geographical variation in cone volatile composition among...

Geographical variation in cone volatile compositionamong populations of the African cycadEncephalartos villosus

TERENCE N. SUINYUY1,3,4*, JOHN S. DONALDSON1,2,4 and STEVEN D. JOHNSON3

1Kirstenbosch Research Centre, South African National Biodiversity Institute, P/Bag X7, Claremont7735, Cape Town, South Africa2Research Associate, Fairchild Tropical Botanic Garden, 10901 Old Cutler Road, Miami, FL 33156,USA3School of Life Sciences, University of KwaZulu Natal, P/Bag X01, Scottsville 3201,Pietermaritzburg, South Africa4Department of Botany, University of Cape Town, P/Bag Rondebosch 7701, Cape Town, South Africa

Received 20 December 2011; revised 6 February 2012; accepted for publication 6 February 2012bij_1905 514..527

Variation in traits across species distribution ranges is often indicative of diversifying evolution that can lead tospeciation. Of particular interest is whether traits vary clinally or abruptly because the latter pattern can beindicative of incipient speciation. Understanding of intraspecific variation in chemical traits is still in its infancybecause studies of population variation have tended to focus on morphology or neutral genetic markers. To addressthese issues, the composition of cone volatile odours was examined in ten populations of the South African cycadEncephalartos villosus across its range in the Eastern Cape and KwaZulu Natal using headspace sampling andanalysis by gas chromatography-mass spectrometry. Because volatiles play a key role in attracting pollinators tocones of Encephalartos cycads and may thus reflect local adaptation to pollinators, pollinator assemblages were alsoinvestigated in the ten populations of E. villosus. Volatile compounds from populations in the north of thedistribution range were dominated by unsaturated hydrocarbons, whereas, in the southern populations, nitrogen-containing compound and terpenoids were the major compounds. A shift between southern and northern popula-tions appeared to occur at the Umtamvuna River, where populations had odour profiles with components of boththe northern and southern populations. However, one population in the north (Vernon Crookes Nature Reserve)had a quantitatively similar odour profile to the populations in the extreme south of the range. These results revealstrong interpopulation variation in the cone scent of E. villosus, including variation in the relative emission ofdominant compounds that may play key functional role in this pollination system. However, pollinator assemblagesdid not differ across the different populations, which suggest that these patterns were produced by co-evolution ordrift, rather than by pollinator shifts. © 2012 The Linnean Society of London, Biological Journal of the LinneanSociety, 2012, 106, 514–527.

ADDITIONAL KEYWORDS: Eastern Cape – gas chromatography-mass spectrometry – insect pollinators –KwaZulu-Natal – nitrogen-containing compounds – odour profile – unsaturated hydrocarbons.

INTRODUCTION

Floral odour can play a key role in plant–pollinatorinteractions through its influence on the compositionand behaviour of pollinators that use this trait as acue (Dobson, 2006; Raguso, 2008). Odours are usuallyblends of compounds belonging to several chemical

classes, typically fatty acid derivatives, benzenoids,terpenoids, and sometimes nitrogen-containing com-pounds, and may vary in the number, composition,and relative amounts of the different constituents,and in their temporal and spatial emission patterns(Raguso, 2004; Knudsen et al., 2006). The particularconstituents and pattern of odour emission comprisethe signals that influence the composition and behav-iour of pollinators (Pellmyr, 1986a; Raguso, 2008). In*Corresponding author. E-mail: [email protected]

Biological Journal of the Linnean Society, 2012, 106, 514–527. With 3 figures

bs_bs_banner

© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2012, 106, 514–527514

some cases, highly specialized plant–pollinator inter-actions are mediated by floral odours (Pellmyr, 1992;Jürgens, 2009) and these can provide a species spe-cific signal (‘private channel’) that elicits the requiredbehavioural response from the pollinating organism(Schiestl & Peakall, 2005; Raguso, 2008; Chen et al.,2009). The implication is that floral odour should be aspecies-wide attribute with the expectation thatchanges in floral odours may be associated with shiftsin pollinators that may lead to speciation in plants(Johnson, 1996; Johnson & Steiner, 1997). It has beenargued that such evolutionary changes in floralodours played a role in the diversification of bothangiosperms and pollinating insects (Pellmyr &Thien, 1986).

Although the majority of studies of floral odourhave focused on flowering plants, work on cycadsshows that cone odour is also an important factorinfluencing behaviour of pollinators in this ancientplant lineage (Terry et al., 2007a, b). Limited avail-able information shows that differences in odour pro-files between cycad species can be associated withdifferences in pollinators, as illustrated by Mac-rozamia species being pollinated by either thrips(Cycadothrips) or weevils (Tranes) (Terry et al., 2004a,b). The expectation may therefore be that the vola-tiles which influence pollinator behaviour will also beinvariant species-wide attributes in cycads. To date,there have been no studies of volatiles across the fulldistribution range of any cycad species to test thishypothesis.

Preliminary investigations of the African cycadEncephalartos villosus showed that cone odoursappeared to be inconsistent with the model of a stableodour profile throughout the species distribution.Plants originating from populations in the EasternCape (EC) had cone odour profiles characterizedby eucalyptol and 2-isopropyl-3-methoxypyrazine,whereas those from KwaZulu Natal (KZN) werecharacterized by (3E)-1,3-octadiene and (3E,5Z)-1,3,5-octatriene (T. N. Suinyuy, unpubl. data). Thesedifferences raise questions about possible shifts inpollination systems across the range of E. villosus andthe relative influence of different compounds on thebehaviour of pollinating insects. The present studytherefore aimed to determine the extent and patternof variation in cone odour chemistry across the fulldistribution range of E. villosus and to analyzewhether cone volatile variation between populationsis associated with different insect visitors. Twoalternative hypotheses were tested to explain theobserved variation in cone volatile composition amongpopulations of E. villosus.

The hypotheses were: (1) that changes in volatilecomposition reflect a change in pollinators and mayindicate that E. villosus, as currently circumscribed,

comprises at least two cryptic species and (2) thatvariation is related to geographical separationbetween populations, such that plants in populationsthat are furthest apart would differ the most inemitted volatile compounds.

MATERIAL AND METHODSPLANT MATERIAL AND LOCALITY

Encephalartos villosus is distributed in a relativelynarrow band along the east coast of South Africa(Fig. 1) with a linear distance of approximately900 km between the southern-most and northern-most populations. There is some morphological varia-tion in E. villosus: plants from the EC tend to haveshorter heavily spined leaflets and cones with toothededges, whereas those from KZN tend to have longeralmost entire leaflets and cones with lightly toothededges (Goode, 1989). Across its distribution, E. villo-sus occurs in patches of Scarp Forest includingEastern Scarp Forest, Pondoland Scarp Forest, andTranskei Coastal Scarp Forest. The forest patches areembedded within three different vegetation types: theKZN Coastal Belt, Pondoland-Ugu Sandstone CoastalSourveld, and Transkei Coastal Belt (Mucina et al.,2006). Volatile odour samples were collected from tenlocalities spread across the range of the species,including populations near to the southern and north-ern limits of its distribution (Fig. 1). In total, sampleswere obtained from 59 male plants and 14 femaleplants. The sampling intensity for each localitydepended on the availability of cones because E. vil-losus does not cone regularly and cones may be scarceor absent in particular populations (Donaldson, 1997).Cones that were sampled were either elongated andshedding pollen in the case of males or receptive topollen with clearly open sporophylls in the case offemales.

SAMPLING OF VOLATILE COMPOUNDS

Headspace sampling was used to collect volatiles frommale and female cones during pollen release andreceptivity respectively. Polyacetate bags (Nalo Brat-folie Kalle GmbH) were placed over the entire cone justprior to sampling to concentrate the volatile com-pounds. Air from inside the bags was suctioned for30 min into an adsorbent trap using a portable battery-operated pump (Spectrex Personal Air Sampler PAS500) calibrated at 200 mL min-1. Air samples weresimultaneously collected from empty polyacetate bagsplaced way from the plant as controls to identifybackground contamination. The trap samples werestored at -20 °C in a sealed vial until analysis. Thetraps contained 2 mg of a 50 : 50 mixture of Tenax TA(Alltech Associates) and activated charcoal (Carbotrap,

VARIATION IN CONE VOLATILE COMPOSITION 515


Supelco) in a glass tube closed on both ends with glasswool. The activated charcoal is highly retentive andsmall quantities can be used as a result of its highadsorbing capacity (Millar & Sims, 1998; Tholl & Röse,2006). The adsorbent Tenax TA used in the presentstudy is commonly used to trap volatile compounds andhas a high thermal stability up to 350 °C, which allowsfor thermal desorption in the gas chomatographyanalysis.

