Herbivore-Mediated Effects of Glucosinolates on Different
Natural Enemies of a Specialist Aphid
Martine Kos & Benyamin Houshyani &
Buddhi B. Achhami & Rafal Wietsma & Rieta Gols &
Berhane T. Weldegergis & Patrick Kabouw &
Harro J. Bouwmeester & Louise E. M. Vet &
Marcel Dicke & Joop J. A. van Loon
Received: 20 October 2011 /Revised: 22 November 2011 /Accepted: 28 December 2011 /Published online: 19 January 2012
#The Author(s) 2012. This article is published with open access at Springerlink.com
Abstract The cabbage aphid Brevicoryne brassicae is a spe-
cialist herbivore that sequesters glucosinolates from its host
plant as a defense against its predators. It is unknown to what
extent parasitoids are affected bythissequestration.We inves-
tigated herbivore-mediated effects of glucosinolates on the
parasitoid wasp Diaeretiella rapae and the predator Episyr-
phus balteatus. We reared B. brassicae on three ecotypes of
Arabidopsis thaliana that differ in glucosinolate content and
on one genetically transformed line with modified concentra-
tions of aliphatic glucosinolates. We tested aphid performance
and the performance and behavior of both natural enemies.
We correlated this with phloem and aphid glucosinolate con-
centrations and emission of volatiles. Brevicoryne brassicae
performance correlated positively with concentrations of both
aliphatic and indole glucosinolates in the phloem. Aphids
selectively sequestered glucosinolates. Glucosinolate concen-
tration in B. brassicae correlated negatively with performance
of the predator, but positively with performance of the
parasitoid, possibly because the aphids with the highest glu-
cosinolate concentrations had a higher body weight. Both
natural enemies showed a positive performance-preference
test combination, whereas the parasitoid wasp preferred the A.
thaliana ecotype with the highest emission of these volatiles.
The study shows thatthere are differential herbivore-mediated
effects of glucosinolates on a predator and a parasitoid of a
its host plant.
to herbivores (Karban and Baldwin, 1997; Schoonhoven et
al., 2005). Specialists that are adapted to feeding on plants
containing specific secondary metabolites, however, often use
these compounds for their own benefit, e.g., as oviposition or
feeding stimulants (van Loon et al., 1992; Gabrys and
Tjallingii, 2002). Some concentrate metabolites actively
taken up from host plants in special tissues or organs.
This sequestration can make these herbivores unpalatable
to natural enemies (Duffey, 1980; Müller, 2009).
Brassicaceous plants contain glucosinolates (GLS) that,
upon damage by chewing herbivores, become exposed to
the plant enzyme myrosinase that hydrolyzes GLS, resulting
in toxic compounds such as (iso)thiocyanates and nitriles
that negatively affect a wide variety of generalist herbivores
Electronic supplementary material The online version of this article
(doi:10.1007/s10886-012-0065-2) contains supplementary material,
which is available to authorized users.
M. Kos:B. B. Achhami:R. Wietsma:R. Gols:
B. T. Weldegergis:L. E. M. Vet:M. Dicke:J. J. A. van Loon
Laboratory of Entomology, Wageningen University,
P.O. Box 8031, 6700 EH Wageningen, The Netherlands
B. Houshyani:H. J. Bouwmeester
Laboratory of Plant Physiology, Wageningen University,
P.O. Box 658, 6700 AR Wageningen, The Netherlands
M. Kos (*):P. Kabouw:L. E. M. Vet
Department of Terrestrial Ecology,
Netherlands Institute of Ecology (NIOO-KNAW),
P.O. Box 50, 6700 AB Wageningen, The Netherlands
J Chem Ecol (2012) 38:100–115
(Halkier and Gershenzon, 2006; Hopkins et al., 2009).
Phloem-feeding herbivores, however, can ingest GLS without
bringing these compounds into contact with plant myrosinases
(Andreasson et al., 2001). Thus, aphids prevent the formation
of toxic hydrolysis products of most GLS (de Vos et al., 2007;
Kim and Jander, 2007). The cabbage aphid Brevicoryne bras-
sicae is a specialist that uses GLS as feeding stimulants
(Gabrys and Tjallingii, 2002), and sequesters GLS from its
food plants (Francis et al., 2001; Kazana et al., 2007; Pratt,
2008; Kos etal.,2011). It contains anendogenous myrosinase,
which is stored separately from the GLS (Jones et al., 2001;
sequestered GLS come in contact with the aphid myrosinase,
resulting in the formation of toxic hydrolytic products. Nega-
tive effects of this sequestration have been reported for aphid
predators, such as ladybird beetles, hoverflies, and lacewings
(Francis et al.,2001; Kazana etal., 2007; Pratt, 2008; Chaplin-
Kramer et al., 2011; Kos et al., 2011). Most predators kill their
preyimmediately and feedon multipleindividuals duringtheir
development. Parasitoids, in contrast, develop inside a single
host individual, and koinobiont parasitoids allow the host to
continue to growand feed after parasitization (Godfray, 1994).
of GLS in B. brassicae compared to predators, but this rarely
has been investigated. Le Guigo et al. (2011) compared the
fitness of the solitary endoparasitoid Diaeretiella rapae when
developing in B. brassicae that were feeding on host plant
species with different foliar GLS concentrations. Parasitoid
performance did not correlate with foliar GLS concentrations.
In that study, GLS concentrations in the aphids were not
analyzed. It has been shown previously that B. brassicae
sequesters GLS selectively (Kabouw et al., 2011; Kos et al.,
2011). It is still unknown to what extent D. rapae performance
is affected by GLS sequestration in B. brassicae.
GLS not only affect performance of natural enemies of
herbivores feeding on brassicaceous plants, but also their
behavior. Formation of volatile GLS breakdown products,
resulting from herbivore feeding, increase attraction of sev-
eral specialist parasitoids (Bradburne and Mithen, 2000;
Blande et al., 2007; Mumm et al., 2008). Effects of volatile
GLS breakdown products on the behavior of generalist
predators are largely unknown.
Our objective was to investigate herbivore-mediated effects
of GLS on the performance and the behavior of the parasitoid
D. rapae and the predacious hoverfly Episyrphus balteatus.
These species represent two different groups within carnivo-
rous insects and are two of the most important natural enemies
GLS concentrations, we reared them on three ecotypes of
in their GLS content (Houshyani et al., in press). Additionally,
a genetically transformed line was created to produce higher
concentrations of foliar aliphatic (methionine-derived) GLS
compared to the wild-type plants. We compared the perfor-
mance of B. brassicae on these different ecotypes/lines, ana-
lyzed the GLS in the phloem and in the aphids feeding on it,
and determined the performance of E. balteatus and D. rapae
when feeding on these aphids. Additionally, we studied para-
sitoid and predator preference behavior in response to aphid-
induced volatile organic compounds emitted by the different
Methods and Materials
Plant Material and Growth Conditions Three Arabidopsis
thaliana (L.) Heynh. ecotypes were selected, based on their
maximal divergence in metabolite profiles (qualitative and
et al., in press). Columbia (Col)-0 was provided by Dr. P.
Reymond (Lausanne, Switzerland); Cape Verde Island (Cvi)
was obtained from the European Arabidopsis Stock Centre
(http://nasc.nott.ac.uk/, Cvi 0 N8580); and Eringsboda (Eri)
was collected in Sweden by members of the Laboratory of
Genetics, Wageningen University (Eri-1 0 CS22548).
To produce plants with higher foliar levels of aliphatic
GLS, we over-expressed the transcription factor HAG1/
MYB28 in A. thaliana ecotype Col-0 (Houshyani et al.
unpublished data, see also Supplemental Material Online
Resource 1). This transcription factor represents a key com-
ponent in the regulation of aliphatic GLS biosynthesis in A.
thaliana (Gigolashvili et al., 2007). T2 generation seeds of
one successfully transformed line (hereafter named Col-0-
MYB28) were used in the experiments.
Arabidopsis thaliana seeds were surface-sterilized over-
night by vapor phase sterilization and inoculated on a
growth medium (purified agar 0.8%+2.2 gl−10.5 MS+
vitamins; pH 6; containing 30 μg ml−1kanamycin to select
transformed seedlings). After 4 d of stratification at 4°C,
plates were transferred to a growth chamber at 21±2°C, 50–
70% relative humidity (RH) and a 8:16L:D photo regime,
with a light intensity of 200 μmol m−2s−1photosynthetic
photon flux density (PPFD).
