Defense by Volatiles in Leaf-Mining Insect Larvae

Jean-Luc Boevé & Gontran Sonet & Zoltán Tamás Nagy &

Françoise Symoens & Ewald Altenhofer &

Christopher Häberlein & Stefan Schulz

Received: 2 February 2009 /Revised: 27 March 2009 /Accepted: 2 April 2009 /Published online: 24 April 2009# Springer Science + Business Media, LLC 2009

Abstract The defense strategy of an insect toward naturalenemies can include a trait that appears at first sight tocontradict its defensive function. We explored phylogeny,chemistry, and defense efficiency of a peculiar group ofhymenopteran sawfly larvae where this contradiction isobvious. Pseudodineurini larvae live in leaf mines thatprotect them from some enemies. Disturbed larvae alsoemit a clearly perceptible lemon-like odor produced byventral glands, although the mine hampers the evaporationof the secretion. The mine could also lead to autointoxica-tion of a larva by its own emitted volatiles. Citral was themajor component in all Pseudodineurini species, and itefficiently repels ants. We conclude that full-grown larvaethat leave their mine to pupate in the soil benefit from citralby avoiding attacks from ground-dwelling arthropods suchas ants. In some species, we also detected biosyntheticallyrelated compounds, two 8-oxocitral diastereomers (i.e.,

(2E,6E)- and (2E,6Z)-2,6-dimethylocta-2,6-dienedial). Syn-thetic 8-oxocitral proved to be a potent fungicide, but notan ant repellent. The discrete distribution of 8-oxocitral wasunrelated to species grouping in the phylogenetic tree. Incontrast, we discovered that its presence was associatedwith species from humid and cold zones but absent inspecies favoring warm and dry environments. The formershould be protected by 8-oxocitral when faced with afungal infestation while crawling into the soil. Our workshows the importance of integrating knowledge aboutbehavior, morphology, and life history stages for under-standing the complex evolution of insects and especiallytheir defense strategies.

Keywords Sawfly larvae . Nematinae . Ventral glands .

Pseudodineura . Leaf-mining insect .Monoterpenes .

Citral . 8-Oxocitral . Ant repellent and antifungal activities .

Abiotic factors

Introduction

Predation pressure is a potent evolutionary force especiallyon insects, which are a basic element in most terrestrialecosystems. As a consequence, defense strategies arewidespread and diversified in insects and often rely on theuse of noxious chemicals (Eisner 1970; Blum 1981).Chemical defense mechanisms have a profound impact onthe entire biology of an insect species, as defense is coupledwith other adaptations in morphology, physiology, nicheuse, behavior, etc. (Whitman et al. 1990). Other nonchem-ical defensive traits also interact in the evolution of insectspecies (Evans and Schmidt 1990). A trait in isolation mayappear at first glance to be in contradiction to the others.For instance, conspicuousness renders insects more visible,

J Chem Ecol (2009) 35:507–517DOI 10.1007/s10886-009-9627-3

J.-L. Boevé (*) :G. Sonet : Z. T. NagyRoyal Belgian Institute of Natural Sciences,Rue Vautier 29,B-1000 Brussels, Belgiume-mail: Jean-Luc.Boeve@naturalsciences.be

F. SymoensMycology Section, Scientific Institute of Public Health,Rue Juliette Wytsman 14,B-1050 Brussels, Belgium

E. AltenhoferEtzen 39,A-3920 Groβ Gerungs, Austria

C. Häberlein : S. SchulzInstitute of Organic Chemistry,Technische Universität Braunschweig,Hagenring 30,38106 Braunschweig, Germany

thus, more vulnerable; but this trait is used in defensestrategies such as aposematism and mimicry, by relying onthe predator’s associative learning capabilities (Guilford1990; Ruxton and Sherratt 2006).

We detected such a paradoxical situation in the biologyof a group of phytophagous insects that not only live withina plant tissue but also emit defensive volatiles. The tribePseudodineurini belongs to the hymenopteran nematine(Tenthredinidae) sawflies (see Boevé (2008) for an intro-duction to the family). The larvae of this tribe are specializedleaf miners on plants mainly belonging to the buttercupfamily Ranunculaceae (Altenhofer 2003; Altenhofer andPschorn-Walcher 2006; Table 1, Fig. 1). Compared to free-living sawfly larvae, endophytic larvae presumably aremechanically protected from some natural enemies (Priceand Pschorn-Walcher 1988; Connor and Taverner 1997).Additionally, almost all larvae of the 1,200 highly diversenematine species possess ventral glands that emit a volatileantipredator secretion (Boevé and Pasteels 1985). A typicallemon-like odor is perceived easily when Pseudodineurinilarvae are disturbed. The odor intrigued us since volatilesoften repel (at a distance) approaching and/or attackingsmall arthropods (Pasteels et al. 1983). For example, theodor of lemon eucalyptus repels insects such as mosquitoes(Moore et al. 2007). Further, several nematine speciesamong the free-living Nematus and galling Pontania show asecondary reduction in gland size (Boevé and Pasteels1985), no perceptible odor, only a trace of volatiles such aslong chain hydrocarbons (Boevé et al. 1992), and thusprobably a more or less functionless secretion. Pseudodi-nerini larvae are endophytic and emit repellents, and ouraim was to understand this apparent redundancy ofdefensive traits. The latter trait is especially puzzling sincethe closed environment in which the larva lives hampers theevaporation of volatiles and could lead to autointoxication.

Pseudodineurini is a small tribe that comprises about 15described Holarctic species (Taeger and Blank 2008). Adozen belong to the genus Pseudodineura, and larvae areknown for seven Western Palaearctic species. Adults areabout 5 mm in body length, and larvae up to 1 cm. Severalspecies of the tribe are restricted locally and occur onlyduring a short season (Altenhofer and Pschorn-Walcher2006). However, we were able to gather enough materialfor chemical and phylogenetic analyses. Preliminary chem-ical analyses revealed the occurrence of two acyclicmonoterpene diastereomers (E)-3,7-dimethyl-2,6-octadienal(geranial) and (Z)-3,7-dimethyl-2,6-octadienal (neral; themixture of the two compounds is called citral) in theglandular secretion of Pseudodineura larvae. These com-pounds also occur in lemon, lemon grass, etc. and areimportant in the food and perfume industry. Their occur-rence was expected in Pseudodineura due to the perceptibleodor and because free-living Nematinus larvae also emit

them (Boevé et al. 1984). However, two isomers of anotherunidentified monoterpene occurred in only some Pseudo-dineura species. In the present study, the chemical profileof the glandular secretion of this group was investigated inrelation to phylogeny. Moreover, bioassays were performedto reveal the defensive efficiency of live larvae and tocompare the bioactivity of citral and the other monoter-penes. The results lead us to an explanatory framework thatbrings together the evolution and ecology that includesbiotic and abiotic factors. As a case study, it allows us tobetter comprehend how multifunctional traits can evolve.