CHEMICAL ANALYSIS AND COMPOUND

IDENTIFICATION

Volatile samples were analyzed by gaschomatography-mass spectrometry using a coupledVarian 3800 gas chomatograph (Varian Palo Alto,

California, USA) and a Varian 1200 mass spectrom-eter. The gas chomatograph was equipped with aCarbowax column (DB-wax) of 30 ¥ 0.32 mm internaldiameter ¥ 0.25 mm film thickness (Alltech). Heliumwas used as the carrier gas at a flow rate of1 mL min-1. After sampling, traps were placed in aVarian 1079 injector by means of a ‘Chromatoprobe’fitting and thermally desorbed. After a 3 min hold at40 °C, the gas chomatograph oven was ramped upto 240 °C at 10 °C min-1 and held there for 12 min.Compound identification was carried out using theNIST05 mass spectral library and comparisons withretention times of chemical standards, where avail-able, as well as comparisons between calculatedKovats retention indices and those published in theliterature. An homologous series of alkanes (C8-C20)

Figure 1. Geographical variation in cone odour composition in 10 populations (green patches with labels) across therange of Encephalartos villosus (dotted area) in South Africa. Pie charts depict the percentage of total emission for eachpopulation. A, Umtiza Nature Reserve (UNR). B, Ocean View Guest Farm (OVGF). C, Dwesa Nature Reserve (DNR).D, Mpande area, Port St Johns (MPD). E, Mount Sullivan area, Port St Johns (MTS). F, Umtamvuna Nature Reserve(UMNR). G, Oribi Gorge Reserve (OGNR). H, Vernon Crookes Nature Reserve (VCNR). I, Kranzkloof Nature Reserve(KKNR). J, Nkandla Forest Reserve (NFR).

516 T. N. SUINYUY ET AL.


was used to determine Kovats retention indices.All reference compounds used for retention time com-parisons were obtained from Sigma Aldrich Inc.GmbH, except (3E)-1,3-octadiene, which was obtainedfrom ChemSampco. Compounds present at higher orsimilar percentages in controls were considered ascontaminants and excluded from the analysis.

INSECT VISITORS TO MALE AND FEMALE CONES

To determine insect pollinator assemblages on differ-ent populations of E. villosus, male and female coneswere sampled during pollen shed and pollen receptiv-ity. Cones were surveyed at the same time that thevolatiles were sampled. Insects from male cones weresurveyed by placing a beating sheet under the cone,which was then tapped to dislodge all the insects ontothe sheet. Only insects crawling on the surface offemale cones were sampled because these could becollected and stored in alcohol without damaging thefemale cone. The insects were counted and recordedbefore storing in alcohol and later identified based onkeys provided in Endrödy-Younga (1991), Donaldson(1991), and Oberprieler (1996). Voucher specimens ofinsects have been deposited in the entomological col-lection of the school of Biological and ConservationSciences of the University of KwaZulu Natal, Pieter-maritzburg campus, South Africa.

STATISTICAL ANALYSIS

PRIMER 6 (Clarke & Gorley, 2006) was used to assessthe variation in odour between different populations.Relative proportions of different compounds in eachsample were used for analyses. Non-metric multidi-mensional scaling (NMDS), based on Bray–Curtissimilarities of square root transformed data, was usedto detect similarities among samples. The stress valueis given to indicate how well the distance matrix isreproduced. The significance of differences in chemi-cal composition in samples from plants in differentpopulations was assessed using one-way analysis ofsimilarities (ANOSIM) with 10 000 permutations(factor: population) and the resulting test statisticR was taken as a relative measure of separationbetween defined groups, based on mean ranksbetween and within groups (0 means no separation,whereas 1 indicates complete separation) (Clarke &Gorley, 2006).

Simple Mantel tests were performed using ZT soft-ware to determine whether cone odour composition iscorrelated with the actual geographical distancesbetween populations (Bonnet & van der Peer, 2002).The odour similarity matrices were calculated usingthe Bray–Curtis similarity coefficient (Clarke &Warwick, 2001). The geographical distance matrix

was calculated from actual geographical distancesbetween the different populations sensu Hughes et al.(2006). Mantel tests with 10 000 permutations wereperformed for the complete data set.

RESULTSCOMPOUND CLASS PATTERNS IN CONE ODOURS

The chemical composition of cone odours emitted bymale and female E. villosus plants is given in Table 1.In total, 88 compounds were detected in all the odoursamples and 87 were identified (Table 1). The identi-fied compounds included 19 fatty acid derivatives (fourunsaturated hydrocarbons, six aldehydes, two ketones,five alcohols, and two esters), nine benzenoids, 54terpenoids (48 monoterpenes and six sesquiterpenes),and four nitrogen-containing compounds.

The populations of E. villosus can be distinguishedclearly from the composition of volatile emissions atthe level of compound class. The most dominant com-pound classes were monoterpenes and unsaturatedhydrocarbons which contributed 43.7% and 31% of thetotal emissions, respectively (Fig. 1, Table 1). Monot-erpenes were present in almost all E. villosus popula-tions but dominated the volatile profiles of populationsin the southern part of the distribution range in theEC, including Umtiza Nature Reserve (UNR), OceanView Guest Farm (OVGF), Dwesa Nature Reserve(DNR), Mpande area (MPD), and Mount Sullivanarea (MTS). By contrast, unsaturated hydrocarbonsoccurred exclusively in populations in the northernpart of the distribution range in KZN, notably atUmtamvuna Nature Reserve (UMNR), Oribi GorgeNature Reserve (OGNR), Vernon Crookes NatureReserve (VCNR), Kranzkloof Nature Reserve (KKNR),and Nkandla Forest Reserve (NFR) and were domi-nant components of volatiles in four populations.

Other compound classes comprised < 10% of totalemissions, although there were notable differences inthese compound classes between populations. Twopopulations in the north (NNR, KNR) had almostno minor compounds (G and I in Fig. 1), whereasnitrogen-containing compounds made up between 5%and 18% of emissions in three populations in the south(C, D, E in Fig. 1) that were otherwise dominatedby monoterpenes. The remaining five populations(A, B, F, H, I in Fig. 1) all had a greater diversity ofcompound classes typically with higher proportions ofaldehydes, benzenoids, and alcohols (Fig. 1), with ben-zenoids comprising almost 50% of emissions from UNR(A in Fig. 1) and aldehydes making up approximnately35% of emissions in the VCNR population (H in Fig. 1).A greater diversity of compound classes was observedin populations dominated by monoterpenes than thosedominated by unsaturated hydrocarbons.



Tab

le1.

Ave

rage

rela

tive

amou

nts

(%)

ofco

mpo

un

dsin

the

con

eod

ours

ofE

nce

phal

arto

svi

llos

us

from

ten

popu

lati

ons

acro

ssth

esp

ecie

sra

nge

inth

eE

aste

rnC

ape

and

Kw

aZu

luN

atal

Com

pou

nd

CA

SK

RI

UN

RM

OV

GF

MO

VG

FF

DN

RM

MP

DM

MT

SM

MT

SF

UM

NR

MO

GN

RM

VC

NR

MK

KN

RM

KK

NR

FN

FR

M

(N=

5)(N

=6)

(N=

4)(N

=8)

(N=

5)(N

=8)

(N=

5)(N

=4)

(N=

6)(N

=5)

(N=

8)(N

=5)

(N=

4)

AL

IPH

AT

ICS

Alk

enes

(3E

)-1,

3-O

ctad

ien

ea10

02-3

3-1

1062

––

––

––

–7.

04(4

)9.

76(6

)0.

