Generalised Pollination Systems for Three Invasive Milkweeds in Australia M

Plant Biology ISSN 1435-8603

RESEARCH PAPER Generalised pollination systems for three invasive milkweeds in Australia M. Ward1 & S. D. Johnson2 1 School of Biological Sciences, The University of Queensland, St Lucia, Qld, Australia 2 DST-NRF Centre for Invasion Biology, School of Biological and Conservation Sciences, University of KwaZulu-Natal, Pietermaritzburg, South Africa

Keywords ABSTRACT Asclepiadoideae; floral rewards; generalisation; invasive plants; pollination. Because most plants require pollinator visits for seed production, the ability of an introduced plant species to establish pollinator relationships in a new ecosystem may Correspondence have a central role in determining its success or failure as an invader. We investigated M. Ward, University of Queensland, School of the pollination ecology of three milkweed species – Asclepias curassavica, Gomphocar- Biological Sciences, St Lucia, Qld 4072, pus fruticosus and G. physocarpus – in their invaded range in southeast Queensland, Australia. Australia. The complex floral morphology of milkweeds has often been interpreted as E-mail: [email protected] a general trend towards specialised pollination requirements. Based on this interpreta- tion, invasion by milkweeds contradicts the expectation than plant species with spec- Editor ialised pollination systems are less likely to become invasive that those with more J. Arroyo generalised pollination requirements. However, observations of flower visitors in natural populations of the three study species revealed that their pollination systems are Received: 22 April 2012; Accepted: 10 Octo- essentially specialised at the taxonomic level of the order, but generalised at the spe- ber 2012 cies level. Specifically, pollinators of the two Gomphocarpus species included various species of Hymenoptera (particularly vespid wasps), while pollinators of A. curassav- doi:10.1111/j.1438-8677.2012.00700.x ica were primarily Lepidoptera (particularly nymphalid butterflies). Pollinators of all three species are rewarded with copious amounts of highly concentrated nectar. It is likely that successful invasion by these three milkweed species is attributable, at least in part, to their generalised pollinator requirements. The results of this study are dis- cussed in terms of how data from the native range may be useful in predicting pollination success of species in a new environment.

genetically self-compatible, are dependent on pollinators for INTRODUCTION reproductive output (e.g. van Kleunen & Johnson 2005; Jesse Invasive plant species are of critical concern on a worldwide et al. 2006; Liu et al. 2006; Hong et al. 2007; Lafuma & Maurice scale, and have consequently become the subject of intensive 2007; Gross et al. 2010; Ward et al. 2012). Furthermore, it has research (e.g. Henderson et al. 2006). However, a central ques- been documented that a lack of pollinator visitation can restrain tion that remains unanswered is: why do some introduced the rate of spread of invasive plant species (Parker 1997). plant species become successful invaders, whereas other species Introduced plant species generally arrive in a new ecosystem do not (e.g. Pysˇek & Richardson 2007)? In an attempt to without the pollinator species that serve them in their native understand factors that control the process of biological plant range, and consequently require formation of new relationships invasions, researchers have mostly focused on antagonistic with pollinator species that are already established in the local interactions between introduced plant species and organisms ecosystem (Richardson et al. 2000). Therefore, it is logical to in their new ecosystem, such as competition (e.g. Blossey & predict that introduced plant species with specialised pollina- No¨tzold 1995) and herbivory (e.g. Keane & Crawley 2002; tion systems are less likely to become invasive due to difﬁculties Wolfe 2002). However, data on positive interactions such as in establishment of pollinator relationships in their invaded reproductive mutualisms may be critical to provide insight into ecosystem, whereas introduced plant species that can be served how biological plant invasions are facilitated (Richardson et al. by numerous pollinator species have a greater chance of suc- 2000). Indeed, an introduced plant species will not be able to cessful reproduction (Baker 1974; Richardson et al. 2000). colonise new habitats unless reproductive barriers are over- Indeed, recent studies have shown that a number of invasive come. Therefore, an understanding of pollination ecology is plant species achieve reproduction through visits by a broad important for assessing the invasive potential of plant species. assemblage of pollinator species (e.g. Memmott & Waser 2002; Plant species that rely on sexual reproduction generally Jesse et al. 2006; Stout et al. 2006). Furthermore, other studies require pollinators, or are wind-pollinated or autonomously have demonstrated that pollinators may increase seed-set even self-fertilising. Although a large proportion of invasive plants in autonomously self-pollinating invasive species, thereby were originally expected to be autonomously self-fertilising enhancing the invasive ability of these species, despite their (Baker 1955, 1967, 1974; Stebbins 1957), there is growing evi- ability to set some seed in the absence of pollinators (e.g. Jacobi dence that many invasive plant species, even those that are et al. 2005; Bartomeus & Vila 2009; Rodger et al. 2010).

