Biological Activity of Marmesin and Demethylsuberosin against a Generalist Herbivore, Spodoptera exigua (Lepidoptera: Noctuidae)

Article abstract of DOI:10.1021/jf960156b

No significant effects were observed in bioassays designed to measure impact of the furanocoumarin precursors demethylsuberosin and marmesin on typical physiological parameters (survival and development rate) of the generalist herbivore, Spodoptera exigua (Huebner). The linear furanocoumarin psoralen did increase developmental times and reduce survival. All of these compounds were demonstrated to have significant behavioral (feeding deterrence) effects against both first and third instars of S. exigua in diet-incorporation bioassays. In tests initiated with first instars, significantly (P < 0.05) more larvae preferred control diet to diet containing marmesin or psoralen on all five sample dates; control diets were significantly preferred to diets with demethylsuberosin on three of five sample dates. Similar results were observed for tests initiated with third instar larvae, but the avoidance of demethylsuberosin-containing diets was stronger. Therefore, assaying furanocoumarin precursors just for effects on growth and survival may not provide an accurate picture of the ecological importance of these compounds.

Full text of DOI:10.1021/jf960156b

J. Agric. Food Chem. 1996, 44, 2859
−2864
2859
Biologica l Activity of Ma r m esin a n d Dem eth ylsu ber osin a ga in st a
Gen er a list Her bivor e, Spod opter a exigu a (Lep id op ter a : Noctu id a e)
J ohn T. Trumble* and J ocelyn G. Millar
Department of Entomology, University of California, Riverside, California 92521
No significant effects were observed in bioassays designed to measure impact of the furanocoumarin
precursors demethylsuberosin and marmesin on typical physiological parameters (survival and
development rate) of the generalist herbivore, Spodoptera exigua (Hu¨bner). The linear furanocou-
marin psoralen did increase developmental times and reduce survival. All of these compounds were
demonstrated to have significant behavioral (feeding deterrence) effects against both first and third
instars of S. exigua in diet-incorporation bioassays. In tests initiated with first instars, significantly
(P < 0.05) more larvae preferred control diet to diet containing marmesin or psoralen on all five
sample dates; control diets were significantly preferred to diets with demethylsuberosin on three of
five sample dates. Similar results were observed for tests initiated with third instar larvae, but
the avoidance of demethylsuberosin-containing diets was stronger. Therefore, assaying furanocou-
marin precursors just for effects on growth and survival may not provide an accurate picture of the
ecological importance of these compounds.
Keywor ds: Furanocoumarin; psoralen; marmesin; demethylsuberosin; feeding deterrents; Spodoptera
exigua; Apium graveolens
INTRODUCTION
The linear furanocoumarins psoralen, 5-methoxypso-
ralen (bergapten), and 8-methoxypsoralen (xanthotoxin)
have considerable biological activity against herbivores,
including humans (Ashwood-Smith et al., 1985; Trumble
et al., 1990; Beier et al., 1993). At low concentrations
they have proven deadly to many insect species (Ber-
enbaum, 1991), thus providing significant protection
against insect herbivory. Even arthropods that have
developed some tolerance to these compounds can show
reduced performance upon ingestion, including delayed
development (Diawara et al., 1993; Hadacek et al., 1994)
that may be associated with antifeedant responses
(Yajima et al., 1977; Muckensturm et al., 1981) and
reduced survival (Diawara et al., 1994).
Figu r e 1. Chemical structures of demethylsuberosin, marmes-
in, and the linear furanocoumarin psoralen.
is subsequently transformed to marmesin (Brown,
1978). Experiments investigating the related coumarin
derivatives ostruthin and osthol found effects on insect
survival or growth (Hadacek et al., 1994). The experi-
ments reported here were designed to examine potential
physiological and behavioral effects of demethylsuber-
osin and marmesin on a generalist insect herbivore,
Spodoptera exigua (Hu¨bner).
The highly polyphagous Spodoptera species are often
used as test animals for studies of coumarins and
furanocoumarins. However, antifeedant responses to
these compounds are not always consistent across
instars nor equivalent between species. Berenbaum
(1978) reported that xanthotoxin acted as a feeding
deterrent to sixth instar S. eridania (Cramer) but did
not reduce feeding by first or second instars. Yajima
et al. (1977) showed that coumarin, umbelliferone, two
7-hydroxycoumarins, and psoralen had little or no
antifeeding activity against S. litura (F). In contrast,
isopimpinellin and bergapten demonstrated strong an-
tifeedant activity.
