Context-dependent induction of allelopathy in plants under competition

Abstract

Some plants use allelopathy to compete against neighbouring plants, and the ability to induce allelopathic compound production in response to competition is hypothesized to be adaptive, as plants can save costs of metabolite production in the absence of competitors. However, whether plants induce allelopathy has rarely been explored so far. We studied the inducibility of polyacetylenes – putative allelopathic compounds in Solidago altissima – in response to competition. Polyacetylenes were found in natural soil surrounding S. altissima patches within the range of concentration known to inhibit competitor growth. Individual S. altissima plants with higher polyacetylene concentration in roots suppressed the growth of the competitor plants more, suggesting that root polyacetylene levels proximate plants’ allelopathic capacity. Competition induced polyacetylenes in a context‐dependent manner: Whereas introduced Japanese and Australian populations of S. altissima had higher constitutive concentration of polyacetylenes than the native North American populations, inducibility was observed only in Australian plants, where the population is still at an early stage of invasion. Also, induction became more prominent under nutrient depletion, where enhanced allelopathy may be particularly beneficial for suppressing a competitor's exploitative capacity. Finally, we found weak evidence for a tradeoff between constitutive and induced polyacetylenes. The observed patterns suggest that allelopathic plants could respond to competition by inducing allelochemical production, but the benefit of such plasticity may vary across time and space. Shifts in competitor communities in introduced range over time may shape plant's plastic responses to competition, while variation in resource availability may alter competitive environment to influence the degree to which plants induce allelopathy. This article is protected by copyright. All rights reserved.

Full text

www.oikosjournal.org
OIKOS
Oikos
1
––––––––––––––––––––––––––––––––––––––––
© 2019 e Authors. Oikos © 2019 Nordic Society Oikos
Subject Editor: Martijn Bezemer
Editor-in-Chief: Dries Bonte
Accepted 4 June 2019
00: 1–11, 2019
do i: 10.1111/oik. 06389
doi: 10.1111/oik.06389 00 1–11
Some plants use allelopathy to compete against neighbouring plants, and the ability
to induce allelopathic compound production in response to competition is hypoth-
esized to be adaptive, as plants can save costs of metabolite production in the absence
of competitors. However, whether plants induce allelopathy has rarely been explored
so far.
We studied the inducibility of polyacetylenes – putative allelopathic compounds
in Solidago altissima – in response to competition. Polyacetylenes were found in natu-
ral soil surrounding S. altissima patches within the range of concentration known to
inhibit competitor growth. Individual S. altissima plants with higher polyacetylene
concentration in roots suppressed the growth of the competitor plants more, suggest-
ing that root polyacetylene levels proximate plants’ allelopathic capacity. Competition
induced polyacetylenes in a context-dependent manner: Whereas introduced Japanese
and Australian populations of S. altissima had higher constitutive concentration of
polyacetylenes than the native North American populations, inducibility was observed
only in Australian plants, where the population is still at an early stage of invasion.
Also, induction became more prominent under nutrient depletion, where enhanced
allelopathy may be particularly benecial for suppressing a competitor’s exploitative
capacity. Finally, we found weak evidence for a tradeo between constitutive and
induced polyacetylenes.
e observed patterns suggest that allelopathic plants could respond to competition
by inducing allelochemical production, but the benet of such plasticity may vary
across time and space. Shifts in competitor communities in introduced range over
time may shape plant’s plastic responses to competition, while variation in resource
availability may alter competitive environment to inuence the degree to which plants
induce allelopathy.
Keywords: allelochemicals, invasive plants, novel weapons hypothesis, plasticity,
polyacetylenes, Solidago altissima
Context-dependent induction of allelopathy in plants
under competition
AkaneUesugi, RobertJohnson and AndréKessler
A. Uesugi (https://orcid.org/0000-0003-3363-5312) ✉ (akane.uesugi@monash.edu), School of Biological Sciences, Monash Univ., Building 18, Victoria
3800, Australia. – R. Johnson, Dept of Science, Mathematics and Technology, Medaille College, Bualo, NY, USA. – A. Kessler, Dept of Ecology and
Evolutionary Biology, Cornell University, Ithaca, NY, USA.
Research

2
Introduction
Plants induce secondary metabolites in response to ever-
changing biotic environments (Karban and Baldwin 1997),
and these plastic responses are often thought to be adaptive
as they can reduce costs of metabolite production in the
absence of stress (Karban and Myers 1989, Agrawalet al.
1999). For instance, plants may increase resistance to her-
bivory by inducing defensive compound production when
damaged, while expressing relatively low levels of constitutive
resistance in the absence of herbivory. While plant-induced
responses to herbivores and pathogens have been extensively
studied (Karban and Baldwin 1997), less is known about
the induction of secondary metabolites in response to other
important biotic interactions, such as competition (Rasher
and Hay 2014).
Besides exploiting limited resources by growing faster
and larger, some plants compete through allelopathy – an
interference of competitor growth through release of chemi-
cal compounds into the environment (Duke 2010). ese
allelopathic compounds that are released as root exudates,
volatile organic compounds, leaf leachate or leaf litter, may
suppress neighbour growth directly, or indirectly by impact-
ing soil microbial mutualists and altering the nutrient
availability (Roberts and Anderson 2001, Inderjitetal. 2011).
Production of allelopathic compounds is likely to be ener-
getically costly (Uesugi and Kessler 2013), and the balance
of benets and costs of allelopathy are likely to vary across
time and space due to variation in competitor communities.
For example, allelopathy may have little impact in disturbed
habitat where plants encounter no or little interspecic com-
petition, but may be benecial in habitats with intense com-
petition. In an introduced range, allelopathic plants may
gain additional advantage during the early phase of invasion,
because their allelochemicals may be highly eective against
naïve recipient communities that did not coevolve with the
novel compounds (the novel weapons hypothesis; Callaway
and Ridenour 2004). e same allelochemicals, however,
may become less eective as the recipient communities evolve
to tolerate the compounds (Callaway and Ridenour 2004,
Lankauetal. 2009). In such heterogeneous competitive envi-
ronments, selection may favour the evolution of induced pro-
duction of allelolopathic compounds, rather than constitutive
(i.e. constant) production of costly allelochemicals at high
levels (Novoplansky 2009, Kegge and Pierik 2010). Despite
its adaptive potential, inducibility of allelopathy in response
to competition remains poorly understood (Songetal. 2008,
Lankau and Kliebenstein 2009, Kato-Noguchi 2011, Rasher
and Hay 2014).