Production of abundant floral rewards is likely to enhance natural populations over peak flowering season from Decem- the ability of an invasive plant species to attract pollinators ber 2005 to April 2006. Study populations encompassed a vari- (Ghazoul 2002). It has been observed that invasive plant species ety of habitat types and population sizes, with six populations achieve this through characteristics such as prolific production surveyed for each of A. curassavica and G. physocarpus, and of flowers (Ghazoul 2002; van Kleunen & Johnson 2005), eight populations surveyed for G. fruticosus (see Table S1). secreting profuse volumes of nectar (Chittka & Schurkens Flower visitor observations involved recording the presence 2001) and making floral resources available over a prolonged of all visitors foraging on flowers during a standardised time period of time (Memmott & Waser 2002). When examining period of 30 min per observation event. Where possible, indi- pollination systems, it is relevant to note that different species vidual insects were closely observed in order to determine of pollinator often display variation in their importance for whether they deposited a pollinium into a stigmatic chamber pollination of a given plant species (e.g. see O’Brien 1980; Lind- of the visited flower. Insect visitors were then captured with a say 1984; Jennersten & Morse 1991; Ivey et al. 2003). The butterfly net and placed in ethyl acetate killing jars. Stigmatic importance of particular pollinators is determined by their rel- chambers of the visited flowers were subsequently inspected to ative abundance (i.e. common pollinator species may contrib- confirm the presence of deposited pollinia, as newly deposited ute more towards pollination than less common pollinator pollinia can generally be differentiated from old pollinia by species), as well as factors that influence their per-visit pollina- their brighter colour (M. Ward, personal observation). Insects tion effectiveness, such as pollen load. were later pinned and identified to species where possible, but Three milkweed species – Asclepias curassavica L., Gompho- otherwise to genus or family, and the number and position of carpus fruticosus (L.) W.T. Aiton and Gomphocarpus physocar- pollinia attached to the body of each insect recorded. Insects pus E. Mey. – have become invasive in Australia. All three were examined under magnification to determine whether any species are non-clonal, self-compatible and dependent on poll- residual corpuscula (i.e. corpuscula without pollinia) were inators for seed-set, and the two Gomphocarpus species have present. Voucher specimens of each insect species have been shown potential for hybridisation (Ward et al. 2012). Due to deposited in the University of Queensland Insect Collection. the characteristically highly evolved floral morphology of milk- Flower visitor observations were repeated at the various weeds, these species form an interesting case study of the polli- study populations at different times of the day throughout the nation systems of invasive plants. Pollen grains of milkweeds flowering season. Overall, 18 diurnal observation periods were are aggregated in discrete packages termed pollinia, which are conducted for A. curassavica (total of 540 min), 25 diurnal joined pair-wise to form a pollinarium. Each milkweed flower observation periods were conducted for G. fruticosus (total of holds five pollinaria, and a single pollinium contains enough 750 min) and 17 diurnal observation periods were conducted pollen grains to ensure full seed-set (Wyatt et al. 2000). Polli- for G. physocarpus (total of 510 min). The distribution of sam- nation is typically achieved when a pollinarium is removed by pling effort across populations is indicated in Table S1. Addi- a pollinator from the anther sac of the donor flower and, after tionally, two evening flower visitor observation events were reconfiguration as the pollinarium dries while attached to the conducted for each of G. fruticosus and G. physocarpus (total of pollinator, one of the pollinia is accurately inserted into an 60 min for each species) to investigate whether these white- internal stigmatic chamber of the recipient flower (Wyatt flowered species attract nocturnal insects. The relative impor- 1976). The specialised floral morphology and complex pollina- tance of each insect species that was captured was assessed in tion mechanism of milkweeds has often been interpreted as a terms of two measures of abundance, specifically: (i) the fre- general trend towards specialised pollination requirements quency of floral visits undertaken by a particular insect species (Wolff et al. 2008). Indeed, a compilation of pollination data as a percentage of the total number of floral visits by all insect for milkweeds documented specialised pollinator interactions species, to provide an assessment relative abundance of pollina- for 70% of the species examined, with 38% of the species hav- tor species; and (ii) the number of populations at which a par- ing only a single pollinator (Ollerton & Liede 1997). However, ticular insect species was captured, to provide an assessment of generalised pollination systems have also been documented in how widespread various pollinator species are. Furthermore, milkweed species (Ollerton & Liede 1997). rarefaction curves (Gotelli & Colwell 2001) were used to inves- Our study investigated the pollination ecology of the three tigate the relationship between sampling effort and number of afore-mentioned invasive milkweed species in southeast recorded visitors for each of the study species. Regression anal- Queensland, Australia, by determining: (i) the degree of ecolog- ysis was undertaken to examine the relationship between popu- ical specialisation in their pollination requirements, and (ii) the lation size (number of flowering individuals, log-transformed) relative importance of the various pollinator species in terms of and the mean rate of pollinator visitation (number of pollina- their relative abundance and pollinia loads. Furthermore, floral tors captured per observation period) for each of the study spe- rewards were quantified to gain an understanding of how cies. rewarding the study species are to pollinators, and floral morphology was measured to provide an assessment of the potential Floral rewards and floral morphology for overlap in pollinator species between the study species. To obtain an assessment of the floral morphology and floral rewards of the study species, 20 flowers per species were MATERIAL AND METHODS randomly selected and their nectar characteristics and floral dimensions measured. This was undertaken at one natural Flower visitors and pollinators population for each of the study species (Table S1). For the To determine the assemblages of pollinators for each of the nectar measurements, flowers were bagged with fine-mesh study species, flower visitor observations were conducted in cloth netting in order to exclude potential visits by pollina-