02(5

)64

.89

(8)

48.0

7(5

)47

.46

(4)

(3E

,5Z

)-1,

3,5-

Oct

atri

eneb

4008

7-61

-411

48–

––

––

––

30.4

6(4

)89

.53

(6)

0.02

(5)

33.2

0(8

)12

.60

(5)

51.3

9(4

)(E

,E,E

)-2,

4,6-

Oct

atri

eneb

1519

2-80

-011

98–

––

––

––

–0.

36(6

)16

.95

(5)

tr(4

)tr

(4)

0.22

(4)

1,2-

Dim

eth

yl-1

,4-c

yclo

hex

adie

neb

1735

1-28

-912

15–

––

––

––

–0.

08(4

)5.

70(5

)tr

(3)

tr(4

)0.

41(4

)A

ldeh

ydes

Hex

anal

a66

-25-

111

25–

––

––

–tr

(2)

––

––

3.34

(5)

–H

epta

nal

b11

1-71

-712

0910

.12

(4)

30.5

0(5

)14

.58

(2)

––

––

––

34.5

8(5

)–

7.50

(4)

–(Z

)-2-

Hep

ten

ala

5726

6-86

-113

40–

––

––

––

––

–tr

(1)

––

(E)-

2-O

cten

alb

2548

-87-

014

46–

––

––

––

–0.

99(5

)–

––

(2E

,4E

)-H

epta

-2,4

-die

nal

b43

13-5

-315

13–

––

––

––

tr(4

)–

0.01

(2)

0.27

(1)

–2,

4-O

ctad

ien

alb

3036

1-28

-516

10–

––

––

––

tr(5

)–

tr(2

)0.

04(1

)–

Ket

ones

3-O

ctan

oneb

106-

68-3

1274

–1.

97(2

)–

––

––

––

3.13

(5)

0.02

(6)

–2,

2,6-

Trim

eth

yl-6

-vin

yldi

hyd

ro-

2H-p

yran

-3(4

H)-

onec

3393

3-72

-114

89–

––

––

––

–0.

05(3

)–

––

Alc

ohol

(Z)-

3-H

exen

-1-o

lb92

8-96

-113

90–

––

––

––

0.41

(4)

––

––

–3-

Oct

anol

a58

9-98

-013

86–

3.67

(3)

––

––

––

0.02

(5)

2.66

(5)

0.35

(6)

––

1-O

cten

-3-o

la33

91-8

6-4

1456

––

7.09

(4)

––

––

0.65

(3)

tr(5

)2.

70(5

)0.

35(6

)16

.04

(3)

0.02

(4)

Lav

andu

lolb

498-

16-8

1690

––

–tr

(1)

tr(4

)–

––

––

––

–1,

7-O

ctad

ien

-3-o

lc30

385-

19-4

2086

––

––

––

––

–0.

59(5

)–

––

Est

er2-

Ph

enet

hyl

hex

anoa

tec

6290

-37-

512

70–

––

––

––

––

––

1.88

(4)

–M

eth

yl2,

4-h

exad

ien

oate

c15

15-8

0-6

1358

––

––

––

––

–1.

08(5

)–

––

BE

NZ

EN

OID

An

isol

ea10

0-66

-313

57–

––

––

––

––

–0.

02(2

)1.

40(1

)–

Ben

zald

ehyd

ea10

0-52

-715

5315

.78

(5)

14.6

2(6

)16

.09

(4)

0.02

(8)

0.54

(5)

––

1.03

(4)

0.01

(6)

2.62

(5)

0.01

(8)

2.89

(5)

0.03

(4)

1-Is

opro

pyl-

2-m

eth

oxy-

4-m

eth

ylbe

nze

neb

1076

-56-

816

16–

––

0.01

(4)

––

––

––

––

–

Met

hyl

ben

zoat

ea93

-58-

316

461.

02(5

)0.

65(3

)0.

70(4

)–

tr(2

)–

––

––

––

–M

eth

ylsa

licy

late

a11

9-36

-818

08–

––

––

––

–tr

(1)

––

a-M

eth

yl-b

enzy

lal

coh

olb

98-8

5-1

1832

––

––

––

––

tr(5

)0.

30(5

)–

––

Ben

zyl

alco

hol

a10

0-51

-618

961.

12(4

)4.

51(6

)8.

40(4

)–

0.01

(1)

––

–tr

(2)

–tr

(4)

––

Ph

enol

a10

8-95

-220

322.

04(5

)1.

54(5

)1.

46(4

)–

0.11

(3)

0.03

(8)

0.43

(3)

–tr

(3)

–tr

(3)

0.16

(2)

0.01

(4)

p-A

nis

alde

hyd

ea12

3-11

-520

6129

.27

(5)

15.1

7(6

)2.

27(4

)0.

23(7

)0.

79(5

)0.

62(8

)tr

(2)

5.58

(4)

0.03

(6)

7.20

(5)

0.02

(8)

tr(2

)0.

09(4

)

TE

RP

EN

OID

SM

onot

erp

enes

a-P

inen

ea77

85-7

0-8

1095

tr(4

)–

–7.

35(8

)23

.48

(4)

14.0

7(8

)21

.11

(5)

––

0.01

(2)

0.03

(1)

tr(1

)–

b-T

hu

jen

ea28

634-

89-1

1102

––

––

–tr

(2)

––

––

–0.

22(1

)–

Cam

phen

ea79

–92-

511

12–

––

7.92

(8)

10.3

9(4

)8.

99(8

)6.

43(4

)–

––

––

–U

nkn

own

1154

––

–0.

29(7

)0.

09(3

)–

–0.

09(3

)–

––

––

b-P

inen

ea12

7-91

-311

9434

.05

(4)

0.01

(4)

10.0

6(2

)16

.55

(8)

2.09

(3)

5.60

(5)

7.01

(4)

2.98

(4)

0.15

(5)

–tr

(1)

tr(1

)–

b-M

yrce

nea

123-

35-3

1199

tr(4

)–

––

–0.

22(4

)tr

(4)

tr(3

)–

–0.

14(2

)1.

23(4

)–

Un

know

n12

13–

––

10.3

3(4

)2.

43(5

)–

–6.

50(4

)–

–0.

04(3

)–

–a-

Terp

inen

ea99

-86-

512

20–

––

24.9

6(8

)11

.16

(5)

12.2

5(7

)–

21.4

7(4

)–

–0.

09(1

)–

–L

imon

enea

138-

86-3

1224

––

7.25

(2)

–0.

51(5

)tr

(3)

tr(2

)–

––

–0.

59(1

)–

Eu

caly

ptol

a47

0-82

-612

312.

55(5

)11

.87

(5)

11.5

0(3

)13

.88

(8)

17.9

4(5

)31

.63

(8)

30.9

4(5

)1.

80(4

)–

1.27

(5)

0.76

(2)

0.02

(1)

0.04

(2)

tran

s-b-

Oci

men

ea50

2-99

-812

670.

83(2

)–

––

–tr

(1)

–1.

06(4

)–

–0.

02(1

)–

0.08

(2)

g-Te

rpin

enea

99-8

5-4

1269

––

–2.

66(8

)3.

17(5

)2.

37(6

)–

1.50

(4)

––

––

–ci

s-b-

Oci

men

ea33

38-5

5-4

1275

––

––

tr(1

)–

–0.

05(4

)–

0.02

(1)

tr(1

)–

p-C

ymen

ea99

-87-

612

941.

56(3

)4.

30(2

)12

.65

(3)

5.14

(8)

4.87

(5)

––

8.14

(4)

––

––

–a-

Terp

inol

enea

586-

62-9

1304

––

–0.

71(8

)–

––

––

––

––

All

o-oc

imen

eb67

3-84

-713

91–

––

tr(1

)–

––

––

––

––

(3E

,5E

)-2,

6-D

imet

hyl

-1,3

,5,7

-oc

tate

trae

nec

460-

01-5

1419

––

–tr

(1)

––

––

––

––

–



Per

ille

neb

539-

52-6

1428

––

––

tr(4

)–

––

––

––

–tr

ans-

Lin

aloo

lox

ide

(fu

ran

oid)

a59

89-3

3-3

1453

–1.

