Abstract The evolution of a showy floral display as anadvertisement to pollinators could simultaneously adver-tise the availability of resources to pre-dispersal seed-predators. The hypotheses tested here are that the inci-dence of seed predation by bud-infesting insect larvae incapitula of Asteraceae is positively related to (1) capitu-lum size among species, (2) capitulum size within spe-cies, (3) capitulum lifespan, and (4) the degree of flower-ing asynchrony on individual plants. Three populationsof each of 20 common herbaceous species of Asteraceaefrom disturbed ground and grassland habitats were moni-tored for the presence of pre-dispersal, seed-eating insectlarvae. Mean capitulum size (receptacle width) of eachspecies was measured. In a sub-set of eight species, indi-vidual capitula were tagged to determine their floweringphenology and lifespan (from anthesis to seed shedding).From these data an index of flowering synchrony on in-dividual plants was derived. Among species, the inci-dence of larval infestation increased with capitulum size.Small-flowered species such as Achillea millefoliumwere largely free of bud-infesting larvae, whilst large-flowered species such as Arctium minus were heavily in-fested. In three cases investigated in greater detail, budinfestation was found to increase with capitulum sizewithin species, suggesting a potential for natural selec-tion to favour smaller capitula. No relationship wasfound between infestation levels and either capitulumlifespan or degree of flowering synchrony, and there wasno evidence that the relationship between capitulum size

and infestation was confounded by correlations withthese other features. The results support hypotheses 1and 2, but not 3 and 4. It is suggested that the character-istic capitulum size of each species may represent atrade-off between the opposing selection pressures ofpollinators and pre-dispersal seed predators.

Keywords Bud-infesting larvae · Flower lifespan ·Flower size · Flowering synchrony · Tephritidae

Introduction

Many species of animal-pollinated plants have evolvedshowy floral displays as an advertisement to pollinators(Proctor et al. 1996). In a range of families, large flowersor inflorescences may be advantageous for attractingmore pollinators, for example in Epilobium angustifoli-um (Onagraceae; Schmid-Hempel and Speiser 1988),and inflorescence structure and individual flower size areusually considered to be largely the result of selection bypollinators. However, advertisements for resources forpollinators may simultaneously act as advertisements forresources for pre-dispersal seed predators.

Seeds are at their most accessible to predators whilestill on the mother plant. They represent a concentratedsource of protein and minerals which can be exploited bya wide range of animals, especially insects. Crawley(1992) lists sixty studies showing pre-dispersal seed pre-dation by insects (mostly larvae of Diptera, Coleopteraand Lepidoptera) in a wide range of plant families. Per-centage losses of seeds in this phase vary widely be-tween species and populations, but are frequently greaterthan 90% (e.g. Mattson 1980; Randall 1986; Crawleyand Gillman 1989; Turner et al. 1996). Louda and Potvin(1995) showed in experiments in which insects were ex-cluded during seed maturation that fitness is markedlyaffected in Cirsium canescens by seed predation. Thepresence of insects reduced seed production, and resultedin fewer seedlings and flowering adults in the next gen-eration. These studies suggest that seed eaters may exact

M. Fenner (✉ )School of Biological Sciences, University of Southampton,Bassett Crescent East, Southampton, SO16 7PX, UKe-mail: M.Fenner@soton.ac.ukTel.: +44-23-80592033, Fax: +44-23-80594269

J.E. CresswellSchool of Biological Sciences, University of Exeter,Hatherly Laboratories, Prince of Wales Road, Exeter, EX4 4PS,UK

R.A. Hurley · T. BaldwinCentre for Environmental Sciences, University of Southampton,Southampton, SO17 1BJ, UK

Oecologia (2002) 130:72–77DOI 10.1007/s004420100773

M. Fenner · J. E. Cresswell · R. A. HurleyT. Baldwin

Relationship between capitulum size and pre-dispersal seed predationby insect larvae in common Asteraceae

Received: 12 December 2000 / Accepted: 3 July 2001 / Published online: 4 August 2001© Springer-Verlag 2001