Plant Biology 15 (2013) 566–572 © 2012 German Botanical Society and The Royal Botanical Society of the Netherlands 567 Pollination of invasive milkweeds Ward & Johnson tors. After a 24-h nectar accumulation period, the volume A of nectar produced per flower was measured using a micro- litre syringe to extract nectar from each of the five coronal hoods comprising the flower. The sugar concentration of the extracted nectar was then determined using a hand-held Brix refractometer to measure sucrose equivalents. Floral dimensions measured included flower diameter, flower depth, petal length, corona diameter and corona depth. Additionally, the colours and scent of each individual were noted. To determine whether significant differences exist between species, analysis of variance (ANOVA) was applied to the morphological and nectar data, and post-hoc tests were performed using Tukey’s procedure for pair-wise compari- B sons. These analyses were undertaken using R version 2.14.1 (R Development Core Team 2011).

RESULTS Flower visitor assemblages Over the duration of the flowering season, a diverse range of insect species was observed foraging on milkweed flowers at various times throughout the day, with a total of 134 individuals captured during the study. No flower visitors were recorded during the nocturnal observation periods. Overall, 26 different insect species were captured, representing 12 families from four Fig. 1. Composition of insect assemblages recorded on the three invasive orders. A complete list of flower visitor species recorded on milkweed species in terms of (A) number of insect species visiting flowers each of the three invasive milkweed species is provided in Table and (B) their relative abundances. Data are presented as all flower visitors S2. captured during the study (V), as well as the sub-set of species that were Gomphocarpus fruticosus attracted the highest diversity of considered to be legitimate pollinators (P). flower visitors at the taxonomic levels of order and species, while A. curassavica attracted the least diverse flower visitor assemblage at both taxonomic levels (Fig. 1A). It is noteworthy that an analysis of the cumulative observations of new flower visitors over time indicated that an asymptote had been reached by the ninth observation period for A. curassavica (Fig. 2), indicating that the sampling had sufficiently encompassed the expected floral visitor diversity for this species. Dis- tinctive asymptotes were not observed for the two Gomphocarpus species (Fig. 2), suggesting that further sampling may have identified a higher diversity of insect species. With regard to the relative abundances of the various flower visitors, the largest proportion of visits to A. curassavica flowers were made by Lepidoptera, while the largest proportion of floral visits to the two Gomphocarpus species were made by Hymenoptera (Fig. 1B). Floral visits by Diptera comprised a Fig. 2. Relationships between sampling effort and the cumulative number very small component of all visits to both Gomphocarpus spe- of insect species recorded on A. curassavica (●), G. fruticosus (○) and cies, with Hemiptera also making a minor contribution to flo- G. physocarpus (□). ral visits of G. fruticosus (Fig. 1B).