Methods and Material

Sample Collection Sawflies were collected in the field,mainly in the provinceNiederösterreich of Austria (Table 1).The Pseudodineurini larvae were identified by EA, andvoucher specimens of all sawfly samples are kept by JLB atthe Royal Belgian Institute of Natural Sciences.

Phylogenetic Analyses Twenty-five specimens representingall West Palearctic Pseudodineurini species except Pseudo-dineura heringi were used for phylogenetic analyses. Wealso added one North American species (Pseudodineuraparva). Six outgroup species were chosen on the basis ofthe subfamilial phylogeny of Nematinae proposed byNyman et al. (2006).

Total genomic DNA was extracted by using DNeasyMini Kits (Qiagen). The posterior half of the larvae (or thewhole specimen when larvae were smaller than 5 mm)stored in 96% ethanol was used. Fragments of the mito-chondrial cytochrome b gene (cob; aligned length: 433 bp)were amplified and sequenced by using the primers CB-J-10933: 5′-TATGTACTACCATGAGGACAAATATC-3′,and CB-N-11367: 5′-ATTACACCTCCTAATTTATTAGGAAT-3′ (Simon et al. 1994). The cytochrome c oxidasesubunit I gene (COI; aligned length: 874 bp) was amplifiedand sequenced with the primers sym-C1-J-1718 (Nyman etal. 2006) and A2590 (Normark et al. 1999). A fragment ofthe nuclear 28S rRNA gene (28S; aligned length: 579 bp)was amplified and sequenced with the primers D2F: 5′-CGTGTTGCTTGATAGTGCAGC-3′ and D2R: 5′-TTGGTCCGTGTTTCAAGACGG-3′ (Schmidt et al. 2006).PCR products were sequenced on an ABI Prism 3130XLcapillary sequencer. For detailed list of samples andGenBank accession numbers, see Table 1. DNA sequencesof 28S rRNA were aligned with MAFFT 6 (Katoh et al.2002; Katoh and Toh 2008) by using the Q-INS-i option.Mitochondrial gene sequences were aligned manually.

Phylogenetic analyses were carried out with PAUP*v4.0b10 (Swofford 2002), MrBayes v3.1.2 (Ronquist andHuelsenbeck, 2003), and BEAST v1.4.8. (Drummond and

508 J Chem Ecol (2009) 35:507–517

Rambaut 2007). Trees were calculated with parsimony(gaps in 28S sequences were considered as fifth character),maximum likelihood (ML), and Bayesian (BI) methodsbased on each gene separately, on the concatenatedsequences of two genes (COI and 28S), and on the con-catenated sequences of three genes. The Akaike and theBayesian information criteria implemented in jModeltestv0.1.1 (Posada 2008) were used to find appropriatenucleotide substitution models, and the recommendedsettings were used in the subsequent ML and BI analyses.In order to infer branch supports, we performed 2,000bootstrap replicates under parsimony and 100 replicatesunder ML criteria using PAUP*. BI analyses were carriedout with MrBayes running 2×106 generations. For theprotein-coding genes (i.e., COI as in Fig. 2), we partitionedthe dataset according to codon positions (first and secondpositions vs. third positions), and performed a Bayesiananalysis in BEAST running 107 generations. In all Bayesiananalyses, the first 25% of the sampled trees were discarded(“burn-in”).

Chemical Analyses and Syntheses Full-grown Pseudodi-neurini larvae were taken out of their leaf mines, andsamples containing 13 to 30 individuals were extracted for5 min in approximately 0.5 ml hexane. For Pseudodineurafuscula, larvae as well as eonymphs were extracted in thisway. Extracts were kept at −80°C until gas chromatogra-phy/mass spectroscopy (GC–MS) analysis.

Most analyses were performed on a Fisons 8060 GCsystem equipped with a nonpolar capillary column (DB-5, 30 m×0.32 mm I.D., film thickness 0.25μm) andcoupled to a Fisons MD800 quadrupole mass spectrom-eter. The extract was concentrated, up to 0.25 ml, whenrelatively few larvae were extracted. From each extract,1μl was injected, with a solvent delay of 3 min. The tem-perature program was 2 min at 60°C, then up to 280°C with3°C min−1, and then 30 min at 280°C. Spectra wereobtained in EI mode.

To confirm the chemical composition of some extractsand to elucidate the structure of a monoterpene thatremained unidentified, additional analyses were performedon a Hewlett Packard GC 6890 with split/splitless-injectorand a fused silica capillary column (BPX5, SGE Inc.,25 m×0.22 mm I.D., film thickness: 0.25μm) connected toa Hewlett Packard MSD 5973 quadrupole mass spectrom-eter operating in the EI mode. The temperature programwas 1 min at 50°C, then up to 320°C with 10°C min−1.

Neral (c1) and geranial (c2) were synthesized from neroland geraniol via Swern oxidation (Omura and Swern 1978).The obtained c1 and c2 contained 1.4% of c2 and 1.1% ofc1, respectively. The 8-oxocitrals also were synthesizedfrom nerol and geraniol. First, an allylic oxidation wasperformed by using catalytic amounts of selenium dioxide

and tert-butyl hydroperoxide as cooxidant (Li et al. 1997).The resulting dialcohols then were oxidized via Swernoxidation (Omura and Swern 1978). (2E,6Z)-2,6-Dimethy-locta-2,6-dienedial (c3) contained 25% of (2E,6E)-2,6-dimethylocta-2,6-dienedial (c4), and c4 contained 10% ofc3. As far as we know, no selective synthesis exists thatwould furnish these compounds in the same diastereomericpurity as obtained for the citrals. The 8-oxocitrals are rela-tively unstable, and they slowly isomerize upon standing.Only the C-6 double bond isomerizes, thus, showing thehigher reactivity of 8-oxocitral at this position (Chan et al.1968). Before performing the bioassays (see below), therelative proportion of the mixtures c3 + c4 had alreadychanged from 3:1 to 1:1 and from 1:9 to 1:4. After thesebioassays, which required several months, the mixtureswere reanalyzed, and their relative proportions remainedstable for the mixture 1:1, whereas the other mixture waslargely degraded.

Bioassays Three types of bioassays were performed: sawflylarvae alive vs. ants, synthesized compounds vs. ants, andthese same compounds vs. fungi.