Two-week-old seedlings with two true leaves were trans-
planted to pots (5 cm diam) containing autoclaved soil (80°C
for 4 h; Lentse potgrond, Lent, The Netherlands). Plants were
watered three times a week, and the soil was treated weekly
with entomopathogenic nematodes (Steinernema feltiae;
Koppert Biological Systems, Berkel en Rodenrijs, The
Netherlands) to control infestation by larvae of sciarid flies.
Plants used were 6–7 wk-old, and remained in the vegeta-
tive state during experiments.
Insect Rearing Brevicoryne brassicae L. (Hemiptera: Aphi-
didae) were reared on Brussels sprouts (Brassica oleracea
L. var. gemmifera cv. Cyrus). Episyrphus balteatus de Geer
J Chem Ecol (2012) 38:100–115101
(Diptera: Syrphidae) pupae were provided by Koppert Bio-
logical Systems and kept in gauze cages (67 × 50 × 67 cm).
Adults emerging from the pupae were provided with water,
a B. brassicae-infested B. oleracea plant, organic sugar
grains, and bee-collected pollen provided by Koppert Biolog-
ical Systems. Diaeretiella rapae McIntosh (Hymenoptera:
Braconidae) was reared in gauze cages (30 × 40 × 60 cm)
containing B. brassicae-infested B. oleracea plants. Wasps
were provided with water and honey. The B. brassicae and
D. rapae culture originated from individuals obtained from B.
oleracea in the vicinity of Wageningen (The Netherlands) in
2008. All insect species were reared at 22±2°C, 60–70% RH
and a 16:8 h L:D photo regime.
GLS and Primary Metabolites in Phloem of Aphid-Infested
Plants After the aphid performance experiment ended,
phloem of aphid-infested plants was collected for chemical
analysis. We used 8 mM EDTA, following the procedure
described in Kos et al. (2011). Four fully-grown leaves of
each plant were placed with their petiole for 5 min in the
EDTA solution to remove any plant chemicals from the
incision. Then, leaves were placed for 4 h in a new vial
with 200 μl EDTA solution, under dark conditions. Using
this method, a small amount of mesophyll fluids was inher-
ently collected as well. Following incubation, the EDTA
solution was collected from the vials, and each vial was
rinsed with 50 μl EDTA, resulting in a sample of 250 μl per
leaf. The EDTA solution of four plants (16 leaves) was
pooled to form one sample of 4 ml, resulting in five repli-
cates per ecotype/line. Leaves were dried at 80°C for 3 d and
weighed on an analytical balance (Mettler-Toledo PM200,
Tiel,The Netherlands). Phloemsamples werefrozenat−80°C
immediately after collection, freeze-dried, and re-suspended
amino acid extraction. To extract GLS from the phloem, we
used the protocol described by Kos et al. (2011). GLS were
separated using high-performance liquid chromatography
(HPLC) as described previously by van Dam et al. (2004)
and Kabouw et al. (2010). GLS detection was performed with
a photodiode array detector set at 229 nm as the integration
wavelength. Different concentrations of sinigrin (2-prope-
nylGLS; Acros, NJ, USA) were used as external standard.
The retention times of the GLS compounds can be found in
Supplemental Material Online Resource 2. The correction
factors at 229 nm from Buchner (1987) and the European
Community (1990) were used to calculate the concentrations
provided by M. Reichelt (Max Planck Institute for Chemical
Ecology, Jena, Germany) and a certified rapeseed standard
(Community Bureau of Reference, Brussels, Belgium, code
Soluble carbohydrates (50 μl from the one ml sample) and
amino acids (50 μl) were extracted and analyzed as described
previously by van Dam and Oomen (2008). For the carbohy-
drates, we used a “10 ppm” reference solution containing
54.9 μM sorbitol and mannitol, 29.21 μM trehalose, sucrose,
and melibiose, and 55.51 μM glucose and fructose. This
reference solution was diluted to obtain 7.5, 5, and 2.5 ppm
calibration standards to obtain a reference curve. To obtain a
the Sigma AA S 18 amino acid standard (Sigma, St Louis,
MO, USA) containing 17 amino acids was supplemented with
asparagine, glutamine, and tryptophane (2.5 μmoles ml−1
each). This reference solution was diluted to obtain calibration
standards ranging from 1 to 8 μM for each amino acid, except
for cysteine, which had a range of 0.5–4 μM. For both the
carbohydrates and the amino acids, an additional standard was
injected after every 10 samples to check for deviations of
retention times and the calibration curve.
Dynamic Headspace Collection of Volatiles from Aphid-
Infested Plants Six-to-seven week-old A. thaliana plants
were infested with 100 B. brassicae nymphs of mixed
instars 3 d prior to headspace collection. Dynamic head-
space collection was carried out in a climate chamber at 20±
2°C. Plants were removed from pots, and the soil was
wrapped with aluminum foil. Three plants were placed
together in a 2.5 l glass jar. Volatiles were collected by
sucking air out of the jar at a rate of 90 ml min−1for 3 h
through a stainless steel cartridge (Markes, Llantrisant, UK)
containing 200 mg Tenax TA (20/35 mesh; Grace-Alltech,
Deerfield, MI, USA). Foliar fresh weight of the plants in
each pot was measured after volatile collection. For each
ecotype/line, 8–11 replicate samples were collected (8 Cvi,
10 Eri, 9 Col-0, 11 Col-0-MYB28).
Headspace samples were analyzed by using a Thermo
Trace Gas Chromatography Ultra (Thermo Fisher Scientific,
Waltham, MA, USA) coupled to a Thermo Trace DSQ
(Thermo Fisher Scientific, Waltham, MA, USA) quadrupole
mass spectrometer (MS) (Supplemental Material Online
Resource 3). The peak area of each compound was
expressed per unit plant fresh weight.
Identification of Compounds Identification of compounds
was based on comparison of mass spectra with those in the
Natural Products MS libraries. Experimentally calculated lin-
ear retention indices (LRI) also were used as additional crite-
rion for confirming the identity of the compounds. Relative
quantification (peak areas of individual compounds) was per-
formed using a single (target) ion, in selected ion monitoring
detailed information on the identification methods for each
102J Chem Ecol (2012) 38:100–115
Plant Morphology and Foliar GLS Concentrations of
Uninfested Plants We quantified two plant morphological
characteristics, trichome density and plant biomass, because
these might influence aphid and natural enemy performance
and behavior. For 10 uninfested plants per ecotype/line, we
measuredfoliarbiomass and countedthenumberoftrichomes
in a 25 mm2area in the central part of the abaxial side of the
6th or 7th youngest leaf by using a microscope (Leitz Dialux
20 EB, Wetzlar, Germany; magnification 40×).
For foliar GLS analysis, we harvested all leaf material of
10 uninfested plants per ecotype/line. Samples were frozen
at −80°C immediately after collection, freeze-dried,
weighed (approximately 100 mg) into micro-centrifuge
tubes, and ground to a fine powder.
GLS were extracted and purified from the leaves by using
a methanol extraction (van Dam et al., 2004; Kabouw et al.,
2010). GLS were separated and detected as described above
for the phloem samples.
Insect Performance Individual plants with insects were con-
fined to cylindrical plastic containers (height 13 cm; diam
11 cm) with a gauze lid. Experiments were performed in a
climate chamber at 21±2°C, 50–70% RH and a 8:16 L:D
photo regime for B. brassicae and 16:8 L:D for E. balteatus
and D. rapae. The light intensity at plant level was
200 μmol m−2s−1PPFD. Plants were watered once a week.
Aphid Performance Several 6-wk-old plants of each eco-
type/line were inoculated with 10 adult aphids per plant.
After 24 h, adult aphids were removed, and the produced
offspring were allowed to develop for 3 d until they reached
the second instar (L2). Three L2 nymphs were transferred to
each of 20 A. thaliana plants per ecotype/line, the same
ecotype/line as the one on which these nymphs had been
feeding before. Until the adult stage, survival of nymphs
was recorded daily. The fastest developing adult was kept on
the plant, while the other adults were removed. Alate
(winged) adults (ca. 5% of all adults) were excluded from
the experiment as these contain lower concentrations of
GLS than apterous (wingless) aphids (Kazana et al., 2007).
The development time until first reproduction (0Td) of the
remaining adult was recorded, and the adult fresh weight
was measured on a microbalance (Sartorius CP2P, Göttin-
gen, Germany). Adults were allowed to feed on plants and
produce offspring, and after a certain number of days
(equivalent to Td), the number of offspring (0N) produced
by the adult was counted. The estimated intrinsic rate of
population increase (rm) was calculated for each aphid using
the formula: rm¼ 0:738 ? ðlnNÞ=Td(Karley et al., 2002).