Exceptions to the assumption that pollinia-carrying insects are Pollinator assemblages legitimate pollinators include the Diptera and Hemiptera, as A total of 81% of the captured flower visitor species were found these species were not observed to deposit pollinia. Other to carry pollinia. Observations in the field indicated that insects exceptions include two butterfly species (Lepidoptera, Nymp- that carried pollinia generally appeared to successfully deposit halidae) that did not have any residual corpuscula, and ant spe- pollinia within stigmatic chambers. The presence of residual cies (Hymenoptera, Formicidae) that are likely too small to be corpuscula on floral visitors added evidence to the assumption effective pollinators. Overall, approximately two-thirds of the that insect species that carry pollinia are legitimate pollinators. flower visitor assemblages for each of the three invasive milk- Furthermore, species typically carried pollinia on all or most weed species were considered to be legitimate pollinators occasions on which they were captured (Table S2), thereby sug- (Fig. 1A, Table S2), while the remainder of species were con- gesting that the presence of pollinia is not merely incidental. sidered to be opportunistic nectar thieves.

The pollinator assemblage of A. curassavica consisted of For all three study species, a high degree of variability was equal numbers of Hymenoptera and Lepidoptera species observed between populations in term of the pollinator species (Fig. 1A), although Lepidoptera visited A. curassavica flowers present, with most pollinator species captured at only one or more frequently than Hymenoptera (Fig. 1B). In particular, two populations (Table S2). With regard to the number of pol- Nymphalidae (Lepidoptera) were well represented within the linator species present at each population, A. curassavica popu- A. curassavica pollinator assemblage. Although present at only lations had either one or two pollinator species, while one population, Danaus petilia (Lepidoptera, Nymphalidae) G. fruticosus had up to six different pollinator species captured was the most frequently captured pollinator species on A. per population and G. physocarpus had up to seven different curassavica (Table S2, Fig. 3). Polistes humilis synoecus was the pollinator species captured per population. Body lengths of most widespread pollinator of A. curassavica (Table S2), and pollinators ranged from 15 to 25 mm for A. curassavica, and was the only species common to the pollinator assemblages of from 7 to 30 mm for the Gomphocarpus species. All pollinator all three milkweed species. The pollinator assemblages of the species carried pollinia on their tarsi, and pollinia loads ranged two Gomphocarpus species were comprised exclusively of from one pollinium per individual to a maximum of ten polli- Hymenoptera (Fig. 1), with four Hymenoptera species com- nia for A. currasavica, 12 pollinia for insects on G. fruticosus mon to both pollinator assemblages. Vespid wasps (Hymenop- and 15 pollinia for insects on G. physocarpus (Table S2). tera, Vespidae) accounted for the greatest representation in both pollinator assemblages in terms of the number of species Floral rewards and floral morphology as well as their relative abundances (Table S2), and Polistes humilis synoecus (Hymenoptera, Vespidae) was the most wide- All three study species produced large volumes of remarkably spread and most frequently captured pollinator for G. physo- concentrated nectar, with all nectar concentrations exceeding carpus and G. fruticosus (Table S2, Fig. 3). 60% sugar by mass (Table 1). Flowers of A. curassavica had the Overall, pollinators were recorded at three of the six A. cu- smallest mean nectar volume, while flowers of G. physocarpus rassavica populations, seven of the eight G. fruticosus popula- produced a mean nectar volume that was significantly larger tions and four of the six G. physocarpus populations. With than that produced by the two other species (Table 1). Simi- regard to abundances of pollinators at the various populations, larly, A. curassavica nectar had the lowest mean sugar concen- the mean number of pollinators recorded per observation per- tration, and the mean sugar concentration of G. physocarpus iod ranged from 0.33–3.0 for A. curassavica, 0.33–5.0 for nectar was significantly higher than that of A. curassavica G. fruticosus and 0.75–4.67 for G. physocarpus. A marginally (Table 1). significant relationship between population size and mean pol- Flowers of the two Gomphocarpus species were generally lar- linator visitation rate was observed for G. fruticosus (R2 = 0.39, ger than those of A. curassavica, with flowers of G. physocarpus P < 0.059, df = 8), with larger populations attracting more larger than those of G. fruticosus (Table 1). Comparisons of pollinators than smaller populations; whereas no relationship the morphological characteristics revealed that all three study was detected between population size and mean pollinator visi- species were significantly different from each other in terms of tation rate for A. curassavica (R2 = 0.15, P < 0.451, df = 5) or flower diameter, petal length and corona diameter, while dif- G. physocarpus (R2 = 0.01, P < 0.900, df = 5). ferences in flower depth were only significant between A.