Expression of inducibility may also depend on abi-
otic environmental conditions in which individual plants
compete. For example, the ability to induce production and
release allelochemicals may be particularly benecial under
resource limitation (e.g. low light, nutrient and water), as
suppression of competitor growth via enhanced allelopathy
would limit the exploitative capacity of neighbouring plants
(Songet al. 2008). us, induction may be predicted to be
higher in low-resource environments than in resource-rich
environments. Such context-dependent induction would
require that plants can detect the competitor presence inde-
pendently of resource depletion caused by competition.
Plants may do so by perceiving plant-specic chemical cues
(e.g. volatile infochemicals and root exudates, Pierik et al.
2013) and/or change in light environment due to shading
(i.e. a change in red to far-red ratio of light, Izaguirreetal.
2006). Previous studies in Oryza sativa indicated that nutri-
ent depletion and exposure to competitor’s root exudates can
independently induce allelopathy in a hydroponic system
(Songetal. 2008, Kato-Noguchi 2011). However, whether
plant’s induction behaviour depends on nutrient availability
has not been explored.
If induction of allelochemical production is a cost-saving
mechanism, we expect inducible allelopathy to trade o with
constitutive allelopathy (Korichevaetal. 2004, Morrisetal.
2006). Within a single population, a tradeo should arise if
there is an upper limit to the level of allelopathy – genotypes
that constitutively express high allelopathy may have less room
to induce further, and vice versa (Lankau and Kliebenstein
2009). Consequently, at a geographic scale, plant populations
that colonize heterogeneous competitive landscapes should
be selected for high induced and low constitutive allelopathy,
whereas in populations that consistently face severe compe-
tition, the opposite may be favoured. Alternatively, theories
of adaptive allocation suggest a variety of ways in which
tradeos between energetically costly traits may be masked
by individual-, and population-level variation in allocation
hierarchies (van Noordwijk and de Jong 1986). For instance,
introduced populations that evolved in the absence of her-
bivory may allocate a larger pool of resources into allelopathy
than native populations, given that allelopathy and herbivore
defense trade o (Blossey and Notzold 1995, Uesugiet al.
2017). In this case, introduced populations may exhibit
higher constitutive and induced allelopathy than native pop-
ulations if both forms of allelopathy are favoured in the novel
range (Callaway and Ridenour 2004). While the presence of
tradeos has been explored extensively for plant resistance
against herbivores and pathogens (Koricheva et al. 2004,
Morriset al. 2006, Kempelet al. 2011), no study, thus far,
has explored tradeos for allelopathy.
Despite its adaptive potential, studies examining the
inducibility of allelochemicals are rare, in part due to a di-
culty in providing concrete evidence for the allelopathic func-
tion of compounds in nature (Duke 2010). To demonstrate
allelopathy, one must rst identify the candidate allelochemi-
cals through bioassays, as any crude extract at a high-enough
concentration can show inhibitory eects (Duke 2010).
Second, we must demonstrate that the compounds in
question are present in the soil at the concentration known to
inhibit competitor growth, since many secondary metabolites
can rapidly degrade and/or be adsorbed to soil particles and
become unavailable (Blairetal. 2005, Kauretal. 2009, Duke
2010). ird, bioactivity of the putative allelochemicals in

3
soil must be demonstrated to show the ecological relevance
of the compounds in competitive interactions. Studies have
applied activated charcoal to neutralize secondary metabolites
in soil in order to tease apart the allelopathic and the direct
exploitative eects of target plants on competitor growth
(Inderjit and Callaway 2003). However, this technique can
produce large experimental artefacts, as charcoal can alter
nutrient availability in the soil that directly inuence plant
growth in some cases (Lau et al. 2008). Alternatively, one
might contrast plant genotypes that vary in allelochemical
production (Belz and Hurle 2005), or transgenic plants that
are silenced for the production of allelochemicals, for their
ability to inhibit neighbour growth in the eld (Baldwin
2003). However, studies using transgenic approaches are rare
so far (but see Inderjitetal. 2009). Due to these methodolog-
ical challenges, ‘good’ examples of allelopathy are currently
limited to a few cases (e.g. juglone in Juglans nigra; Jose and
Gillespie 1998, sorgoleone in Sorghum spp.; Nimbaletal.
1996, and DIBOA/BOA in wheat; Huang et al. 2003,
glucosinolates in Alliaria petiolata; Cantoretal. 2011).
e tall goldenrod, Solidago altissima (Asteraceae),
provides a model system to study induced allelopathy in
response to competition in ecological context. Several lines of
evidence suggest that putative allelopathic compounds (poly-
acetylenes) produced in S. altissima roots have roles in com-
petition under experimental and natural environments. First,
only the fraction of root extracts containing polyacetylenes
inhibited germination and seedling growth of Oryza sativa
(Kobayashi et al. 1980). Four polyacetylene compounds,
including a major compound cis-dehydromatricaria ester
(DME), were subsequently identied, and demonstrated
to have similar inhibitory eects on several other competi-
tor species in petri-dish experiments (Kobayashietal. 1980,
Sawabeetal. 1999, Johnsonetal. 2010, Uesugi and Kessler
2013) and soil experiments (Kobayashietal. 1980). Second,
DME is found in natural soil within a S. altissima patch
(Kobayashietal. 1980), although its spatial variation in soil
concentration is unclear, as the retention of the compound
could heavily depend on the soil type (Kobayashietal. 2004,
Itoet al. 1998). Finally, a long-term removal of herbivory
drove a rapid evolution of increased competitive ability, as
well as increase in root concentration of polyacetylenes in
S. altissima (Uesugi and Kessler 2013). is evolutionary
shift suggests that selection favours increased resource alloca-
tion towards competition via allelopathy in the absence of
herbivory (Uesugietal. 2017).
