09 0.52 0.006* 0.46 "
0.00 268.21 <0.001***
Significance of P
MCMC
: *P < 0.05, **P < 0.01, ***P < 0.001.
© 2016 European Weed Research Society
Bracken control strategies 5
However, it has also been established that P. aquilinum
recovers from single initial asulam applications (Robin-
son, 1986; Lowday & Marrs, 1992). The Bamford site
used here was treated with asulam by helicopter in
1990 (Marrs et al., 1992), and two of the results from
the untreated plots reported here suggest that they
Fig. 1 Modelled mean response and raw data of (A) bracken cover (%), (B) frond density (fronds m
!2
), (C) frond length (cm) and (D)
diversity (Shannon–Weiner) to bracken control treatments through time at Bamford Edge, Derbyshire. Treatment codes: Untr, untreated
experimental control; Cut92 and Cut93, cut twice and cut thrice yearly; Bruise92 and Bruise93, Bruise twice and thrice yearly; Asu-
lam, initial asulam application followed by annual spot spraying of all emergent fronds.
© 2016 European Weed Research Society
6 G Milligan et al.
conform to the generalisations of Alday et al. (2013).
At the start of the present study, 15 years after the
1990 asulam application, the P. aquilinum cover was
relatively low (~50%) but increased continuously to
~100% by 2013, and there was at least some species in
the understory (cover 5%) and diversity did not show
a significant decline through time. These results suggest
that the alternative stable state described as possible
for at least a 10-year period under experimental
conditions (Alday et al., 2013) might be happening
here, but that it has started to break down after about
15–20 years after the P. aquilinum was sprayed. This
suggests that the understory species are relatively resili-
ent, in that they maintain low cover values for at least
23 years. This suggests that the understory flora can
persist for a long time at low cover and provide a
reservoir for revegetation, if the P. aquilinum cover is
reduced in future control programmes. These conclu-
sions are speculative and it is possible that the reten-
tion of the understory species in our experimental
plots is a function of elevated side light and other
disturbance associated with the management of these
experimental plots. Nevertheless, our hypotheses must
be tested on the basis that the P. aquilinum infestation
is increasing in the untreated controls and the underly-
ing species diversity is low, but stable.
Here, we demonstrated that three P. aquilinum con-
trol treatments were extremely effective (asulam once
followed by annual spot spraying of individual fronds,
and cutting twice or thrice yearly), producing a com-
plete frond clearance within 8 years from the majority
of the experimental plot area; the remaining fronds
were restricted to the periphery of the treated areas,
Fig. 2 Change in the herbicide application intensity through time
at the whole-plot scale between 2009 and 2012 in the asulam
treatment at Bamford Edge, Derbyshire; intensity was assessed
by counting the number of squirts applied per plot during spot
spraying the entire plot (one squirt per frond). Mean values " SE
are presented.
Fig. 3 Frond distribution map of the
three most successful treatments in 2013
at the Bamford Edge, Derbyshire; maps
of each replicate blocks (A, B, C) are pre-
sented. Treatment codes: Cut92 and
Cut93, cut twice and thrice yearly; Asu-
lam, initial asulam application followed
by annual spot spraying of all emergent
fronds.
© 2016 European Weed Research Society
Bracken control strategies 7
growing in from the plot edges (Hypothesis 1 accepted
for these three treatments). The two bruising treat-
ments were ineffective (Hypothesis 1 rejected, see
Hypothesis 2 below). At this site, the asulam treatment
produced the fastest impact on P. aquilinum. A single
asulam application is usually very effective in reducing
initial P. aquilinum frond density and cover, but there
is often a relatively rapid recovery within 5–10 years
(Lowday & Marrs, 1992; Marrs et al., 1998). In previ-
ous long-term experimental studies of P. aquilinum
control where the cut twice yearly option and single
asulam treatments have been tested, complete eradica-
tion has not been achieved in 10 years (Alday et al.,
2013) and 18 years (Marrs et al., 1998); in these two
studies, the P. aquilinum frond reduction was to a very
low level but it was never zero. From a land manage-
ment perspective, Alday et al. (2013) suggested that
mechanical methods were preferred on sites where
there was deep litter layer and a depauperate under-
story, but herbicide use produced faster results and
was less expensive to implement where a diverse grass-
land understory community was present.