73

considerable fitness penalties on plants, which should fa-vour individuals whose floral traits reduce such preda-tion. Previous studies have suggested a link between flo-ral display and seed predation, for example in Ipomopsisaggregata (Polemoniaceae; Brody 1992; Brody andMitchell 1997), Lathyrus vernus (Fabaceae; Ehrlén1996), and Trollius europaeus (Ranunculaceae; Hemborgand Després 1999). Among the Asteraceae, it has beenshown that the number of eggs laid by the weevil Rhino-cyllus conicus in flowerheads of Cirsium vulgare andSilybum marianum is positively related to the capitulumdiameter (Zwölfer and Preiss 1983). In Carduus nutansthe number of weevil larvae maturing per flowerhead hasalso been shown to be a function of capitulum size(Zwölfer and Harris 1984). It is thus possible that themean size of the flowerhead (or individual flower) that ischaracteristic of each species may be a trade-off betweenthe simultaneous conflicting selection pressures imposedby pollinators and seed predators.

This study investigates the relationships betweenflowerhead size and level of infestation in a group of 20common members of the Asteraceae (Compositae). Inthis family the functional unit which serves as the floweris the capitulum, a group of florets borne on a commonreceptacle surrounded by sepal-like involucral bracts.Within the family, the size of the capitulum varies in dif-ferent species by about two orders of magnitude in lineardimensions. Many Asteraceae are hosts to bud-infesting,seed-eating insect larvae which can belong to a numberof insect orders but are predominantly species of tephridflies (Diptera: Tephritidae; Straw 1991; English-Loeband Kabran 1992; Sheppard et al. 1994; Milton 1995;Scott 1996; Edwards and Brown 1997). Typically, theadult inserts one or more eggs into the capitulum while itis still in bud. The sedentary larvae eat the developingseeds and pupate in situ, and are thus completely depen-dent on the resources supplied by one capitulum to attainthe adult stage. There is a high level of host specificitybetween plant and insect. Representative case-historiesare provided by the detailed studies of the insects that in-fest Chrysanthemoides monilifera (Edwards and Brown1997), Centaurea solstitialis (Sobhian and Zwölfer1985) and Carduus nutans (Zwölfer and Harris 1984).Many of the investigations of pre-dispersal seed preda-tion in the Asteraceae reported in the literature have beencarried out in the context of using the insects as agentsfor the biocontrol of invading weeds (e.g. Zwölfer andPreiss 1983; Muller-Scharer and Schroeder 1993; Maysand Kok 1996; Scott 1996; Turner et al. 1996).

In addition to capitulum size, we recognise that otherfeatures of the floral display such as capitulum longevityand the degree of flowering synchrony on individualplants, may also influence seed predation, and may them-selves be associated with capitulum size. Longevity mayexpose the flowers to higher risks of predation, whilesynchronous flowering might help to reduce the level ofseed-eating by means of a predator-satiation (“masting”)effect. Capitulum longevity and flowering synchronywere therefore examined in a sub-group of eight of the

species studied to see if these features are also associatedwith capitulum size or levels of infestation.

The following hypotheses were tested here: thatamongst a selection of common herbaceous Asteraceae,a higher incidence of pre-dispersal seed predation by in-sect larvae is associated with (1) larger capitula amongspecies, (2) larger capitula within populations of individ-ual species, (3) long-lived capitula, and (4) staggeredflowering among capitula on individual plants.

Materials and methods

The Asteraceae family was chosen for this study because of theavailability of numerous species, and because many of them arecommonly observed in the field to have flower-heads infestedwith seed-eating insect larvae. The 20 plant species used (Table 1)are all common herbaceous species found in grasslands, arablefields or waste ground. They include 11 perennials, 3 biennialsand 6 annuals. For each species three populations within 25 km ofSouthampton University, Hampshire, UK, were sampled. Eachsample consisted of 200 capitula collected at random. To avoidany unconscious bias towards size, all fully-open capitula on eachsampled plant were collected before proceeding to the next plant.Infestation by insect larvae is not visible externally. All were col-lected in the summer of 1997 except for Hieracium pilosellawhich was sampled the following year. After collection, the sam-ples were stored in a freezer at –80°C for later dissection.