Using this system, we rst ask if DME is present in nat-
ural S. altissima patches at ecologically relevant concentra-
tions, and how the concentrations vary spatially. Second,
in a pot experiment, we test how the root polyacetylenes
concentration in S. altissima correlates with the growth of a
common grass competitor (Poa pratensis). ird, we test if S.
altissima responds to interspecic competition by inducing
root polyacetylene production. Induction was tested in two
separate common garden experiments that used S. altissima
from a single population in the native North American
range, and from the native and introduced (Japanese and
Australian) populations to sample a wide range of plant
phenotypes. Fourth, we test how nutrient availability inu-
ences the induction behaviour of root polyacetylenes in
response to competition in a factorial experiment. Finally,
we test if constitutive and induced expression of polyacety-
lenes negatively covary, as predicted under a model of strong
tradeos (Korichevaetal. 2004, Morrisetal 2006).
Material and methods
Study system
Solidago altissima is a perennial forb that dominates mid- to
late-successional stages of old elds (Werneret al. 1980).
It is native to eastern North America, but has been intro-
duced globally, including widespread and highly invasive
populations in Japan (Fukuda 1982), and established but
sparse populations in Australia (Atlas of Living Australia).
In its native range, S. altissima commonly competes with an
understory grass, Poa pratensis. Field experiments show that
S. altissima strongly reduces the density of P. pratensis (Carson
and Root 2000), and the laboratory assays demonstrate that
DME extracted from S. altissima inhibits germination and
seedling growth of P. pratensis in agar media (Uesugi and
Kessler 2013). Introduced populations of S. altissima are also
likely to compete with P. pratensis, which is common in Japan
and Australia (Holm et al. 1979). Based on its likelihood
of being a major competitor of S. altissima in both native
and introduced ranges, and its susceptibility to S. altissima
polyacetylenes, we chose P. pratensis as a model interspecic
competitor in the following experiments.
Field soil DME survey
We tested for the spatial variation in soil DME concentra-
tion in ten S. altissima sites, spanning 186 km of its native
range in western and central New York, USA. Soils across this
range are typically silty loams of glacial origin. To examine
microscale variation in DME concentration within each site,
we collected soil samples in September, 2007 at two distances
– 0 and 25 cm away from the margin of S. altissima clones.
Two replicated soil samples were collected from each distance
per site using a hand-held soil corer to a depth of 10 cm. Field
samples were maintained on cold packs and stored at −80°C
in the laboratory. In April 2008, samples were dried at room
temperature overnight in a fume hood to provide a uniform
consistency and limit thermal binding of metabolites to soil
particles. Samples were sifted to remove organic matter; a
2000 mg fraction was taken from each and transferred to a
capped scintillation vial for DME analysis.
Induction experiments
We conducted two experiments to estimate the eect of
competition on polyacetylene concentration in roots using
dierent sets of S. altissima genotypes: one from a single
eld in the native range (hereafter called a ‘single-population

4
experiment’), and the other from multiple populations across
its native and introduced ranges (‘multi-range experiment’).
e single-population experiment was conducted as a
part of study by Uesugi and Kessler (2013) that showed a
rapid evolution of constitutive polyacetylene production
in response to a long-term removal of herbivory. Here, we
examine if the inducibility of polyacetylenes has evolved in
response to herbivore removal. e eld experiment was
conducted in Whipple Farm, Tompkins Co., Ithaca, NY,
USA, where half of 12 plots (5 × 5 m2 each) were sprayed
with insecticide to remove herbivory for 12 years, while the
other half were exposed to ambient insect herbivory (see
Uesugi and Kessler 2013 for details). Twenty-nine geno-
types originated from insecticide sprayed plots, while 30
genotypes were taken from plots that experienced ambient
herbivory (control plots). In April 2010, we conducted a
common garden experiment on the roof top of Corson Hall,
Cornell University, NY. Four ramets from each genotype
were clonally propagated, and each ramet was grown in a
20-cm pot containing 1:3 ratio of sand and potting soil. e
four clones from each genotype were split into two compe-
tition treatments (no-competition control and inter-specic
competition with P. pratensis). In control, a single S. altissima
plant was placed in the centre of each pot, whereas in com-
petition treatment, a S. altissima plant was surrounded by
P. pratensis plants grown from ca 200 seeds sprinkled evenly
on the soil surface. Poa pratensis plants grew within 10 cm of
the target S. altissima plants in all pots.
Root samples were collected for secondary metabolite
analyses two months after the start of common garden exper-
iments. Approximately 200 mg of root tissue was removed
from living target plants by clipping a few strands of lateral
roots, ash-frozen with liquid nitrogen, and stored at −80°C
for later analyses. At the end of the ve-month growing
season, both S. altissima and the competitor plants were har-
vested, and dried at 45°C for 48 h. Biomass of shoots, ramets,
roots and rhizomes of S. altissima were separately measured.
For P. pratensis, we measured total aboveground biomass of all
individuals growing within each pot. Belowground biomass
of P. pratensis could not be measured accurately, as its ne
roots were easily lost during harvest.
e multi-range experiment was conducted in 2016 to
sample a wider range of plant phenotypes. We used S. altissima
genotypes originating from across the species’ geographic
distribution along the latitudinal gradients of native North
American range and along the introduced ranges of Japan
and Australia (Supplementary material Appendix 1 Table
A1). Eight to nine populations per range, and four genotypes
per population were sampled (total of 105 genotypes), and
each genotype was clonally replicated by four. A common
garden experiment (at ambient temperature of 22–28°C, and
light/dark cycle of L:D = 12:12) started in March 2016 in the
Plant Science Complex greenhouse at Monash University,
VIC, Australia. Each plant grew in a 20-cm pot with 1:3
ratio of sand and potting soil. e competition treatments
mirrored those described above. Two replicates per genotype
grew alone (as no-competition controls), and the other two
grew under competition from P. pratensis. Root samples were
collected and stored for polyacetylene analyses two months
after the start of the common garden experiments.
Nutrient experiment
To determine if soil nutrient depletion induces polyacety-
lenes, and if nutrient availability inuence plastic responses to
competitor presence, we examined S. altissima plant responses
in a 2 × 2 factorial experiment with two nutrient treatments
(high and low), and two competition treatments (with and
without P. pratensis). e experiment was conducted in July
2015, under the same growing conditions as the multi-range
experiment. Eight S. altissima genotypes originating from
New York, USA, were each clonally replicated eight times,
and spread evenly across the factorial treatments. Nutrient
treatments started two weeks after planting, where high- and
low-nutrient-treatment pots received 200 and 10%, respec-
tively, of full-strength Hoagland solution, 100 ml twice a
week until harvest. After two months of growth, root samples
were collected and stored as above.