Little evidence is available on the biological activity
of marmesin and demethylsuberosin, the immediate
precursors to the furanocoumarins (Figure 1). Accord-
ing to Berenbaum (1991, and references therein), fura-
nocoumarins are derived from umbelliferone via pre-
nylation. In Apiaceae and Rutaceae, two plant families
in which furanocoumarins are commonly found, umbel-
liferone is first converted to demethylsuberosin, which
MATERIALS AND METHODS
Syn th esis of Dem eth ylsu ber osin a n d Ma r m esin . NMR
spectra were recorded with a GE 300 instrument, at 300 MHz
(protons) or 75 MHz (13C). Mass spectra were recorded either
with a Hewlett-Packard 5970B mass selective detector (elec-
tron impact, 70 eV) coupled to an H-P 5890 gas chromatograph
equipped with a DB5-MS column (20 m × 0.2 mm i.d.; J &W
Scientific, Folsom, CA) or with a VG 7070E instrument (Fisons
International, Beverly, MA), using a direct insertion probe.
The temperature program used for all GC-MS runs was 100
°C/0 min, 15 °C/min to 275 °C for 20 min. Melting points were
measured with a Fisher-J ohns melting point apparatus and
are uncorrected.
The synthetic route to demethylsuberosin and marmesin is
shown in Figure 2. Thus, a mixture of 7-hydroxycoumarin 1
(16.2 g, 100 mmol), benzyl bromide (19 g, 111 mmol), 50 g of
K2CO3, and 600 mL of acetone was refluxed for 48 h. An
additional 3 g of benzyl bromide was then added, and the
mixture was refluxed for a further 24 h. The mixture was then
cooled to 20 °C and filtered, and the filtrate was concentrated.
The residue was added to water, and the aqueous mixture was
extracted 3 times with CHCl3. The combined CHCl3 solutions
were washed with water and brine, dried, and concentrated.
The residue was recrystallized from EtOH (1 L), yielding pale
yellow plates (23.4 g, 98%), mp 155-6 °C. NMR and mass
spectra matched those previously reported (Trumble et al.,
1992).
* Author to whom correspondence should be ad-
dressed [telephone and fax (909) 787-5624; e-mail
J ohn@ucrac1.ucr.edu].
S0021-8561(96)00156-2 CCC: $12.00
© 1996 American Chemical Society

2860 J. Agric. Food Chem., Vol. 44, No. 9, 1996
Trumble and Millar
material. A portion of the crude material was recrystallized
from benzene, mp 131-2 °C (lit. 131-3 °C; Massanet et al.,
1987). NMR: δ 7.63 (d, 1 H, J ) 9.41 Hz), 7.20 (s, 1 H), 6.91
(s, 1 H), 6.25 (d, 1 H, J ) 9.47 Hz), 5.32 (m, 1 H), 3.38 (br d,
2 H, J ) 8.18 Hz), 1.80 (br s, 3 H), 1.77 (br s, 3 H). MS: 230
(M+, 45), 215 (14), 187 (5), 176 (12), 175 (100), 159 (5), 147
(19), 115 (6), 91 (8), 77 (7), 69 (21), 51 (10), 41 (11).
Crude demethylsuberosin 6 (1.07 g, 4.7 mmol) in CHCl3 (10
mL) was added dropwise to an ice-cooled slurry of m-chloro-
perbenzoic acid (3.36 g of 55% pure; 2 equiv) and NaHCO3 (3.36
g, 40 mmol) in CHCl3 (150 mL). The mixture was stirred for
3 h at 0 °C and then quenched by slow addition of a solution
of 10 g NaHSO3 in 150 mL water (Foams!). The mixture was
stirred 20 min and then the layers were separated. The CHCl3
layer was extracted twice with 0.2 M NaOH (to remove traces
of overreduced demethylsuberosin from the previous step) and
brine, then dried, and concentrated. The residue (1.01 g) was
recrystallized from 30 mL of toluene, yielding pure marmesin
7, mp 153 °C (lit. 150-152 °C, from benzene; Murray et al.,
1971). 1H NMR: δ 7.60 (d, 1 H, J ) 9.46 Hz), 7.22 (s, 1 H),
6.76 (s, 1 H), 6.22 (d, 1 H, J ) 9.56 Hz), 4.74 (t, 1 H, J ) 8.84
Hz), 3.14-3.3 (m, 2 H), 1.76 (s, 1 H), 1.37 (s, 3 H), 1.23 (s, 3
H). 13C NMR: δ 24.32, 26.08, 29.47, 71.62, 91.14, 97.87,
112.18, 112.73, 123.40, 125.10, 143.72, 155.60, 161.48, 163.18.
MS: 246 (M+, 47), 213 (23), 188 (72), 187 (100), 175 (14), 160
(28), 131 (20), 102 (8), 77 (15), 69 (14), 59 (61), 51 (16), 50 (10),
44 (14), 43 (34), 41 (12).
