Temporal Patterns in Pollination of Deceptive Aristolochia Rotunda L

Plant Biology ISSN 1435-8603

RESEARCH PAPER Spatio-temporal patterns in pollination of deceptive Aristolochia rotunda L. (Aristolochiaceae) B. Oelschlagel€ 1, M. von Tschirnhaus2, M. Nuss3, T. Nikolic4, S. Wanke1,†,S.Dotterl€ 5,† & C. Neinhuis1,† 1 Institut fur€ Botanik, Technische Universitat€ Dresden, Dresden, Germany 2 Fakultat€ Biologie, Universitat€ Bielefeld, Bielefeld, Germany 3 Senckenberg Naturhistorische Sammlungen Dresden & Museum fur€ Tierkunde, Dresden, Germany 4 Department of Botany, Faculty of Science, University of Zagreb, Zagreb, Croatia 5 Department of Ecology & Evolution, University of Salzburg, Salzburg, Austria

Keywords ABSTRACT Aristolochia rotunda; autogamy; Chloropidae; deceptive pollination; floral life span; flower biology; pollinator variability. • Pollination success of highly specialised flowers is susceptible to fluctuations of the pollinator fauna. Mediterranean Aristolochia rotunda has deceptive trap flowers Correspondence exhibiting a highly specialised pollination system. The sole pollinators are kleptopara- B. Oelschlagel,€ Institut fur€ Botanik, Technische sitic flies in search of food. This study investigates these pollinators on a spatio-tem- Universitat€ Dresden, Zellescher Weg 20b, poral scale and the impact of weather conditions on their availability. Two potential 01062 Dresden, Germany. strategies of the plants to cope with pollinator limitation, i.e. autonomous selfing and E-mail: [email protected] an increased floral life span, were tested. • A total of 6156 flowers were investigated for entrapped pollinators in 10 Croatian † Authors contributed equally to this study. populations. Availability of the main pollinator was correlated to meteorological data. Artificial pollination experiments were conducted and the floral life span was recorded Editor in two populations according to pollinator availability. A. Dafni • Trachysiphonella ruficeps (Chloropidae) was identified as dominant pollinator, along with less abundant species of Chloropidae, Ceratopogonidae and Milichiidae. Pollina- Received: 16 March 2016; Accepted: 22 tor compositions varied among populations. Weather conditions 15–30 days before August 2016 pollination had a significant effect on availability of the main pollinator. Flowers were not autonomously selfing, and the floral life span exhibited considerable plasticity doi:10.1111/plb.12503 depending on pollinator availability. • A. rotunda flowers rely on insect pollen vectors. Plants are specialised on a guild of kleptoparasitic flies, rather than on a single species. Pollinator variability may result in differing selection pressures among populations. The availability/abundance of pollinators depends on weather conditions during their larval development. Flowers show a prolonged trapping flower stage that likely increases outcrossing success during periods of pollinator limitation.

short-term weather conditions (Dunkeloh€ & Jacobeit 2003; INTRODUCTION Martin-Vide & Lopez-Bustins 2006). It has been reported that Deceptive pollination has evolved several times independently weather conditions during flowering influence the frequency of in flowering plants (Proctor et al. 1996; Renner 2006). Decep- pollinator visits and therefore pollination success (Baker et al. tive flowers advertise a reward to their pollinators that they do 2000; Jacquemyn et al. 2009). Such strong variability of polli- not provide in the end, and exploit the instinctive behaviour of nator abundance has earlier been shown for Mediterranean often non-typical flower visitors (e.g. Brodmann et al. 2008; Aristolochia species (Berjano et al. 2009). Stokl€ et al. 2010). This results in highly specialised pollination The genus Aristolochia (Aristolochiaceae) comprises over that is usually mediated by only one to a few pollinator species 450 species and is distributed in the tropics and in temperate (Dafni 1984; Proctor et al. 1996). Highly specialised plants (in- regions (Wagner et al. 2014). In the Mediterranean about 60 dependent of whether deceptive or not) benefit from highly species are known, representing a diversity hotspot for the specific pollen delivery and high outcrossing rates compared to genus (Wanke 2007). Virtually all Aristolochia species exhibit less specialised rewarding species (Scopece et al. 2010). How- deceptive pollination strategies and it was assumed that flowers ever, they are counter-balanced by a high dependence on spa- mimic the brood sites of their fly pollinators (Vogel 1978; tiotemporal fluctuation of the pollinator fauna (Waser et al. Proctor et al. 1996). However, detailed knowledge on the 1996; Armbruster et al. 2000) and overall lower reproductive mechanism of pollinator attraction and deception is presently success (Scopece et al. 2010). The Mediterranean climate is only available for A. rotunda. A. rotunda flowers are klepto- characterised by strong seasonality and high variability of myiophilous, a pollination strategy recently described