Aphid GLS Concentrations After the aphid performance
experiment ended, aphids on the 4 plants that were used to
obtain one phloem sample (described above) were removed
and pooled into one sample. GLS were extracted similarly to
the method used for the leaves.
Predator Performance Female E. balteatus from the stock
rearing were allowed to lay eggs on Brussels sprouts plants
infested with B. brassicae. After hatching, neonate larvae
were transferred to A. thaliana plants that had been infested
by 10 adult B. brassicae from the stock rearing 1 wk earlier.
Larvae were allowed to develop on the plants until pupation.
Pupae were checked once a day for eclosion of adults.
Survival, larva-to-adult development time, sex, and adult
dry weight were determined. Newly eclosed adults were
frozen, dried to constant weight at 80°C for 3 d, and then
weighed on a microbalance. We determined the perfor-
mance of 35 larvae per A. thaliana ecotype/line, one larva
Parasitoid Performance Aphid mummies containing a D.
rapae pupa were collected from the stock rearing, and
reared until adult parasitoid eclosion. Adult parasitoids were
provided with water and honey, allowed to mate, and used
for parasitisation when they were 2–4 d-old. Second instar
(3-d-old) B. brassicae nymphs that had been feeding on one
of the A. thaliana ecotypes/lines were exposed individually
to mated female parasitoids on an aphid-infested leaf until
parasitisation was observed (i.e., when the female inserted
her ovipositor into the nymph). Four parasitized nymphs
were transferred to one A. thaliana plant of the same eco-
type/line as the one on which these aphids had been feeding
before. In total we tested 22 plants per A. thaliana ecotype/
line. Mummies were collected from plants, and after eclo-
sion, parasitoid sex was determined, and egg-to-adult devel-
opment time and adult dry weight were measured, as
described for the predator. The percentage of successful
parasitism of B. brassicae by D. rapae was calculated per
plant by dividing the number of D. rapae adults by the total
number of B. brassicae nymphs that survived (either until
the adult stage or until D. rapae eclosion) on each plant.
Predator and Parasitoid Preference The preference of pred-
ators and parasitoids for volatiles from an ecotype/line was
investigated in two-choice bioassays. We tested the ecotypes
against each other, and Col-0-MYB28 against Col-0. Plants
were treated similarly as described above under Dynamic
Headspace Collection of Volatiles from Aphid-infested
Predator Oviposition Preference Mated female hoverflies
from the stock rearing were used in the behavioral assays
when they were 2–3-wk-old. Females were transferred to a
plastic cage (30 × 30 × 30 cm) containing one aphid-infested
plant of two different ecotypes/lines, and 10% sugar solu-
tion. Females were allowed to oviposit on the plants for
J Chem Ecol (2012) 38:100–115 103
Replicates with females that did not lay any eggs were elimi-
replicates with ovipositing females were obtained.
Parasitoid Preference for Aphid-Induced Plant Volatiles Para-
sitoid behavior was assessed in a Y-tube olfactometer in a
climatized room at 22±2°C as described by Bukovinszky et
al. (2005). Compressed air was filtered over charcoal and
split into two air streams each at a flow of 2 lmin−1. Each air
stream was led through a 5 l glass jar that contained 4 aphid-
infested plants of one of the two ecotypes/lines of a test com-
bination. Each air stream was then led into one of the two arms
of the Y-tube. The olfactometer was illuminated from above
with artificial light at an intensity of 60 μmol m−2s−1PPFD.
Naïve, mated 2-d-old D. rapae females were allowed to
oviposit for 1 h in aphids feeding on one of the two ecotypes/
lines of a combination (equally divided among the tested
wasps) to increase their host-searching behavior. Experienced
their preference for one of both odor sources was recorded. A
choice was recorded when a wasp crossed a finish line drawn
one cm before the end of each arm, and did not return to the
junction within 15 s. Wasps that did not make a choice within
15 min were considered as non-responsive and were omitted
from the statistical analysis. Four or five new sets of plants
were used for each test combination. For every new set of
plants, 20 wasps were tested. After every 10 wasps, the
position of the odor sources was exchanged to compensate
for any asymmetry in the set-up.
Statistical Analyses Analyses were performed in SPSS for
Windows (18th edition, Chicago, IL, USA), unless indicated
ity and equal variance, this is indicated in the relevant table of
the Results section. To test the effect on the continuous vari-
ables, such as development time and body weight, we used
ANOVA followed by post-hoc Tukey-tests for pair-wise eco-
type comparisons and t-tests for comparisons between Col-0
and Col-0-MYB28. If assumptions on normality and equal
variance were violated, Kruskal-Wallis tests with post-hoc
Mann–Whitney U tests with a Holm’s sequential Bonferroni
correction were used for pair-wise ecotype comparisons, and
Mann–Whitney U tests for comparisons between Col-0 and
Col-0-MYB28. Survival and the percentage of successful
parasitism were calculated per plant (plant was used as the
experimental unit), and differences in these variables among
ecotypes and between Col-0 and Col-0-MYB28 were ana-
lyzed by logistic regression in GenStat (13th edition, VSN
International, UK). If over-dispersion was observed, the data
were corrected for this by using estimated dispersion instead
wise differences between means.
To test whether an equal number of predator eggs was laid
pairs signed-rank tests were used. To test whether an equal
number of parasitoid wasps chose either ecotype/line in a test
combination in the Y-tube olfactometer, Chi-square tests were
used. Effects of parasitoid experience on the preference of the
wasps was tested by logistic regression in GenStat.
To determine whether there were differences in volatile
profiles and aphid GLS profiles among the ecotypes and be-
tween Col-0 and Col-0-MYB28, we used multivariate discrim-
inant analysis Projection to Latent Structures-Discriminant
Analysis (PLS-DA) in SIMCA-P (12th edition, Umetrics,
Umeå, Sweden) (Eriksson et al., 2006). For the volatiles, the
variable importance in the projection (VIP) was calculated.
Variables with a VIP value higher than 1 are most influential
2006). For volatile compounds with a VIP higher than 1, the
difference among the ecotypes and between Col-0 and Col-
0-MYB28 was analyzed as described above for the continuous
(PLS) in SIMCA-P, a multivariate method for regression anal-
ysis, was used to test the relationship between a) metabolite
profiles in the phloem and performance of aphids feeding on
those plants, and b) the GLS profile in the phloem and in the
the GLS compounds that were found in both the phloem and
the aphids, as well as the total GLS, total aliphatic GLS and
total indole GLS, were included. To pre-process data, metabo-
lite concentrations were log-transformed, mean-centered, and
scaled to unit variance. To test whether concentrations of
were correlated, we used Spearman’s correlation test. PLS
analyses of the relationships between aphid GLS concentra-
tions and predator/parasitoid performance could not be per-
formed, as we measured these variables in separate
experiments. Note that in the Results section ‘aliphatic GLS’
refers to the total of all aliphatic GLS compounds that were
detected, ‘indole GLS’ refers to the total of all indole GLS
compounds, and ‘total GLS’ refers to the total of all GLS
compounds (aliphatic and indole GLS combined).
GLS and Primary Metabolites in Phloem of Aphid-Infested
Plants Ecotype effect: Total, aliphatic and indole GLS con-
centrations in the phloem of aphid-infested plants differed
among ecotypes (Kruskal-Wallis H, df02, total: χ206.48,
P00.039; aliphatic: χ208.07, P00.018; indole: χ2010.82,
P00.004), due to both qualitative and quantitative differ-
ences (Table 1). Phloem of Cvi had the highest total and
aliphatic GLS concentrations, whereas phloem of Eri plants
had the highest indole GLS concentrations. Phloem of Col-0
104J Chem Ecol (2012) 38:100–115
Table 1 Mean (± SE) concentrations of metabolites in the phloem of aphid-infested plants of three Arabidopsis thaliana ecotypes and the
transformed COL-0-MYB28 line, and in Brevicoryne brassicae aphids reared on these plants
Metabolite Arabidopsis thaliana ecotypeTransformed
Cvi EriCol-0 Col-0-MYB28
Total amino acidsc
Total aliphatic GLS
Total indole GLS
Total aliphatic GLS
J Chem Ecol (2012) 38:100–115 105
plants had the lowest concentration of all GLS classes
(Table 1). Ecotypes did not differ in total concentrations of
carbohydrates and amino acids in the phloem (ANOVA,
P>0.05 for both analyses), although there were small qualita-
tive and quantitative differences in the concentrations of the
individual compounds (Table 1).