It is worth speculating on the possible cause of this
success on the Bamford site. One reason might be dif-
ferential sheep grazing pressure. Whilst the grazing
pressure was at the ESA prescription level of 0.5 sheep
ha
!1
, it was obvious that the sheep spent a great deal
of time in the cut and asulam-treated plots) and grazed
the vegetation to a relatively low height (<5 cm, Fig-
ures S1 and S3). It is also very difficult to prove com-
plete eradication, especially for weedy species like
P. aquilinum that persist and reproduce mainly though
rhizomes. Rhizomes can remain dormant for up to
10 years (Marrs & Watt, 2006), but there is no infor-
mation on the length of time that rhizomes persist
when they have been subject to considerable control
treatments. Nevertheless, true eradication cannot be
proven until a relatively long ‘P. aquilinum -free’ per-
iod has been demonstrated. An assessment of future
P. aquilinum recolonisation of these treated plots at
Bamford Edge might help in separating the relative
role of edge invasion versus recovery from extant dor-
mant rhizomes. The plot separators in this study were
bruised thrice annually and responded in the same way
as the bruised treatment plots, that is poor P. aquil-
inum control. Edge invasion was detected in both cut-
ting treatments from P. aquilinum in the bruised
pathways, but this was absent from the asulam plots
(Fig. 3) and this suggests that where cutting is used to
treat a subarea of P. aquilinum, continued asulam
treatment along the border will help keep it in check
and prevent re-invasion. This requires further testing.
These results also demonstrate that to get very good
P. aquilinum control, a long-term strategy needs to be
Fig. 4 Ordination biplots of the first two canonical axes (CC1,
CCA2 respectively) derived from the CCA showing: (A) five
treatments and their interactions with time (vectors) and (B) spe-
cies. The origin (0,0) represents the untreated controls in 2005.
(A) Treatments: Asulam, Bruise2 9 is twice-yearly bruising,
Bruise3 9 is thrice-yearly bruising, Cut29 is twice-yearly cut-
ting, and Cut39 is thrice-yearly cutting. (B) Main species group
descriptors in each quadrant are denoted in bold; species abbrevi-
ations: Ac, Agrostis capillaris; Ao, Anthoxanthum odoratum; Ap,
Aira praecox; Av, Agrostis vinealis; Cc, Carex caryophyllea; Cf,
Cerastium fontanum; Ci, Campylopus introflexus; Cp, Campylopus
pyriformis; Cpi, Carex pilulifera; Cq, Cladonia squamosa; Cs,
Carex spp.; Cv, Calluna vulgaris; Df, Deschampsia flexuosa; Ds,
Dicranum scoparium; Ea, Epilobium angustifolium; Fe, Fraxinus
excelsior; Fo, Festuca ovina; Fr, Festuca rubra; Fu, Filipendula
ulmaria; Gs, Galium saxatile; Hc, Hypnum cupressiforme; Hj,
Hypnum jutlandicum; Hl, Holcus lanatus; Lb, Lophocolea biden-
tata; Lc, Luzula campestris; Lm, Luzula multiflora; Ns, Nardus
stricta; Pa, Pteridium aquilinum; Pc, Ptilidium ciliare; Pe, Poten-
tilla erecta; Pf, Polytrichum formosum; Po, Poa annua; Pp, Pseu-
doscleropodium purum; Ps, Pleurozium schreberi; Ra, Rumex
acetosella; Rs, Rhytidiadelphus squarrosus; Rt, Rumex acetosa;
Sm,
Stellaria media; Vm, Vaccinium myrtillus; Vo, Veronica offici-
nalis. Species nomenclature: Atherton et al. (2010) for bryo-
phytes, Stace (2010) for vascular plants. Treatment codes: Untr,
Untreated experimental control; Cut92 and Cut93, cut twice
and thrice yearly; Bruise92 and Bruise93, Bruise twice and thrice
yearly; Asulam, initial asulam application followed by annual
spot spraying of all emergent fronds.
© 2016 European Weed Research Society
8 G Milligan et al.
adopted; here, an 8-year programme was needed, with
16 and 24 cuts were applied in the twice- and thrice-
yearly treatments, and an initial overspray plus seven
annual spot-sprays for the asulam treatment. To
achieve these results requires intensive management,
which is both expensive in time and resources (cutters,
fuel, labour, herbicides).
The effectiveness of the asulam treatment plus fol-
low-up spot spraying is further evidence of its value in
P. aquilinum control; it is the only effective, selective
herbicide available which can be applied from the air
over rough terrain at the present time (Robinson,
1986, 2000). Application techniques have been devel-
oped specifically to spot-spray P. aquilinum on rough
terrain, but they require a relatively large number of
operatives (Robinson, 2000; Figure S4). Moreover,
with spot spraying, the reduction in number of squirts
required per plot may follow the law of diminishing
returns. It may be easy to enforce this treatment under
experimental conditions, but it may be harder to main-
tain operator enthusiasm under practical land manage-
ment conditions.