Bioa ssa ys for Su r viva l, Weigh t Ga in , a n d Develop -
m en ta l Ra te. For diets containing psoralen (Aldrich Chemi-
cal Co., Milwaukee, WI), demethylsuberosin, or marmesin, the
chemicals were dissolved in ethanol and adsorbed onto al-
phacel (ICN Biochemicals, Cleveland, OH), removing the
ethanol by vacuum as described by Chan et al. (1978). The
amount of alphacel constituted 5% of the entire diet media.
The ethanol and alphacel procedure was used for all treat-
ments including the control. Diet medium (Patana, 1969) then
was added to the alphacel and blended for 5 min prior to
dispensing into 30 mL plastic cups.
Diet cups were covered with Teflon FEP fluorocarbon film
(E. I. DuPont de Nemours & Co., Wilmington, DE). This film
is more transparent to ultraviolet light than glass. For
treatments exposed to UV radiation, cups with larvae were
placed beneath UV-producing fluorescent lamps (40W Sylvania
350 blacklight, Inland Lighting Supplies, Riverside, CA). UV
lamps with a peak intensity at 350 nm were chosen because
wavelengths between 300 and 400 nm are believed to be
critical for activation of the psoralens (Musajo and Rodighiero,
1962). The lamps were adjusted in height such that the
intensity of UV light beneath a layer of the Teflon film was
20 µmW/cm2, a value equivalent to the irradiance observed
beneath a celery leaf in coastal southern California at mid-
morning during the late summer (Trumble, personal observa-
tion). All UV measurements were made with a System 371
Optical Power Meter equipped with a Model 268 detector head
with a UV optimization filter peaked at 365 nm (United
Detector Technology, Hawthorne, CA). Our goal was neither
to determine the extent of potential activation at varying levels
of UV nor to prove that marmesin and demethylsuberosin are
or are not activated by UV. Rather, we wanted to document
if the level of UV occurring in foliage, in conjunction with the
test compounds, would have significant developmental effects
on S. exigua.
F igu r e 2. Schematic for synthesis of demethylsuberosin and
marmesin.
Protected coumarin 2 (23 g, 97 mmol) was refluxed 20 h in
a solution of NaOMe (prepared from 23 g of Na in 750 mL of
dry MeOH). The solution was cooled to 20 °C, quenched with
glacial acetic acid (65 mL), and poured into 2 L of cold 0.1 M
HCl. The mixture was stirred thoroughly and filtered, and
the filter cake was rinsed thoroughly with cold water. The
solids were air dried overnight, then pumped under vacuum
for 6 h, yielding 25 g (96%) of the ring-opened product 3 as a
pale brown powder, which was used without further purifica-
tion. An analytical sample was recrystallized from benzene,
yielding 3 as a white solid, mp 185-6 °C. 1H NMR: δ 7.89
(d, 1 H, J ) 16.07 Hz), 7.48-7.3 (m, 6 H), 6.58 (dd, 1 H, J )
8.63, 2.25 Hz), 6.46 (d, 1 H, J ) 16.46 Hz), 6.42 (d, 1 H, J )
2.64 Hz), 5.6 (br s, 1 H, OH), 5.07 (s, 2 H), 3.79 (s, 3 H).
Compound 3 reverted to 2 upon attempted GC-MS analysis
(see: Cairns et al., 1986c). MS (DIP, 50 eV): 284 (13), 253
(2.7), 133 (2.9), 91 (100), 65 (7.5).
Methyl ester 3 (13.5 g, 50 mmol) was refluxed with 1-bromo-
3-methyl-2-butene (15 g, 100 mmol) and 40 g of K2CO3 in
acetone (500 mL) for 2 h. The mixture was cooled and filtered,
and the filtrate was concentrated. The residue was taken up
in hexane (500 mL) at room temperature and filtered to remove
unreacted starting material, and the filtrate was concentrated,
yielding 12.53 g (74%) of ester 4 as a viscous yellow oil, which
was used without further purification. NMR: δ 7.93 (d, 1 H,
J )16.12 Hz), 7.48-7.3 (m, 6 H), 6.57 (m, 2 H), 6.44 (d, 1 H,
J ) 16.14 Hz), 5.49 (m, 1 H), 5.08 (s, 2 H), 4.55 (br d, 2 H, J
) 6.48 Hz), 3.28 (s, 3 H), 1.80 (s, 3 H), 1.73 (s, 3 H).
4
rearranged to 5 and other isomeric products during attempted
GC-MS analysis. MS (DIP, 50 eV): 352 (13), 284 (74), 253
(10), 224 (5), 193 (3), 162 (2), 134 (3), 91 (100), 69 (11).
Ester 4 (12.5 g, 37 mmol) was refluxed in diethylaniline (120
mL) under Ar for 3 h. The solution was cooled, poured into 1
L of 1.5 M HCl, and extracted 3 times with EtOAc. The
combined EtOAc extracts were washed with 1.5 M HCl and
brine, then dried, and concentrated. The off-white solid
residue was purified in two batches by flash chromatography
(25 cm x 5 cm i.d.), eluting with toluene:EtOAc, 95:5, yielding
8.43 g (71%) of substituted coumarin 5. An analytical sample
was recrystallized from EtOH (white needles, mp 108-9 °C).