Over-expression effect: The phloem of aphid-infested
Col-0-MYB28 plants had lower concentrations of total
and aliphatic GLS than Col-0 plants, and similar con-
centrations of indole GLS (Mann–Whitney U-test: total:
U<0.001, P00.008; aliphatic: U<0.001, P00.008; in-
dole: U07.00, P00.310; Table 1). This was unexpected,
as foliar tissue of Col-0-MYB28 plants had higher concen-
trations of aliphatic GLS than Col-0 plants (Supplemental
Material Online Resource 5). Total concentrations of carbo-
hydratesand amino acids inthe phloemdid not differ between
Col-0 and Col-0-MYB28 plants (Table 1).
Dynamic Headspace CollectionofAphid-InfestedPlants Eco-
type effect: The three A. thaliana ecotypes differed in vol-
atile profiles of aphid-infested plants (4 PLS-DA principal
components, R2Xcum00.906, R2Ycum00.836, Q2cum00.682).
The volatile profile of Cvi was high in breakdown products
of GLS such as 3-butenyl isothiocyanate, 3-butene nitrile,
and 3-methyl-3-butene nitrile. Col-0 plants emitted larger
amounts of the sesquiterpenes δ-selinene and daucene,
whereas the headspace of Eri plants was high in the ester
methyl salicylate (Fig. 1a, b; Table 2).
Over-expression effect: Volatile profiles of Col-0 and
Col-0-MYB28 plants could be separated by PLS-DA (4
PLS-DA principal components, R2Xcum00.861, R2Ycum0
0.932, Q2cum00.612; Fig. 1c,d). Of the compounds that
had a VIP-value higher than 1 in the PLS-DA model, only
one compound was emitted in significantly different
amounts by Col-0 and Col-0-MYB28 plants: the GLS
breakdown product 3-butene nitrile, which was emitted in
larger amounts by Col-0-MYB28 plants (Table 2).
Plant Morphology Ecotype effect: Eri plants had a higher
biomass than plants of the other ecotypes. Cvi plants had the
highest, and Eri plants the lowest trichome density (Supple-
mental Material Online Resource 5). Over-expression effect:
There was no difference in biomass or trichome density
between Col-0 and Col-0-MYB28 plants (Supplemental
Material Online Resource 5).
Aphid Performance Ecotype effect: Aphid survival did not
significantly differ among ecotypes (logistic regression,
P00.051, Table 3). Aphid development time, adult weight,
number of offspring, and estimated intrinsic rate of population
increase (rm) differed among ecotypes (ANOVA, respectively
F2,57015.10, P<0.001; F2,57030.24, P<0.001; F2,57018.62,
Table 1 (continued)
Metabolite Arabidopsis thaliana ecotypeTransformed
Total indole GLS
N05 for each sample. For every sample, phloem or aphids collected from four plants were pooled
aμmol g−1dry weight leaf
cParameter was log-transformed in statistical analysis to obtain normality
dnmol g−1dry weight leaf
eGlucosinolates (GLS) are grouped according to their biosynthetic origin into indole and aliphatic GLS, and analyses were performed separately for
total GLS, aliphatic GLS and indole GLS
fμmol g−1dry weight aphids
Statistical tests were performed only for the total carbohydrate, amino acid, aliphatic GLS, indole GLS and total GLS concentrations, not for
individual compounds. Different letters denote differences in means among the three ecotypes as analyzed by Mann–Whitney U-tests with
sequential Bonferroni correction (for GLS) or ANOVA and post-hoc Tukey tests (for carbohydrates and amino acids)
*denotes significant difference and ns denotes non-significant difference between Col-0 and Col-0-MYB28 as analyzed by Mann–Whitney U-tests
(for GLS) or t-tests (for carbohydrates and amino acids)
Carbohydrates and amino acids have been identified and quantified based on calibration lines for the corresponding authentic standards. The
retention times used for identification of each GLS compound can be found in Supplemental Material Online Resource 2. For quantification of GLS
sinigrin (2-propenylGLS) was used as the external standard
106 J Chem Ecol (2012) 38:100–115
P<0.001; F2,57045.66, P<0.001). All aphid performance
parameters were higher (development time shorter) on Cvi
plants than on plants of the other ecotypes (Table 3).
Over-expression effect: There was no difference in any of
the measured performance parameters of B. brassicae be-
tween Col-0 and Col-0-MYB28 plants (P>0.05 for any
comparison, Table 3).
All measured aphid performance parameters were signif-
icantly positively correlated (development time inversely
correlated) with total and aliphatic GLS and several carbo-
hydrates (4 PLS principal components, R2Xcum00.757,
R2Ycum00.735, Q2cum00.343; Fig. 2). Aphid performance
parameters were, to a lesser extent, also seemingly positively
correlated with indole GLS, but this was significant only for
Fig. 1 Projection to Latent Structures-Discriminant Analysis (PLS-
DA) score and loading plots of the first two components based on
the volatile emission of aphid-infested plants of three Arabidopsis
thaliana ecotypes (a and b) and of ecotype Col-0 and the transformed
Col-0-MYB28 (c and d). Plant ecotypes investigated are Cvi (filled
boxes), Eri (open diamonds) and Col-0 (filled triangles); the
transformed line is Col-0-MYB28 (open triangles). The score plots (a
and c) show the distinction in volatile profiles of the ecotypes/lines. In
brackets the percentage of variation explained is indicated. The loading
plots (b and d) show the contribution of the volatile compounds to the
discrimination among the ecotypes/lines. Numbers refer to the volatile
compounds listed in Table 2
J Chem Ecol (2012) 38:100–115 107
development time (inversely correlated). Aphid performance
parameters were in general not correlated with amino acids,
sucrose, or total carbohydrates.
Aphid GLS Concentrations Ecotype effect: Aphids reared
on different ecotypes differed in total and aliphatic GLS
concentrations (Kruskal-Wallis H, df02, total: χ206.98,
P00.031; aliphatic: χ206.98, P00.031), due to both quali-
tative and quantitative differences (Table 1). Aphids reared
on Cvi plants contained the highest, and aphids reared on
Col-0 plants the lowest total and aliphatic GLS concentra-
tions (Table 1). There were no differences in indole GLS
Table 2 Mean (± SE) amount ofvolatilesemitted byaphid-infested plantsofthree Arabidopsis thaliana ecotypes and the transformedCOL-0-MYB28
No.Compound Arabidopsis thaliana ecotype Transformed
Cvi EriCol-0 Col-0-MYB28
5731 ±1059 b
N08–11 for each ecotype/line. Unit is peak area mg−1fresh weight
Numbers correspond to the numbers in Fig. 1
aStatistical tests were performed only for the compounds that had a VIP-value higher than 1 in the PLS-DA model shown in Fig. 1 that included
either the three ecotypes (Fig. 1a and b) or Col-0 and Col-0-MYB28 (Fig. 1c and d). The compounds with a VIP higher than 1 are most influential
for the discrimination among the ecotypes/lines. Different letters denote differences in means among the three ecotypes as analyzed by Mann–
Whitney U-tests with sequential Bonferroni correction. * denotes significant difference and ns denotes non-significant difference between Col-0
and Col-0-MYB28 as analyzed by Mann–Whitney U-tests. Compounds have been identified based on the linear retention index (LRI) and mass
spectrum, or mass spectrum only (compounds 4, 5 and 22). See Supplemental Material Online Resource 4 for details on identification methods
108 J Chem Ecol (2012) 38:100–115
among aphids reared on the different ecotypes (Kruskal-
Wallis H, P>0.05).
Over-expression effect: Aphids reared on Col-0-MYB28
plants had similar concentrations of total, aliphatic and
indole GLS to aphids reared on Col-0 plants (Mann–Whitney
U, P>0.05 for every analysis; Table 1).
Correlations Between GLS Profiles in Phloem and in B.
Brassicae Aphids In both the univariate Spearman’s correla-
tion tests, as well as the multivariate PLS model, concentra-
tions of most of the GLS compounds or classes in the aphids
were not significantly positively correlated with their concen-
trations in the phloem. Positive correlations were, however,
significant for total and aliphatic GLS (PLS model: 1 PLS
principal component, R2X00.486, R2Y00.373, Q200.280;
Spearman’s correlation: 3-butenylGLS: rs00.53; P00.015;
aliphatic: rs00.72; P<0.001; total: rs00.82; P<0.001). The
concentration of the indole 4-methoxy-3-indolylmethylGLS
seemed to be negatively correlated, although not significantly,
between aphids and the phloem they were feeding on (Fig. 3;
Spearman’s correlation, P00.650).