41(2

)–

0.01

(3)

––

––

–0.

14(1

)–

–ci

sL

inal

ool

oxid

e(f

ura

noi

d)a

3499

5-77

-214

67–

–0.

55(4

)0.

01(5

)tr

(1)

––

0.17

(2)

–0.

24(5

)0.

01(2

)–

–N

erol

oxid

eb17

86-0

8-9

1488

––

–0.

01(3

)0.

01(1

)–

––

––

––

Cam

phor

a46

4-48

-215

43–

––

tr(2

)0.

01(3

)–

––

––

––

–L

inal

oola

78-7

0-6

1562

0.37

(2)

1.68

(5)

3.42

(4)

0.02

(8)

0.15

(5)

––

0.49

(4)

0.01

(6)

20.8

3(5

)0.

02(8

)0.

01(2

)0.

01(4

)4-

Isop

ropy

l-1-

met

hyl

-2-

cycl

ohex

en-1

-ola

2980

3-82

-515

760.

57(5

)0.

69(3

)0.

01(3

)tr

(2)

tr(4

)–

––

––

––

–

Cam

phen

eh

ydra

tec

465-

31-6

1615

––

––

––

––

–tr

(1)

––

–4-

Terp

ineo

la56

2-74

-316

13–

–0.

28(1

)0.

17(4

)0.

42(4

)–

–0.

12(2

)–

0.15

(5)

tr(2

)–

–p-

Men

th-1

-en

-4-o

lc20

126-

76-5

1618

––

–0.

16(4

)–

––

––

––

––

b-C

yclo

citr

alb

432-

25-7

1647

––

–tr

(7)

––

–0.

25(4

)tr

(5)

0.15

(5)

––

Myr

ten

ala

564-

94-3

1650

––

––

tr(2

)–

–0.

28(5

)0.

21(6

)–

–b-

Cit

ralb

106-

26-3

1703

––

–tr

(1)

0.01

(3)

––

––

––

––

a-Te

rpin

eola

1048

2-56

-117

200.

72(5

)0.

93(4

)1.

55(2

)0.

11(6

)0.

22(5

)3.

01(7

)–

0.03

(2)

1.36

(5)

0.08

(2)

––

tran

s-p-

Men

th-2

-en

-7-o

lc19

898-

87-4

1717

––

––

––

––

––

0.05

(4)

––

Bor

neo

la50

7-70

-017

25–

0.47

(3)

–tr

(1)

0.01

(4)

––

––

0.29

(5)

––

a-C

yclo

gera

nio

lc66

27-7

4-3

1737

––

––

––

––

–0.

86(5

)0.

27(4

)–

–C

itra

lb53

92-4

0-5

1732

––

––

tr(4

)0.

01(2

)–

––

––

0.30

(1)

–tr

ans-

Lin

aloo

lox

ide

(pyr

anoi

d)a

5989

-33-

317

55–

0.22

(2)

––

––

––

–0.

11(5

)0.

02(3

)–

–

Pip

erit

one

oxid

ec52

86-3

8-4

1760

––

––

–0.

04(1

)–

––

––

––

cis-

Lin

aloo

lox

ide

(pyr

anoi

d)a

5989

-33-

317

81–

––

––

––

––

–0.

14(4

)–

–ci

s-G

eran

iolb

106-

25-2

1803

––

0.02

(1)

0.19

(4)

0.28

(6)

––

––

––

–D

ihyd

ro-b

-ion

onec

1728

3-81

-718

07–

––

––

––

–tr

(4)

–tr

(1)

–0.

01(2

)M

yrte

nol

a51

5-00

-418

15–

––

––

––

––

0.78

(5)

0.08

(6)

––

Ner

olb

106-

24-1

1817

––

–0.

05(3

)–

––

––

––

––

Gra

ndi

solb

2653

2-22

-918

21–

––

––

––

––

1.03

(5)

0.29

(3)

––

p-C

ymen

-8-o

lb11

97-0

1-9

1872

––

–0.

01(7

)0.

01(5

)–

––

––

tr(1

)–

–a-

Ion

oneb

127-

41-3

1875

––

––

––

––

tr(4

)1.

48(5

)tr

(1)

–0.

01(4

)E

xo-2

-Hyd

roxy

cin

eole

b92

999-

78-5

1877

––

–tr

(3)

––

––

–0.

11(5

)–

––

cis-

Myr

tan

olc

514-

99-8

1893

––

––

––

––

–0.

30(5

)–

––

Dih

ydro

-a-i

onon

ec31

499-

72-6

1989

––

––

––

––

tr(2

)0.

23(5

)–

––

p-C

ymen

-3-o

la89

-83-

822

25–

––

0.02

(3)

0.03

(4)

0.90

(7)

––

–1.

16(5

)–

––

p-C

ymen

-2-o

lb49

9-75

-222

32–

––

0.02

(3)

0.03

(4)

0.06

(1)

––

––

––

–S

esq

uit

erp

enes

Dih

ydro

edu

lan

Ic63

335-

66-0

1538

––

–tr

(3)

tr(2

)–

––

tr(2

)0.

02(5

)–

––

a-B

erga

mot

eneb

1769

9-05

-716

04–

––

–tr

(1)

––

0.21

(5)

––

–b-

Car

yoph

ylle

neb

87-4

4-5

1636

––

––

––

––

––

tr(1

)–

0.01

(2)

b-C

ube

ben

eb13

744-

15-5

1738

––

––

––

––

––

tr(1

)–

0.01

(2)

(E,E

)-a-

Far

nes

eneb

502-

61-4

1760

––

––

tr(2

)–

––

–0.

21(5

)tr

(1)

––

a-C

urc

um

enec

644-

30-4

1796

––

tr(4

)tr

(3)

––

–tr

(4)

0.21

(5)

0.02

(3)

––

NIT

RO

GE

N-C

ON

TA

ININ

GC

OM

PO

UN

DS

2,5-

Dim

eth

ylpy

razi

neb

123-

32-0

1336

––

–tr

(3)

0.01

(5)

––

––

–0.

01(6

)–

–2-

Met

hox

y-3-

met

hyl

pyra

zin

ec28

47-3

0-5

1392

––

––

tr(3

)–

––

––

––

–2-

Isop

ropy

l-3-

met

hox

ypyr

azin

ea25

773-

40-4

1452

0.01

(5)

5.80

(6)

1.59

(4)

9.31

(8)

17.5

9(5

)12

.56

(8)

3.16

(5)

9.58

(4)

––

––

–2-

Sec

-bu

tyl-

3-m

eth

oxyp

yraz

inec

2416

8-70

-515

19–

––

0.01

(7)

0.04

(5)

––

––

––

––

Com

pou

nds

are

iden

tifi

edby

com

mon

nam

esan

dC

AS

(Ch

emic

alA

bstr

act

Ser

vice

s)re

gist

ryn

um

ber,

and

are

list

edac

cord

ing

toth

eK

ovat

sre

ten

tion

inde

x(K

RI)

wit

hin

each

com

pou

nd

clas

s.T

he

nu

mbe

rof

sam

ples

inw

hic

hth

eco

mpo

un

dw

asid

enti

fied

isgi

ven

inpa

ren

thes

es.

a Ide

nti

fica

tion

base

don

mas

ssp

ectr

um

,K

RI

and

auth

enti

cst

anda

rd.

b Ide

nti

fica

tion

base

don

mas

ssp

ectr

um

and

KR

I.c I

den

tifi

cati

onba

sed

onm

ass

spec

tru

mon

ly.

tr,

trac

eam

oun

ts(<

0.01

%).

FF

emal

epl

ants

.MM

ale

plan

ts.U

NR

,Um

tiza

Nat

ure

Res

erve

;OV

GF,

Oce

anV

iew

Gu

est

Far

m;D

NR

,Dw

esa

Nat

ure

Res

erve

;MP

D,M

pan

dear

ea;M

TS

,Mt

Su

lliv

anar

ea;U

MN

R,U

mta

mvu

na

Nat

ure

Res

erve

;O

GN

R,

Ori

biG

orge

Nat

ure

Res

erve

;V

CN

R,

Ver

non

Cro

okes

Res

erve

;K

KN

R,

Kra

nzk

loof

Nat

ure

Res

erve

;N

FR

,N

kan

dla

For

est

Res

erve

.