Bruising was an effective means of P. aquilinum
control (Hypothesis 2)
Bruising is a relatively old technique that was com-
monly used before the development of suitable cutting
equipment and selective herbicides. As rugged cutters
and herbicides became available, bruising fell out of
favour, but is now seeing a resurgence in popularity,
especially for conservation use (Lewis et al., 1997;
Bacon et al., 2001). In the bruised plots, P. aquilinum
cover and frond density increased through time and
there was no reduction in frond length. Indeed, the
bruised plots resembled the untreated experimental
controls (Hypothesis 2 rejected, Figure S3). Neverthe-
less, there are accounts of bruising providing better
results than obtained here (Braid, 1959; Lewis et al.,
1997) and it is difficult to reconcile this.
There are, however, some advantages to bruising
over cutting, in that bruisers are more robust and can
be towed at faster speeds than cutters, and hence, more
land can be covered in the same time. Moreover, they
can be towed by heavy horses and this allows access to
some land that is inaccessible even to ATV vehicles
(Figure S4). Nevertheless, experience here shows that it
is not possible to guarantee good P. aquilinum control,
even with double and triple passes each year over an
8-year period. A further drawback to bruising is that
the bruising bars/flanges can (like cutters) damage and
uproot stones (R.H. Marrs pers. comm.). This is not
necessarily a problem, but clearly both are inappropri-
ate where there is archaeological interest (Pakeman
et al., 2005) and an herbicidal approach would be pre-
ferred. The variability in efficacy of bruising was noted
by Braid (1959), who stated ‘recently the maker of a
well-known bracken bruiser confessed his preference
for cutting (Henderson, 1954)’.
One reason for the poor performance reported here
is may be the lack of damage to the vascular strands
during bruising (Fig. 5), with physiological processes
(photosynthesis, transpiration) remaining at similar
rates above the bruise to untreated stands for several
Vascular strand
Vascular strand
Vascular strand
A
B
C
Fig. 5 Scanning electron microscope photographs illustrating sec-
tions of stem from three different bruised fronds showing damage
to external and soft tissue layers but intact vascular strands
within an endodermis-like structure (abstracted from Cox, 2007).
© 2016 European Weed Research Society
Bracken control strategies 9
weeks (Cox, 2007). We can, therefore, assume that car-
bohydrate and hormone flows to the rhizome continue,
where the latter should enforce bud dormancy and pre-
vent the new waves of frond emergence so obvious
after cutting (Lowday et al., 1983). This reduced frond
production will, therefore, result in much less removal
of rhizome resources. These preliminary speculations
require confirmation using tracers to study carbohy-
drate/nutrient fluxes between fronds and in intact and
bruised P. aquilinum.
Differing control treatments produce different
trajectories of change in the underlying plant
community? (Hypothesis 3)
The species detected here are representative of many
depauperate, calcifuge vegetation communities that
are widely distributed in the UK uplands at altitudes
below 600 m (Rodwell, 1992). The treatments applied
here did not affect species richness, but the Shannon–
Weiner diversity index was increased only by cutting
twice and thrice yearly. Both cutting and asulam
treatments produced a grassland community contain-
ing at least some species important for sheep grazing,
especially the grasses A. capillaris and F. ovina (Rawes
& Welch, 1969). The multivariate analyses, however,
indicated divergence in community composition within
the subordinate species through time between the
treatments, with the untreated control and both bruis-
ing treatments moving towards greater P. aquilinum
cover with a lower abundance of all other species.
The asulam and cutting treatments were most
effective, in that they both retained a greater species
complement and showed an increasing divergence
from both untreated plots and bruising treatments.
Nevertheless, whilst the asulam and cutting treatments
had many species in common, the asulam treatment
produced a sward with a greater component of annu-
als and ruderals (Aira praecox , Poa annua, Epilobium
angustifolium) with a relatively low grazing value,
whereas the cutting treatments had a greater cover of
typical upland, acid grassland species which will
enhance the grassland quality for sheep grazing
(Rawes & Welch, 1969). Hypothesis 3 is, therefore,
accepted.
Conclusions
This study has shown that under some circumstances,
it is possible to clear P. aquilinum from heavily infested
land using either cutting or herbicide, but it required a
considerable time (8 years here), and substantive
investment. However, a different vegetation trajectory
resulted, depending on the control strategy adopted.