928 Plant Biology 18 (2016) 928–937 © 2016 German Botanical Society and The Royal Botanical Society of the Netherlands Oelschlagel,€ Tschirnhaus, Nuss, Nikolic, Wanke, Dotterl€ & Neinhuis Spatio-temporal patterns in pollination

(Oelschl€agel et al. 2015). Representatives of Chloropidae and populations and years. The impact of weather conditions on Ceratopogonidae were found pollinating this species A. rotunda’s highly specialised pollination system is discussed (Oelschl€agel et al. 2015), and members of these families are and potential mechanisms promoting reproductive success known kleptoparasites, stealing food of invertebrate predators during pollinator limitation are evaluated. With respect to the (e.g. Sivinski et al. 1999). The flowers of A. rotunda employ an highly specialised pollinator attraction we test the following olfactory mimicry of the heteropteran prey to attract and hypothesis: (i) A. rotunda flowers are pollinated by the same deceive their kleptoparasitic pollinators (Oelschl€agel et al. pollinator species in different populations; (ii) pollinator avail- 2015). Due to this particular attraction mechanism, A. rotunda ability is strongly influenced by weather conditions during is a highly specialised plant species that could make the plant’s flowering of A. rotunda, and unfavourable weather conditions, reproductive success susceptible to fluctuations in pollinator such as high precipitation or low temperatures, result in peri- fauna. ods with pollinator absence; (iii) autonomous selfing is a strat- Known strategies of plants to cope with pollinator limitation egy of A. rotunda to ensure reproductive success in periods of and increase reproductive fitness are autonomous selfing and pollinator limitation; and (iv) flowers of A. rotunda show a increased floral longevity (e.g. Ashmann & Schoen 1994, 1996; prolonged trapping flower stage (female stage and interphase) Fenster & Marten-Rodrıguez 2007; Castro et al. 2008). as compared to tropical Aristolochia species. Although flowers of Aristolochia show strong adaptations for cross-pollination, the ability for delayed autonomous selfing has been proven for several of its species (Razzak et al. 1992; MATERIAL AND METHODS Hall & Brown 1993; Sakai 2002; Berjano et al. 2006; Bliss et al. Sampling sites 2013). However, it is yet unknown if A. rotunda is outcrossing or selfing. Aristolochia flowers possess an elaborated perianth Aristolochia rotunda is endemic to the northwestern Mediter- morphology evolved for trapping, retention and release of pol- ranean region and occurs from western Turkey to eastern linators (Correns 1891; Oelschl€agel et al. 2009; Fig. 1A). The Spain. It colonises damp grassy areas in woodlands, on river- perianth consists of three main parts: the limb, the tube and banks and boulders (Nardi 1984, 1991). The study was the utricle that contains the gynostemium. The flowers show performed at ten localities along the Croatian coast (Fig. 1G): pronounced protogyny, with functional adaptations in the dif- Dragonja (45°27051.7″ N, 013°37020.3″ E, 0 m a.s.l.), Most Rasa ferent sexual stages. Erect and receptive stigmas, closed pollen (45°03046.5″ N, 014°02044.4″ E, 2 m a.s.l.) both Istria; sacks (Fig. 1B) and functional trapping devices characterise the Caska (44°32059.7″ N, 014°55016.3″ E, 2 m a.s.l.), Dinjisko female flower stage (i.e. bendable trichomes inside the flower polje (44°23.5940 N, 015°07.9120 E, 73 m a.s.l.), Mt. Sveti tube; Fig. 1C). Pollinators are attracted and trapped during this Vid (44°30045.6″ N, 014°57025.6″ E, 40 m a.s.l.), Stari stage. During the interphase