NMR: δ 7.63 (d, 1 H, J ) 9.41 Hz), 7.48-7.3 (m, 5 H), 7.22 (s,
1 H), 6.84 (s, 1 H), 6.24 (d, 1 H, J ) 9.48), 5.31 (m, 1 H), 5.16
(s, 2 H), 3.38 (br d, 2 H, J ) 8.2 Hz), 1.78 (s, 3 H), 1.67 (s, 3
H). MS: 320 (M+, 11), 229 (17), 187 (4), 175 (6), 92 (8), 91
(100), 65 (10).
Coumarin 5 (2.85 g, 8.4 mmol) in EtOH (25 mL) was added
to an aqueous slurry of Raney nickel (10 g; Aldrich Chemical
Co. no. 22-167-8), and the mixture was stirred for 28 min at
room temperature under nitrogen. The mixture was then
rapidly filtered through a plug of Celite and concentrated. The
residue was taken up in CHCl3, and the CHCl3 solution was
extracted twice with 0.2 M aqueous NaOH. The CHCl3
solution was then washed with brine, dried, and concentrated,
yielding 0.37 g (1.2 mmol) of unreacted starting material.
The combined NaOH solutions were acidified with 1 M HCl
and extracted 3 times with EtOAc. The EtOAc extracts were
washed with brine and dried, yielding 1.66 g (86%; quantita-
tive based on recovered starting material) of crude demethyl-
suberosin 6, contaminated with a small amount of overreduced
The potential influence of demethylsuberosin, marmesin,
and psoralen on S. exigua larval survival, weight gain, and
development rate was examined by using diet bioassay.
Twenty neonate S. exigua/diet were placed in individual cups
on diets containing demethylsuberosin (400 µg/g of diet),
marmesin (100, 250, and 400 µg/g of diet), psoralen (375 µg/g
of diet), and two control diets containing no furanocoumarins
or precursors (one control diet was exposed to UV light, the
other was not). Selection of relatively high test concentrations
was based on LC50 values reported for psoralen (Diawara et
al., 1993). Foliar levels of furanocoumarins in some celery
plants have been found to exceed 400 µg/g of fresh weight
(Trumble et al., 1990). Demethylsuberosin has been isolated
from umbelliferous and rutaceous plants (Brown and Steck,

Activity of Marmesin and Demethylsuberosin
J. Agric. Food Chem., Vol. 44, No. 9, 1996 2861
1972, and references therein), but in vivo concentrations were
not reported. Because demethylsuberosin and marmesin must
be produced before the furanocoumarins can accumulate, the
levels of these compounds may be quite high. The speed of
the biosynthetic transformation in A. graveolens from dem-
ethylsuberosin to marmesin to the linear furanocoumarins is
not known. However, in Ruta graveolens, transformation from
demethylsuberosin to marmesin appears to be fairly rapid (a
matter of hours), while marmesin accumulates and only slowly
transforms to furanocoumarins (Brown and Steck, 1972). In
one of the few assays of marmesin in vivo (for A. graveolens),
Afek et al. (1993) reported that 24 µg/g of fresh weight of
marmesin were present concurrently with 10 µg/g of fresh
weight of the linear furanocoumarins in the cultivar “Tender
Crisp”. Thus, the rates used in the experiments reported here
represent the high end of concentrations, as might be found
in resistant cultivars (Trumble et al., 1990; linear furanocou-
marin levels >400 µg/g of fresh weight in A. graveolens) or
stressed celery (Dercks et al., 1989; >165 µg/g of fresh weight
in A. graveolens).
All insects except those on the second control diet were
exposed to UV as previously described. A black cloth was used
to separate the second control group from the UV lights but
not the indirect room lighting controlling photoperiod (40-W
Sylvania Cool Daylight). All insects were kept in an environ-
mental chamber at 27 ( 2 °C and a 16:8 (L:D) photoperiod
and examined daily. The entire test was replicated three
times. For insects on each diet, larval weights at 9 days, pupal
weights, time to pupation and adult eclosion, and survival to
the pupal and adult stages were recorded.
ation of phenol 3 with 1-bromo-3-methyl-2-butene did
not go to completion. However, the product 4 was
readily separated from unreacted 3 by trituration of the
mixture with hexane and filtration to remove the
hexane-insoluble 3. Purified 4 was then refluxed in
diethylaniline, producing a mixture of rearrangement
products from which the desired demethylsuberosin
benzyl ether 5 was separated by flash chromatography.