Secondary metabolite extraction
Soil samples from the Field soil DME survey (2000 mg
sample–1) were each extracted into 3.0 ml dichloromethane
(DCM), with 100 µl of 1.0 mM abietic acid in MeOH as
internal standard. Each sample was sonicated for 6 min and
a 2.0 ml aliquot was ltered through glasswool + NaSO4 to
clarify and dry; the DCM aliquots were then evaporated
under a fume hood and residue resuspended into 0.5 ml
MeOH for HPLC analysis.
Root samples from Induction and Nutrient experiments
(200 mg fresh mass per sample) were extracted in 1.0 ml of
90% methanol using a tissue homogenizer. Samples were
sonicated in extraction buer for 6 min, and left in the dark
at room temperature for 24 h. Samples were centrifuged and
0.5 ml aliquot was ltered with 0.45 µm syringe lter.
HPLC analysis of polyacetylenes
Soil and root samples prepared in 2008 and 2010 were
analysed at Cornell University with HPLC with C18 reverse-
phase column (3 µm, 150 × 4.6 mm). e elution system
consisting of solvents (A) 0.25% H3PO4 in water (pH 2.2)
and (B) acetonitrile was: 0–5 min, 0–20% of B, 5–35 min,
20–95% of B and 35–45 min, 95% of B, with a ow rate
of 0.7 ml min–1 and injection volume of 15 µl (Uesugietal.
2013). Root samples collected in 2015 and 2016 were
analysed at Monash University using Agilent Innity
1260 equipped with C18 reverse-phase column (EC-C18,
2.7 μm, 150 × 3.0 mm). e elution method was modied
from the above, and was: 0–5 min, 0–20% of B; 5–25 min,
20–95% of B and 25–30 min, 95% of B, with a ow rate
of 0.5 ml min–1 and injection volume of 2 µl. We identied
peaks of polyacetylene compounds using UV spectra with

5
quantication at 254 nm (Sawabe et al. 2000). e abietic
acid internal standard was quantied at 230 nm.
Statistical analysis
All analyses were conducted using RStudio (ver. 1.0.44,
< www.r-project.org >), except when stated otherwise.
Does soil DME concentration vary spatially?
DME concentration in soil sample was calculated as peak area
of DME relative to that for abietic acid internal standard,
and expressed as ppm abietic acid equivalent. e value was
square-root transformed to improve normality, and tested
for the site, distance from the S. altissima plants, and their
interaction using ANOVA.
Does polyacetylene concentration correlate with
competitor growth?
Using data from the single-population experiments, we
examined the relationship between root polyacetylene con-
centration and the growth of P. pratensis. Relative concentra-
tion of all polyacetylene compounds in a root sample was
expressed as total polyacetylene peak areas relative to fresh tis-
sue mass. e total biomass of S. altissima plant was included
to control for the eect of biomass on competitive outcome,
as biomass is thought to be a major driver of competition in
many plants (Goldberg and Landa 1991).
In a linear mixed model (lmer function in package lme4;
Bates et al. 2015), total concentrations of polyacetylenes
and S. altissima biomass were modelled as xed eects, and
S. altissima genotype was entered as random eect. Semi-
partial R2 for each xed predictor was determined using
r2beta function in r2glmm package (Jaeger 2017). For visu-
alisation of the results, we extracted the partial eect of each
xed factor on P. pratensis biomass using remef function in
the piecewiseSEM package (Lefcheck 2016).
Does competition induce root polyacetylene production?
To allow comparisons between single-population and
multi-range experiments that used dierent conditions for
HPLC analyses, values of total polyacetylene were scaled
to unit variance (mean = 0, SD = 1). We tested the eect
of competition on root polyacetylene concentration using
a linear mixed model with competition treatment, plant
origin and their interactions as xed eects, and plant
genotype as a random eect. p-values were estimated using
the Anova function in package car (Fox and Weisberg
2011). Post hoc analyses using lsmeans function (package
emmeans, Lenth 2019) were conducted to test induction
within the origin.
Do soil nutrient levels affect plant induced-response to
competitors?
Eects of nutrient and competition on root secondary
metabolites were examined with a linear mixed model with
both treatments and their interaction as xed eects, and
plant genotype as a random eect. Post hoc analyses using the
lsmeans function were conducted to test induction within
each nutrient treatment.
Do constitutive and induced levels of secondary metabolites
trade off?
Tradeos between constitutive and induced expression of
polyacetylenes were tested using the data from the single-
population and multi-range experiments. In the multi-range
experiment, the analyses were done separately for each range,
as well as across the ranges. For each analysis, we conducted
a regression analysis while controlling for potential spurious
correlations due to measurement error and sampling variation
(outlined and scripted for MATLAB by Morrisetal. 2006).
Data deposition
Data are available from the Dryad Digital Repository: < http://
dx.doi.org/10.5061/dryad.7b90m0b > (Uesugietal. 2019).
Results
Does soil DME concentration vary spatially?
DME concentration in natural soil varied across sites
(F9,23 = 3.2, p = 0.011), ranging from undetectable level
to 17.3 ppm (Fig. 1). Within a site, the concentra-
tion was greater at the margin of Solidago altissima
clones (mean = 6.3 ± 4.8 ppm, range = 0–17.3 ppm) than
25 cm away from the clone edge (mean = 1.8 ± 2.1 ppm,
range = 0–7.8 ppm, F1,23 = 28.6, p < 0.0001). No interaction
between site and location was observed (F9,23 = 0.9, p = 0.5).
Does root polyacetylene concentration correlate
with competitor growth?
e biomass of Poa pratensis was negatively correlated with
the biomass of the target S. altissima plant (coef = −0.18,
χ2 = 15.4, p < 0.0001, partial R2 = 0.14, Fig. 2b), and the
total polyacetylene concentration (coef = −0.79, χ2 = 6.2,
p = 0.013, partial R2 = 0.08, Fig. 2a). Overall, the target
S. altissima biomass and polyacetylene production explained
21.4% of variance in the competitor biomass.
Does competition induce root polyacetylenes?