stigmas wilt and bend together grad (44°25048″ N, 015°03052″), Vlasici’ (44°19021.5″ N, while pollen sacs open (Fig. 1D). The trapping devices remain 015°12031.7″ E, 20 m a.s.l.) all Isle of Pag; Omisalij lake turgescent, enabling the flower to cover the trapped flies with (45°10022.3″ N, 14°34027.3″ E 6 m a.s.l., Isle of Krk), Veli its pollen. During the male flower stage, the trapping devices Losinj (44°31017.1″ N, 014°30059.8″ E, 5 m a.s.l., Isle of Los- wilt (Fig. 1E) and enable the flies, loaded with pollen, to leave inje) and Dracevac Ninski (44°10054.9″ N, 015°18036.6″ E, the flower (Correns 1891; Daumann 1971; Gonzalez & Steven- 49 m a.s.l., Ravni Kotari, Croatia mainland coast). Climate of son 2000; Oelschl€agel et al. 2009). the study region corresponds to the typical seasonal climate of Stotz & Gianoli (2013) hypothesised an adaptation of the the Mediterranean, with dry summer and wet winter seasons. floral longevity within the genus to pollinator limitation factors. Species in habitats with low net primary productivity and Collection of flower visitors and pollinators, pollinator low precipitation that are more affected by pollinator limita- identification tion evolved longer-living flowers compared to species in habitats of high net primary productivity and high precipitation The different stages of anthesis were determined in all investi- (Stotz & Gianoli 2013). Longer floral lifespan increases floral gated flowers based on the current state of the gynostemium fitness as longer-lived flowers have a higher likelihood for polli- and the turgescence of trapping trichomes. Insects that carried nator visitation than shorter-lived flowers. In contrast to tropi- pollen and were trapped in a female stage flower must have pre- cal Aristolochia species with a typical 1 day female flower stage viously been trapped in another flower. Only these insects have (Cammerloher 1923; Hall & Brown 1993; Sakai 2002; Burgess proven to repeatedly visit Aristolochia flowers and thus are et al. 2004; Trujillo & Sersic 2006), a prolonged duration of the potential pollinators (Rulik et al. 2008). Therefore, only arthro- female flower stage has been observed in A. baetica (3.5– pods found in the utricle of female stage flowers were subject to 4.2 days), A. chilensis (2–3 days) and A. paucinervis (5.3 days), subsequent analysis. Potential pollinators, henceforth for sim- all of which occur in Mediterranean and/or arid climates (Ber- plicity of reading called ‘pollinators’, were discriminated from jano et al. 2009; Stotz & Gianoli 2013). A prolonged female accidental flower visitors through the presence of a pollen load. flower stage may be constrained by an increase in insect mor- Flowers were collected during annual field trips in May tality within the flowers, especially in hot and dry climates 2009–2012, the main flowering period of A. rotunda. In 2009 (Stotz & Gianoli 2013). Besides the matter of whether the to 2011 collections were conducted once per population and flower is pollinated or not, the maximum duration of the trap- year; in 2012 populations were sampled twice or three times to ping flower stage is therefore also likely limited by the surviva- observe changes in pollinator abundance during the flowering bility of the pollinators (Stotz & Gianoli 2013). period. Flowers were either frozen and dissected during the The present study focuses on the pollinators of A. rotunda field trip or preserved in 100% ethanol for later laboratory (Fig. 1F) and their spatio-temporal variability among multiple investigation. All pollinators were identified to species level,