GENERAL OVERVIEW OF PATTERN OF

VOLATILE EMISSIONS

The number of compounds emitted by cones variedmarkedly between populations, ranging from 16 inUNR to 45 in KKNR (Table 1) and also betweensexes. In female plants, as few as ten compoundswere emitted in the MTS population and up to 24compounds were emitted in KKNR. In male andfemale plants, the most commonly occurring com-pounds in sampled populations were p-anisaldehyde(all ten populations); benzaldehyde, eucalyptol andlinalool in nine populations; b-pinene and a-terpinene in eight populations; and phenol anda-pinene in seven and six populations, respectively.Fatty acid derivatives in the EC populations werecomposed of only six compounds compared to 19compounds in the KZN populations (Table 1). It is

noteworthy that all the nitrogen-containing com-pounds were pyrazine derivatives (Table 1).

CONE ODOUR VARIATION BETWEEN POPULATIONS

A Bray–Curtis NMDS analysis of cone odour com-pounds of E. villosus (Fig. 2) showed a significantseparation between the different populations (NMDSstress value 0.12; one-way ANOSIM, factor popula-tion: global R = 0.835, P < 0.01). Out of 45 pairwisecomparisons between the different populations,significant separation was found between 17 of them(Table 2). The clustering of populations based onchemical profiles (Fig. 2) tended to follow the overallgeographical pattern of separation of EC andKZN populations. The only exceptions were indivi-duals from VCNR in KZN in which volatiles were

Figure 2. Non-metric multidimensional scaling (NMDS) based on Bray–Curtis similarities of male cone odour com-position comprising 89 compounds from 73 samples of Encephalartos villosus from ten populations across the speciesrange in the Eastern Cape and KwaZulu Natal. [two-dimensional stress value: 0.12; one-way analysis of similarities,global R (population) = 0.835, P < 0.01]. FR, Forest Reserve; GF, Guest Farm; NR, Nature Reserve; PSJ, Port StJohns.



characterized by heptanal, linalool, p-anisaldehyde,and benzaldehyde, and had closer affinities to popu-lations from the south of the range (OVGF and UNR;Fig. 2). Individuals from the UMNR population, situ-ated at the boundary between EC (south) and KZN(north), had volatile compounds characterized by(3E)-1,3-octadiene and (3E,5Z)-1,3,5-octatriene thatwere typically dominant in KZN populations (OGNR,KKNR, and NFR), as well as a-terpinene and2-isopropyl-3-methoxypyrazine that occurred in ECpopulations. In the EC, populations in the extremesouth (UNR, OVGF) clustered together as a resultof the presence of eucalyptol, benzaldehyde, heptanal,b-pinene, linalool, and p-anisaldehyde. By contrast,other EC populations (DNR, MPD, and MTS)clustered together as a result of the prevalence of2-isopropyl-3-methoxypyrazine, a-pinene, camphene,and a-terpinene.

Mantel tests were performed to compare cone odourmatrices calculated using the Bray–Curtis similaritycoefficient (Clark and Warwick, 2001) with the matrixof geographical distance (km) between populations.The results showed a significant correlation (r = 0.39;P = 0.001) between changes in cone volatiles and geo-graphical distance across the full range of E. villosus.When the data were analyzed in subsets, changesin volatile composition were strongly correlatedwith distance between populations in EC (r = 0.78;P = 0.008) and might be indicative of restricted geneflow between the populations. This is observed in thechange in composition across populations from UNRto MTS (Fig. 1). There was no correlation betweenvolatile composition and geographical separation inKZN populations (r = 0.05; P = 0.51). This is indicatedby the lack of variation in the dominant compoundsacross the populations, especially at OGNR, KKNR,and NFR (Fig. 1).

INSECT POLLINATORS ON MALE AND FEMALE

CONES OF E. VILLOSUS

Pollen dehiscent cones of E. villosus in all the popu-lations were visited by an undescribed species ofPorthetes (Fig. 3C, Table 3). Oberprieler (1996) gavethe latter species a manuscript name of P. pearsoniibut the description has not been formally published.Two other beetle species (Coleoptera), namely anundescribed Erotylidae sp. nov. and Metacucujusgoodei Endrödy-Younga (Boganiidae) (Fig. 3A, B),were collected from all the populations except UMNRand NFR (Table 3). Their absence from UMNR andNFR may be a result of the small number of cones(four male cones) in each of these populations. Por-thetes sp. occurred in between the sporophylls, alongthe cone axis, and on the cone surface, whereas Ero-tylidae sp. nov. and M. goodei were moving inbetween the sporophylls and along the cone axis.Female cones of E. villosus in the three populationssampled were visited by Erotylidae sp. nov. andPorthetes sp. (the same taxa as on male cones)(Table 3) and Antliarhinus zamiae (Thunberg)(Fig. 3D, E, Table 3). Female A. zamiae were observedpiercing the cone sporophylls with their rostrums andcrawling over the cone surface. Similarly, maleA. zamiae were seen crawling on the cone surface,with some forcing their way in between the tightmegasporophylls. In some cases, male and femaleA. zamiae were observed mating on the cone surface(Fig. 3F). A few Erotylidae sp. nov. and Porthetes sp.were actively crawling on the female cone surface,whereas some were forcing their way in betweenthe tightly packed megasporophylls. Overall, themean ± SE number of Porthetes sp. individuals perpopulation (833.8 ± 115.) was significantly greaterthan that for Erotylidae sp. nov. (294.0 ± 58.9),

Table 2. Test statistics (R) of pairwise comparisons (one-way analysis of similarities) between populations of Enceph-alartos villosus

UNRa OVGF DNR MPD MTS UMNR OGNR VCNR KKNR NFR

UNROVGF 0.41DNR 1.00 0.98b**MPD 0.97 0.92* 0.41MTS 1.00** 0.97** 0.42 0.13UMNR 1.00 0.96 1.00 0.88 0.95*OGNR 1.00 1.00** 1.00* 1.00 1.00* 1.00VCNR 0.90 0.76* 1.00 1.00 1.00** 1.00 1.00KKNR 1.00** 0.98** 1.00** 1.00** 1.00** 0.70 0.44 1.00**NFR 1.00 1.00 1.00 1.00 1.00* 1.00 1.00 1.00 -0.08

aNames of populations are the same as given in Table 1. bBold values indicate populations that are significantly different.**P < 0.01; *P < 0.05.



M. goodei (171.2 ± 32.3), and A. zamiae (43.8 ± 9.7)(analysis of variance: F3,16 = 26.8, P < 0.01). This trendwas evident in all of the study populations (Table 3).

DISCUSSION

The present study provides a detailed investigationof geographical variation in volatile composition forE. villosus and is the first detailed geographicalanalysis of volatile composition in any cycad species.The results obtained show that there is considerablevariation in the chemical composition of cone volatileemissions between populations of E. villosus includ-ing a shift from dominance of monoterpenes in thesouthern part of the range (e.g. DNR, MPD, and MTSaround Port St Johns) to dominance of unsaturatedhydrocarbons in the northern part of the range (e.g.

KKNR, NFR). The UMNR population, situated in thecentre of the range, appears to be the transition point(Fig. 1). The results further show that despite thevariation in volatile composition, the same insectspecies, namely A. zamiae, Erotylidae sp. nov.,M. goodei, and Porthetes sp., are common across allthe sampled E. villosus populations.

Studies of geographical variation in plant traitshave shown several outcomes, including lack of struc-tured variation across the range (Svensson et al.,2005), clinal variation (Knudsen, 2002), and discreteor saltational variation reflecting adaptation to differ-ent pollinators (Schlumpberger & Raguso, 2008). TheMantel test for E. villosus data provides statisticalevidence for volatile profiles being associated withgeographical separation, whereas the cluster analysisindicates that this variation is more consistent with

Figure 3. Insects that visit male and/or female cones of Encephalartos villosus. A, an undescribed Erotylidae sp. nov.(male and female cones). B, Metacucujus goodei (male cones). C, Porthetes sp (male and female cones). D, femaleAntliarhinus zamiae (female cones). E, male Antliarhinus zamiae (female cones). F, male and female Antliarhinus zamiaemating. Scale bars = 1000 mm.



discrete or saltational changes than with clinalchanges across the range. Further studies arerequired to explore genetic structure across the dif-ferent populations of E. villosus that may be linked todifferences in cone volatiles.