A C

Fig. 3. Abundant pollinator species recorded during the study included (A) Danaus petilia on A. curassavica, (B) Polistes humilis synoecus on G. physocarpus and (C) Lophocheilus sp. on G. fruticosus. Scale bar = 10 mm.

Plant Biology 15 (2013) 566–572 © 2012 German Botanical Society and The Royal Botanical Society of the Netherlands 569 Pollination of invasive milkweeds Ward & Johnson curassavica and G. physocarpus, and no significant differences dence that generalisation in reproductive mutualisms is a sig- between species were detected for corona depth (Table 1). Flo- nificant facilitator of biological invasions (sensu Richardson ral scent was completely absent in the red-flowered A. curas- et al. 2000). Nevertheless, it is important to acknowledge that savica, whereas the two white-flowered Gomphocarpus species other reproductive characteristics are also important in facili- both had sweet, honey-like scents (Table 1). tating biological invasions, particularly breeding system attri- butes (see Ward et al. 2012). In accordance with the expectation that small populations are less attractive to pollina- DISCUSSION tors than are large, showy populations (A˚ gren 1996; Morgan Diverse assemblages of insects were attracted to flowers of the 1999), a positive relationship between population size and pol- three invasive milkweed species, with this likely due to the linator visitation was detected for G. fruticosus. The lack of accessibility of large volumes of concentrated nectar in shallow such a relationship for A. curassavica and G. physocarpus is coronal cups. The success of the three study species in attract- likely attributed to sampling effort. ing pollinators is evidenced by high levels of fruit-set in the Consistent with our results, several other studies provide invasive range, with even isolated single plants being suffi- support for the original expectation that invasive species ciently rewarding to attract pollinators and produce abundant should have pollinator assemblages comprised of several polli- fruit (M. Ward, unpublished data). Floral scent appears to be nator species (e.g. Memmott & Waser 2002; Jesse et al. 2006; important for pollinator attraction by the two Gomphocarpus Liu et al. 2006; Stout et al. 2006). In a recent study that is par- species, while A. curassavica appears to primarily rely on flower ticularly relevant here, Herrera & Nassar (2009) showed that colour for pollinator attraction. Although most floral morphol- the South African asclepiad Stapelia gigantea reproduces prolif- ogy parameters differed between the study species (see ically in its invaded range in Venezuela by virtue of a fly polli- Table 1), two Polistes species are able to pollinate all three nation system that is highly generalised at the species level, yet milkweed species, indicating that the differences in floral mor- specialized at the order level. In cases where relatively specia- phology are not sufficient to inhibit overlap in pollinator spe- lised pollination systems have been documented for invasive cies between study species. plant species, invasion has typically been facilitated by the Pollinators of the two Gomphocarpus species included vari- apparent functional equivalence of pollinators in the native ous species of Hymenoptera (particularly vespid wasps), while and introduced ranges (e.g. Stout et al. 2002; Rodger et al. pollinators of A. curassavica were predominantly Lepidoptera 2010). Since pollinator functional groups (pollinators that (particularly nymphalid butterflies). The results of this study behave in similar ways on a flower and exert similar selection show that the pollinator assemblages of the study species were pressures, sensu Fenster et al. 2004) can be distributed across observed to be essentially specialised at the taxonomic level of continents, specialisation for pollination by one of these func- order – and family to a degree – but generalised at the species tional groups may less frequently inhibit biological invasions level. The generalisation of the pollination system at the species than might be expected according to Baker’s (1974) original level is evidenced through the high number of pollinator spe- predictions. cies recorded, the lack of obvious saturation in the pollinator Despite the apparent generalisation in the pollination sys- species rarefaction curves for the two Gomphocarpus species, tems of our study species within the level of pollinator func- the variation in pollinator species between populations, and tional groups, data on functional groups from the native range the large range of pollinator body lengths. It is likely that suc- may be useful in predicting pollination success of species in a cessful invasion by these three milkweed species in Australia is new environment. This is because the absence of an entire pol- attributable, at least in part, to their generalised pollinator linator functional group in the new environment may limit requirements. Therefore, this study adds to the growing evi- plant invasions and, conversely, its presence may promote invasion. The pollinator assemblage of G. physocarpus in its native range in South Africa was recently investigated (Coombs Table 1. Floral morphology measurements of the study species. et al. 2009). As found in our study (and see Forster 1994), the characteristic A. curassavica G. fruticosus G. physocarpus assemblage of pollinators was dominated by Hymenoptera, especially numerous vespid wasp species (Coombs et al. 2009). nectar volume (ll) 6.2 ± 2.6a 8.2 ± 1.7a 11.2 ± 2.6b It is particularly noteworthy that South African pollinators nectar 61.1 ± 7.1a 66.4 ± 8.7a,b 72.7 ± 9.1b included several Polistes species, and that P. humilis synoecus concentration (%) was documented in the current study to be one of the most flower diameter (mm) 16.7 ± 1.2a 18.9 ± 0.7b 20.6 ± 0.8c important pollinators of G. physocarpus in the invaded south- flower depth (mm) 12.5 ± 1.3a 13 ± 1.4a,b 14.0 ± 1.9c east Queensland range. Vespid wasp assemblages are found on a b c petal length (mm) 9.7 ± 0.7 8.9 ± 0.9 10.8 ± 0.8 every continent except Antarctica, meaning that a plant a b c corona 7.6 ± 0.6 8.2 ± 0.4 9.3 ± 0.7 adapted for pollination by this functional group would diameter (mm) encounter few pollination barriers to establishment outside its ± a ± a ± a corona depth (mm) 5.0 0.7 4.4 0.5 4.7 0.7 native range. The wasp-dominated pollination system of petal colour red white to cream white to cream G. physocarpus appears to be highly generalised at the species corona colour bright yellow cream to green cream to purple level, even in the native range, and this facilitated the successful scent absent faint, sweet strong, sweet establishment of mutualisms involving functionally equivalent Values are presented as means ± 1SD(n= 20). Hymenoptera in the introduced range. Similarly, our observa- Different superscript letters represent significant differences based on post- tion that A. curassavica is predominantly pollinated by butter- hoc Tukey tests for pair-wise comparisons of morphological and nectar char- flies in Australia is consistent with reports of butterfly acteristics. pollination of this species in the native range in Costa Rica