The contribution of indole GLS to the total concentration
of GLS was lower in aphids than in the phloem of the aphid-
infested plants (10% indole GLS in aphids compared to 42%
indole GLS in the phloem, as averaged over all four eco-
types/lines; see also Table 1). Additionally, the ratio of the
indole compounds was different in the aphids from that in
the phloem: in the phloem the concentration of 4-methoxy-
3-indolylmethylGLS was higher than the concentration of 3-
indolylmethylGLS, whereas this was reverse in the aphids
Predator Performance Ecotype effect: Survival of E. bal-
teatus to the adult stage did not differ among ecotypes
(logistic regression, P>0.05, Table 3). Larva-to-adult devel-
opment time of the hoverflies was affected by plant ecotype
and hoverfly sex (ANOVA, ecotype: F2,23027.11, P<0.001;
sex: F1,2308.10, P00.009). Hoverflies developed slowest
on aphids fed on Cvi plants and fastest on aphids fed on Eri
plants (Table 3). Averaged over ecotypes, male hoverflies
took longer (18.7±0.8 d) to develop into adults than females
(17.3±0.5 d). However, the difference in development time
between males and females was only significant on Cvi, and
not on Col-0 and Eri, resulting in a significant interaction
between ecotype and sex (F2,2304.65, P00.020). Adult dry
weight was affected by hoverfly sex (ANOVA, F1,2306.63,
P00.017), as male hoverflies (3.18±0.23 mg) were heavier
than females (2.49±0.13 mg), but not by plant ecotype
(Table 3) or the interaction between ecotype and sex
(ANOVA, P>0.05 for both analyses).
Over-expression effect: There was no difference in sur-
vival, development time, or adult dry weight of E. balteatus
Table 3 Mean (± SE) performance characteristics of Brevicoryne brassicae, Episyrphus balteatus and Diaeretiella rapae reared on three
Arabidopsis thaliana ecotypes and the transformed COL-0-MYB28 line
Insect species Performance parameterA. thaliana ecotypea
B. brassicaeSurvival until adult stage (%)c,d
Development time until first reproduction in days (Td)e,f
Adult fresh weight in mgf
Number of offspring (N) in time period equivalent to Tde,f
Estimated intrinsic rate of population increase (rm)f
Survival until adult stage (%)c
Larva-to-adult development time in dayse,f,g
Adult dry weight in mge,f,g
Successful parasitism (%)c,d
Larva-to-adult development time in dayse,f,g
Adult dry weight in mgf,g
aDifferent letters denote differences in means among the three ecotypes
bns denotes no significant difference between Col-0 and Col-0-MYB28
cAnalyzed by logistic regression and post-hoc T-probability tests
dPerformance parameter was averaged per plant before statistical analysis
ePerformance parameter was log-transformed in statistical analysis to obtain normality
fAnalyzed by ANOVA and post-hoc Tukey tests (for the ecotypes), or t-test (for the wild-type and transformed Col-0 line)
gThe data for males and females were combined
J Chem Ecol (2012) 38:100–115109
between Col-0 and Col-0-MYB28 (P>0.05 for all parame-
ters; Table 3).
Parasitoid Performance Ecotype effect: Plant ecotype did
not affect the percentage of successful parasitism of B.
brassicae by D. rapae (logistic regression, P>0.05), nor
did it affect egg-to-adult development time (ANOVA,
P>0.05; Table 3). Only adult dry weight was affected by
plant ecotype (ANOVA, ecotype: F2,169010.16, P<0.001).
Adult dry weight was higher on Cvi plants than on plants of
the other ecotypes (Table 3). There was no effect of parasit-
oid sex or the interaction between ecotype and sex for any of
the performance parameters (P>0.05 for all parameters).
Over-expression effect: There was no difference in the
percentage of successful parasitism, development time, or
adult dry weight of D. rapae between Col-0 and Col-0-
MYB28 (P>0.05 for all parameters; Table 3).
Predator Oviposition Preference Female E. balteatus pre-
ferred to oviposit on aphid-infested Eri plants over aphid-
infested Col-0 and Cvi plants, and aphid-infested Col-0
plants over aphid-infested Cvi plants (Wilcoxon: Eri vs.
Col-0: Z02.63, N027, P00.008; Eri vs. Cvi: Z03.44, N0
22, P00.001; Col-0 vs. Cvi: Z02.12, N029, P00.034).
Females did not differentiate between aphid-infested plants
of Col-0 and Col-0-MYB28 (Wilcoxon, P>0.05; Fig. 4).
Parasitoid Preference for Aphid-Induced Plant Volatiles Fe-
male D. rapae preferred volatiles from aphid-infested Cvi
plants over volatiles from aphid-infested Col-0 plants (Chi-
square, χ205.69, P00.017). Females neither differentiated
between volatiles from any of the other ecotype combina-
tions, nor between Col-0 and Col-0-MYB28 (Chi-square,
P>0.05 for every combination; Fig. 5). There was no effect
of previous oviposition experience on the preference of the
wasps (logistic regression, P>0.05 for any combination).
Fig. 3 Loading plot of the first two components of PLS showing the
relationship of the concentration of each glucosinolate (GLS) compound
(squares, label in italics) with the concentration of these compounds or
classes in the aphid Brevicoryne brassicae (triangles, label underlined)
feeding on the phloem. In brackets the percentage of variation explained
is indicated. Note that only the first component is significant according to
the multivariate model, whereas two components were included to en-
hance the clarity of the figure. Compound abbreviations: Aliphatic GLS:
GNA 0 3-butenylGLS (gluconapin), IBV 0 3-methylthiopropylGLS
(glucoiberverin), SIN 0 2-propenylGLS (sinigrin). Indole GLS: GBC 0
3-indolylmethylGLS (glucobrassicin), 4MeOH 0 4-methoxy-3-indolyl-
Fig. 2 Loading plot of the first two components of Projection to Latent
Structures showing the contribution of each individual compound or
compound class measured in the phloem, i.e., glucosinolates (GLS),
carbohydrates and amino acids, to the performance of the aphid Bre-
vicoryne brassicae in terms of survival, development time, adult
weight, number of offspring and estimated intrinsic rate of population
increase (rm). In brackets the percentage of variation explained is
indicated. Compound abbreviations: Aliphatic GLS: EPRO 0 2-(S)-2-
hydroxy-butenylGLS (epiprogoitrin), GNA 0 3-butenylGLS (glucona-
pin), IBV 0 3-methylthiopropylGLS (glucoiberverin), SIN 0 2-
propenylGLS (sinigrin). Indole GLS: GBC 0 3-indolylmethylGLS
(glucobrassicin), 4MeOH 0 4-methoxy-3-indolylmethylGLS (4-
methoxyglucobrassicin). Carbohydrates: Fru 0 fructose, Gluc 0 glu-
cose, Man 0 mannitol, Raf 0 raffinose. Sor 0 sorbitol, Suc 0 sucrose,
Tre 0 trehalose. Amino acids: Ala 0 alanine, Arg 0 arginine, Asn 0
asparagine, Asp 0 aspartate, Glu 0 glutamate, Gln 0 glutamine, His 0
histidine, Leu 0 leucine, Lys 0 lysine, Met 0 methionine, Phe 0
phenylalanine, Ser 0 serine, Thr 0 threonine, Tyr 0 tyrosine, Val 0
110 J Chem Ecol (2012) 38:100–115
The performance of B. brassicae was best on the A. thaliana
ecotype with the highest concentrations of aliphatic GLS in
the phloem. Furthermore, we found a positive correlation
between aliphatic GLS and aphid performance in the multi-
variate regression analysis. Due to the intercellular path
taken by the aphid stylet to the phloem (Tjallingii and
Hogen Esch, 1993), aphids can ingest aliphatic GLS from
the phloem without bringing these compounds into contact
Fig. 4 Oviposition preference of the aphid predator Episyrphus bal-
teatus in a two-choice assay with aphid-infested plants of three Arabi-
dopsis thaliana ecotypes (Col-0, Cvi and Eri) and one transformed line
(Col-0-MYB28). The boxes span the first to third quartile range with
the line across the box indicating the median. The whiskers represent
the range. Open circles represent outliers. An asterisk indicates a
significant difference (P<0.05) between the number of eggs deposited
on each ecotype/line as analyzed by the Wilcoxon matched-pairs
signed-rank test; NS 0 not significant
Fig. 5 Responses of Diaeretiella rapae females to volatile blends
emitted by aphid-infested Arabidopsis thaliana ecotypes/lines in a Y-
tube olfactometer. Ecotypes investigated are: Col-0, Cvi and Eri; the
transformed line is Col-0-MYB28. Each bar represents the percentage
of females that made a choice for the indicated odor sources. The
percentage of no choice in each experiment and the total number of
tested females are indicated on the right. An asterisk indicates a
significant preference for one of the two ecotypes/lines in a combina-
tion, as analyzed by Chi-square tests
J Chem Ecol (2012) 38:100–115111
with plant myrosinases that are stored in cells adjacent to the
phloem (Andreasson et al., 2001). Thus, aphids can prevent
the formation of toxic hydrolytic products of aliphatic GLS
(de Vos et al., 2007; Kim and Jander, 2007). Together with
the observation that B. brassicae uses GLS as feeding
stimulants (Gabrys and Tjallingii, 2002), our finding of a
positive correlation between aliphatic GLS and aphid per-
formance is expected. In contrast to aliphatic GLS, indole
GLS are hydrolyzed by aphids into toxic products indepen-
dently of myrosinase activity (Kim and Jander, 2007; Kim et
al., 2008), and negative correlations between the concentra-
tions of indole GLS and performance of B. brassicae and
other aphid species have been reported (Cole, 1997; Mewis
et al., 2005; Kim and Jander, 2007; Kim et al., 2008). Our
observation of a slight, but significant, positive correlation
between aphid performance and total indole GLS concen-
trations in the phloem are in disagreement with these latter
studies. A possible explanation for this discrepancy is that
specific indole GLS may affect aphid performance more
strongly than others. The difference in the abundance of
specific indole GLS between our study and studies from
the literature might be the explanation of differences in
effects on aphid performance.