To investigate the relationship between capitulum size and in-cidence of insect attack, each capitulum was first cut in half verti-cally and its diameter measured at the widest point of the recepta-cle. The halves were then cut again to reveal the presence (or ab-sence) of any seed-feeding insects. Each capitulum was recordedas either infested or not infested. The number of larvae or the pro-portion of seeds affected within single capitula was not recorded.As the focus of the study was the effects of seed predation on theplants, the insect larvae themselves were not identified. The meancapitulum diameter and the mean percentage infestation was cal-culated for each species and the relationship between these vari-ables was tested by correlation analysis after logarithmic transfor-mation. To determine whether a similar relationship exists withinas well as among species, 600 additional capitula of three species(Arctium minus, Leucanthemum vulgare and Tripleurospermuminodorum) were collected from single locations and recorded inthe same way. The incidence of infestation was determined in eachof ten categories of capitulum size, each defined by intervals of10% along the percentiles of the capitulum size distribution (e.g.category 1=0–10th percentile, category 2=10–20th percentile, etc).Association between the incidence of infestation and capitulumsize category was tested by Spearman’s correlation analysis.

The degree of flowering synchrony within individual plantsand the lifespan of individual capitula were determined in a sub-set of eight species to investigate their relationship to the inci-dence of infestation. Between three and ten plants (depending onthe number of capitula per plant) were grown individually in flow-erpots (20 cm in diameter) outdoors. As each plant came intoflower, the “population dynamics” of all its capitula were moni-tored. As the capitula on each plant opened they were numberedby labelling with jewellers’ tags (10×5 mm) and their dates ofopening and seed dispersal recorded. This was carried out for thewhole flowering period of each plant. The total number of capitulaper species varied from 50 (Centaurea nigra) to 324 (Crepis ca-pillaris). From these data an index of flowering synchrony on in-dividual plants was devised as follows:

74

A plant whose capitula opened all at once (complete synchro-ny) would have an index value of 100%. Low percentages indicatea more staggered flowering phenology. Using the size and infesta-tion data in Table 1, relationships were sought between capitulumattributes (size, longevity, synchrony) and the incidence of infesta-tion. The capitulum attributes were an inter-correlated set, so weused partial correlation analysis (Field 2000) to test these pair-wise associations. All statistical analyses were conducted usingSPSS for Windows version 9.0 (SPSS, Chicago, Ill.).

Results

Table 1 shows the mean capitulum sizes and mean per-centage infestation in the 3 populations of each of the 20species, as well as their life-history type (annual, bienni-al or perennial). The largest and smallest mean capitu-lum diameters in the species studied differed by a factorof 6.8. Variation in capitulum size between populationsof the same species was small (5.8% of the mean, over-all). The incidence of infestation varies greatly amongspecies, with small-flowered species such as Achilleamillefolium being largely free of bud-infesting larvae,whilst large-flowered species such as Arctium minuswere heavily infested. The incidence of bud infestationincreases significantly with mean capitulum size (corre-lation analysis, r=0.68, df=18, P=0.001; Fig. 1). The ten-dency for larger capitula to have a higher level of infes-tation was also found to hold within the three speciestested: Arctium minus (Spearman’s correlation analysis,r=0.76, df=8, P<0.05); Leucanthemum vulgare (r=0.96,df=8, P<0.001); and Tripleurospermum inodorum(r=0.94, df=8, P<0.001: see Fig. 2).