(Wyatt 1980; Bierzychudek 1981) and Brazil (Fuhro et al. further case studies, it is anticipated that generalisations 2010). As the pollination of G. fruticosus has not yet been stud- regarding the biological characteristics of invasive plants can be ied in its native range in southern Africa, we are not able to developed and incorporated into screening attribute profiles establish if populations of G. fruticosus in the native range use for species prior to their introduction. the same functional pollinator groups that we observed in Aus- tralia. ACKNOWLEDGEMENTS In conclusion, this study contributes evidence that supports the prediction that introduced plant species that have general- Appreciative thanks to Greg Daniels, curator of the University ised pollination systems are able to overcome some of the of Queensland Insect Collection, for assistance with insect reproductive barriers to invasion. Interestingly, the complex identification. The authors thank two anonymous reviewers pollination mechanism of the three milkweed species has evi- who provided comments on a previous version of the manu- dently not inhibited establishment of pollinator interactions in script. Funding for this research was provided by CSIRO Sus- the invasive range. The quest to determine why certain species tainable Ecosystems. become successful invaders while others do not is complicated, but is likely to benefit greatly from further case studies that SUPPORTING INFORMATION investigate the pollination ecology of invasive species. In particular, investigating the pollination systems of other milkweed Additional Supporting Information may be found in the online species that have been introduced to a new range but have not version of this article: become invasive would provide valuable information. With Table S1. Details of populations at which floral visitor obser- regard to whether generalisations concerning wasp pollination vations were undertaken, including their location, population systems can be made, it is noteworthy that studies of other size in terms of the number of flowering plants and habitat native milkweed species in South Africa have shown that most type. species are typically pollinated by only one or two pompilid Table S2. Floral visitors recorded on the three study species, wasp species (Shuttleworth & Johnson 2012), therefore sug- including details regarding their distribution, relative abun- gesting that other wasp systems may not be as generalised as dances, pollinia loads and body size. the vespid wasp pollination systems reported here. Following