The A. thaliana ecotypes did not differ significantly in
the concentrations of carbohydrates and amino acids in the
phloem, and we did not observe a consistent correlation
between aphid performance and concentrations of individu-
al or total carbohydrates and amino acids in the regression
analysis. Aphids did not seem to be affected by trichomes,
as their performance was best on the A. thaliana ecotype
with the highest trichome density. We note that we measured
trichome density of uninfested plants. It has been demon-
strated that feeding by leaf chewers can increase trichome
density in A. thaliana (Traw and Dawson, 2002), but wheth-
er this also is true for aphids is, to our knowledge, unknown.
Brevicoryne brassicae sequestered GLS from the phloem,
and total aliphatic GLS concentrations in the phloem were
significantly positively correlated with their concentrations in
the aphids. In contrast, whereas in the phloem 4-methoxy-
3-indolylmethylGLS was the most abundant indole GLS,
aphidssequestered thiscompoundinlow concentrationscom-
pared to its precursor, 3-indolylmethylGLS. This is in accor-
dance with what we reported previously in aphids feeding on
B. oleracea plants (Kos et al., 2011). Although the concen-
trations of most aliphatic GLS in phloem were positively
correlated with their concentrations in aphids, selective se-
whereas 3-butenylGLS was the least abundant GLS in the
phloem of Cvi plants, it dominated in the aphids feeding on
these plants. These findings suggests that B. brassicae selec-
tively sequestered GLS from the phloem, a phenomenon we
previously reported (Kos et al., 2011). The mechanism under-
lying the selective sequestration of GLS could be that
transporters for GLS in the aphid gut wall may be specific.
In other GLS-sequestering species, such as the sawfly Athalia
rosae, selective sequestration of aliphatic GLS has been
reported (Müller and Wittstock, 2005; Müller, 2009). Little
is known about the specificity of GLS transporters or the
specific mechanisms underlying GLS sequestration (Opitz et
sequestered GLS, in agreement with previous work (Kos et
al., 2011). Because hydrolysis into toxic products requires
myrosinase activity circumvented by aphids (de Vos et al.,
likely little affected by high sequestration of aliphatic GLS.
Interestingly, aliphatic GLS are degraded more by purified
aphid myrosinase, whereas the lowest activities of the aphid
myrosinase are observed with indole GLS (Francis et al.,
2002). Thus, higher sequestration of aliphatic GLS by B.
brassicae may lead to higher toxicity to predators, without
affecting aphid performance itself. We note that not all GLS
detected in the aphids were detected in the phloem, probably
because the concentrations of some GLS in the phloem were
below the detection limit of the HPLC. However, we cannot
rule out that the aphids converted certain GLS from the
phloem into other compounds that were subsequently stored
in their body.
As expected, over-expressing Col-0-MYB28 plants pro-
duced higher foliar concentrations of aliphatic GLS and sim-
ilar concentrations of indole GLS compared to the wild-type
plants.Unexpectedly, we observedthataliphaticGLSconcen-
trations in the phloem were lower in Col-0-MYB28 plants
than in Col-0 plants. Probably, GLS biosynthesis in Col-
0-MYB28 plants occurred mainly in the mesophyll, and phlo-
em loading of GLS was limited.
Performance of the generalist aphid predator E. baltea-
tus was lowest in terms of development time when fed B.
brassicae aphids that contained the highest aliphatic GLS
concentrations (i.e., aphids reared on Cvi). It is likely that
this led to highest concentrations of GLS hydrolysis prod-
ucts after breakdown by the aphid myrosinase, as purified
aphid myrosinase quickly degrades aliphatic GLS (Francis
et al., 2002), although we did not quantify GLS and their
hydrolytic products separately. Our results are in agree-
ment with other studies that have reported negative effects
of GLS sequestration by B. brassicae on the performance
of E. balteatus (Vanhaelen et al., 2002; Kos et al., 2011)
and other aphid predators (Francis et al., 2001; Kazana et
al., 2007; Pratt, 2008; Chaplin-Kramer et al., 2011; Kos et
al., 2011). There are several other Brassicaceae-feeding
insects that sequester GLS for defense against predators.
Similarly to B. brassicae, the turnip aphid (Lipaphis ery-
simi) sequesters GLS from phloem into the haemolymph
and contains an endogenous myrosinase, a mechanism
that is expected to affect negatively predators (Bridges et
112 J Chem Ecol (2012) 38:100–115
Episyrphus balteatus performance did not differ between
Col-0 and Col-0-MYB28 plants. This was expected as aphid
GLS concentrations did not differ between these plant lines.
Hoverfly performance in terms of development time was
better when their prey had been feeding on ecotype Eri than
on Col-0. However, Eri-fed aphids had higher but not sta-
tistically different total GLS concentrations in their phloem
from Col-0-fed aphids. This points to the importance of
qualitative effects of GLS profiles on hoverfly performance.
It was reported previously that differences in B. brassicae
GLS profiles affect the performance of E. balteatus (Kos et
al., 2011). Furthermore, haemolymph of A. rosae larvae,
containing a mix of several GLS compounds, deterred ants
and predatory wasps more strongly than the individual major
GLS compounds (Müller et al., 2002; Müller and Brakefield,
2003). This suggests that GLS profiles rather than total con-
centrations influenced these predators, although the stronger
deterrence also could have been due to completely different
compounds fromGLSpresent inthe haemolymph. We cannot
rule out effects of other plant or aphid traits on predator
performance. Epicuticular plant characteristics, such as leaf
waxes and trichomes, may affect predator attachment to the
plant (Eigenbrode, 2004). Trichomes have been shown to
negatively affect the performance of hoverfly larvae due to
entrapment by glandular trichomes, reduced mobility, or fall-
ing off the plant (Verheggen et al., 2009). Although we do not
know whether this is also true for non-glandular trichomes on
A.thaliana, the lower trichomedensity ofEri plantsmay have
contributed to the better performance of hoverfly larvae on
Parasitoid performance in terms of adult weight was best
when developing in the largest aphids, containing the highest
and parasitoid size is in agreement with other studies (Harvey,
2005; Bukovinszky et al., 2008). Our results suggest that the
performance of D. rapae is not negatively affected by GLS
concentrations inthehost,supporting the findings of Le Guigo
et al. (2011). Although D. rapae parasitizes several aphid
species, it is the main parasitoid of B. brassicae (Bukovinszky
et al., 2008), and may be relatively tolerant to GLS. We do not
know, however, how D. rapae copes with GLS in its host. In
fact, there is not much known about detoxification of plant
secondary metabolites by parasitoids in general (Ode, 2006;
Gols and Harvey, 2009). Negative effects of breakdown prod-
ucts of GLS on D. rapae might be prevented by the feeding
strategy of the parasitoid larvae. Brevicoryne brassicae stores
GLS in the haemolymph, but the aphid’s myrosinases are
stored in the non-flight muscles (Jones et al., 2001; Bridges
et al., 2002; Francis et al., 2002). Tissue-feeding endoparasi-
toids, such as D. rapae, consume host haemolymph during
most of their larval development and only consume other host
tissues shortly before egression (Godfray, 1994; Harvey et al.,
toxic products during the major part of their development.