Plant Biology 18 (2016) 928–937 © 2016 German Botanical Society and The Royal Botanical Society of the Netherlands 929 Spatio-temporal patterns in pollination Oelschlagel,€ Tschirnhaus, Nuss, Nikolic, Wanke, Dotterl€ & Neinhuis

(A) (B) (C)

(D) (E)

(F) (G)

Fig. 1. Aristolochia flower morphology and study sites. A: Floral morphology and characteristics of the female and male flower stage. B: Female stage gynostemium showing receptive stigmata and closed pollen sacs. C: Trapping trichomes at the junction between tube and utricle during the female flower stage. D: Male stage gynostemium showing dried stigmata and open pollen sacs. E: Wilted trapping trichomes at the junction of tube and utricle in the male flower stage. F: Flower of A. rotunda. G: Study sites in Croatia (dots) and position of meteorological stations used (asterisks). a – anther, g – gynostemium, l – limb, p – peduncle, s – stigma, t – tube, tr – trapping trichome, u – utricle. Images (C) and (E) are from Oelschlagel€ et al. (2009).

and in the case of Ceratopogonidae to family level (Collin Monthly precipitation was compared to a reference period 1946; Dely-Draskovits 1981; Beshovski 1985; Narchuk et al. (1961–2000) using the online facility of DHMZ (http://kli- 1989; Oosterbroek 2006; Nartshuk & Andersson 2013). To con- ma.hr/klima_e.php?id=SPI; DHMZ starting page – Climate – firm that pollen load was indeed Aristolochia pollen, 49% of all Drought monitoring, accessed 10/23/2014). pollen samples were subject to scanning electron microscopy using a Supra 40 VP SEM (Carl Zeiss, Oberkochen, Germany). Coupling meteorological data with pollinator availability During our fieldwork we observed high variability in pollina- Meteorological data tor availability and abundance during the flowering period in The Meteorological and Hydrological Service Croatia (DHMZ) May as well as among years. The observed variation could be provided meteorological data for April and May 2009–2012. potentially explained by fluctuations in weather conditions. Daily data of mean, minimum and maximum temperature, To test for a possible correlation, meteorological data were precipitation, sunshine hours and wind velocity were collected analysed in dependence to temporary availability of the main from two meteorological stations close to the study localities: pollinator, Trachysiphonella ruficeps (Macquart, 1835) station Zadar (44°080 N, 15°130 E; 5 m) and station Pula air- (Oelschl€agel et al. 2015), in all investigated populations where port (44°540 N, 13°550 E; 63 m). As data were not normally dis- T. ruficeps occurred (Tables 1 and S1). Other pollinators were tributed Kruskal–Wallis test (software SigmaPlot 12; Systat not included because of the limited number of collected speci- Software, San Jose, CA, USA) was employed to test for differ- mens. The irresistible nature of pollinator attraction and ences of the single parameters among years and stations. In deception of A. rotunda flowers has been shown in a previous case of statistical significance, the Student–Newman–Keuls study (Oelschl€agel et al. 2015). We therefore assumed that the method (SigmaPlot 12) was used for pair-wise comparisons. pollinators start visiting A. rotunda flowers as soon as they are