REFERENCES Asclepias curassavica (Apocynaceae), Epidendrum (Rosaceae), in Iowa, USA. Weed Biology and Man- fulgens (Orchidaceae), and Lantana camara (Verben- agement, 6, 235–240. A˚ gren J. (1996) Population size, pollinator limitation, aceae) in Southern Brazil? Revista Brasileira de Botaˆ- Keane R.M., Crawley M.J. (2002) Exotic plant inva- and seed set in the self-incompatible herb Lythrum nica, 33, 589–598. sions and the enemy release hypothesis. Trends in salicaria. Ecology, 77, 1779–1790. Ghazoul J. (2002) Flowers at the front line of invasion? Ecology & Evolution, 17, 164–170. Baker H.G. (1955) Self compatibility and establishment Ecological Entomology, 27, 638–640. van Kleunen M., Johnson S.D. (2005) Testing for eco- after long distance dispersal. Evolution, 9, 347–349. Gotelli N.J., Colwell R.K. (2001) Quantifying biodiver- logical and genetic allee effects in the invasive shrub Baker H.G. (1967) Support for Baker’s law – as a rule. sity: procedures and pitfalls in the measurement and Senna didymobotrya (Fabaceae). American Journal of Evolution, 21, 853–856. comparison of species richness. Ecology Letters, 4, Botany, 92, 1124–1130. Baker H.G. (1974) The evolution of weeds. Annual 379–391. Lafuma L., Maurice S. (2007) Increase in mate avail- Review of Ecology and Systematics, 7,1–24. Gross C.L., Gorrell L., MacDonald M.J., Fatemi M. ability without loss of self-incompatibility in the Bartomeus I., Vila M. (2009) Breeding system and pol- (2010) Honeybees facilitate the invasion of Phyla invasive species Senecio inaequidens (Asteraceae). Oi- len limitation in two supergeneralist alien plants canescens (Verbenaceae) in Australia – no bees, no kos, 116, 201–208. invading Mediterranean shrublands. Australian Jour- seed! Weed Research, 50, 364–372. Lindsay A.H. (1984) Reproductive biology of Apiaceae nal of Botany, 57, 109–115. Henderson S., Dawson T.P., Whittaker R.J. (2006) Pro- I. Floral visitors to Thaspium and Zizia and their Bierzychudek P. (1981) Asclepias, Lantana, and Epiden- gress in invasive plants research. Progress in Physical importance in pollination. American Journal of Bot- drum: a floral mimicry complex? Biotropica, 13,54– Geography, 30,25–46. any, 71, 375–387. 58. Herrera I., Nassar J.M. (2009) Reproductive and Liu H., Pemberton R.W., Stiling P. (2006) Native and Blossey B., No¨tzold R. (1995) Evolution of increased recruitment traits as indicators of the invasive introduced pollinators promote a self-incompatible competitive ability in invasive non-indigenous potential of Kalanchoe daigremontiana (Crassula- invasive woody vine (Paederia foetida L.) in Florida. plants: a hypothesis. Journal of Ecology, 83, 887–889. ceae) and Stapelia gigantea (Apocynaceae) in a neo- The Journal of the Torrey Botanical Society, 133, 304– Chittka L., Schurkens S. (2001) Successful invasion of a tropical arid zone. Journal of Arid Environments, 73, 311. floral market – an exotic Asian plant has moved in 978–986. Memmott J., Waser N.M. (2002) Integration of alien on Europe’s river-banks by bribing pollinators. Nat- Hong L., Shen H., Ye W.H., Cao H.L., Wang Z.M. plants into a native flower–pollinator visitation web. ure, 411, 653. (2007) Self-incompatibility in Mikania micra- Proceedings of the Royal Society of London Series B – Coombs G., Peter C.I., Johnson S.D. (2009) A test for ntha in South China. Weed Research, 47, 280–283. Biological Sciences, 269, 2395–2399. Allee effects in the self-incompatible wasp-pollinated Ivey C.T., Martinez P., Wyatt R. (2003) Variation in Morgan J.W. (1999) Effects of population size on seed milkweed Gomphocarpus physocarpus. Austral Ecol- pollinator effectiveness in swamp milkweed, Asclepi- production and germinability in an endangered, ogy, 34, 688–697. as incarnata (Apocynaceae). American Journal of fragmented grassland plant. Conservation Biology, Fenster C.B., Armbruster W.S., Wilson P., Dudash Botany, 90,214–225. 13, 266–273. M.R., Thomson J.D. (2004) Pollination syndromes Jacobi C.M., Ramalho M., Silva M. (2005) Pollination O’Brien M.H. (1980) The pollination biology of a and floral specialization. Annual Review of Ecology, biology of the exotic rattleweed Crotalaria retusa L. pavement plain: pollinator visitation patterns. Oeco- Evolution and Systematics, 35, 375–403. (Fabaceae) in NE Brazil. Biotropica, 37, 357–363. logia, 47, 213–218. Forster P.I. (1994) Diurnal insects associated with the Jennersten O., Morse D.H. (1991) The quality of polli- Ollerton J., Liede S. (1997) Pollination systems in the flowers of Gomphocarpus physocarpus E. Mey. nation by diurnal and nocturnal insects visiting Asclepiadaceae: a survey and preliminary analysis. (Asclepiadaceae), an introduced weed in Australia. common milkweed, Asclepias syriaca. American Mid- Biological Journal of the Linnean Society, 62, 593–610. Biotropica, 26, 214–217. land Naturalist, 125,18–28. Parker I.M. (1997) Pollinator limitation of Cytisus Fuhro D., de Arau´ jo A.M., Irgang B.E. (2010) Are there Jesse L.C., Moloney K.A., Obrycki J.J. (2006) Insect scoparius (Scotch broom), an invasive exotic shrub. evidences of a complex mimicry system among pollinators of the invasive plant, Rosa multiflora Ecology, 78, 1457–1470.