CHEMICAL COMPOSITION OF CONE VOLATILES

Many of the volatile compounds and compoundclasses identified in this study are known to occurin other cycads (Pellmyr et al., 1991; Terry et al.,2004a, b; Azuma & Kono, 2006, Proches & Johnson,2009; Suinyuy, Donaldson & Johnson, 2010) and otherplants (Knudsen et al., 2006). Out of 87 identifiedcompounds, only five compounds occurred in highrelative amounts (� 30%) in at least one population(Table 1). The majority of the compounds wereemitted in small relative amounts ranging from traceamounts to just above 20%. Schlumpberger & Raguso(2008) suggest that compounds that occur in smallrelative amounts should not be ignored because theycan serve critical functions in plant–pollinator rela-tions. Terpenoids, particularly the monoterpenes, arethe most numerous compounds in the volatile blend ofE. villosus and some of them together with somebenzenoids occur in almost all the populations(Table 1). Their occurrence in almost all populationssuggests that they could be critical compounds thatserve different functions and require further investi-gation. Sampling pollen dehiscent and receptive conesat different times of the day will establish whetherthese compounds are emitted in different concentra-tions that can affect insect behaviour in a similar

manner to b-myrcene in some Macrozamia cycads(Terry et al., 2004a, 2007a, b). There were relativelyfew unsaturated hydrocarbons but they included (3E)-1,3-octadiene and (3E,5Z)-1,3,5-octatriene, the mostabundant compounds emitted by plants from KZNpopulations. These compounds have been recorded inthe volatile profile of few plants and have been iden-tified as possible insect attractants (Skubatz et al.,1996; T. N. Suinyuy, unpubl. data). Four pyrazinecompounds occurred in varying amounts, mostly inplants from EC populations, with 2-isopropyl-3-methoxypyrazine as a dominant compound. Generallypyrazines have distinct sensory properties and havebeen associated with warning signals, alerting signalsto predators, aggregation pheromones of insects, ovi-position stimulants (Rothschild, Moore & Brown,1984; Abassi et al., 1998), and insect attractants(Ervik, Tollsten & Knudsen, 1999). The differentpyrazines could therefore fulfill different functions incycads.

GEOGRAPHICAL VARIATION OF E. VILLOSUS

CONE VOLATILES

Although E. villosus exhibited geographical variationin cone volatile emissions, the present study showedthat eight of the 88 compounds occurred in almost allthe populations (Table 1) and may be of critical impor-tance in influencing insect behaviour because thesame insect assemblages occurred in all populations(Table 3). It is noteworthy that the pattern ofgeographical variation in E. villosus cone odour is

Table 3. Mean ± SE number of insects of different species collected from male and female cones of Encephalartos villosusin different populations across the distribution range in the Eastern Cape and KwaZulu Natal

Populationa Cone sexNumberof cones

Insect species and number of individuals

Antliarhinuszamiae

Erotylidaesp. nov.

Metacucujusgoodei Porthetes sp.

UNR Male 5 – 1.4 ± 0.5 1.8 ± 0.7 7.0 ± 0.9OVGF Male 5 – 2.6 ± 1.5 2.0 ± 0.9 15.8 ± 3.6

Female 5 15.5 ± 3.0 2.8 ± 1.2 – 1.4 ± 0.6DNR Male 4 – 86.0 ± 23.0 60.8 ± 11.3 168.0 ± 36.3MPD Male 4 – 101.0 ± 16.8 47.8 ± 9.6 191.8 ± 9.1MTS Male 5 – 11.0 ± 3.4 7.4 ± 2.3 37.6 ± 9.7

Female 3 18.0 ± 5.3 1.0 ± 0.6 – 3.0 ± 1.0UMNR Male 4 – – – 22.5 ± 4.3OGNR Male 5 – 43.4 ± 4.3 31.2 ± 3.1 93.4 ± 3.5VCNR Male 4 – 64.5 ± 4.8 31.7 ± 3.4 186.3 ± 15.5KKNR Male 5 – 34.4 ± 7.2 16.6 ± 4.3 213.2 ± 21.1

Female 4 17.8 ± 2.0 2.5 ± 0.6 – 11.0 ± 1.8NFR Male 4 – – – 65.5 ± 6.6

aNames of populations are the same as given in Table 1.



explained best by changes in two compounds: 2-isopropyl-3-methoxypyrazine and (3E)-1,3-octadiene.The nitrogen-containing compound 2-isopropyl-3-methoxypyrazine occurs in all southern populationsfrom UNR to UMNR and increases in relativeamounts from UNR to MTS. By contrast (3E)-1,3-octadiene occurs in all northern populations startingfrom UMNR and tends to increase along the northerngradient to NFR (Fig. 1).

Samples from UMNR in the centre of the distribu-tion range were characterized by compounds fromboth the southern and northern populations. TheUMNR is situated in the eastern part of the Umtam-vuna river gorge and occurs within the Maputaland-Pondoland centre of endemism. Pondoland isdominated by ancient outcrops of nutrient-poorquartzite sandstone that appear to have acted asedaphic barriers to plant migration (Carbutt &Edwards, 2001) and support a resident flora that isapparently trapped by these barriers. It is not clearexactly why the UMNR population of E. villosus con-tains odour compounds that otherwise occur sepa-rately in populations to the north and south, althoughthe sharp transition in cone volatile composition inE. villosus across this region (Fig. 1) supports the ideaof a biogeographical barrier in the Umtamvuna areaas suggested by Carbutt & Edwards (2001).

The VCNR population, sandwiched between OGNRand KKNR, has a suite of volatile compounds similarto that of plants from the KKNR but present indifferent relative quantities (Fig. 1, Table 1). Thedominant compounds are monoterpenes, aldehydes,and benzenoids, and these are closest to those emittedby plants from the EC region (Fig. 1, Table 1). Thissuggests that these compounds occurred more widelyacross the range but that only plants from the VCNRpopulation have retained and expressed the genesfor biosynthesis of all the volatile compounds oncepresent in the different populations. Long distancedispersal of seeds from the southern part of the rangeis highly unlikely to account for the volatile compo-nents of the VCNR population. There is no knownlong distance dispersal mechanism and cycad dis-persal is typically within a short distance of theparent plant (Snow & Walter, 2007). Long rangepollen dispersal (> 5 km) also appears to be unlikelybecause Donaldson (1997) discovered that Porthetessp., Erotylidae sp. nov., M. goodei, and A. zamiae losta substantial amount of pollen within a few hoursafter they left the pollen shedding cones of E. villosus,and also that plants situated > 5 km from sourcepopulations never had insect pollinators present (J. S.Donaldson, unpubl. data).

Intraspecific variation in plant morphology iswell documented, although an increasing numberof studies are revealing similar variation in che-

mical traits (Dobson et al., 1997; Azuma, Toyota &Asakawa, 2001; Dötterl, Wolfe & Jürgens, 2005;Chess, Raguso & LeBuhn, 2008; Jhumur, Dötterl &Jürgens, 2008; Schlumpberger & Raguso, 2008).Hypotheses for intraspecific trait variation includephenotypic plasticity, neutral processes such as drift,adaptive processes such as co-evolution or pollinatorshifts, and local hybridization. The evidence for eachof these hypotheses is weighed up in relation tothe geographical variation in the cone odour ofE. villosus.

Phenotypic plasticity is highly unlikely to accountfor the geographical odour patterns observed in E.villosus because plants from populations in EC thathave been growing under different environmentalconditions in the Kirstenbosch Botanic Garden inCape Town for close to 100 years emitted the samevolatile compounds as those from the natural popu-lations (T. N. Suinyuy, unpubl. data).