930 Plant Biology 18 (2016) 928–937 © 2016 German Botanical Society and The Royal Botanical Society of the Netherlands Oelschlagel,€ Tschirnhaus, Nuss, Nikolic, Wanke, Dotterl€ & Neinhuis Spatio-temporal patterns in pollination available and conditions allow it. As for the majority of Permanova+ (version 1.0.5; Clarke 1993; Clarke & Gorley dipterans, the biology and ecology of T. ruficeps, i.e. duration 2006) to visualise similarities and differences in meteorologi- of larval stages, the occurrence of imagines during the year, cal parameters among the collected samples and according to preconditions to complete the life cycle, the substrate of larval pollinator availability. The contribution of single meteorologi- development, etc., remain unknown. We therefore linked cal parameters to the ordination of the samples is plotted. meteorological data to periods of trapping events within Aris- Permutation test (PERMANOVA) as implemented in Primer 6 tolochia flowers (see Collection of flowers visitors and pollina- and the Permanova+ add on was conducted on Euclidean dis- tors, pollinator identification) and up to 1 month before tances of normalised meteorological parameters of individual pollinator collection. The following time spans were analysed: time spans using the factor pollinator availability (presence/ (a) the day of collection in a flower, (b) the day before the absence) and controlling for year. Additional two-way Simi- collection date, (c) days 1–5 before collection, (d) days 5–15 larity percentages (SIMPER) analyses (Primer 6) using the fac- before collection, and (e) days 15–30 before the collection tor pollinator availability, and again controlling for year, was date. Time spans (a) and (b) represent the period of attraction performed to determine the contribution of individual meteo- to and trapping in the investigated flower. Time span (c) rological parameters to the differences among pollinator avail- likely fits the period when the pollinators were attracted, ability groups. trapped, covered with pollen and released by (an)other flower (s). Hence the chosen time spans are based on the variable Breeding system floral trapping stage, as outlined in the Results section. Finally, time spans (d) and (e) represent most likely weather Aristolochia rotunda’s ability for autonomous selfing, geitonoga- conditions during periods of pollinator development (i.e. lar- mous and xenogamous pollination was tested on 24 plants col- val and pupal s stages). lected from 15 natural populations in France, Italy and Croatia. For all collection dates of the populations Caska and Vlasici To avoid resource limitation that has been reported for natural (Isle of Pag) meteorological data were obtained from station Aristolochia populations in the Mediterranean (e.g. Berjano et al. Zadar. Meteorological data for the Istrian populations Dra- 2006, 2010) pollination experiments were conducted under gonja and Most Rasa were obtained from the station Pula air- greenhouse conditions. Flower buds were enclosed with small port. Data on daily precipitation, daily sunshine hours, average gauze bags to exclude visitors. Test for autonomous autogamy daily temperature, maximum daily temperature and minimum was performed on 15 plants employing up to nine replicates daily temperature were recorded for the time spans (a) to (e), (flowers) per plant. For this experiment, flowers were left and data on wind velocity for the time spans (a) to (c). For the untreated within the gauze bags. Artificial geitonogamous polli- time spans (a) and (b) absolute daily values were used. For the nation was performed on 15 plants with up to 13 replicates, and time spans (c) to (e) cumulative values for precipitation artificial xenogamous pollination on 17 plants with up to 13 and sunshine hours and average daily values for temperature replicates. At the beginning of anthesis the perianth was cut in and wind velocity were calculated for subsequent analyses, the middle of the utricle and pollen from the same or another respectively. plant was transferred to the stigmatic lobes using a toothpick. Scatter plots were drawn on the individual meteorological Subsequently, flowers were re-bagged till abortion to exclude fur- parameters using R 3.1.1 (R Core Team 2014) and RStudio ther pollination. Fruit and seed set were recorded for each flower. 0.98.978 (RStudio 2014). Normalised values of meteorological Average fruit and seed set per plant were calculated. Kruskal– parameters underwent principal components analysis (PCA) Wallis rank sum test was employed for global and Wilcoxon rank as implemented in the software Primer 6 (version 6.1.15) and

Table 1. Number of pollinators entrapped in Aristolochia rotunda ﬂowers at female ﬂower stage in seven Croatian populations on the Isle of Pag and in Istria.