Pysˇek P., Richardson D.M. (2007) Traits associated ized plant guild. Botanical Journal of the Linnean Wolfe L.M. (2002) Why alien invaders succeed: sup- with invasiveness in alien plants: where do we stand? Society, 168, 278–299. port for the escape-from-enemy hypothesis. Ameri- In: Nentwig W. (Ed.), Biological invasions. Springer, Stebbins G.L. (1957) Self fertilization and population can Naturalist, 160, 705–711. Berlin, Germany, pp 97–125. variability in the higher plants. The American Natu- Wolff D., Meve U., Liede-Schumann S. (2008) Pollina- R Development Core Team (2011) R: a language and ralist, 91, 337–354. tion ecology of Ecuadorian Asclepiadoideae (Apo- environment for statistical computing. R Foundation Stout J.C., Kells A.R., Goulson D. (2002) Pollination of cynaceae): how generalized are morphologically for Statistical Computing, Vienna, Austria. URL the invasive exotic shrub Lupinus arboreus (Faba- specialized ﬂowers? Basic and Applied Ecology, 9,24– http://www.R-project.org. ceae) by introduced bees in Tasmania. Biological 34. Richardson D.M., Allsopp N., D’Antonio C.M., Milton Conservation, 106, 425–434. Wyatt R. (1976) Pollination and fruit-set in Asclepias:a S.J., Rejma´nek M. (2000) Plant invasions – the role Stout J.C., Parnell J.A.N., Arroyo J., Crowe T.P. (2006) reappraisal. American Journal of Botany, 63, 845–851. of mutualisms. Biological Reviews, 75,65–93. Pollination ecology and seed production of Rhodo- Wyatt R. (1980) The impact of nectar-robbing ants on Rodger J.G., van Kleunen M., Johnson S.D. (2010) dendron ponticum in native and exotic habitats. Bio- the pollination system of Asclepias curassavica. Bulle- Does specialised pollination impede plant invasions. diversity and Conservation, 15, 755–777. tin of the Torrey Botanical Club, 107,24–28. International Journal of Plant Sciences, 171, 382–391. Ward M., Johnson S.D., Zalucki M.P. (2012) Modes of Wyatt R., Broyles S.B., Lipow S.R. (2000) Pollen– Shuttleworth A., Johnson S.D. (2012) The Hemipepsis reproduction in three invasive milkweeds are consis- ovule ratios in milkweeds (Asclepiadaceae): an wasp-pollination system in South Africa: a compara- tent with Baker’s Rule. Biological Invasions, 14, 1237 exception that probes the rule. Systematic Botany, tive analysis of trait convergence in a highly special- –1250. 25, 171–180.