Table 2 gives the mean lifespans of capitula (bud-break to seed shedding) of eight species plus their indi-ces of flowering synchrony. Both capitulum lifespan(range: 13.2–40.1 days) and synchrony (range:25.9–67.2%) varied among species. There was no detect-

Table 1 Mean capitulumwidths (mm) and mean percent-age of capitulum infestations in3 populations (sample size 200)of 20 species of Asteraceae.The species are in rank order ofcapitulum size. The ranges inparentheses give the two ex-treme values (A annual, B bien-nial, P perennial)

Species Life-form Mean Percent capitulum capitulum width (mm) infestation

1 Cirsium vulgare B 15.7 (±1.1) 12.0 (±12.0)2 Arctium minus B 11.1 (±1.05) 46.5 (±15)3 Leucanthemum vulgare P 11.0 (±0.15) 26.0 (±11.8)4 Pulicaria dysenterica P 7.4 (±0.35) 2.8 (±2.5)5 Centaurea nigra P 7.0 (±0.3) 12.7 (±4.0)6 Cirsium arvense P 6.2 (±0.45) 0.5 (±0.5)7 Tripleurospermum inodorum A 6.2 (±0.7) 33.0 (±12.3)8 Cirsium palustre B 6.0 (±0.6) 10.0 (±8.0)9 Taraxacum officinalea P 6.0 (±0.15) 2.0 (±1.5)

10 Hieracium pilosellaa P 4.9 (±0.25) 0.17 (±0.25)11 Matricaria recutita A 4.8 (±0.3) 30.5 (±4.5)12 Hypochaeris radicataa P 4.1 (±0.5) 5.2 (±3.8)13 Leontodon taraxacoidesa P 3.7 (±0.2) 2.0 (±2.5)14 Sonchus oleraceusa A 3.6 (±0.19) 0.17 (±0.25)15 Bellis perennis P 3.5 (±0.1) 2.2 (±2.5)16 Senecio jacobaea P 3.5 (±0.25) 1.3 (±1.3)17 Senecio vulgaris A 3.2 (±0.25) 0 (±0)18 Achillea millefolium P 2.5 (±0.15) 0 (±0)19 Crepis capillarisa A 2.4 (±0.05) 2.5 (±3.0)20 Lapsana communisa A 2.3 (±0.15) 0 (±0)

a Sub-family Cichorioideae;all others are in sub-familyCarduoideae

Fig. 1 Relationship between mean capitulum size (width of recep-tacle, mm) and mean level of infestation (% of capitula) in threepopulations of each of 20 species (r=0.68; P=0.001). The speciesare numbered as in Table 1

Table 2 Mean lifespan of individual capitula (days) ± SD (thenumber of capitula varied with species from 50 to 324); and meanindex of flowering synchrony (%) ± SD (the number of plants var-ied from 3 to 10)

Mean lifespan Mean synchrony(days) index (%)

Centaurea nigra 26.3±3.53 59.3±8.12Crepis capillaris 21.6±3.47 52.5±6.95Hypochaeris radicata 22.1±4.68 34.2±12.5Leucanthemum vulgare 34.2±13.4 64.0±5.88Pulicaria dysenterica 40.1±12.3 67.2±8.17Senecio jacobaea 27.5±2.87 54.6±6.70Sonchus oleraceus 13.2±3.06 25.9±5.68Tripleurospermum inodorum 28.6±5.27 50.3±13.7

75

able pairwise relationship between the incidence of in-festation and either capitulum lifespan (partial correla-tion =0.10, df=5, P>0.05) or synchrony (partial correla-tion =0.26, df=5, P>0.05). Capitulum size showed no as-sociation with either longevity (partial correlation =0.53,df=5, P>0.05), or synchrony (partial correlation =0.58,df=5, P>0.05).

The first hypothesis tested here, namely that specieswith larger capitula have a higher incidence of infesta-tion, is confirmed. The second hypothesis – that within apopulation of a single species, the larger capitula are

Fig. 2 Incidence of infestation (%) in capitula of different sizecategories (receptacle width, mm) in populations of three species:Artium minus (Spearman’s correlation analysis, r=0.76; P<0.05);Leucanthemum vulgare (r=0.96; P<0.001); and Tripleurospermuminodorum (r=0.94; P<0.001). Total samples of 600 capitula ineach species are arranged in 10 size categories, with 60 capitula ineach category

more likely to be attacked – is also confirmed, at leastfor the three species tested. The third and fourth hypoth-eses – that long-lived capitula and staggered floweringare associated with higher levels of attack – were notconfirmed.