The interactions between larvae of Pseudodineuramentiens and ant workers of Myrmica rubra—a commonant species in Europe, including Austria (Seifert 1988)—were observed in 9-cm diam Petri dishes. First, one leaf ofHepatica nobilis with a full-grown larva still in its minewas deposited in a Petri dish containing ten ants. Sixreplications were made. After 24 h, we observed whetherthe ants opened the mine. Second, ten ants per Petri dishwere confronted with one larva just taken out of its mine.Seven full-grown and six penultimate instar larvae weretested. We counted after 15 min the number of ants thatsurrounded the larva, i.e., clearly making a mandibularcontact with it. After the bioassay, we determined the instarby recording the head coloration and measuring the headcapsule width (HCW). Full-grown larvae have a whitishhead and HCW=1.0 to 1.18 mm, whereas larvae atpenultimate instar have a dark head and HCW=0.85 to1.0 mm.

To measure the repellent effect of the synthesizedmonoterpene isomers (see above) we used the ant Crema-togaster scutellaris in an experimental setup described inBoevé (1988) and in which this ant species performedbetter than M. rubra. Forty ant workers were taken from alaboratory colony and left for 30 min in a 14-cm diam Petridish coated on its inner border with Fluon® to preventescape. Then, a 5×5-cm glass plate with a metallic podiumof 1 cm diam and 0.5 cm in height fixed at its center wasplaced in the uncovered Petri dish. The plate had 75μl of a1:1 water/honey solution deposited on the glass around thepodium. For 5 min, ants were allowed to find the solutionand feed. The number of feeding ants then was counted for

J Chem Ecol (2009) 35:507–517 509

Tab

le1

Saw

flysamples

used

inchem

ical

(GC-M

S)andgenetic

(DNA)analyses

Taxon

Field

host-plant

Vou

cher,stagea

Locality

bDate

Collector

cGC-M

Sd

DNAe

COIf

Cob

f28

Sf

Pseud

odineura

clem

atidis

(Hering)

Clematisalpina

P16

25,L

Umballfälle

08.06.19

99EA

+

P22

52,L

Umballfälle

31.07.20

02EA

.1FJ858

810

FJ858

866

FJ858

842

.2FJ858

811

FJ858

867

FJ858

843

P.clem

atidisrectae

Hering

Clematisrecta

P15

84,L

Langenlois

20.05.19

98EA

+

P21

69,L

Langenlois

25.05.20

02EA

.1FJ858

809

FJ858

861

FJ858

837

.2FJ858

815

FJ858

862

FJ858

838

.3FJ858

820

FJ858

863

FJ858

839

P29

01,L

Langenlois

04.06.20

08EA

.1–

FJ858

873

FJ858

852

.2FJ858

828

FJ858

874

FJ858

853

.3–

FJ858

875

FJ858

854

P.enslini(H

ering)

Trolliu

seuropa

eus

P15

88,L

Allentsteig

02.06.19

98EA

+

P21

90,L

Ötscher

05.07.20

02EA

.1FJ858

817

FJ858

865

FJ858

841

P29

00,L

Allentsteig

04.06.20

08EA

.1FJ858

825

–FJ858

849

.2FJ858

826

–FJ858

850

.3FJ858

827

–FJ858

851

P.fuscula(K

lug)

Ran

unculusspp.

P15

87,Lon

R.b.

Hernstein

01.06.19

98EA

+

P16

32,Lon

R.p.

Arbesbach

13.06.19

99EA

+

P21

67,Lon

R.p.

GroβGerun

gs09

.06.20

02EA

.1FJ858

812

FJ858

859

FJ858

835

.2–

––

.3FJ858

819

FJ858

860

FJ858

836

P28

99,Lon

R.p.

GroβGerun

gs04

.06.20

08EA

.1FJ858

822

FJ858

870

FJ858

846

.2FJ858

823

FJ858

871

FJ858

847

.3FJ858

824

FJ858

872

FJ858

848

P28

48,A

NanaAseme(E)

21.04.20

02MH

.1FJ858

818

FJ858

868

FJ858

844

P.hering

i(Enslin

)Anemon

esylvestris

P15

89,L

Langenlois

01.06.19

98EA

+

P.mentiens

(Tho

mson)

Hepaticano

bilis

P16

35,L

Neuleng

bach

29.06.19

99EA

+

P28

42,L

Langenlois

24.08.20

08EA

+

P21

72,L

Neuleng

bach

03.07.20

02EA

.1FJ858

816

FJ858

864

FJ858

840

P.pa

rvula(K

lug)

Pulsatilla

spp.

P15

90,Lon

P.v.

Langenlois

01.06.19

98EA

+

P21

68,Lon

P.p.

Langenlois

25.05.20

02EA

.1FJ858

813

––

.2FJ858

814

––

P.pa

rva(N

orton)

(Hepatica)

P28

49,A

Petersham

(M)

05.200

2TN

.1FJ858

821

FJ858

869

FJ858

845

End

ophytusan

emon

es(H

ering)

Anemon

enemorosa

P25

73,L

Etzen

08.05.20

05EA

+

P20

98,L

Etzen

05.05.20

02.1

FJ858

808

–FJ858

834

Dineura

pullior

Schmidt&

Walter

Betula

P18

09,Eo

labrearing(F)

04.200

1AK

.1FJ858

807

–FJ858

831

510 J Chem Ecol (2009) 35:507–517

the first time (t=0), and simultaneously, a piece of 5×5-mmfilter paper embedded with 0.25μl of the tested monoter-pene was placed on the podium. The number of feedingants was recounted, once per minute, up to t=10 min. Thisexperiment, performed at 25°C, was replicated four timesper monoterpene. Distilled water instead of a monoterpenewas used as a control.

Antifungal testing of c1, c2, c3 + c4 (1:1 and 1:4) wasperformed with the yeast Candida albicans (strain: IHEM9559) and the filamentous fungus Aspergillus fumigatus(IHEM 18963). These strains are referenced in the BCCM/IHEM Collection catalog (http://bccm.belspo.be/db/ihem_search_form.php). These two human pathogenic specieswere chosen because standardized reference methods areavailable for antifungal susceptibility testing (i.e., CLSIM27A (2002a) for yeasts and CLSI M38A (2002b) forfilamentous fungi). These tests are performed in 96-wellplates. We used an individual microplate per sample toavoid interferences due to the volatility of the compounds.Stock inoculum suspensions were prepared from a 24-h-oldculture of C. albicans grown on Sabouraud medium, andfrom a 7-day-old culture of A. fumigatus grown on potatodextrose agar. As starting sample amounts, we used 3.75,4.5, 5.0, and 3.0 mg of c1, c2, c3 + c4 (1:1), and c3 + c4(1:4), respectively. The sample was solubilized in 200μl ofsterile saline. The microplate contained 100μl of twofoldserial dilutions of the sample and 100μl of the fungalinoculum in twice concentrated RPMI medium (2%glucose, 0.165M MOPS). The final concentration of theinoculum was 2.5 103CFU/ml for C. albicans and 5 103