Clean removal of the benzyl protecting group from 5
proved problematic, as previously reported (Cairns et
al., 1986a,b). Treatment of 5 with Raney nickel and
hydrogen resulted in extensive overreduction of the
prenyl side chain in addition to the desired removal of
the benzyl ether. Monitoring the progress of the reac-
tion was further complicated because 5 and the product
6 had very similar Rf values on TLC. However, it was
found that careful treatment of 5 with an aqueous slurry
of Raney nickel, terminating the reaction at approxi-
mately 75% conversion, minimized overreduction. The
debenzylated product 6 was readily separated from
unreacted 5 by extraction of a CHCl3 solution of the
crude mixture with aqueous base, with 6 being recov-
ered from the aqueous phase after acidification and
unreacted 5 being recovered for recycling from the
CHCl3 solution. Crude 6 was purified by recrystalliza-
tion to remove traces of overreduced material, giving
demethylsuberosin.
Bioa ssa ys for F eed in g Deter r en ce. Choice tests between
control diet and either demethylsuberosin, marmesin, or
psoralen were performed with neonates and third instar S.
exigua, using arenas adapted from Gould et al. (1991). For
neonates, arenas were constructed from 30 mL plastic cups
lined with 4% agar (w/v) with two holes at opposite sides of
the cup where 1.5 mL microcentrifuge cups were placed. One
of these tubes contained control diet, and the other contained
treated diet. Demethylsuberosin and marmesin were tested
at 400 µg/g of diet, and psoralen was tested at 375 µg/g of diet.
For each combination of treated diet and control diet, positions
of larvae (5/arena) were monitored for 25 arenas for 5 days at
approximately 24 h intervals. Throughout the test, larvae
were held at 27 ( 2 °C and a 16:8 (L:D) photoperiod. This
study was replicated 3 times.
Preferences of third instars for control or treated diets were
tested at the same concentrations as in the neonate studies.
The arenas consisted of 150 mL containers lined with 4% agar
(w/v) with four holes in a cardinal arrangement with four, 1.5-
mL microcentrifuge tubes. Opposite tubes contained the same
treated or control diet. Two newly-molted third instars were
placed in each of 25 arenas and monitored for position at 24-h
intervals for 5 days. Throughout the test, larvae were held
at 27 ( 2 °C and a 16:8 (L:D) photoperiod. This study was
replicated 3 times.
However, crude 6 could be used directly in the
subsequent cyclization to marmesin, from which the
overreduced material was readily removed. Thus, one-
step epoxidation and cyclization under buffered condi-
tions (Murray et al., 1971; Steck, 1971) produced
marmesin in excellent yield. Overreduced material,
containing a free phenolic functionality, was extracted
into aqueous NaOH, and the marmesin product remain-
ing was purified by recrystallization from benzene.
Bioa ssa ys for Su r viva l, Weigh t Ga in , a n d Devel-
op m en ta l Ra te. Even at high dosages (up to 400 µg/g
of diet), demethylsuberosin and marmesin did not
significantly affect larval or pupal survival (Figure 3).
Psoralen, however, was toxic (F6,12 ) 6.08, P ) 0.004;
Fisher’s Protected LSD Test), allowing only 30% of the
larvae to survive 9 days. Approximately 10% of the
larvae successfully pupated, but only half of these
emerged as adults (Figure 3c). These results are
consistent with previous studies reporting no mortality
from ingestion of other precursors (Hadacek et al., 1994)
but substantial mortality from high concentrations of
psoralen (Diawara et al., 1993).
Ingestion of the precursors did not increase develop-
mental time in the larval or pupal stages or decrease
larval or pupal weight as compared with control larvae
exposed to UV (Table 1). Developmental times were
longer for all larvae (including controls) exposed to UV
Sta tistica l An a lyses. Because the data for the behavioral
bioassays were not normally distributed, the Wilcoxon signed
rank test was used (Lance, 1992). All other data were
analyzed by using one-way ANOVA (SuperANOVA, 1989).
Percent data were transformed with the arcsin square-root
transformation prior to analysis. Means were separated at
the 5% significance level using the Tukey-Kramer test or
Fisher’s protected LSD test (SuperANOVA, 1989).
than for larvae not subjected to UV radiation (F6,19
)
3.253, P < 0.01; Tukey-Kramer test). Similarly, larval
weights were reduced for all larvae exposed to UV
radiation, regardless of treatment (F6,19 ) 6.175, P <
0.01; Tukey-Kramer test). Like previous studies, inges-
tion of psoralen significantly increased the number of
RESULTS
Syn th esis of Dem eth ylsu ber osin a n d Ma r m esin .
Demethylsuberosin was synthesized essentially as out-
lined in a previous communication (Cairns et al., 1986b).