Discussion

The results of this investigation indicate that larger capit-ula are more prone to infestation by insect larvae, bothamong and within species. This is probably due to thegreater apparency (sensu Feeny 1976) of larger flower-heads. The preference for larger capitula may represent amore economic foraging behaviour on the part of the in-sect because of the greater mass of food provided for agiven amount of search effort. In other studies, larger ca-pitula have been found to have more individual larvae,but not necessarily a greater number of different insectspecies (Zwölfer 1979, 1987; Zwölfer and Brandl 1989).

The possibility that capitulum size is an evolutionarycompromise between selection by pollinators and seedpredators depends on the pollinators showing a prefer-ence for larger flowers or inflorescences. This preferencehas been demonstrated in a number of species from arange of families, for example Phacelia linearis (Hy-drophyllaceae; Eckhart 1991), Corydalis ambigua (Papa-veraceae; Ohara and Higashi 1994), Raphanus raphani-strum (Brassicaceae; Conner and Rush 1996) and Jasmi-num fruticans (Oleaceae; Thompson 2001). Larger capit-ula attracting more pollinators may be beneficial if theindividuals thereby set more seed, or set higher qualityseed, or obtain greater paternity on other plants. Thoughall of the species in this study are regularly visited by in-sects, some may be independent of insects for pollina-tion, either because their seeds are produced by apomixis(e.g. Hieracium pilosella, Taraxacum officinale) or be-cause they are largely self-pollinated (e.g. Sonchus ole-raceus, Senecio vulgaris). Many can be self-pollinated ifinsect pollination fails (e.g. Arctium minus, Bellis pere-nnis, Lapsana communis, Leucanthemum vulgare). How-ever, others are largely self-incompatible (e.g. Achilleamillefolium, Centaurea nigra, Hypochaeris radicata, Tri-pleurospermum inodorum) and are therefore dependenton insects for cross pollination (details of the pollinationof these species are given by Grime et al. 1988). Apartfrom the apomictic species, most are likely to benefit byattracting insects that bring about cross pollination, andso it is reasonable to invoke selection for pollination suc-cess as a cause of capitulum evolution in the Asteraceae.

The index of flowering synchrony used here is similarto that used by Eriksson and Ehrlén (1991) to quantifyfruiting synchronization in individuals in a study of 34species of fleshy fruited plants. Other measures of flow-ering synchrony have been proposed that incorporate thestandard deviation of the onset of flowering (Gorchov1990; Bolmgren 1998). However, the index used herehas the advantage that it is also a measure of the meanpercentage of the total number of an individual’s flowers

76

that are open per day during its flowering period. Therange of values (from 25.9% to 67.2%) indicate that spe-cies differ markedly in this feature. A high degree ofsynchrony might be interpreted as a means of predatoravoidance (by means of a predator satiation effect), butin the event no relationship was found between flower-ing synchrony and degree of infestation.

The use of comparisons among groups of species is in-creasingly recognised as a powerful method for studyingevolutionary patterns (Futuyma 1998). The results ob-tained from investigations of this type may be biased bythe phylogenetic affinities among the species studied.However, in the present study, there is a reasonablespread of species between different genera and betweenthe two major subfamilies within the Asteraceae: 13 ofthe species belong to the Carduoideae, 7 to the Cichoriod-eae. The latter group are characterized by the presence oflatex within the tissues which may serve as a chemicaldefence against insect predation. It is notable that all themembers of Cichorioideae here (identified in Table 1)have very low levels of predispersal insect infestation, re-gardless of the size of their capitula. The presence of achemical defence in this group may make the capitulumsize-reduction strategy irrelevant for these species.