CFU/ml for A. fumigatus. Growth and sterility controlswere included in each experiment. Microplates wereT

able

1(con

tinued)

Taxon

Field

host-plant

Vou

cher,stagea

Locality

bDate

Collector

cGC-M

Sd

DNAe

COIf

Cob

f28

Sf

Hem

ichroa

crocea

(Geoffroy)

Alnus

glutinosa

P17

75,L

Grimminge

(B)

04.09.20

00JLB

.1FJ858

804

FJ858

856

FJ858

830

Nem

atinus

fuscipennis(Serville)

(Alnus)

P20

78,A

Illfeld(G

)20

.05.20

01JLB

.1FJ858

806

FJ858

858

FJ858

833

Stau

ronematus

compressicornis

(Fabricius)

Pop

ulus

trem

ula

P17

52,L

Wellin

(B)

01.08.20

00JLB

.1FJ858

803

FJ858

855

FJ858

829

Cladius

pectinicornis(G

eoffroy)

Rosasp.

P20

72,L

Delém

ont(S)

21.08.20

00US

.1FJ858

805

FJ858

857

FJ858

832

Llarva,

Eoeony

mph

,Aadult,R.b.Ran

unculusbu

lbosus,R.p.R.platan

ifoliu

m,P.v.Pulsatilla

vulgaris,P.p.

P.pratensis,

EEston

ia,M

Massachusetts,FFinland

,BBelgium

,G

Germany,

SSwitzerland

,MH

MikeHeidemaa,TNTom

miNym

an,AKAnttiKause,USUrs

Schaffner

aThe

correspo

ndinglarval

host-plant

givenbetween(parentheses),and,

whennecessary,

theplantspecies

bIfno

tAustria,thecoun

tryof

thelocalityismentio

ned

cCollectors,ifno

ttheauthors,areMH,TN,AK,andUS

dThe

Pseud

odineurini

larvae

where

alwayscollected,persample,

inasm

allarea,thus,clearlybelong

ingto

asing

lepo

pulatio

n.Theirho

st-plantsoccurred

asasm

allpatch,

especially

Clematis

recta,

Ran

unculusplatan

ifoliu

mfrom

GroβGerun

gs,andPulsatilla

vulgaris.Larvaeused

inchem

ical

analyses

aremarked(+)

eFor

genetic

analyses,weoftenused

morethan

oneindividu

alfrom

onesawflypo

pulatio

n,each

onebeingnu

mbered(.1,

.2,.3)as

DNA

individu

al.Thisspecim

ennu

mbering

isreused

inFig.2

fGenBankaccessionnu

mbers

aregivenforCOI,cob,

and28

S;sequ

ence

couldno

tbe

obtained

(–)

Fig. 1 Sawfly larva of Pseudodineura fuscula on Ranunculusplatanifolium. All Pseudodineura species are leaf miners ofRanunculaceae plants. Both leaf epidermis layers being translucent,the larva is visible as is the feces that accumulate in the mine.Photographed by J-L Boevé

J Chem Ecol (2009) 35:507–517 511

incubated at 35°C for 48 h and analyzed by spectropho-tometry at 405 nm. The minimum inhibitory concentration(MIC)100 corresponds to a total growth inhibition, andMIC50 to a 50% growth inhibition. The minimum fun-gicidal activity (MFC) is the lethal concentration deter-mined by inoculation on an agar plate of 20μl aliquots fromeach well and that showed 100% growth inhibitioncompared to the growth control. Thus, the MFC is thelowest concentration resulting in no growth.

Results

Phylogenetic Analyses Complete datasets of COI, cob, and28S sequences (total aligned length: 1866 bp) wereobtained for representatives of ten species. For Pseudodi-

neura parvula, only the amplification and sequencing ofCOI was successful; for Endophytus anemones andDineura pullior we obtained COI and 28S sequences.

All combined and single-gene analyses (except 28S)supported monophyly of the genus Pseudodineura withhighest bootstrap and posterior probability values (Fig. 2A).Endophytus anemones appeared to be the sister taxon ofPseudodineura. Phylogenetic relationships among Pseudodi-neura species could not be completely resolved despite thedifferent evolutionary characteristics and amount of phylo-genetic information of the three selected markers. Nonethe-less, the sister-species relationship of Pseudodineuraclematidis and Pseudodineura clematidisrectae consistentlyreceived highest support. Furthermore, COI analysesrevealed a probable sister-species relationship betweenPseudodineura parvula and Pseudodineura enslini (Fig. 2B).

S. compressicornisP1752.1

H. croceaP1775.1

D. pulliorP1809.1

N. fuscipennisP2078.1

C. pectinicornisP2072.1

E. anemonesP2098.1

P2169.1

P2169.3

P2169.2

P2901.2

P2252.1

P2252.2

P. mentiensP2172.1

P. parvaP2849.1

P2190.1

P2900.1

P2900.2

P2900.3

P2167.1

P2899.1

P2899.2

P2167.3

P2899.3

P2848.1

0.10

91/91/*

64/64/93

-/-/65

94/96/96

-/-/90

91/94/*

-/-/64

87/86/*

*/*/*

P. fuscula

P. enslini

P. clematidis

P. clematidisrectae

*/*/**/*/*

*/*/*

*/*/*

*/*/*

*/*/*

*/*/*

*/*/*

©M

rEnt

AFig. 2 Phylogenetic relation-ships of the Pseudodineurini(A) and Pseudodineura (B). InA, the tree was reconstructedusing Bayesian inference forconcatenated COI and 28Ssequences (1,453 bp). Branchesin bold show clades supportedby the combined analyses ofCOI, cob, and 28 S sequences(1,886 bp). Taxa marked withblack inverted triangles werenot included in the latter analy-sis. Clade support values are:parsimony bootstraps (2,000replicates)/maximum likelihoodbootstraps (100 replicates)/Bayesian posterior probabilities(2×106 generations). Values un-der 60% are not shown. Boot-strap values of 97–100% andposterior probabilities of 100%are replaced by asterisks. In B,the tree from COI sequenceswas reconstructed usingBEAST, and posterior probabil-ities were calculated by running107 generations. Asterisks re-place values between 99% and100%. Presence (plus sign) orabsence (minus sign) of 8-oxocitral (c3 and c4). Speciesdocumented as xerotherm (Yes)or not (No) in Altenhofer andPschorn-Walcher (2006).Unknown or unclear(question mark)

512 J Chem Ecol (2009) 35:507–517

Chemical Analyses The glandular secretion was composedmainly of monoterpenes (Fig. 3). Neral (c1) and geranial(c2) were the major compounds in all studied Pseudodi-neurini. In several Pseudodineura species, an uncommonmonoterpene occurred as two isomers in medium amounts.The mass spectra showed a molecular mass 14 amu higherthan that of citral, which is consistent with an additionalmethylene group or carbonyl function. The tabular data ofthe mass spectrum were comparable with those for (2E,6E)-2,6-dimethylocta-2,6-dienedial (c4) as described in litera-ture (Veith et al. 1996). The structure then was confirmedby synthesis, followed by comparison of mass spectra andretention indices of the natural and synthetic compounds.The analogous (2E,6Z)-2,6-dimethylocta-2,6-dienedial (c3)was identified in the same way.