Polyacetylene levels were strongly aected by plant origin in
both experiments, with higher overall concentration in the
insecticide-sprayed populations than in the control popula-
tions within the single-population experiment (χ2 = 19.3,
p < 0.0001; results reported in Uesugi and Kessler 2013,
Fig. 3a), and introduced Japanese and Australian popula-
tions than native North American populations in the multi-
range experiment (χ2 = 32.3, p < 0.0001; Fig. 3b). In contrast,
competition treatment had varying eect on polyacetylene
concentration depending on experiments and plant origins.
Overall eect of competition was signicant in the multi-
range experiment (χ2 = 4.6, p = 0.031; Fig. 3b), but not in

6
single-population experiment (Fig. 3a). Competition treat-
ment and plant origin interacted in the multi-range experi-
ment (χ2 = 9.9, p = 0.007). A post hoc test revealed that
induction was limited to the Australian populations (t = 3.39,
p = 0.01, Fig. 3b).
Do soil nutrient levels affect plant induced-response
to competitors?
Nutrient treatment did not have any eect on polyacetylene
concentration (χ2 = 0.26, p = 0.12), but competition treat-
ment increased the concentration (χ2 = 10.57, p = 0.001,
Fig. 4). Although the interaction between nutrient and com-
petition was not signicant (χ2 = 2.44, p = 0.12), the post
hoc tests revealed that the induced response to competition
was signicant under the low nutrient treatment (t = 3.42,
p = 0.007), but not in the high nutrient treatment (t = 1.16,
p = 0.65).
Do constitutive and induced levels of root
secondary metabolites trade off?
Overall, constitutive and induced levels of polyacetylenes
tended to negatively correlate in both experiments (Fig. 5), but
after controlling for spurious correlation (Morrisetal. 2006),
the negative correlation was marginally signicant in the
single-population experiment (observed correlation = −0.51,
one-sided p = 0.078, Fig. 5a). In the multi-range experi-
ment, Australian populations showed negative correlation
(observed correlation = −0.56, one-sided p = 0.049), but
0
5
10
15
20
34B1 AL BL1 BL2 BR1 BR2MZ10WE1 WENC1WENC2 WRD
Site
Soil DME concentration (ppm)
Figure1. DME concentration in soil samples within 11 sites across western and central NY, USA. Light grey boxes indicate DME concen-
tration at the margin of S. altissima clone (0 cm from the plant) and dark grey boxes represent concentration 25 cm away from the clone
margin. Box plots show median, upper and lower quartiles and extremes.
coef = − 0.79
p = 0.013
(a) (b)
10
15
20
25
−3 −2 −1 012
Polyacetylene [ ]
Partial effects
coef = − 0.18
p < 0.0001
10
15
20
25
−2 −1 012
Solidago biomass
Figure2. e partial eects of (a) total polyacetylene concentration and (b) Solidago plant biomass on P. pratensis aboveground biomass.
Partial eects were calculated by controlling for other xed eects, and were plotted against standardised values of each xed term for
comparison.

7
North American and Japanese populations did not (North
America: observed correlation = −0.50, one-sided p = 0.12,
Japan: observed correlation = −0.05, one-sided p = 0.78),
resulting in non-signicant correlation across the ranges
(observed correlation = −0.19, one-sided p = 0.56, Fig. 5b).
Discussion
Do root polyacetylenes have allelopathic effect
on P. pratensis?
Polyacetylenes have been previously suggested to have an alle-
lopathic function in Solidago altissima based on their inhibi-
tory eects in petri dish and soil assays (Kobayashietal. 1980,
Johnson et al. 2010, Uesugi and Kessler 2013), and their
evolutionary response to the relaxation of herbivory that pre-
sumably intensied selection for competitive ability (Uesugi
and Kessler 2013). Our study provides additional evidence
for their allelopathic eect in nature, by rst conrming
the presence of a major polyacetylene, DME, in natural soil
surrounding S. altissima plants. On average, soil DME con-
centration at the margin of S. altissima plants was 6.3 ppm,
similar to the level previously reported elsewhere (~6 ppm,
Kobayashietal. 1980). However, the concentration varied
among our study sites, with some samples containing DME
as high as 17.3 ppm. DME concentration rapidly decreased
with distance away from S. altissima plants, but some sites
retained a relatively high concentration (up to 7.8 ppm) at
the distance of 25 cm. ese results indicate that polyacety-
lenes found in the natural soil are associated with S. altissima
plants, and that these compounds are available at the range
of concentrations known to inhibit competitor growth in
laboratory experiments (5–20 ppm: Kobayashi et al. 1980,
Sawabeetal. 1999, Johnsonetal. 2010, Uesugi and Kessler
2013). us, polyacetylenes are likely to have an ecological
function in natural environments, although its availabil-
ity may vary spatially depending on the soil properties and
microbial activities (Itoetal. 1998, Kobayashi et al. 2004,
Kauretal. 2009, Duke 2010).
Experimentally, we further provide evidence for the alle-
lopathic function of polyacetylene compounds: We found a
signicant negative correlation between root polyacetylene
(a) (b)
aa
bb
0
2
4
ctrl spray
Polyacetylene [ ]
aa
bc
bc bc
0
2
4
6
AMRAUS JPN
Figure3. Induction of total polyacetylenes in response to competition from P. pratensis. Boxplots show standardised concentration of poly-
acetylenes in the single-population experiment (a; ‘ctrl’ = populations exposed to ambient herbivory, ‘spray’ = insecticide-sprayed popula-
tions) and multi-range experiment (b; ‘AMR’ = North American, ‘AUS’ = Australian and ‘JPN’ = Japanese populations). Light grey boxes
represent polyacetylene concentration in no-competition control, and dark grey boxes represent that under competition treatment. Box
plots show median, upper and lower quartiles and extremes. Dots represent outliers. Letters above the box indicate signicance with post
hoc tests.
ab ab
a
b
−1
0
1
2
3
High nutrient Low nutrient
Polyacetylene [ ]
Figure 4. Eects of competition (light grey = control, dark
grey = competition with P. pratensis) and nutrient (high and low)
treatments on root polyacetylene concentration. Box plots show
median, upper and lower quartiles and extremes. Dots represent
outliers. Letters above the box indicate signicance with post
hoc tests.