Pag Istria

total (%) Caska Dinjisko polje Mt. Sveti Vid Stari Grad Vlasici Dragonja Most Rasa no. flowersa 4424 375 74 895 173 1777 377 753 no. pollinators 303 12 6 6 2 20 96 161 no. pollinators/100 flower 6.8 3.2 8.1 0.7 1.2 1.1 25.5 21.4 Chloropidae Aphanotrigonum femorellumb 2(<1%) 1 ––––1 – Oscinimorpha koeleriae 2(<1%) –– 2 –––– Oscinimorpha minutissimab 14 (5%) 2 1 4 1 – 24 Trachysiphonella ruficepsb 242 (80%) 9 –––16 61 156 Tricimba humeralis 6 (2%) – 5 ––1 –– Ceratopogonidaeb 35 (12%) –– – 1 1 32 1 Milichiidae Leptometopa niveipennis 1(<1%) –– – – 1 –– Neophyllomyza acyglossa 1(<1%) –– – – 1 –– aOnly pollinator-bearing collections considered (see Table S1). bTaxa revealed as pollinators by Oelschlagel€ et al. (2015).

Plant Biology 18 (2016) 928–937 © 2016 German Botanical Society and The Royal Botanical Society of the Netherlands 931 Spatio-temporal patterns in pollination Oelschlagel,€ Tschirnhaus, Nuss, Nikolic, Wanke, Dotterl€ & Neinhuis sum test for pair-wise comparisons of fruit set among treatments abundance of 68%, Chloropidae represented both the main (both executed in R 3.1.1 and RStudio 0.98.978). flower visitors and pollinators of A. rotunda. Additionally, a diverse set of other arthropods (e.g. Acari, Aphididae, Collem- bola, Colleoptera, Formicidae, Lepidoptera), mostly larvae, was Duration of trapping stage also found trapped within the flowers (Table S1). The trapping flower stage is characterised as the ability to trap and retain flower visitors by means of fully turgescent trapping Weather conditions during the study years trichomes within the floral tube (Oelschl€agel et al. 2009). This stage includes the female flower stage and the interphase. As Main weather trends were similar between meteorological sta- flower damage might shorten the floral lifespan (Berjano et al. tions Pula airport and Zadar, although minor differences were 2009), the turgescence of the trapping trichomes, which were found in temperature (slightly lower monthly mean values in visible from the entrance of the floral tube, was used as indica- Pula airport; Table S2), wind velocity (variation in daily values) tor of the trapping stage in order to avoid flower damage dur- and precipitation (stations show similar precipitation pattern ing the investigation. Investigations were performed in Vlasici over time, but considerable differences in daily precipitation (2010/05/09–2010/05/18) and Most Rasa (2012/05/27–2012/ rates; Fig. 2). In contrast, considerable variation in weather 06/01). A total of 64 and 27 flower buds, respectively, were conditions was detected among years (Table S2). Total labelled and, following flower opening, the flower stage was monthly precipitation values varied strongly among the study recorded daily in the morning. Flowers that did not enter the years (Table S2). In comparison to the reference period 1961– male stage during the observation period (30 in Vlasici and 2000 (data not shown) extremely wet periods were observed in one in Most Rasa) were excluded from statistical analysis. Wil- May 2010 for both stations and April 2012 in station Zadar coxon rank sum test (R 3.1.1) was employed to compare trap- only (outside 90% percentile), and very dry periods in May ping flower stages among sites and years. 2009 and April 2011 for both stations (outside 10% percentile). Both 2009 and 2011 were characterised by a significantly war- mer May than the other years (Table S2). May 2012 showed RESULTS alternating sunny weather and rainfall events, in contrast to the more steady weather conditions of the remaining years that Flower visitors, pollinators and pollinator variability among were mostly sunny (2009 and 2011) or very wet (2010; Fig. 2). populations Altogether 6156 female-stage flowers were investigated on 31 Temporal pollinator variability and dependence on climate different collection dates and sites. Arthropods were found factors entrapped in about 9% of the flowers, but only 3% contained arthropods carrying pollen. A total of 1081 arthropods was The period of pollinator availability and the pollinator abun- found, 84% of which were Diptera. Among the arthropods, dance