Diaeretiella rapae performance did not differ between Col-0
and Col-0-MYB28 plants, which was expected as both aphid
Aphid-infested plants of the three A. thaliana ecotypes
differed in their volatile profiles. Both the predator (E.
balteatus) and the parasitoid wasp (D. rapae) preferred the
ecotype on which their offspring performed best. This dem-
onstrates that preference and performance of these natural
enemies are positively correlated, in agreement with other
studies (Soler et al., 2007; Gols et al., 2009). The predator
always laid fewest eggs on the ecotype within a test combi-
nation that had the highest emission of volatile GLS hydro-
lysis products, suggesting that volatile breakdown products
of GLS were repellent for the predators. Episyrphus baltea-
tus had access to the aphid-infested plants in the bioassays.
We do not know if other plant characteristics or aphid cues
also played a role in the selection of an oviposition site by E.
balteatus. In particular, the preferred ecotype in each test
combination had the lowest trichome density. It has been
shown previously that adult hoverflies have problems with
landing on plants with high trichome densities (Verheggen
et al., 2009). In contrast to the predator, the parasitoid
preferred volatile cues from the ecotype with the highest
emission of volatile GLS hydrolysis products (Cvi), but
only when offered in combination with ecotype Col-0. A
preference for a high emission of volatile GLS hydrolysis
products was expected, as D. rapae is known to be attracted
to host plants emitting volatile breakdown products of GLS
(Read etal.,1970;Bradburneand Mithen,2000; Blande etal.,
2007). Neither the predator nor the parasitoid wasp differen-
tiated between cues from Col-0 and Col-0-MYB28 plants.
The relatively small difference in volatile profiles between
these lines might not allow olfactory discrimination.
In summary, the four main findings of our study are: 1) The
performance of the specialist cabbage aphid B. brassicae is
positively correlated with concentrations of both aliphatic and
brassicae selectively sequestered GLS from the phloem; 3)
The performance of the aphid predator E. balteatus is nega-
tively correlated with aphid GLS concentrations. The perfor-
mance of the aphid parasitoid D. rapae is positively correlated
with aphid GLS concentrations, probably because the aphids
with the highest GLS concentrations have a higher body
weight; 4) Both natural enemies prefer the A. thaliana ecotype
on which their offspring perform best, indicating a positive
performance-preferencecorrelation. The predatorpreferred the
A. thaliana ecotype with the lowest emission of volatile break-
down products of GLS in each test combination, whereas the
parasitoid wasp preferred the A. thaliana ecotype with the
highest emission of these volatiles, but only in one test com-
bination. Our study shows that there are differential herbivore-
J Chem Ecol (2012) 38:100–115113
mediated effects of GLS on a predator and a parasitoid of a
specialist aphid that selectively sequesters GLS from its host
structive comments on an earlier version of the manuscript; Ana
Pineda, Meindert van der Wielen, and Qianjue Wang for practical
assistance; Prof. Flügge (University of Cologne, Germany) for provid-
ing the MYB28 cDNA and Dr. Beekwilder for contacting Prof. Flügge;
Koppert Biological Systems for providing E. balteatus, and Unifarm
for rearing the Brussels sprouts plants. This work was supported by a
grant from the Earth and Life Sciences Council of the Netherlands
Organization for Scientific Research (NWO-ALW) under the ERGO
program (number 838.06.010). Publication 5186 Netherlands Institute
of Ecology (NIOO-KNAW).
We thank two anonymous reviewers for con-
tive Commons Attribution Noncommercial License which permits any
noncommercial use, distribution, and reproduction in any medium,
provided the original author(s) and source are credited.
This article is distributed under the terms of the Crea-
ANDREASSON, E., JORGENSEN, L. B., HOGLUND, A. S., RASK, L., and
MEIJER, J. 2001. Different myrosinase and idioblast distribution
in Arabidopsis and Brassica napus. Plant Physiol. 127:1750–
BLANDE, J. D., PICKETT, J. A., and POPPY, G. M. 2007. A comparison
of semiochemically mediated interactions involving specialist and
generalist Brassica-feeding aphids and the braconid parasitoid
Diaeretiella rapae. J. Chem. Ecol. 33:767–779.
BRADBURNE, R. P., and MITHEN, R. 2000. Glucosinolate genetics and
the attraction of the aphid parasitoid Diaeretiella rapae to Bras-
sica. Proc. R. Soc. Lond. B Biol. Sci. 267:89–95.
BRIDGES, M., JONES, A. M. E., BONES, A. M., HODGSON, C., COLE,
R., BARTLET, E., WALLSGROVE, R., KARAPAPA, V. K., WATTS, N.,
and ROSSITER, J. T. 2002. Spatial organization of the
glucosinolate-myrosinase system in brassica specialist aphids is
similar to that of the host plant. Proc. R. Soc. Lond. B Biol. Sci.
BUCHNER, R. 1987. Approach to determination of HPLC response
factors for glucosinolates, pp 50–58, in J. P. Wathelet (ed.),
Glucosinolates in Rapeseeds. Martinus Nijhoff Publishers, Dor-
drecht, The Netherlands
BUKOVINSZKY, T., GOLS, R., POSTHUMUS, M. A., VET, L. E. M., and
VAN LENTEREN, J. C. 2005. Variation in plant volatiles and attrac-
tion of the parasitoid Diadegma semiclausum (Hellen). J. Chem.
BUKOVINSZKY, T., VAN VEEN, F. J. F., JONGEMA, Y., and DICKE, M.
2008. Direct and indirect effects of resource quality on food web
structure. Science 319:804–807.
CHAPLIN-KRAMER, R., KLIEBENSTEIN, D. J., CHIEM, A., MORRILL, E.,
MILLS, N. J., and KREMEN, C. 2011. Chemically mediated tritro-
phic interactions: opposing effects of glucosinolates on a special-
ist herbivore and its predators. J. Appl. Ecol. 48:880–887.
COLE, R. A. 1997. The relative importance of glucosinolates and
amino acids to the development of two aphid pests Brevicoryne
brassicae and Myzus persicae on wild and cultivated brassica
species. Entomol. Exp. Appl. 85:121–133.
DE VOS, M., KIM, J. H., and JANDER, G. 2007. Biochemistry and
molecular biology of Arabidopsis-aphid interactions. BioEssays
DUFFEY, S. S. 1980. Sequestration of plant natural products by insects.
Annu. Rev. Entomol. 25:447–477.
EIGENBRODE, S. D. 2004. The effects of plant epicuticular waxy
blooms on attachment and effectiveness of predatory insects.
Arthropod. Struct. Dev. 33:91–102.
ERIKSSON, L., JOHANSSON, E., KETTANEH-WOLD, N., TRYGG, J.,
WIKSTRÖM, C., and WOLD, S. 2006. Multi- and Megavariate
Data Analysis. Part I: Basic Principles and Applications.
Umetrics Academy, Umeå, Sweden.
European Community. 1990. Oilseeds-determination of glucosinolates
High performance liquid chromatography. Official Journal of the
European Communities L 170/28 Annex VIII: 03.07.27–34.
FRANCIS, F., LOGNAY, G., WATHELET, J. P., and HAUBRUGE, E. 2001.
Effects of allelochemicals from first (Brassicaceae) and second
(Myzus persicae and Brevicoryne brassicae) trophic levels on
Adalia bipunctata. J. Chem. Ecol. 27:243–256.
FRANCIS, F., LOGNAY, G., WATHELET, J. P., and HAUBRUGE, E. 2002.
Characterisation of aphid myrosinase and degradation studies of
glucosinolates. Arch. Insect Biochem. Physiol. 50:173–182.
GABRYS, B., and TJALLINGII, W. F. 2002. The role of sinigrin in host
plant recognition by aphids during initial plant penetration. Ento-
mol. Exp. Appl. 104:89–93
GIGOLASHVILI, T., YATUSEVICH, R., BERGER, B., MÜLLER, C., and
FLUGGE, U. I. 2007. The R2R3-MYB transcription factor
HAG1/MYB28 is a regulator of methionine-derived glucosinolate
biosynthesis in Arabidopsis thaliana. Plant J. 51:247–261.
GODFRAY, H. C. J. 1994. Parasitoids: Behavioral and Evolutionary Ecol-
ogy. Princeton University Press, Princeton, New Jersey, USA.
GOLS, R., and HARVEY, J. A. 2009. Plant-mediated effects in the
Brassicaceae on the performance and behaviour of parasitoids.
Phytochem. Rev. 8:187–206.