8
concentration in S. altissima plants and the biomass of com-
peting Poa pratensis. e result suggests that polyacetylene
production provides a competitive advantage to S. altissima
plants. Biomass of S. altissima plants was also negatively cor-
related with P. pratensis biomass, indicating that exploitative
competition through increased biomass was also an impor-
tant mode of competition in S. altissima (Goldberg and
Landa 1991). However, despite the negative association, S.
altissima biomass and polyacetylene concentration explained
only 14 and 8% of variance in P. pratensis biomass, respec-
tively. Other unmeasured factors are likely to contribute to
the variation: For example, the rate at which polyacetylenes
in roots are released into the environment may vary with S.
altissima genotypes and may inuence how the plants sup-
press their competitors. Examining the relationship between
root concentration and the amount exuded from the roots
across multiple genotypes will test this hypothesis. e inter-
pretation of our results based on trait correlation also requires
caution, as it does not show causal relationship. Nevertheless,
our study provided a rst possible link between polyacety-
lene production in S. altissima roots and the plant’s ability to
suppress neighbouring competitors in soil environment.
Does competition induce root secondary
metabolites?
Because production of polyacetylenes is expected to incur
metabolic costs, we predicted that these compounds could
be induced under competition when allelopathy is benecial
(Novoplansky 2009, Kegge and Pierik 2010). Total poly-
acetylene concentration in roots increased under competi-
tion treatments in two out of three experiments (multi-range
experiment and nutrient experiment, but not in single-popu-
lation experiment), suggesting that plants can actively induce
the production of polyacetylenes in response to competition.
A few studies have explored inducibility of allelopathy
so far. Some demonstrated inducibility of allelochemicals in
response to changes in abiotic conditions, such as light, tem-
perature, and water and nutrient availability (Dayan 2006,
Konget al. 2006, Song et al. 2008, Kato-Noguchi 2011).
Volatile organic compounds from Artemisia tridentata that
inhibit the growth of neighbouring plants were induced by
herbivory, presumably because the same compounds also
deter herbivores (Karban 2007). But studies that tested the
direct eect of competition on allelopathy induction are
limited: Lankau and Kleibenstein (2009) showed in Brassica
nigra that sinigrin, which was found to have both defensive
and allelopathic functions, was induced by a combination of
herbivory and competition, but less so by competition alone.
A rare demonstration of induced allelochemicals in response
to competitor presence was found in Oryza sativa (Kongetal.
2006, Songet al. 2008, Kato-Noguchi 2011), where plants
grown hydroponically with competing plants exuded more
allelochemicals than when grown without competitors
(Kato-Noguchi 2011). Our observation in S. altissima pro-
vides rare evidence that competition can induce allelopathy
in a non-crop plant that is subjected to natural selection.
Nevertheless, the induction of polyacetylenes was incon-
sistent among experiments, which may reect the dierences
in the source of S. altissima genotypes used in the experi-
ments. Under a normal nutrient condition, induction was
not observed in the single-population experiment that
exclusively used genotypes from the native range in North
America, nor was in American populations from the multi-
range experiment. In the latter experiment, both introduced
Japanese and Australian populations had higher constitu-
tive levels of polyacetylenes than the native populations,
corroborating the results by Uesugi and Kessler (2016) that
used fewer populations. However, a signicant induction of
polyacetylenes was observed only in Australian populations.
(a) (b)
−1
0
1
2
−1 012
Constitutive (control)
Induction (competition−control)
−1
0
1
2
−1 012
Constitutive (control)
origin
AMR
AUS
JPN
Figure5. e relationship between constitutive (control) and induced (competition – control) expression of total polyacetylenes in the
single-population experiment (a) and multi-range experiment (b). In the multi-range experiment, the correlation was tested separately for
each population origin (closed circles = North America, open circles = Australia and crosses = Japan). Lines represent linear regressions for
each population.

9
is geographic variation in constitutive and inducible poly-
acetylenes may suggest rapid evolution of allelopathy in novel
ranges. Allelopathy is expected to be more eective in plant’s
introduced ranges, particularly in its early stage of invasion,
where recipient communities are naïve to novel allelochemi-
cals (Callaway and Ridenour 2004). Colonization of the
novel competitive environments may favour increased abil-
ity to induce secondary metabolites (Zangerl and Rutledge
1996), and may drive a transient evolutionary increase in
plasticity (Lande 2015). is plastic response, however, is
expected to decrease eventually following the genetic assimi-
lation that increases constitutive expression of allelopathy
(Lande 2015). e lack of inducibility in Japanese S. altissima
populations may be explained by the fact that the popula-
tions are rmly established, and are in the later phase of
invasion (Fukuda 1982). Alternatively, founder eects may
explain the dierences observed among introduced ranges,
but the genetic structures of these introduced populations are
currently unknown.
Does soil nutrient level affect plant induced-
response to competitors?
e benet of inhibiting neighbour growth through increased
allelopathy is expected to increase when the competition for
limited resource becomes severe (Songet al. 2008). us,
assuming that induction of allelopathy is adaptive, we pre-
dicted that inducibility of polyacetylenes would be greater in
nutrient depleted environments than in high nutrient envi-
ronments. Consistent with the hypothesis, we found poly-
acetylene induction in response to competition only under
low nutrient treatment. is decoupling of plant response to
competition and nutrient availability suggests that plants are
able to perceive the presence of competitors via plant-specic
chemical (Pierik et al. 2013) or light cues (Izaguirre et al.
2006), rather than indirectly detecting competitors via loss
of nutrients.
e eect of nutrient depletion on allelopathic ability
has been previously studied in rice in articial medium
(Songet al. 2008, Kato-Noguchi 2011). Songet al. (2008)
found that nitrogen limitation caused up-regulation of gene
expression in enzymes associated with allelochemical syn-
thesis, and an increase in allelopathic eects. Kato-Noguchi
(2011) showed that the exposure to root exudates from the
competitor plants also induces rice allelochemical produc-
tion. However, they do not test whether the inducibility
of allelochemicals vary across nutrient environments. Our
results, showing context-dependent induced responses, sug-
gest that S. altissima may be able to respond to competition
adaptively by inducing polyacetylenes more strongly when it
is most needed (i.e. under severe exploitative competition).
Do constitutive and induced levels of polyacetylenes
trade off?