Future work to elucidate the relative roles of pollina-tion and seed predation in the evolution of capitulumsize in the Asteraceae should attempt to quantify boththe fitness penalties and the fitness benefits of larger ca-pitula. These studies should include an investigation ofthe levels of seed set in capitula of different sizes underfield conditions, both within and among species. Estima-tions of seed loss to predators could be refined by notingthe percentage loss within individual capitula as well asthe percentage of capitula infested. The importance ofthis is indicated by Zwölfer (1994) in his survey on arange of thistles. The role of inflorescence structure inboth pollination and seed predation could be investigatedby manipulating capitulum numbers in whole racemes. Acomplementary strategy would be the investigation ofevolutionary changes in populations of plants whichhave existed in the field in the absence of pre-dispersalseed predators for a number of generations. For example,many of the European Asteraceae that have become es-tablished as naturalised aliens in New Zealand over thelast 150 years may owe their success to the virtual ab-sence of the bud-infesting insect larvae that attack themin their native range (Fenner and Lee 2001). These popu-lations could thus provide a critical test of the idea thatpre-dispersal seed predators exert a selective influence infavour of smaller capitula because they might be expect-ed to have evolved larger flowers in response to selec-tion by the pollinators alone in recent generations. Theapomictic species would provide a crucial test of the the-ory because they would be expected to be unaffected bypollinator selection.

Acknowledgements The authors thank the British Ecological So-ciety for the award of a Small Ecological Project grant which part-ly funded this work. We also thank Raymond Cornick and MichaelCotton for technical assistance.

References

Bolmgren K (1998) The use of synchronization measures in stud-ies of plant reproductive phenology. Oikos 82:411–415

Brody AK (1992) Oviposition choices by a pre-dispersal seedpredator (Hylema sp.). 1. Correspondence with hummingbirdpollinators, and the role of plant sizes, density and floral mor-phology. Oecologia 91:56–62

Brody AK, Mitchell RJ (1997) Effects of experimental manipula-tion of inflorescence size on pollination and pre-dispersal seedpredation in the hummingbird-pollinated plant Ipomopsis ag-gregata. Oecologia 110:86–93

Conner JK, Rush S (1996) Effects of flower size and number onpollinator visitation to wild radish, Raphanus raphanistrum.Oecologia 105:509–516

Crawley MJ (1992) Seed predators and plant population dynamics.In: Fenner M (ed) Seeds: the ecology of regeneration in plantcommunities. CAB International, Wallingford, pp 157–191

Crawley MJ, Gillman MP (1989) Population dynamics of cinnabarmoth and ragwort in grassland. J Anim Ecol 58:1035–1050

Eckhart VM (1991) The effects of floral display on pollinator visi-tation vary among populations of Phacelia linearis (Hydrop-hyllaceae). Evol Ecol 5:370–384

Edwards PB, Brown EM (1997) Mesoclanis seed flies (Diptera:Tephritidae) on Chrysanthemoides (Asteraceae) in South Afri-ca: distribution, attack strategy and impact on seed production.Bull Entomol Res 87:127–135

Ehrlén J (1996) Spatiotemporal variation in predispersal seed pre-dation intensity. Oecologia 108:708–713

English-Loeb GM, Kabran R (1992) Consequences of variation inflowering phenology for seed head herbivory and reproductivesuccess in Erigeron glaucus (Compositae). Oecologia89:588–595

Eriksson O, Ehrlén J (1991) Phenological variation in fruit charac-teristics in vertebrate-dispersed plants. Oecologia 86:463–470

Feeny P (1976) Plant apparency and chemical defense. In: WallaceJW, Mansell RL (eds) Biochemical interactions between plantsand insects. Plenum Press, New York, pp 1–40

Fenner M, Lee WG (2001) Lack of pre-dispersal seed predators inintroduced Asteraceae in New Zealand. N Z J Ecol 25:95–99

Field A (2000) Discovering statistics using SPSS for Windows.Sage Publications, London

Futuyma DJ (1998) Evolutionary biology, 3rd edn. Sinauer, Sun-derland, Mass.