For each species, the relative abundance (in percentage)of c1, c2, c3, and c4 was as follows (see also Fig. 2B): P.clematidisrectae (29, 53, 0, 0, respectively), P. clematidis(24, 67, 2, 3), P. parvula feeding on Pulsatilla vulgaris (34,57, 0, 0), P. enslini (15, 63, 3, 10), P. fuscula feeding onRanunculus bulbosus (35, 56, 1, 3), P. fuscula feeding onRanunculus platanifolium (10, 55, 2, 13), P. heringi (31, 46,0, 0), P. mentiens (25, 53, 1, 2 for a sample from June; 25,32, 0, 0 for a sample from August; 25, 52, 0, 0 foreonymphs from this sample), and E. anemones (35, 65, 0,0). Thus, the relative abundance varied between 10% and37% for c1, 46–72% for c2, 1–3% for c3, and 3–13% forc4, which means that c2 was always more abundant thanc1, and c4 more than c3.

Bioassays The six P. mentiens larvae left in their leaf minesurvived well their 24 h confrontation with ant workers ofM.rubra, since no mine was opened during this period. In thebioassay where larvae were experimentally taken from theirmine, the mean number (± SD) of ants surrounding a singlepenultimate instar larva was 1.7 (± 0.8), but none surroundeda full-grown larva (P<0.01, Mann–Whitney U test).

The two isomers of citral (c1 and c2) were significantlyrepellent for C. scutellaris ant workers, with compound c1

Fig. 2 (continued)

Fig. 3 Chromatograms of the volatiles emitted by two closely relatedPseudodineura species, P. parvula and P. enslini. The monoterpeneisomers correspond to citral (c1 and c2) and 8-oxocitral (c3 and c4).Other, smaller peaks correspond to alkanes: from dodecane with tR=10.13 to octadecane with tR=44.95

J Chem Ecol (2009) 35:507–517 513

being slightly more repellent than c2. In contrast, the mix-ture of the two isomers of 8-oxocitral (c3 and c4) showedonly a weak nonsignificant activity (Fig. 4).

The antifungal susceptibility tests revealed that all foursynthesizedmonoterpeneswere active against fungi (Table 2).The mixture c3 + c4 (1:1) showed the strongest fungicidalactivity (with MFC=0.781 mg/ml), for both C. albicans andA. fumigatus, compared to the other tested monoterpenes.We consider activity values as similar between the two fungispecies, taking into account that such values obtained fromterpenes can vary even between strains of the same species(Sabini et al. 2006).

GC–MS analyses, performed after the bioassays, re-vealed that only the mixture c3 + c4 (1:1) remained stableover time, whereas c3 + c4 (1:4) was degraded. This mayexplain why a relatively low antifungal activity wasobserved with c3 + c4 (1:4), which is, however, closer to

the natural mixture (see Fig. 3). It is possible that c3 is moreactive than c4, which would increase the potency of c3 + c4(1:1) mixture where c3 is proportionally more concentrated.

Discussion

There are both advantages and disadvantages for an insectto feed as a leaf miner, as compared to an external leaffeeder (Connor and Taverner 1997). From the point of viewof aggression from natural enemies, the risk of externaldisease infection is lower, whereas parasitoids inflict highermortality due to a lower mobility of the leaf miner. Gallingnematines may be less susceptible to this kind of parasitismthan free-living species, since the mean number ofparasitoid species is 4.0 for shoot-galling nematines, incontrast to almost 16 for free-living colonial nematines(Price and Pschorn-Walcher 1988; but see Nyman et al.2007). Among the Pseudodineurini, 0–3 parasitoid speciesper host species were recorded (Altenhofer and Pschorn-Walcher 2006). Thus, it is likely that the leaf-mining habitallows Pseudodineurini larvae to avoid at least someparasitoids, such as tachinid flies. Considering predation,mining vs. external feeding habits seem to be similar inrisks incurred (Connor and Taverner 1997). The mechanicalprotection offered by the mine was effective in our bioassaywhere larvae and mines remained intact for 24 h whilesurrounded by M. rubra workers. Mechanical protectionappears to be insufficient, however, in the birch leaf miningsawfly Fenusa pusilla. It feeds in newly unfolding thinleaves and can be preyed upon by ants (Pezzolesi andHager 1994). This sawfly species does not belong to thenematines and does not possess ventral glands. Thus,predation by ants may have a significant impact on sawflyleaf miners if the leaf epidermal layers are thin and if thelarvae are not defended chemically.

Citral was the major compound in all analyzed Pseudo-dineurini species (Fig. 3), and it was detected in full-grownlarvae still in their mine as well as in eonymphs in their

Fig. 4 Repellent activity of citral (c1 and c2) and related 8-oxocitral(c3 and c4) on the ant C. scutellaris. The monoterpenes were offeredat t=0 on a small podium that was surrounded at its base by feedingants. The number of feeding ants was counted every minute during10 min. Water was used as a control. Standard deviation values are notshown, for clarity. *P<0.05, **P<0.01, Kruskall–Wallis test withpost-test (corrected for ties), by comparing each volatile vs. water

Monoterpenes Fungus MIC50 (mg/ml) MIC100 and MFC (mg/ml)a

c1 C. albicans 1.172 2.344

A. fumigatus 2.343 4.687

c2 C. albicans 1.406 2.812

A. fumigatus 5.625 11.25

c3 + c4 (1:1) C. albicans 0.39 0.781

A. fumigatus 0.39 0.781

c3 + c4 (1:4) C. albicans no MIC50 3.75

A. fumigatus 0.937 1.875

Table 2 In vitro susceptibility oftwo fungal species Candida albi-cans and Aspergillus fumigatusto citral (c1 and c2) and mixturesof 8-oxocitrals (c3 and c4)

aMFC and MIC100 values wereidentical in all tests

MIC50 minimum inhibitory con-centration 50%, MIC100 mini-mum inhibitory concentration100%