Bruising was ineffective here. Nevertheless, we showed
that sustained weed control over a considerable time
period using either cutting or repeated applications of
an appropriate selective herbicide was extremely effec-
tive for P. aquilinum control. Similar approaches might
be appropriate for other intractable perennial weed
species but would need testing.
Acknowledgements
This experiment was set up under a DEFRA-funded
contract (BD1226), and continuation has been sup-
ported by both the Heather Trust and the Ecological
Continuity Trust. We thank J. Archdale and N. Taylor
for site access and logistic support, and M. Harris, M.
O’Connor, J. Ghorbani, K. Hatton and T. Tollitt for
technical support. Peter Gotham provided the bracken
bruiser, and asulam was provided by Bayer CropS-
cience Ltd and United Phosphorus Ltd.
References
ALDAY JG, COX ES, PAKEMAN RJ, HARRIS MPK, LE DUC
MG & MARRS RH (2013) Overcoming resista nce and
resilience of an invaded community is necessary for
effective restoration: a multi-site bracken-control study.
Journal of Applied Ecology 50, 156–167.
Anon (2011) Commission Implementing Regulation (EU)
No. 1045/2011 of 19 October 2011. Available at: http://
eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:
L:2011:275:0023:0024:EN:PDF (last accessed 18/
November 2013).
A
THERTON I, BOSANQUET S&LAWLEY M (2010) Mosses and
Liverworts of Britain and Ireland – A Field Guide. British
Bryological Society, Plymouth, UK.
BACON J, BARNES N, COLESHAW T, ROBINSON T&TITHER J,
(2001) Practical solutions. Equipment, techniques and ideas
for wildlife management. Forum for the Application of
Conservation Techniques, English Nature, Peterborough,
UK.
B
OLKER BM, BROOKS ME, CLARK CJ et al. (2009)
Generalized linear mixed models: a practical guide for
ecology and evolution. Trends in Ecology and Evolution 24,
127–135.
B
RAID KW (1959) Bracken: A Review of the Literature . CAB,
Hurley, UK.
C
HAPMAN DS, TERMANSEN M, QUINN CH et al. (2009)
Modelling the coupled dynamics of moorland management
and upland vegetation. Journal of Applied Ecology 46, 278–
288.
C
OX ES (2007) The implications and effects of bracken
(Pteridium aquilinum (L.) Kuhn) control on species
diversity, re-vegetation and bracken performance. PhD
thesis, University of Liverpool, Liverpool, UK.
C
RAWLEY MJ (2013) The R Book, 2nd edn. Wiley, Chichester,
UK.
G
UREVITCH J&CHESTER JST (1986) Analysis of repeated-
measures experiments. Ecology 67, 255.
© 2016 European Weed Research Society
10 G Milligan et al.
HADFIELD JD (2010) MCMC methods for multi-response
generalized linear mixed models: the MCMCglmm R
Package. Journal of Statistical Software 33,1–22.
H
ENDERSON G (1954) Mechanisation of bracken eradication.
Journal of the Institute of British Agricultural Engineers 3,
3–44.
L
E DUC MG, PAKEMAN RJ & MARRS RH (2003) Changes in
the rhizome system of bracken subjected to long-term
experimental treatment. Journal of Applied Ecology 40,
508–522.
L
EGENDRE P&GALLAGHER ED (2001) Ecologically
meaningful transformations for ordination of species data.
Oecologia 129, 271–280.
L
EWIS N, GOTHAM P, OTTERBURN B, OVERBURY T, SHEPHERD
P&BACON J (1997) Bracken breaking – a bruising battle.
Enact 5, 19–22.
L
OWDAY JE & MARRS RH (1992) Control of bracken and the
restoration of heathland. I. Control of bracken. Journal of
Applied Ecology 29, 195–203.
L
OWDAY JE, MARRS RH & NEVISON GB (1983) Some of the
effects of cutting bracken (Pteridium aquilinum (L.) Kuhn)
at different times during the summer. Journal of
Environmental Management 17, 373–380.
M
ARRS RH & LOWDAY JE (1992) Control of bracken and the
restoration of heathland. II. Regeneration of the heathland
community. Journal of Applied Ecology 29, 204–211.
M
ARRS RH & WATT AS (2006) Biological flora of the British
Isles: Pteridium aquilinum (L.) Kuhn. Journal of Ecology
94, 1272–1321.