Drift also appears unlikely to account for variationin the major compounds as these compounds havebeen shown to play functional roles in pollinatorattraction (T. N. Suinyuy, unpubl. data). However,neutral and adaptive processes could apply to differ-ent compounds. For example, geographical variationin the volatile profile may occur only in compoundsthat are not used by pollinators to find and locate hostplants (Dötterl et al., 2005; Füssel, Dötterl & Jürgens,2007). Pollinator shifts have been invoked for caseswhere there are different pollinators in differentgeographical areas and these can result in quantita-tive shifts involving changes in assemblages forplant species which are visited by a number of differ-ent pollinators (Pellmyr, 1986b; Schlumpberger &Raguso, 2008), or involve complete transitions(Johnson & Steiner, 1997). However, variation infloral traits can also occur without pollinator shifts(Ellis & Johnson, 2009). This was evidently the casefor E. villosus, which showed no change in beetlespecies composition across the distribution range. Thesame insect visitors were recorded from all E. villosuspopulations (i.e. A. zamiae, Erotylidae sp. nov.,M. goodei, and Porthetes sp.) despite the difference involatile compounds. A recent molecular study of phy-logenetic relationships within Porthetes concludedthat the specimens from different E. villosus popula-tions across the range of distribution, as well as fromthe related cycads Encephalartos aplanatus andEncephalartos umbeluziensis (Treutlein, Vorster &Wink, 2005), comprised a single species (Downie,Donaldson & Oberprieler, 2008). This suggests thatchanges in cone volatiles are not associated withdifferent pollinator assemblages. However, it is stillpossible that there has been localized co-evolutionbetween these insects and E. villosus, which is notreflected in molecular markers or morphology of the



beetles. Performing scent bioassays at different siteswill aid in understanding whether beetles, which areostensibly the same species, exhibit regional differ-ences in volatile preferences that could accountfor the geographical variation in the cone odours ofE. villosus.

It is noteworthy that the barrier between northernand southern populations of E. villosus at theUmtamvuna river is also the contact zone for twoother cycad species that are closely related to oneanother (Treutlein et al., 2005): Encephalartos natal-ensis north of the Umtamvuna and Encephalartosaltensteinii to the south. The odour compound (3E)-1,3-octadiene, which dominated the odour profile ofE. villosus north of the Umtamvuna, is also the domi-nant compound in cone volatiles of E. natalensis(Suinyuy et al., 2010). In addition, at least one of thepollinators of E. villosus also occurs on E. natalensis(Donaldson, 1997; T. N. Suinyuy, unpubl. data).Hybridization between E. villosus and other Enceph-alartos species such as E. natalensis could account forsome of the geographical variation in the scent ofE. villosus. If there are no other barriers and back-crosses occur, gene flow across species boundaries ispossible. Stökl et al. (2008) showed that overlap offlowering times in Ophrys iricolor and Ophrys luper-calis, which emit the same compounds and attract thesame insects results in extensive hybridization andintrogression. Although natural hybrids betweenE. villosus and E. natalensis have not been recorded,hybrids have been found to occur between E. villosusand Encephalartos senticosus (previously consideredpart of Encephalartos lebomboensis) (Dyer, 1965;Vorster, 1986), which also emits (3E)-1,3-octadiene(T. N. Suinyuy, unpubl. data). Hybrids have also beenfound in the south of the range between E. villosusand E. altensteinii in the EC where they occur insympatry (Dyer, 1965; Vorster, 1986). Although E. vil-losus and E. altensteinii have different major com-pounds, they emit similar minor volatile compoundsthat could be involved in pollinator attraction. Thegeographical variation in volatile compounds in E. vil-losus could therefore be affected by introgressionof odours traits through hybridization with otherEncephalartos species. Alternatively, the dominanceof (3E)-1,3-octadiene among cycads in KZN may rep-resent convergent evolution resulting from adaptationto a similar local suite of insects. Further investiga-tion of volatile compounds and pollinators in otherEncephalartos species is required to determine howthey vary in relation to E. villosus across a widegeographical area.

In conclusion, the widespread cycad E. villosus con-sists of a number of geographically structured chemo-types. The discontinuities between these chemotypesare not sufficiently pronounced to justify recognition

of distinct taxa, nor does it appear from preliminaryinvestigations that these chemotypes are associatedwith different pollinators. However, the patternssuggest ongoing evolutionary diversification in E. vil-losus, which makes this species suitable for furthermicroevolutionary studies on the role that insect pol-linators played in the evolution of Encephalartos.

ACKNOWLEDGEMENTS

We thank Jacques De Wet Bösenberg for his dedi-cated assistance in the field. We greatly appreciatethe assistance of Andreas Jürgens in the volatileanalysis and Sandy-Lynn Steenhuisen for technicalsupport in the laboratory. Special thanks go toFerozah Conrad for reading through the manuscriptand John Measey for helping with the data analysis.We also thank the two anonymous reviewers for theirhelpful comments on the manuscript. We are gratefulto Ian Smith and Johanne Van Ryneveld who grantedus access to their farm and to The Eastern CapeParks Board and Ezemvelo KwaZulu Natal Wildlifefor issuing permits to sample from plant populationsin nature reserves. The present study was financiallysupported by The Fairchild Tropical Botanic Garden,South African National Biodiversity Institute, theAndrew W. Mellon Foundation, University of CapeTown and the University of KwaZulu Natal.

REFERENCES

Abassi SA, Birkett MA, Petterson J, Pickett JA, Wood-cock CM. 1998. Ladybird beetle odour identified and foundto be responsible for attraction between adults. Cellular andMolecular Life Sciences 54: 876–879.

Azuma H, Kono M. 2006. Estragole (4-allylanisole) is theprimary compound in volatiles emitted from male andfemale cones of Cycas revoluta. Journal of Plant Research119: 671–676.

Azuma H, Toyota M, Asakawa Y. 2001. Intraspecificvariation of floral scent chemistry in Magnolia kobus DC.(Maganoliaceae). Journal of Plant Research 114: 411–422.

Bonnet E, van der Peer Y. 2002. ZT: a software toolfor simple and partial Mantel tests. Journal of StatisticalSoftware 7: 1–12.

Carbutt C, Edwards T. 2001. Cape elements on high-altitude corridors and edaphic islands: historical aspectsand preliminary phytogeography. Systematic and Geogra-phy of Plants 71: 1033–1061.

Chen C, Song Q, Proffit M, Bessière J-M, Li Z, Hossaert-McKey M. 2009. Private channel: a single unusual com-pound assures specific pollinator attraction in Ficussemicordata. Functional Ecology 23: 941–950.

Chess SKR, Raguso RA, LeBuhn G. 2008. Geographicdivergence in floral morphology and scent in Linanthusdichotomus (Polemoniaceae). American Journal of Botany95: 1652–1659.



Clarke KR, Gorley RN. 2006. Primer v6: user manual/tutorial. Plymouth: Primer-E.

Clarke KR, Warwick RM. 2001. Change in marine commu-nities: an approach to statistical analysis and interpretation,2nd edn. Plymouth: PRIMER-E.

Dobson HEM. 2006. Relationship between floral fragrancecomposition and type of pollinator. In: Dudareva N, Pirch-ersky E, eds. Biology of floral scent. Boca Raton, FL: Taylorand Francis Group, CRC Press, 147–198.

Dobson HEM, Arroyo J, Bergström G, Gröth I. 1997.Interspecific variation in floral fragrances within the genusNarcissus (Amaryllidaceae). Biochemical Systematics andEcology 25: 685–706.

Donaldson JS. 1991. Adaptation to the host and evolution ofhost specialisation in cycad weevils (Coleoptera: Brentidae).PhD Thesis, University of Cape Town, Cape Town.

Donaldson JS. 1997. Is there a floral parasite mutualismin cycad pollination? Pollination biology of Encephalartosvillosus (Zamiaceae). American Journal of Botany 84: 1398–1406.

Dötterl S, Wolfe LM, Jürgens A. 2005. Qualitative andquantitative analyses of flower scent in Silene latifolia.Phytochemistry 66: 203–213.

Downie DA, Donaldson JS, Oberprieler RG. 2008.Molecular systematics and evolution in an African cycad–weevil interaction: amorphocerini (Coleoptera: Curculion-idae: Molytinae) weevils on Encephalartos. MolecularPhylogenetics and Evolution 47: 102–116.