More detailed descriptions of the syntheses of demeth-
ylsuberosin and marmesin are provided here to aid
repetition (Figure 2). Thus, 7-hydroxycoumarin 1 was
protected as its benzyl ether 2, followed by opening of
the pyranone ring with NaOMe, giving phenol 3. Alkyl-
days required to reach the pupal stage (F6,19 ) 3.253, P
< 0.01; Tukey-Kramer test) and the adult stage (F6,19
) 2.242, P < 0.05; Tukey-Kramer test) and significantly
decreased larval weight (F6,19 ) 6.175, P < 0.01; Tukey-
Kramer test) (Table 1). Interestingly, pupal weights
were not statistically different among any of the treat-
ments, suggesting there is a critical weight which a
larva must achieve in order to successfully pupate.

2862 J. Agric. Food Chem., Vol. 44, No. 9, 1996
Trumble and Millar
F igu r e 4. Percent of S. exigua choosing diets containing
demethylsuberosin, marmesin, psoralen, or no coumarin-based
compounds. Tests were initiated with first instars. For each
sample date within each graph, asterisks above data points
indicate significance at the P < 0.05 level (*) or the P < 0.01
level (**), Wilcoxon signed rank test.
erosin-containing diet was statistically significant on
three of five sample dates. For marmesin and psoralen,
this trend was substantially stronger, with control diets
strongly preferred to coumarin-containing diets on all
sample dates. Fewer than 25% of the larvae were found
on diets with marmesin or psoralen on any sample date.
Behavioral responses of third instar S. exigua to diets
containing demethylsuberosin, marmesin, or psoralen
were similar but more pronounced (Figure 5). For
demethylsuberosin, significantly more larvae were found
on control diets on four of five sample dates (Wilcoxon
signed rank test, P < 0.05; see Figure 5 for specific
sample dates). Control diets again were strongly pre-
ferred to coumarin-containing diets on every sample
date. Less than 20% of the larvae were found on diets
with marmesin or psoralen, with multiple sample dates
showing 0-5% of the available population on diets
containing these compounds.
F igu r e 3. Survival, pupation, and adult eclosion of S. exigua
exposed to psoralen, demethylsuberosin, or marmesin. (A)
Percent survival of larvae at 9 days. (B) Percent of test
population successfully pupating. (C) Percent of population
successfully eclosing as adults. Rates for the compounds tested
include demethylsuberosin (400 µg/g of diet), marmesin (low
) 100, medium ) 250, and high ) 400 µg/g of diet), psoralen
(375 µg/g of diet), and control diets containing no furanocou-
marins or precursors. All treatments except one control group
were exposed to UV light beneath a layer of Teflon film at
1.023 mW/cm2. Lines above bars delineate standard errors of
the mean. Different letters above bars within a test (A, B, or
C) indicate significant differences (ANOVA, Fisher’s protected
LSD test).
Ta ble 1. Effect of In gestion of Lin ea r F u r a n ocou m a r in
P r ecu r sor s a n d P sor a len on Develop m en ta l Ra te a n d
Weigh t Ga in by S. exigu a a
developmental
time, days
weight, mg
DISCUSSION
treatment (rate, µg/g)
pupal
22.8 c
adult
larvab
pupa
The determination that demethylsuberosin and
marmesin elicit a behavioral response has several
interesting implications. Because the presence of the
furanocoumarin precursors significantly affected feeding
preferences, plants which produce these compounds
could be afforded some protection from generalist insect
herbivores. Thus, early in the evolution of the furano-
coumarins, as well as at the present time, demethylsu-
berosin and marmesin could be expected to provide some
defense against herbivores such as S. exigua. This
result is important because it eliminates the need for
two major biosynthetic steps to occur consecutively and
without any apparent evolutionary driving force (pro-
duction of demethylsuberosin and then marmesin)
before the linear furanocoumarins can be synthesized
by plants.
psoralen (375)
marmesin (100)
marmesin (250)
marmesin (400)
demethylsuberosin (400) 17.4 b
control plus (UV)
control no (UV)
30.0 c
1.9 a
101.2 a
125.8 a
16.6 ab 22.1 ab 29.6 b
16.2 ab 23.3 ab 23.3 ab 106.6 a
17.4 b
23.7 ab 26.6 ab 97.7 a
23.8 bc 28.5 ab 104.2 a
18.0 b
15.2 a
23.8 bc 37.9 b
21.7 a 75.6 c
106.4 a
106.0 a
a
Analyses conducted using Super ANOVA (Abacus Concepts,
Berkeley, CA); different letters within columns indicate significant
differences at the P < 0.05 level, Fishers protected LSD test.
b
Weights measured at 9 days.
Bioa ssa ys for Deter r en cy. Diets containing dem-
ethylsuberosin, marmesin, or psoralen were signifi-
cantly less preferred by first instar S. exigua (Wilcoxon
signed rank test, P < 0.05; see Figure 4 for specific
sample dates). Avoidance by larvae of a demethylsub-

Activity of Marmesin and Demethylsuberosin
J. Agric. Food Chem., Vol. 44, No. 9, 1996 2863
Ashwood-Smith, M. J .; Ceska, O.; Chaudhary, S. K. Mecha-
nism of photosensitivity reactions to diseased celery. Br.