Assuming that allelochemical production is costly, we
expected constitutive and induced allelopathy to trade o.
We found only a weak negative correlation in the single-
population experiment and Australian populations in the
multi-range experiment. Across the ranges, constitutive and
induced polyacetylenes did not correlate. e tradeo may
be masked in the multi-range experiment because of large
population-level variation in resources available for allelopa-
thy (van Noordwijk and de Jong 1986). Exotic populations
that escape herbivory may evolve to allocate a larger pool
of resources to allelopathy than native populations (Blossey
and Notzold 1995, Uesugietal. 2017). us, the relaxation
of herbivory may drive an increased constitutive allelopa-
thy in both exotic popoulations (Uesugi and Kessler 2016).
Further selection for plastic responses may favour coloniza-
tion of heterogeneous competitive environments in early
stages of invasion. Such independent selection for constitu-
tive and induced allelopathy could explain increased levels
of both expression in Australian populations compared to
native North American populations (Fig. 5b). Interestingly,
the negative correlation was most prominent in Australian
populations: these results may indicate that costs of express-
ing allelopathy in the absence of competitors become more
apparent in the novel and heterogeneous environment.
Conclusion
Inducibility of allelochemicals in response to competition was
predicted based on potential metabolic costs of compound
production. We found some evidence that polyacetylenes,
the putative allelopathic compounds in S. altissima, could be
induced in competitive environments. However, inducibility
of polyacetylenes was context-dependent: evolutionary his-
tory of the plant populations, as well as nutrient availability,
seem to inuence the degree to which plants induce poly-
acetylenes. e lack of a strong tradeo between constitu-
tive and induced polyacetylenes may indicate that these two
forms of allelopathy could evolve independently. Further
geographic contrasts of multiple, independently introduced
populations at varying stages of introduction may allow us
to empirically test how the plasticity for allelopathy evolves
in invasive ranges, thereby inuencing invasion dynamics of
the exotic plants.
Acknowledgements – Funding – is research was funded by School
of Biological Sciences at Monash University, Australian Research
Council (DE180101164), National Science Foundation (USA,
NSF-IOS 0950225) and Cornell University.
Author contributions – e rst and second authors led the
designing of experiments, and data collection and analyses. All
authors contributed critically to the drafts and gave nal approval
for publication.
References
Agrawal, A. A.etal. 1999. Polymorphism in plant defense against
herbivory: constitutive and induced resistance in Cucumis
sativus. – J. Chem. Ecol. 25: 2285–2304.

10
Atlas of Living Australia. Occurrence download at < www.ala.org.au >.
Baldwin, I. T. 2003. Finally, proof of weapons of mass destruction.
– Sci. STKE 2003: pe42–pe42.
Bates, D. et al. 2015. Fitting linear mixed-eects models using
lme4. – J. Stat. Softw. 67: 1–48.
Belz, R. G. and Hurle, K. 2005. Dierential exudation of two
benzoxazinoids one of the determining factors for seedling
allelopathy of Triticeae species. – J. Agric. Food Chem. 53:
250–261.
Blair, A. C.etal. 2005. New techniques and ndings in the study
of a candidate allelochemical implicated in invasion success.
– Ecol. Lett. 8: 1039–1047.
Blossey, B. and Notzold, R. 1995. Evolution of increased competitive
ability in invasive nonindigenous plants: a hypothesis. – J. Ecol.
83: 887–889.
Callaway, R. M. and Ridenour, W. M. 2004. Novel weapons:
invasive success and the evolution of increased competitive
ability. – Front. Ecol. Environ. 2: 436–443.
Cantor, A.etal. 2011. Low allelochemical concentrations detected
in garlic mustard-invaded forest soils inhibit fungal growth and
AMF spore germination. – Biol. Invas. 13: 3015–3025.
Carson, W. P. and Root, R. B. 2000. Herbivory and plant species
coexistence: community regulation by an outbreaking
phytophagous insect. – Ecol. Monogr. 70: 73–99.
Dayan, F. E. 2006. Factors modulating the levels of the allelochemical
sorgolene in Sorghum bicolor. – Planta 224: 339–346.
Duke, S. O. 2010. Allelopathy: current status of research and future
of the discipline: a commentary. – Allelopathy J. 25: 17–30.
Fox, J. and Weisberg, S. 2011. An R companion to applied
regression, 2nd edn. Sage, ousand Oaks, CA.
Fukuda, I. 1982. Distributions and population structures of North
American plants, Solidago altissima L. and S. gigantea Ait,
introduced into the Japanese islands. – Sci. Rep. Tokyo Woman’s
Christian Univ. 32: 675–690.
Goldberg, D. E. and Landa, K. 1991. Competitive eect and
response: hierarchies and correlated traits in the early stages of
competition. – J. Ecol. 79: 1013–1030.
Holm, L.etal. 1979. A geographical atlas of world weeds. – Wiley.
Huang, Z. et al. 2003. Correlation between phytotoxicity on
annual ryegrass (Lolium rigidum) and production dynamics of
allelochemicals within root exudates of an allelopathic wheat.
– J. Chem. Ecol. 29: 2263–2279.
Inderjit and Callaway, R. M. 2003. Experimental designs for the
study of allelopathy. – Plant Soil 256: 1–11.
Inderjitetal. 2009. Use of silenced plants in allelopathy bioassays:
a novel approach. – Planta 229: 569–575.
Inderjit et al. 2011. e ecosystem and evolutionary contexts of
allelopathy. – Trends Ecol. Evol. 26: 655–662.
Ito, I.etal. 1998. Fate of dehydromatricaria ester added to soil and
its implications for the allelopathic eect of Solidago altissima
L. – Ann. Bot. 82: 625–630.
Izaguirre, M. M.etal. 2006. Remote sensing of future competitors:
impacts on plant defenses. – Proc. Natl Acad. Sci. USA 103:
7170–7174.
Jaeger, B. 2017. Computes R squared for mixed (multilevel) models.
– R package ver. 0.1.2.
Johnson, R. H.etal. 2010. Simultaneous analysis of tissue- and
genotype-specic variation in Solidago altissima (Asteraceae)
rhizome terpenoids, and the polyacetylene dehydromatricaria
ester. – Chemoecology 20: 255–264.