514 J Chem Ecol (2009) 35:507–517

cocoons. It is an insect repellent (Boevé 1988; Vartak et al.1994), fungicide, and a fungus growth inhibitor (Kuwaharaet al. 1989; Lima et al. 2005; Sabini et al. 2006). Thesebioactivities were corroborated by our bioassays (Fig. 4,Table 2). We believe that citral will be used especially whenthe larva leaves a mine. The larva will then crawl into thesoil to spin a cocoon (Hering 1951; Zinovjev and Vikberg1998; Altenhofer and Pschorn-Walcher 2006) and may beeasily attacked by foraging ground-dwelling arthropodssuch as ants. A similar situation is found in anotherendophytic nematine, the apple sawfly Hoplocampa testu-dinea, which also pupates in the soil. The size of ventralglands increases in an allometric way once the larvabecomes full-grown, the glandular secretion then becomesobvious, and the defense is efficient against ants (Boevé etal. 1997). The ontogenic increase in P. mentiens chemicaldefense was reflected in our bioassays by the significantdifference between full-grown larvae, which were notapproached, and younger ones.

There seems to be no direct influence of the host-planton the biosynthesis of monoterpenes by Pseudodineurinilarvae. Citral is the major compound of the glandular secre-tion not only in the Pseudodineurini but also in the alder-feeding Nematinus species (Boevé et al. 1984), and it isdetected in two spruce-feeding Pristiphora species, Pristi-phora compressa and Pristiphora pallida (Boevé, personalobservation). The independence between host-plant (chem-istry) and chemical composition of the glandular secretionalso is observed in other nematine sawflies: for benzalde-hyde produced by Nematus and Pristiphora species and fordolichodial produced by Nematus and Craesus species(Boevé et al. 1992; Boevé and Heilporn 2009). Thus, thesegenera as well as Pseudodineurini probably produce themajor compounds of their glandular secretion de novo.

Not all phylogenetic relationships among the studiedPseudodineurini species could be resolved. However, it isobvious that P. clematidisrectae and P. clematidis are sisterspecies, as are probably P. parvula and P. enslini. Theinteresting point is that in both species pairs, one speciesproduces the 8-oxocitral isomers, the other does not. Thisimplies that the production of 8-oxocitrals is not asynapomorphic trait for a certain subgroup of Pseudodi-neurini. We hypothesize that one or more gene(s) have beenswitched on/off several times during evolution. Citral mightbe converted by enzymatic reaction into 8-oxocitral, but itis more likely that geraniol or nerol is first enzymaticallyoxidized at the terminal position to the respective 8-hydroxygeraniol or 8-hydroxynerol, which is subsequentlyenzymatically oxidized to the dialdehyde. Since citraloccurred in the secretion of all Pseudodineurini, obviouslyderived from geraniol by oxidation, the same precursor for8-oxocitral is available in each species. Thus, the on/offswitching that leads to the presence or absence of the 8-

oxocitrals would require an additional enzyme, namely anoxidase that acts on geraniol, while the final oxidationmight be performed by the same enzyme in the case ofcitral and 8-oxocitral. A potential subsequent step is thecyclization of 8-oxocitral into dolichodial. This is a majorcompound of the glandular secretion in several nematinesawflies (Boevé et al. 1984, 1992; Boevé and Heilporn2009). Traces of iridoids such as dolichodial were detectedin the glandular secretion of Pseudodineurini larvae, but itis not clear how they are formed. The pathway that involvestwo enzymatic steps from geraniol or nerol to the dialdehydeis described in leaf beetles, which use these dialdehydes asprecursors of defensive iridoids (Veith et al. 1996). 8-Oxocitral is also the precursor of iridoids in plants.

What is the adaptive value of 8-oxocitral, and why doesit occur in only some Pseudodineurini species? This mono-terpene is inefficient as a defense against attacking smallarthropods (Fig. 4), but it is an efficient fungus inhibitor aswell as a fungicide (Table 2). Such antifungal activitieswould be relevant when the larva lives in a mine since fecesaccumulate therein (Fig. 1), offering a substrate for theproliferation of fungi. Additionally, when a full-grown larvacrawls into the soil, it again may be subject to fungalinfestation. Living in leaf mines and pupating in the soil arelife history traits common to all Pseudodineurini larvae, andit does not explain why only some Pseudodineura speciesproduce 8-oxocitral. There is, however, a parallel betweenan abiotic factor in the biotope where species live, asrecorded in Altenhofer and Pschorn-Walcher (2006), andthe presence vs. absence of 8-oxocitral. The humid and coldzones species P. clematidis, P. enslini, and P. fuscula pro-duce 8-oxocitral, whereas the xerotherm species P. clem-atidisrectae, P. heringi, and P. parvula do not (P=0.05,Fisher exact probability test; N=6 species, not consideringP. mentiens that showed variable chemical data, see below).We, thus, think that larvae crawling into a relatively warmand dry soil are less susceptible to fungal infestation, and thatthe production of 8-oxocitral may be less necessary. Indeed,entomopathogenic fungi are rarer in dry zones than in humidones (Hall and Papierok 1982), and they require a highrelative humidity, of at least 92–93%, for sporulation andfor spore germination on insect adults and larvae (Deacon1997). In this context, it is noteworthy that only thegeographic distribution of P. enslini, P. fuscula, and P.mentiens extends into the humid cold zones of NorthEurope (Viramo 1969), and these species produce 8-oxocitral. Once the mining habit and the production ofcitral evolved in the ancestor of the Pseudodineurini, theproduction of 8-oxocitral may have been opted by thosespecies living in relatively humid and cold zones.

We have no clear answer whether the adaptation toproduce 8-oxocitral happened at a specific or subspecificlevel. Two samples of P. fuscula larvae were chemically

J Chem Ecol (2009) 35:507–517 515

analyzed. The larvae came from two different populations,one collected on R. bulbosus and the other on R. plata-nifolium. 8-Oxocitral was detected in both samples. Incontrast, larvae from one population of P. mentiens contained8-oxocitral, whereas another sample did not. Interestingly,these latter larvae were collected from the xerotherm biotope(location “Langenlois”; Table 1). This suggests that, at leastfor P. mentiens, the production of 8-oxocitral is an adaptationat the population level. The intraspecific variation that weobserved at the genetic level in several species should bestudied more carefully to see whether it is linked to the(sub)-specific production of 8-oxocitral.