Gorchov DL (1990) Pattern, adaptation, and constraint in fruitingsynchrony within vertebrate-dispersed woody plants. Oikos58:169–180

Grime JP, Hodgson JG, Hunt R (1988) Comparative plant ecolo-gy: a functional approach to common British species. UnwinHymam, London

Hemborg AM, Després L (1999) Oviposition by mutualistic seed-parasitic pollinators and its effects on annual fitness of single-and multi-flowered host plants. Oecologia 120:427–436

Louda SM, Potvin MA (1995) Effect of inflorescence-feeding in-sects on the demography and lifetime fitness of a native plant.Ecology 76:229–245

Mattson WJ (1980) Cone resources and the ecology of the redpine cone beetle, Conophthorus resinosae (Coleoptera: Scolyt-idae). Ann Entomol Soc Am 73:390–396

Mays WT, Kok LT (1996) Establishment and dispersal of Uro-phora affinis (Diptera: Tephritidae) and Metzneria paucipunc-tella (Lepidoptera: Gelechiidae) in southwestern Virginia. BiolControl 6:299–305

Milton SJ (1995) Effects of rain, sheep and tephritid flies on seedproduction of two arid Karoo shrubs in South Africa. J ApplEcol 32:137–144

Muller-Scharer H, Schroeder D (1993) The biological control ofCentaurea spp. in North America: do insects solve the prob-lem? Pestic Sci 37:343–353

Ohara M, Higashi S (1994) Effects of inflorescence size on visitsfrom pollinators and seed set of Corydalis ambigua (Papave-raceae). Oecologia 98:25–30

77

Proctor M, Yeo P, Lack A (1996) The natural history of pollina-tion. Harper Collins, London

Randall MGM (1986) The predation of predispersed Juncussquarrosus seeds by Coleophora alticolella (Lepidoptera) lar-vae over a range of altitudes in northern England. Oecologia69:460–465

Schmid-Hempel P, Speiser B (1988) Effects of inflorescence sizeon pollination in Epilobium angustifolium. Oikos 53:98–104

Scott JK (1996) Population ecology of Chrysanthemoides monili-fera in South Africa: implications for its control in Australia. JAppl Ecol 33:1496–1508

Sheppard AW, Cullen JM, Aeschlimann JP (1994) Predispersalseed predation on Carduus nutans (Asteraceae) in southernEurope. Acta Oecol 15:529–541

Sobhian R, Zwölfer H (1985) Phytophagous insect species associ-ated with flower heads of yellow starthistle (Centaurea sols-titialis L.). J Appl Entomol 99:301–321

Straw NA (1991) Resource limitation of tephritid flies on lesserburdock, Arctium minus (Hill) Bernh (Compositae). Oecologia86:492–502

Thompson JD (2001) How do visitation patterns vary among pol-linators in relation to floral display and floral design in a gen-eralist pollination system? Oecologia 126:386–394

Turner CE, Piper GL, Coombs EM (1996) Chaetorellia australis(Diptera: Tephritidae) for biological control of yellow starthis-tle Centaurea solstitialis (Compositae) in the western USA:establishment and seed destruction. Bull Entomol Res86:177–182

Zwölfer H (1979) Strategies and counterstrategies in insect popu-lation systems competing for space and food in flower headsand plant galls. Fortschr Zool 25:331–353

Zwölfer H (1987) Species richness, species packing, and evolutionin insect-plant systems. Ecol Stud 61:301–319

Zwölfer H (1994) Structure and biomass transfer in food webs:stability, fluctuations, and network control. In: Schulze ED(ed) Flux control in biological systems from enzymes to popu-lations and ecosystems. Academic Press, London, pp 365–420

Zwölfer H, Brandl R (1989) Niches and size relationships in Cole-optera associated with Cardueae host plants: adaptations to re-source gradients. Oecologia 78:60–68

Zwölfer H, Harris P (1984) Biology and host specificity of Rhino-cyllus conicus (Froel.) (Col., Curculionidae), a successfulagent for biocontrol of the thistle, Carduus nutans L. J ApplEntomol 97:36–62

Zwölfer H, Preiss M (1983) Host selection and oviposition behav-iour in West-European ecotypes of Rhinocyllus conicus Froel.(Col., Curculionidae). J Appl Entomol 95:113–122