GOLS, R., VAN DAM, N. M., RAAIJMAKERS, C. E., DICKE, M., and
HARVEY, J. A. 2009. Are population differences in plant quality
reflected in the preference and performance of two endoparasitoid
wasps? Oikos 118:733–743.
HALKIER, B. A., and GERSHENZON, J. 2006. Biology and biochemistry
of glucosinolates. Annu. Rev. Plant Physiol. Plant Mol. Biol.
HARVEY, J. A. 2005. Factors affecting the evolution of development
strategies in parasitoid wasps: the importance of functional con-
straints and incorporating complexity. Entomol. Exp. Appl. 117:1–
HARVEY, J. A., KADASH, K., and STRAND, M. R. 2000. Differences in
larval feeding behavior correlate with altered developmental strat-
egies in two parasitic wasps: implications for the size-fitness
hypothesis. Oikos 88:621–629.
HOPKINS, R. J., VAN DAM, N. M., and VAN LOON, J. J. A. 2009. Role of
glucosinolates in insect-plant relationships and multitrophic inter-
actions. Annu. Rev. Entomol. 54:57–83.
HOUSHYANI, B., KABOUW, P., MUTH, D., DE VOS, R. C. H., BINO, R. J.,
and BOUWMEESTER, H. J. in press. Characterization of the natural
variation in Arabidopsis thaliana metabolome by the analysis of
metabolic distance. Metabolomics. doi:10.1007/s11306-011-0375-3.
T. 2001. Purification and characterisation of a non-plant myrosinase
from the cabbage aphid Brevicoryne brassicae (L.).Insect Biochem.
Mol. Biol. 31:1–5.
KABOUW, P., BIERE, A., VAN DER PUTTEN, W. H., and VAN DAM, N.
M. 2010. Intra-specific differences in root and shoot glucosinolate
profiles among white cabbage (Brassica oleracea var. capitata)
cultivars. J. Agric. Food Chem. 58:411–417.
KABOUW, P., KOS, M., KLEINE, S., VOCKENHUBER, E. A., VAN LOON,
J. J. A., VAN DER PUTTEN, W. H., VAN DAM, N. M., and BIERE, A.
2011. Effects of soil organisms on aboveground multitrophic
interactions are consistent between plant genotypes mediating
the interaction. Entomol. Exp. Appl. 139:197–206.
114J Chem Ecol (2012) 38:100–115
KARBAN, R., and BALDWIN, I. T. 1997. Induced Responses to Herbiv- Download full-text
ory. The University of Chicago Press, Chicago, Illinois, USA.
KARLEY, A. J., DOUGLAS, A. E., and PARKER, W. E. 2002. Amino acid
composition and nutritional quality of potato leaf phloem sap for
aphids. J. Exp. Biol. 205:3009–3018.
KAZANA, E., POPE, T. W., TIBBLES, L., BRIDGES, M., PICKETT, J. A.,
BONES, A. M., POWELL, G., and ROSSITER, J. T. 2007. The
cabbage aphid: a walking mustard oil bomb. Proc. R. Soc. Lond.
B Biol. Sci. 274:2271–2277
KIM, J. H., and JANDER, G. 2007. Myzus persicae (green peach aphid)
feeding on Arabidopsis induces the formation of a deterrent
indole glucosinolate. Plant J. 49:1008–1019
KIM, J. H., LEE, B. W., SCHROEDER, F. C., and JANDER, G. 2008.
Identification of indole glucosinolate breakdown products with
antifeedant effects on Myzus persicae (green peach aphid). Plant
KOS, M., KABOUW, P., NOORDAM, R., HENDRIKS, K., VET, L. E.
M., VAN LOON, J. J. A., and DICKE, M. 2011. Prey-mediated
effects of glucosinolates on aphid predators. Ecol. Entomol.
LE GUIGO, P., QU, Y., and LE CORFF, J. 2011. Plant-mediated effects on
a toxin-sequestering aphid and its endoparasitoid. Basic Appl.
MEWIS, I., APPEL, H. M., HOM, A., RAINA, R., and SCHULTZ, J. C.
2005. Major signaling pathways modulate Arabidopsis glucosi-
nolate accumulation and response to both phloem-feeding and
chewing insects. Plant Physiol. 138:1149–1162.
MÜLLER, C. 2009. Interactions between glucosinolate- and
myrosinase-containing plants and the sawfly Athalia rosae. Phy-
tochem. Rev. 8:121–134
MÜLLER, C., BOEVE, J. L., and BRAKEFIELD, P. 2002. Host plant
derived feeding deterrence towards ants in the turnip sawfly
Athalia rosae. Entomol. Exp. Appl. 104:153–157.
MÜLLER, C., and BRAKEFIELD, P. M. 2003. Analysis of a chemical
defense in sawfly larvae: Easy bleeding targets predatory wasps in
late summer. J. Chem. Ecol. 29:2683–2694.
MÜLLER, C., and WITTSTOCK, U. 2005. Uptake and turn-over of
glucosinolates sequestered in the sawfly Athalia rosae. Insect
Biochem. Mol. Biol. 35:1189–1198
MUMM, R., BUROW, M., BUKOVINSZKINE’KISS, G., KAZANTZIDOU, E.,
WITTSTOCK, U., DICKE, M., and GERSHENZON, J. 2008. Forma-
tion of simple nitriles upon glucosinolate hydrolysis affects direct
and indirect defense against the specialist herbivore, Pieris rapae.
J. Chem. Ecol. 34:1311–1321.
ODE, P. J. 2006. Plant chemistry and natural enemy fitness: Effects on
herbivore and natural enemy interactions. Annu. Rev. Entomol.
OPITZ, S. E. W., JENSEN, S. R., and MÜLLER, C. 2010. Sequestration of
glucosinolates and iridoid glucosides in sawfly species of the
genus Athalia and their role in defense against ants. J. Chem.
PRATT, C. 2008. Accumulation of glucosinolates by the cabbage aphid
Brevicoryne brassicae as a defense against two coccinellid spe-
cies. J. Chem. Ecol. 34:323–329
READ, D. P., FEENY, P. P., and ROOT, R. B. 1970. Habitat selection by
the aphid parasite Diaeretiella rapae (Hymenoptera: Braconidae)
and hyperparasite Charips brassicae (Hymenoptera: Cynipidae).
Can. Entomol. 102:1567–1578.
SCHOONHOVEN, L. M., VAN LOON, J. J. A., and DICKE, M. 2005.
Insect-Plant Biology. Oxford University Press, Oxford, UK.
SOLER, R., HARVEY, J. A., KAMP, A. F. D., VET, L. E. M., VAN DER
PUTTEN, W. H., VAN DAM, N. M., STUEFER, J. F., GOLS, R.,
HORDIJK, C. A., and BEZEMER, T. M. 2007. Root herbivores
influence the behaviour of an aboveground parasitoid through
changes in plant-volatile signals. Oikos 116:367–376.
TJALLINGII, W. F., and HOGEN ESCH, T. H. 1993. Fine-structure of
aphid stylet routes in plant-tissues in correlation with EPG sig-
nals. Physiol. Entomol. 18:317–328.
TRAW, M. B., and DAWSON, T. E. 2002. Differential induction of
trichomes by three herbivores of black mustard. Oecologia
VAN DAM, N. M., and OOMEN, M. W. A. T. 2008. Root and shoot
jasmonic acid applications differentially affect leaf chemistry and
herbivore growth. Plant Signal. Behav. 3:91–98.
VAN DAM, N. M., WITJES, L., and SVATOS, A. 2004. Interactions
between aboveground and belowground induction of glucosino-
lates in two wild Brassica species. New Phytol. 161:801–810.
VAN LOON, J. J. A., BLAAKMEER, A., GRIEPINK, F. C., VAN BEEK, T.
A., SCHOONHOVEN, L. M., and DE GROOT, A. 1992. Leaf surface
compound from Brassica oleracea (Cruciferae) induces oviposition
byPierisbrassicae (Lepidoptera:Pieridae). Chemoecology3:39–44.
VANHAELEN, N., GASPAR, C., and FRANCIS, F. 2002. Influence of prey
host plant on a generalist aphidophagous predator: Episyrphus
balteatus (Diptera: Syrphidae). Eur. J. Entomol. 99:561–564.
VERHEGGEN, F. J., CAPELLA, Q., SCHWARTZBERG, E. G., VOIGT, D.,
and HAUBRUGE, E. 2009. Tomato-aphid-hoverfly: a tritrophic
interaction incompatible for pest management. Arthropod-Plant
J Chem Ecol (2012) 38:100–115 115