M
ARRS RH, FROST AJ, PLANT RA & LUNNIS P (1992) Aerial
applications of asulam: a bioassay technique for assessing
buffer zones to protect sensitive sites in upland Britain.
Biological Conservation 59, 19–23.
M
ARRS RH, JOHNSON SW & LE DUC MG (1998) Control of
bracken and the restoration of heathland. VIII. The
regeneration of the heathland community after 18 years of
continued bracken control or 6 years of control followed
by recovery. Journal of Applied Ecology 35, 857–870.
M
ARRS RH, GALTRESS K, TONG C et al. (2007) Conflicts
between competing conservation goals, biodiversity or
ecosystem services: quantification of element losses and
species recruitment using a moorland-bracken model
system. Journal of Environmental Management 85, 1034–
1047.
M
IATTO RC, SILVA LA, SILVA-MATO DM & MARRS RH
(2011) Woody vegetation structure of Brazilian Cerrado
invaded by Pteridium arachnoideum (Kaulf.) Maxon
(Dennstaedtiaceae). Flora 206, 757–762.
N
AYLOR REL, ed. (2002) Weed Management Handbook, 9th
edn. Wiley-Blackwell, Oxford, UK.
O
KSANEN J, BLANCHET FG, KINDT R et al. (2011) vegan:
Community Ecology Package, R package version 2.0-2.
Available at: http://CRAN.R-project.org/package=vegan
(last accessed 1 October 2013).
P
AKEMAN RJ & MARRS RH (1992) The conservation value of
bracken Pteridium aquilinum (L.) Kuhn-dominated
communities in the UK, and an assessment of the
ecological impact of bracken expansion or its removal.
Biological Conservation 62, 101–114.
P
AKEMAN RJ, LE DUC MG & MARRS RH (2000) Bracken
distribution in Great Britain: strategies for its control and
the sustainable management of marginal land. Annals of
Botany 85B, 37–46.
P
AKEMAN RJ, LE DUC MG, MARRS RH, CONDLIFFE I,
E
NRIGHT D&SMALLSHIRE D (2005) Bracken control,
vegetation restoration and land management. Rural
Development Services Technical Advice Note No. 23.
Defra, London, UK. Available at: http://
www.defra.gov.uk/rds/publications/technical/tan_23.pdf.
P
INHEIRO JC & BATES DM (2000) Mixed Effects Models in S
and S-Plus . Springer Verlag, New York, NY, USA.
R Development Core Team (2011) R: A Language and
Environment for Statistical Computing. R Foundation for
Statistical Computing, Vienna, Austria. Available at:
http://www.R-project.org/ (last accessed 1 October 2013).
R
AWES M&WELCH D (1969) Upland productivity of vegetation
and sheep at Moor House National Nature Reserves,
Westmorland, England. Oikos Supplementum 11,1–69.
R
OBINSON RC (1986) Practical herbicide use for bracken
control. In: Bracken: Ecology, Land-use and Control (eds
RT S
MITH & JA TAYLOR), 331–340. Parthenon Press,
Carnforth, UK.
R
OBINSON RC (2000) The means to large-scale, effective
bracken clearance in the UK. In: Bracken Fern: Toxicity,
Biology and Control (eds JA T
AYLOR & RT SMITH), 156–
162. International Bracken Group, Aberystwyth, UK.
R
ODWELL JS (1992) British Plant Communities, Volume 3.
Grassland and Montane Communities. Cambridge
University Press, Cambridge, UK.
S
TACE C (2010) New Flora of the British Isles, 3rd edn. CUP,
Cambridge, UK.
USEPA (1995) R.E.D. Facts: Asulam. United States
Environmental Protection Agency. Available at: http://
www/epa/gov/oppsrrd1/REDs/factsheets/0265fact.pdf (last
accessed 1 April 2013).
W
ILLIAMS GH & FOLEY A (1976) Seasonal variations in the
carbohydrate content of bracken. Botanical Journal of the
Linnean Society 73, 87–93.
Supporting Information
Additional Supporting Information may be found in
the online version of this article:
Figure S1 Experiment layout of the Bamford Edge
Experiment.
Figure S2 Bruising in practice: (a) Film of P. aquil-
inum bruising at Bamford Edge; (b–e) various types of
bruising equipment.
Figure S3 Photographs of the six control treatments
in Block C of the Bamford Edge experiment in 2013,
the year after treatments were finished (Photos by W.
Chiba).
Figure S4 Application of asulam by various tech-
niques.
© 2016 European Weed Research Society
Bracken control strategies 11