Jose, S. and Gillespie, A. R. 1998. Allelopathy in black walnut
(Juglans nigra L.) alley cropping. II. Eects of juglone on
hydroponically grown corn (Zea mays L.) and soybean (Glycine
max L. Merr.) growth and physiology. – Plant Soil 203:
199–206.
Karban, R. 2007. Experimental clipping of sagebrush inhibits seed
germination of neighbours. – Ecol. Lett. 10: 791–797.
Karban, R. B. and Baldwin, I. T. 1997. Induced responses to
herbivory. – Univ. of Chicago Press.
Karban, R. and Myers, J. H. 1989. Induced plant responses to
herbivory. – Annu. Rev. Ecol. Syst. 20: 331–348.
Kato-Noguchi, H. 2011. Barnyard grass-induced rice allelopathy
and momilactone B. – J. Plant Physiol. 168: 1016–1020.
Kaur, H. et al. 2009. Taking ecological function seriously: soil
microbial communities can obviate allelopathic eects of
released metabolites. – PLoS One 4: e4700.
Kegge, W. and Pierik, R. 2010. Biogenic volatile organic com-
pounds and plant competition. – Trends Plant Sci. 15: 126–132.
Kempel, A.et al. 2011. Tradeos associated with constitutive and
induced plant resistance against herbivory. – Proc. Natl Acad.
Sci. USA 108: 5685–5689.
Kobayashi, A. et al. 1980. C 10-polyacetylenes as allelopathic
substances in dominants in early stages of secondary succession.
– J. Chem. Ecol. 6: 119–131.
Kobayashi, K. et al. 2004. Relationship between behavior of
dehydromatricaria ester in soil and the allelopathic activity of
Solidago altissima L. in the laboratory. – Plant Soil 259: 97–102.
Kong, C. H.etal. 2006. Allelochemicals released by rice roots and
residues in soil. – Plant Soil 288: 47–56.
Koricheva, J.etal. 2004. Meta-analysis of tradeos among plant
antiherbivore defenses: are plants jacks-of-all-trades, masters of
all? – Am. Nat. 163: E64–E75.
Lande, R. 2015. Evolution of phenotypic plasticity in colonizing
species. – Mol. Ecol. 24: 2038–2045.
Lankau, R. A. and Kliebenstein, D. J. 2009. Competition, herbivory
and genetics interact to determine the accumulation and tness
consequences of a defence metabolite. – J. Ecol. 97: 78–88.
Lankau, R. A. et al. 2009. Evolutionary limits ameliorate the
negative impact of an invasive plant. – Proc. Natl Acad. Sci.
USA 106: 15362–15367.
Lau, J. A.etal. 2008. Inference of allelopathy is complicated by
eects of activated carbon on plant growth. – New Phytol. 178:
412–423.
Lenth, R. 2019. emmeans: estimated marginal means, aka least-
squares means. – R package ver. 1.3.2. <https://cran.r-project.
org/web/packages/emmeans/>.
Lefcheck, J. S. 2016. piecewiseSEM: piecewise structural equation
modelling in R for ecology, evolution and systematics.
– Methods Ecol. Evol. 7: 573–579.
Morris, W. F. et al. 2006. On testing for a tradeo between
constitutive and induced resistance. – Oikos 112: 102–110.
Nimbal, C. I. et al. 1996. Phytotoxicity and distribution of
sorgoleone in grain sorghum germplasm. – J. Agric. Food
Chem. 44: 1343–1347.
Novoplansky, A. 2009. Picking battles wisely: plant behaviour
under competition. – Plant Cell Environ. 32: 726–741.
Pierik, R.etal. 2013. Molecular mechanisms of plant competition:
neighbour detection and response strategies. – Funct. Ecol. 27:
841–853.
Rasher, D. B. and Hay, M. E. 2014. Competition induces
allelopathy but suppresses growth and anti-herbivore defence in
a chemically rich seaweed. – Proc. R. Soc. B 281: 20132615.
Roberts, K. J. and Anderson, R. C. 2001. Eect of garlic mustard
Alliaria petiolata (Beib. Cavara and Grande) extracts on plants

11
and arbuscular mycorrhizal (AM) fungi. – Am. Midl. Nat. 146:
146–152.
Sawabe, A.etal. (1999). Eects of acetylenes and terpenoids from
Solidago altissima L. on seed germination. – J. Japan Oil Chem.
Soc. 48: 139–142.
Sawabe, A.etal. 2000. Characterization of acetylenes and terpenoids
isolated from Solidago altissima L. – Bull. Inst. Comprehens.
Agric. Sci. Kinki Univ. 8: 81–88.
Song, B. et al. 2008. Allelopathic enhancement and dierential
gene expression in rice under low nitrogen treatment. – J.
Chem. Ecol. 34: 688–695.
Uesugi, A. and Kessler, A. 2013. Herbivore exclusion drives the
evolution of plant competitiveness via increased allelopathy.
– New Phytol. 198: 916–924.
Uesugi, A. and Kessler, A. 2016. Herbivore release drives parallel
patterns of evolutionary divergence in invasive plant
phenotypes. – J. Ecol. 104: 876–886.
Uesugi, A.etal. 2013. A test of genotypic variation in specicity
of herbivore-induced responses in Solidago altissima L.
(Asteraceae). – Oecologia 173: 1387–1396.
Uesugi, A.etal. 2017. Relaxation of herbivore-mediated selection
drives the evolution of genetic covariances between plant
competitive and defense traits. – Evolution 71: 1700–1709.
Uesugi, A.etal. 2019. Data from: context-dependent induction of
allelopathy in plants under competition. – Dryad Digital
Repository, <http://dx.doi.org/10.5061/dryad.7b90m0b>.
van Noordwijk, A. J. and de Jong, G. 1986. Acquisition and alloca-
tion of resources: their inuence on variation in life history
tactics. – Am. Nat. 128: 137–142.
Werner, P. A. et al. 1980. e biology of Canadian weeds: 45.
Solidago canadensis L. – Can. J. Plant Sci. 60: 1393–1409.
Zangerl, A. R. and Rutledge, C. E. 1996. e probability of attack
and patterns of constitutive and induced defense: a test of
optimal defense theory. – Am. Nat. 147: 599–608.
Supplementary material (available online as Appendix oik-
06389 at < www.oikosjournal.org/appendix/oik-06389 >).
Appendix 1.