Our study began by asking why a larva that lives con-cealed and protected in a mine also would produce adefensive volatile secretion. By investigating the occurrenceof citral in the larvae, the discrete occurrence of thebiosynthetically related 8-oxocitral was incidentallydetected. This stimulated us to reconstruct the phylogeny ofthe insect group and to compare the bioactivities of bothcompounds. It was surprising that 8-oxocitral was detected,twice, in only one of two closely related species. Further, thebioassays were at first restricted to ant repellence tests. Weexpected a higher activity with 8-oxocitral than citral sincethe former compound possesses two (reactive) aldehydegroups. When the bioassay results showed the oppositeantifungal tests were performed due to the observation thatfeces accumulate in a leaf mine. These results led to theconclusion that abiotic factors play an important role inselecting for 8-oxocitral, since it is present in species that livein cold and humid areas. The efforts of a multidisciplinaryteam were necessary in order to integrate this accumulationof unpredictable results and to construct a more comprehen-sive picture of the evolution of a particular defense strategyin insects. Finally, M. Hering who devoted his life to leaf-mining insects mentions as an epigraph a sentence by J.Swammerdam that is strangely connected to our work: Ichhabe mir sagen lassen, in heißen Ländern fände manzwischen den Blättern daumenlange Würmer (Hering 1935-1937), i.e., “I was informed that in warm countries one canfind inch-long worms between the leaves.”

Acknowledgements JLB performed the GC–MS analyses in the lab ofMonika Hilker (Berlin, Germany), with the technical assistance of FrankMüller. Filip De Block provided JLB with ants. Many thanks, also, to thecollectors of sawfly specimens used in the genetic analyses and to TommiNyman, Herbert R. Jacobson, and two anonymous reviewers for criticaland helpful reading of the manuscript. For phylogenetic analyses,funding was provided by the Joint Experimental Molecular Unit (projectECES_PSEU) that is supported by the Belgian Science Policy Office.

References

ALTENHOFER, E. 2003. Minierende Blattwespen (Hym.: Symphyta):ihre Minenformen, Wirtspflanzen, Ökologie und Biologie.Gredleriana 3:5–24.

ALTENHOFER, E. and PSCHORN-WALCHER, H. 2006. Zur Faunistik,Biologie und Parasitierung der minierenden Blattwespen derTribus Pseudodineurini (Hymenoptera: Tenthredinidae). pp. 73–82, in S. M. Blank, S. Schmidt, A. Taeger (eds). Recent SawflyResearch: Synthesis and Prospects. Goecke & Evers, Keltern(Germany).

BLUM, M. S. 1981. Chemical Defenses of Arthropods. AcademicPress, New York.

BOEVÉ, J.-L. 1988. Stratégies défensives des larves de Nématines(Hymenoptera, Tenthredinidae) vis-à-vis de leurs prédateurs. PhDdissertation. Université Libre de Bruxelles.

BOEVÉ, J.-L. 2008. Sawflies (Hymenoptera: Tenthredinidae), pp.3252-3257, in J. L. Capinera (ed.). Encyclopedia of Entomology,Second Edition. Springer.

BOEVÉ, J.-L. and HEILPORN, S. 2009. Secretion of the ventral glandsin Craesus sawfly larvae. Biochem. Syst. Ecol. 36:836–841.

BOEVÉ, J.-L. and PASTEELS, J. M. 1985. Modes of defense innematine sawfly larvae. J. Chem. Ecol. 11:1019–1036.

BOEVÉ, J.-L., BRAEKMAN, J.-C., DALOZE, D., HOUART, M., andPASTEELS, J. M. 1984. Defensive secretions of Nematinae larvae(Symphyta - Tenthredinidae). Experientia. 40:546–547.

BOEVÉ, J.-L., DETTNER, K., FRANKE, W., MEYER, H., and PASTEELS.J. M. 1992. The secretion of the ventral glands in Nematussawfly larvae. Biochem. Syst. Ecol. 20:107–111.

BOEVÉ, J.-L., GFELLER, H., SCHLUNEGGER, U. P., and FRANCKE, W.1997. The secretion of the ventral glands in Hoplocampa sawflylarvae. Biochem. Syst. Ecol. 25:195–201.

CHAN, K. C., JEWELL, R.A., NUTTING, W. H., and RAPOPORT, H.1968. The synthesis and stereochemical assignment of cis- andtrans-2-methyl-2-pentenoic acid and the corresponding esters,aldehydes, and alcohols. J. Org. Chem. 33:3382–3385.

CLINICAL AND LABORATORY STANDARDS INSTITUTE/NATIONAL COM-

MITTEE FOR CLINICAL LABORATORY STANDARDS. 2002a. Refer-ence method for broth dilution antifungal susceptibility testing ofyeasts. Approved Standard document M27-A2. Clinical andLaboratory Standards Institute/National Committee for ClinicalLaboratory Standards Institute, Wayne P.A.

CLINICAL AND LABORATORY STANDARDS INSTITUTE/NATIONAL COM-

MITTEE FOR CLINICAL LABORATORY STANDARDS. 2002b. Refer-ence method for broth dilution antifungal susceptibility testing offilamentous fungi. Approved Standard document M38-A. Clin-ical and Laboratory Standards Institute/National Committee forClinical Laboratory Standards Institute, Wayne P.A.

CONNOR, E. F. and TAVERNER, M. P. 1997. The evolution andadaptive significance of the leaf-mining habit. Oikos 79:6–25.

DEACON, J. W. 1997. Fungal parasites of humans, insects and nematodes.pp. 254–272, in Modern Mycology. Blackwell Science Ed.

DRUMMOND, A. J. and RAMBAUT, A. 2007. BEAST: Bayesianevolutionary analysis by sampling trees. BMC Evol. Biol. 7:214.

EISNER, T. 1970. Chemical defense against predation in arthropods,pp. 157–217 in E. Sondheimer, J. B. Simeone (eds). ChemicalEcology. Academic Press, New York.

EVANS, D. L. and SCHMIDT, J. O. (eds.). 1990. Insect Defenses:Adaptive Mechanisms and Strategies of Prey and Predators. StateUniversity of New York Press, Albany, New York.

GUILFORD, T. 1990. The evolution of aposematism, pp. 23–61, in D. L.Evans and J. O. Schmidt (eds.). Insect Defenses: AdaptiveMechanisms and Strategies of Prey and Predators. State Universityof New York Press, Albany, New York.

HALL, R. A. and PAPIEROK, B. 1982. Fungi as biological controlagents of arthropods of agricultural and medical importance.Parasitology 84:205–240.

HERING, E. M. 1935-1937. Die Blattminen Mittel- und Nord-Europaseinschließlich Englands. Bestimmungstabellen aller von Insek-tenlarven der verschiedenen Ordnungen erzeugten Minen. Neu-brandenburg 1–6:1-631.