Dyer RA. 1965. The cycads of Southern Africa. Bothalia 8:404–515.

Ellis AG, Johnson SD. 2009. The evolution of floralvariation without pollinator shifts in Gorteria diffusa(Asteraceae). American Journal of Botany 967: 93–801.

Endrödy-Younga S. 1991. Boganiidae (Coleoptera: Cucu-joidea) associated with cycads in South Africa: two newspecies and a new synonym. Annals of the TransvaalMuseum 35: 285–293.

Ervik F, Tollsten L, Knudsen JT. 1999. Floral scentchemistry and pollination ecology in phytelephantoidpalms (Arecaceae). Plant Systematics and Evolution 217:279–297.

Füssel U, Dötterl S, Jürgens A. 2007. Inter- and intraspe-cific variation in floral scent in the genus Salix and itsimplication for pollination. Journal of Chemical Ecology 33:749–765.

Goode D. 1989. Cycads of Africa. Cape Town: StruikPublishers.

Hughes MJB, Bohannan BJM, Brown JH, Colwell RK,Fuhrman JA, Green JL, Horner-Devine MC, Kane M,Krumins JA, Kuske CR, Morin PJ, Naeem S, Øvreås L,Reysenbach A-L, Smith VL, Staley JT. 2006. Microbialbiogeography: putting micro-organisms on the map. NatureReviews Microbiology 4: 102–112.

Jhumur U, Dötterl S, Jürgens A. 2008. Floral odors ofSilene otites: their variability and attractiveness to mosqui-toes. Journal of Chemical Ecology 34: 14–25.

Johnson SD. 1996. Pollination, adaptation and speciationmodels in the Cape flora of South Africa. Taxon 45: 59–66.

Johnson SD, Steiner KE. 1997. Long-tongued fly pollinationand evolution of floral spur length in Disa draconis complex.Evolution 51: 45–53.

Jürgens A. 2009. The hidden language of flowering plants:floral odours as a key for understanding angiosperm evolu-tion? New Phytologist 183: 240–243.

Knudsen JT. 2002. Variation in floral scent compositionwithin and between populations of Geonoma macrostachys(Arecaceae) in the western Amazon. American Journal ofBotany 89: 1772–1778.

Knudsen JT, Eriksson R, Gershenzon J, Ståhl B. 2006.Diversity and distribution of floral scent. Botanical Review72: 1–120.

Millar JG, Sims JJ. 1998. Preparation, cleanup and prelimi-nary fractionation of extracts. In: Millar JG, Haynes KF,eds. Methods in chemical ecology. Norwell, MA: KluwerAcademic Publishers, 1–37.

Mucina L, Scott-Shaw CR, Rutherford MC, Camp KGT,Matthews WS, Powrie LW, Hoare DB. 2006. IndianOcean coastal belt. In: Mucina L, Rutherford MC, eds.The vegetation of South Africa, Swaziland and Lesotho.Strelitzia 19, Pretoria: South African National BiodiversityInstitute.

Oberprieler RG. 1996. Systematics and evolution of thetribe Amorphocerini (Coleoptera: Curculionidae), with areview of the cycad weevils of the world. PhD Thesis,University of the Orange Free State.

Pellmyr O. 1986a. Function of olfactory and visual stimuliin pollination of Lysichiton americanum (Araceae) by aStaphylinid beetle. Madrono 33: 47–54.

Pellmyr O. 1986b. Three pollination morphs in Cimicifugasimplex; incipient speciation due to inferiority in competi-tion. Oecologia 68: 304–307.

Pellmyr O. 1992. Evolution of insect pollination andangiosperm diversification. Trends in Ecology and Evolution7: 46–49.

Pellmyr O, Tang W, Groth I, Bergström G, Thien LB.1991. Cycad cone and angiosperm floral volatiles: inferencesfor the evolution of insect pollination. Biochemical System-atics and Ecology 19: 623–627.

Pellmyr O, Thien LB. 1986. Insect reproduction and floralfragrances: keys to the evolution of angiosperms? Taxon 35:76–85.

Proches S, Johnson SD. 2009. Beetle pollination of thefruit-scented cones of the South African cycad Stangeriaeriopus. American Journal of Botany 96: 1722–1730.

Raguso RA. 2008. Wake up and smell the roses: the ecologyand evolution of floral scent. Annual Review of Ecology,Evolution and Systematics 39: 549–569.

Raguso RA. 2004. Why do flowers smell? The chemicalecology of fragrance-driven pollination. In: Cardé RT, MillarJG, eds. Advances in insect chemical ecology. Cambridge:Cambridge University Press, 141–178.

Rothschild M, Moore BP, Brown WV. 1984. Pyrazinesas warning odour components in the monarch butterfly,Danaus plexippus, and in moths of the genera Zygaena andAmata (Lepidoptera). Biological Journal of Linnean Society23: 375–380.



Schiestl F, Peakall R. 2005. Two orchids attract differentpollinators with the same floral odour compound: ecologicaland evolutionary implications. Functional Ecology 19: 674–680.

Schlumpberger BO, Raguso RA. 2008. Geographic varia-tion in floral scent of Echinopsis ancistrophora (Cataceae);evidence of constraints on hawkmoth attraction. Oikos 117:801–814.

Skubatz H, Kunkel DD, Howald NW, Trenlke R,Mookherjee B. 1996. The Sauromatum guttatum appendixas an osmophore: excretory pathways, composition of vola-tiles and attractiveness to insects. New Phytologist 134:631–640.

Snow EL, Walter GH. 2007. Large seeds, extinct vectors andcontemporary ecology: testing dispersal in a locally distrib-uted cycad, Macrozamia lucida (Cycadales). AustralianJournal of Botany 55: 592–600.

Stökl J, Schlüter PM, Stuessy TF, Paulus H, Assum G,Ayasse M. 2008. Scent variation and hybridization causethe displacement of a sexually deceptive orchid species.American Journal of Botany 95: 472–481.

Suinyuy TN, Donaldson JS, Johnson SD. 2010. Scentchemistry and patterns of thermogenesis in male andfemale cones of the African cycad Encephalartos natalensis(Zamiaceae). South African Journal of Botany 76: 717–725.

Svensson GP, Hickman MO, Bartram S, Boland W,Pellmyr O, Raguso RA. 2005. Chemistry and geographicvariation of floral scent in Yucca filamentosa (Agavaceae).American Journal of Botany 92: 1624–1631.

Terry I, Moore CJ, Forster PI, Walter GH, Machin PJ,Donaldson JS. 2004b. Pollination ecology of the genusMacrozamia: cone volatiles and pollination specificity. In:Lindstrom AJ, ed. Proceedings of the 6th international con-ference on cycad biology. Nong Nooch: Nong Nooch TropicalBotanical Garden, 155–169.

Terry I, Moore CJ, Walter GH, Forster PI, Roemer RB,Donaldson JS, Machin PJ. 2004a. Association of conethermogenesis and volatiles with pollinator specificity inMacrozamia cycads. Plant Systematics and Evolution 243:233–247.

Terry I, Walter GH, Hull C, Moore C. 2007b. Responses ofpollinating thrips and weevils to specific Macrozamia cycadcone volatiles. In: Vovides AP, Stevenson DW, Osborne R,eds. Proceedings of the 7th international conference on cycadbiology. New York, NY: The New York Botanical GardenPress, 346–371.

Terry I, Walter GH, Moore C, Roemer R, Hull C. 2007a.Odour-mediated push–pull pollination in cycads. Science318: 70.

Tholl D, Röse URS. 2006. Detection and identification offloral scent compounds. In: Dudareva N, Pichersky E, eds.Biology of floral scent. Boca Raton, FL: CRC Press, Taylor &Francis Group, 3–25.

Treutlein J, Vorster P, Wink M. 2005. Molecular relation-ships in Encephalartos (Zamiaceae, Cycadales) based onnucleotide sequences of nuclear ITS1 and 2, Rbcl, andgenomic ISSR fingerprinting. Plant Biology 7: 79–90.

Vorster P. 1986. Hybridization in Encephalartos. Excelsa 12:101–106.



geographical variation in cone volatile composition among...

Documents