Med. J . 1985, 290, 1249.
Beier, R. C.; Ivie, G. W.; Oertli, E. H. Psoralens as phytoalexins
in food plants of the family Umbelliferae. In Xenobiotics in
Foods and Feeds; Finley, J ., Schwass, D., Eds.; ACS
Symposium Series 234; Washington, DC, 1993; pp 295-310.
Berdegue, M.; Trumble, J . T. Effects of plant chemistry and
physical characteristics of Apium graveolens and Chenopo-
dium murale on host selection by Spodoptera exigua larvae.
Entomol. Exp. Appl. 1996, 78, 253-262.
Berenbaum, M. R. Toxicity of a furanocoumarin to army-
worms: a case of biosynthetic escape from insect herbivores.
Science 1978, 201, 532-534.
Berenbaum, M. R. Coumarins. In Herbivores: Their Interac-
tions with Secondary Plant Metabolites. Rosenthal, G. A.,
Berenbaum, M. R., Eds.; Academic Press: San Diego, 1991;
Vol. 1, pp 221-250.
Bernays, E. A. Plant secondary compounds deterrent but not
toxic to the grass specialist acridid Locusta migra-
toriasimplications for the evolution of graminivory. Ento-
mol. Exp. Appl. 1990, 54, 53-56.
Brown, S. A. Biochemistry of the coumarins. Recent Adv.
Phytochem. 1978, 12, 249-285.
Brown, S. A.; Streck, W. 7-Demethylsuberosin and osthenol
as intermediates in furanocoumarin biosynthesis. Phy-
ochemistry 1973, 12, 1315-1324.
Cairns, N.; Harwood, L. M.; Astles, D. P. Application of the
Lewis-acid catalyzed Claisen rearrangement of 4′-(1,1-
dimethylallyloxy)coumarates to the synthesis of demethyl-
suberosin. J . Chem. Soc., Chem. Commun. 1986a , 10, 750-
752.
Cairns, N.; Harwood, L. M.; Astles, D. P. Synthesis of linear
coumarins via para-Claisen rearrangement of coumarate
ester derivatives: total synthesis of suberosin, demethyl-
suberosin, and ostruthin. J . Chem. Soc., Chem Commun.
1986b, 16, 1264-1266.
Cairns, N.; Harwood, L. M.; Astles, D. P. Regioselective
preparation of 6-allyl-7-hydroxycoumarin via boron halide
catalyzed ortho-Claisen rearrangement of 4′-allyloxycou-
maric acid derivatives. J . Chem. Soc., Chem. Commun.
1986c, 2, 182-183.
Chan, B. C.; Waiss, A. C. J .; Stanley, W. L.; Goodban, A. E. A
rapid diet preparation method for antibiotic phytochemical
bioassay. J . Econ. Entomol. 1978, 71, 366-368.
Dercks, W.; Trumble, J . T.; Winter, C. Impact of atmospheric
pollution on linear furanocoumarin content in celery. J .
Chem. Ecol. 1990, 16, 443-454.
F igu r e 5. Percent of S. exigua choosing diets containing
demethylsuberosin, marmesin, psoralen, or no coumarin-based
compounds. Tests were initiated with third instars. For each
sample date within each graph, asterisks above data points
indicate significant differences between controls and treat-
ments at the P < 0.05 level (*) or the P < 0.01 level (**),
Wilcoxon signed rank test.
Spodoptera exigua, like many armyworms, is highly
mobile in the latter larval stages and each individual
usually feeds on multiple plants (Smits et al., 1987;
Berdegue and Trumble, 1996). The consequences of
leaving the host plant due to the presence of a feeding
deterrent effect are probably substantial for the small
first instars. Third instars, in contrast, are highly
mobile and may feed on several plants a night (Smits
et al., 1987). These larger instars can readily avoid
plants containing feeding deterrents. Thus, the feeding
deterrent effect is distinctly different for each life stage.
Because food plants for the highly polyphagous S. exigua
are common in both natural and agricultural habitats
(Mitchell, 1979), host plant choice by larvae could be
an important evolutionary force.
Diawara, M. M.; Trumble, J . T.; Quiros, C. F.; White, K. K.;
Adams, C. Plant age and seasonal variations in genotypic
resistance of celery to beet armyworm (Lepidoptera: Noc-
tuidae). J . Econ. Entomol. 1994, 87, 514-522.