516 J Chem Ecol (2009) 35:507–517

HERING, E. M. 1951. Biology of the Leaf Miners. Junk, Gravenhage.KATOH, K. and TOH, H. 2008. Improved accuracy of multiple ncRNA

alignment by incorporating structural information into a MAFFT-based framework. BMC Bioinformatics 9:212.

KATOH, K., MISAWA, K., KUMA, K., and MIYATA, T. 2002. MAFFT: anovel method for rapid multiple sequence alignment based onfast Fourier transform. Nucleic Acids Res. 30:3059-3066.

KUWAHARA, Y., LEAL, W. S., SUZUKI, T., MAEDA, M., andMASUTAMI, T. 1989. Antifungal activity of Caloglyphus poly-phyllae sex pheromone and other mite exudates. Naturwissen-schaften 76:578–579.

LI, Y., HUANG, C., Li, W., and LI, Y. (1997) A facile total synthesis of(3R,7RS)-3,7,11-trimethyl-10-oxododecanoic acid. Synth. Comm.27:4341–4348.

LIMA, I. O., DE ARAÚJO, G., OLIVEIRA, R., DE O. LIMA, E., DE

SOUZA, E. L., PORTO FARIAS, N., and DE FÁTIMA NAVAR, D.2005. Inhibitory effect of some phytochemicals in the growth ofyeasts potentially causing opportunistic infections. Braz. J.Pharmac. Sc. 41:199–203.

MOORE, S. J., LENGLET, A., and HILL, N. 2007. Plant-based insectrepellents, in M. Debboun, S. P. Frances, and D. Strickman(eds.). Insect Repellents: Principles, Methods, and Use. CRCPress, Taylor & Francis Group, Boca Raton.

NORMARK, B. B., JORDAL, B. H., and FARRELL, B. D. 1999. Origin ofa haplodiploid beetle lineage. Proc. R. Soc. Lond. B 266:2253–2259.

NYMAN, T., ZINOVJEV, A. G., VIKBERG, V., and FARRELL, B. D. 2006.Molecular phylogeny of the sawfly subfamily Nematinae(Hymenoptera: Tenthredinidae). Syst. Entomol. 31:569–583.

NYMAN, T., BOKMA, F., and KOPELKE, J.-P. 2007. Reciprocaldiversification in a complex plant-herbivore-parasitoid foodweb. BMC Biology 5:49

OMURA, K. and SWERN, D. 1978. Oxidation of alcohols by“activated” dimethyl sulfoxide. A preparative, steric and mech-anistic study. Tetrahedron 34:1651–1660.

PASTEELS, J. M., GRÉGOIRE, J.-C., and ROWELL-RAHIER, M. 1983.The chemical ecology of defense in arthropods. Annu. Rev.Entomol. 28:263–289.

PEZZOLESI, L. S. W. and HAGER, B. J. 1994. Ant predation on twospecies of birch leaf-mining sawflies. Am. Midl. Nat. 131:156–168.

POSADA, D. 2008. jModelTest: Phylogenetic Model Averaging. Mol.Biol. Evol. 25:1253–1256.

PRICE, P. W. and PSCHORN-WALCHER, H. 1988. Are galling insectsbetter protected against parasitoids than exposed feeders? A testusing tenthredinid sawflies. Ecol. Entomol. 13:195–205.

RONQUIST, F. and HUELSENBECK, J. P. 2003. MrBayes 3: Bayesianphylogenetic inference under mixed models. Bioinformatics19:1572–1574.

RUXTON, G. D. and SHERRATT, T. N. 2006. Aggregation, defence andwarning signals: the evolutionary relationship. Proc. R. Soc. B273:2417–2424.

SABINI, L. I., GABRIELLI, P. S., TORRES, C. V., ESCOBAR, F. M.,CACCIABUE, M., ROVERA, M., and KOLB, N. 2006. Study of thecytotoxic and antifungal activity of the essential oil of Elyonurusmuticus against Candida spp. Mol. Medic. Chem. 11:31–33.

SCHMIDT, S., DRIVERB, F., and DE BARROC, P. 2006. The phyloge-netic characteristics of three different 28 S rRNA gene regions inEncarsia (Insecta, Hymenoptera, Aphelinidae). Organ. Divers.Evol. 6:127–139.

SEIFERT, B. 1988. A taxonomic revision of the Myrmica species ofEurope, Asia Minor, and Caucasia (Hymenoptera, Formicidae).Abh. Naturkundemus. Görlitz 63:1–75.

SIMON, C., FRATI, F., BECKENBACH, A., CRESPI, B., LIU, H., and FLOOK,P. 1994. Evolution, weighting, and phylogenetic utility of mito-chondrial gene sequences and a compilation of conserved poly-merase chain reaction primers. Ann. Entomol. Soc. Am. 87:651–701

SWOFFORD, D. L. 2002. PAUP*. Phylogenetic Analysis UsingParsimony (*and other Methods), Version 4. Sunderland,Massachusetts, Sinauer Associates.

TAEGER, A. and BLANK, S. M. 2008. ECatSym - Electronic WorldCatalog of Symphyta (Insecta, Hymenoptera). Program version3.9, data version 34 (05.09.2008). Digital Entomological Infor-mation, Müncheberg.

VARTAK, P. H., TUNGIKAR, V. B., and SHARMA, R. N. 1994.Comparative repellent properties of certain chemicals againstmosquitoes, house flies and cockroaches using modified tech-niques. J. Comm. Dis. 26:156–160.

VEITH, M., DETTNER, K., and BOLAND, W. 1996. Stereochemistry ofan alcohol oxidase from the defensive secretion of larvae of theleaf beetle Phaedon armoraciae. Tetrahedron 52:6601–6612.

VIRAMO, J. 1969. Zur Kenntnis der Miniererfauna Finnlands. Über dieWirtspflanzen und die Verbreitung der minierenden Blattwespen(Hym., Tenthredinoidea). Ann. Entomol. Fenn. 35:3–44.

WHITMAN, D. W., BLUM, M. S., and ALSOP, D. W. 1990. Allomones:chemicals for defense, pp. 289–351, in D.L. Evans and J.O.Schmidt (eds.). Insect Defenses: Adaptive Mechanisms andStrategies of Prey and Predators. State University of New YorkPress, Albany, New York.

ZINOVJEV, A. G. and VIKBERG, V. 1998. On the biology of theNematinae with hiding larvae (Hymenoptera, Symphyta, Ten-thredinidae). Beitr. Entomol. 48:145–155.

J Chem Ecol (2009) 35:507–517 517