The results of our study indicate that assaying
furanocoumarins and their precursors only for effects
on growth and survival may not provide an accurate
picture of the importance of these compounds. Feeding
deterrents that are not toxic in the traditional sense
have been shown to play significant ecological and
evolutionary roles in herbivory (Bernays, 1990). Thus,
it would be wise to assay for deterrence or other
behavioral modifications before concluding that any
linear furanocoumarin precursor does not have a sig-
nificant effect on insect herbivores.
Diawara, M. M.; Trumble, J . T.; White, K. K.; Carson, W. G.;
Martinez, L. A. Toxicity of linear furanocoumarins to
Spodoptera exigua: evidence for antagonistic interactions.
J . Chem. Ecol. 1993, 19, 2473-2484.
Gould, F.; Anderson, A.; Landis, D.; Van Mellaert, H. Feeding
behavior and growth of Heliothis virescens larvae on diet
containing Bacillus thuringiensis formulations or endotox-
ins. Entomol. Exp. Appl. 1991, 58, 199-210.
Hadacek, F.; Mu¨ller, C.; Werner, A.; Greger, H.; Proksch, P.
Analysis, isolation and insecticidal activity of linear fura-
nocoumarins and other coumarin derivatives from Peu-
cedanum (Apiaceae: Apioideae). J . Chem. Ecol. 1994, 20,
2035-2054.
Lance, D. R. Odors influence choice of oviposition sites by
Diabrotica virgifera virgifera (Coleoptera: Chrysomelidae).
J . Chem. Ecol. 1992, 18, 1227-1237.
ACKNOWLEDGMENT
We thank M. Diawara, K. White, and W. Carson for
assistance on this project. We appreciate the reviews
of S. Reitz, C. Rodriquez-Saona, and R. Redak.
Massanet, G. M.; Pando, E.; Rodriguez-Luis, F.; Salva, J .
Synthesis of 3-(1′,1′-dimethylallyl)coumarins: gravellifer-
one, chapelin, and rutamarin. Heterocycles 1987, 26, 1541-
1548.
Mitchell, E. R. Migration by Spodoptera exigua and Spodoptera
frugiperda- North American style. In Movement of Highly
LITERATURE CITED
Afek, U.; Carmeli, S.; Aharoni, N.; Roiser, L. A suggestion for
new mechanism of celery resistance to pathogens. Acta
Hortic. 1993, 343, 357-360.

2864 J. Agric. Food Chem., Vol. 44, No. 9, 1996
Trumble and Millar
Mobile Insects: Concepts and Methodology; Rabb, R.,
Kennedy, G., Eds.; North Carolina State University: Ra-
leigh, NC, 1979; pp 386-395.
Muckensturm, B.; DuPlay, D.; Robert, P. C.; Simonis, M. T.;
Kienlen, J . C. Substances antiappetantes pour insectes
phytophages presentes dans Angelica et Heracleum sphondy-
lium. Biochem. Syst. Ecol. 1981, 9, 289-292.
Murray, R. D. H.; Sutcliffe, M.; McCabe, P. H. Claisen
rearrangements-IV. Oxidative cyclization of two coumarin
o-isoprenylphenols. Tetrahedron 1971, 27, 4901-4906.
Musajo, L.; Rodighiero, G. The skin photosensitizing furano-
coumarins. Experientia 1962, 18, 153-200.
Patana, R. Rearing cotton insects in the laboratory. U.S.
Department of Agriculture Product Research Report 108,
1969; 6 pp.
Smits, P. H.; van Velden, M. C.; van deVrie, M.; Vlak, J . M.
Feeding and dispersion of Spodoptera exigua larvae and its
relevance for control with a nuclear polyhedrosis virus.
Entomol. Exp. Appl. 1987, 43, 47-49.
Steck, W. New syntheses of demethylsuberosin, xanthyletin,
(()-decursinol, (+)-marmesin, (-)-nodakenetin, (()-decurs
in, and (()-prantschimgin. Can. J . Chem. 1971, 49, 2297-
2301.
SuperANOVA. Abacus Concepts Inc.: Berkeley, CA, 1989.
Trumble, J . T.; Dercks, W.; Quiros, C. F.; Beier, R. C. Host
plant resistance and linear furanocoumarin content of
Apium accessions. J . Econ. Entomol. 1990, 83, 519-525.
Trumble, J . T.; Millar, J . G.; Ott, D. E.; Carson, W. C. Seasonal
patterns and pesticidal effects on the phototoxic linear
furanocoumarins in celery, Apium graveolens L. J . Agric.
Food Chem. 1992, 40, 1501-1506.
Yajima, T.; Kato, N.; Munakata, K. Isolation of insect anti-
feedant principles in Orixa japonica Thunb. Agric. Biol.
Chem. 1977, 41, 1263-1268.
Received for review March 8, 1996. Revised manuscript
received J une 17, 1996. Accepted J une 25, 1996.X
J F960156B
X Abstract published in Advance ACS Abstracts, Au-
gust 15, 1996.