varied strongly among years, populations and in the course 303 individuals (28%) carried Aristolochia pollen and thus of May (Fig. 2). In general, pollinator collections were most suc- proved to have repeatedly visited A. rotunda flowers. All polli- cessfulfrommidandthroughtotheendofMay.Fieldtripsin nators were Diptera belonging to the families Chloropidae 2009 and 2011 were characterised by predominantly warm and (88%), Ceratopogonidae (12%) and Milichiidae (<1%; sunny weather conditions and abundant T. ruficeps pollinators in Table 1). Chloropid pollinators were: Aphanotrigonum femorel- early May, although in considerably different amounts between lum Collin, 1946 (2♀), Oscinimorpha koeleriae Narchuk, 1970 years. In contrast, frequent rainfall and relatively low tempera- (2♀), Oscinimorpha minutissima (Strobl, 1900) (14♀), Trachysi- tures characterised the field period in 2010, during which over phonella ruficeps (221♀20♂, 1 unknown sex) and Tricimba 1250 investigated flowers revealed no T. ruficeps pollinators. Year humeralis (Loew, 1858) (6♀). Milichiid pollinators were identi- 2012 showed alternating sunny weather and rainfall with abun- fied as Leptometopa niveipennis (Strobl, 1898) (1♀) and Neo- dant T. ruficeps pollinators in the second half of May. phyllomyza acyglossa (Villeneuve, 1920) (1♀). Among the Overall, a significant effect of weather conditions on the Ceratopogonid pollinators were 26 females, and nine individu- availability of the main pollinator was found in a period 15– als of unknown sex. 30 days before pollinator collection (Pseudo-F1,22 = 3.65, In seven out of 10 sampled populations, pollinators were P = 0.02; Fig. 3B). This effect was dependent on the year stud- found (Fig. 2, Table S1). Chloropid pollinators were found in ied (pollinator 9 year: Pseudo-F2,22 = 4.29, P = 0.002). Struc- all pollinator-bearing populations, but ceratopogonid and mili- ture of data allowed only for 2012 to test for a weather effect chiid pollinators in only four and one of the populations, on pollinator availability (>100 permutations possible), and respectively (Table 1). Throughout the study period pollinators this test revealed a highly significant effect (Pseudo- were found more consistently in Istrian than in the Pag popu- F1,9 = 15.73, P = 0.005). SIMPER analysis of meteorological lations (Fig. 2). The pollinator with overall highest numbers of parameters during this time span, controlling for factor year, individuals trapped in flowers, T. ruficeps, was found in only revealed a contribution of nearly 80% by maximum tempera- four populations (Table 1). ture, sunshine hours and average temperature to the differences Despite the well-defined pollinator guild, the spectrum of in samples with pollinators present versus absent. flower visitors is much broader (Table S1). Dipterans of alto- In all other time spans weather conditions did not explain gether ten families were identified: besides Chloropidae, Cer- the availability of pollinators (PERMANOVA, P > 0.10), although atopogonidae and Milichiidae, non-pollinating representatives scatter plots point towards a negative correlation between pre- of Cecidomyidae, Chironomidae, Hybotidae, Phoridae, Psy- cipitation and the occurrence of the main pollinator on the col- chodidae, Scatopsidae as well as Sciaridae. However, with an lection day (Fig. 3).

932 Plant Biology 18 (2016) 928–937 © 2016 German Botanical Society and The Royal Botanical Society of the Netherlands Oelschlagel,€ Tschirnhaus, Nuss, Nikolic, Wanke, Dotterl€ & Neinhuis Spatio-temporal patterns in pollination

2009 2010 2011 2012 (A) Meteorological parameter Minimum temperature Maximum temperature Precipitation Pula Sample details Pollinators absent temperature (°C)

Precipitation (mm) Chloropidae pollinators Ceratopogonidae pollinators 0 5 10 15 20 25 5 101520253035 Milichiidae pollinators Dr Study site 85 No. female-stage ﬂowers MR 86

MR O MR O Dr MR Dr MR MR Dr Dr 50 99 Dr 18 391 67 345 MR 199 122 150 85 26 Pollinators per 100 flowers April 20 April 25 April 30 May 05 May 10 May 15 May 20 May 25 May 30 April 20 April 25 April 30 May 05 May 10 May 15 May 20 May 25 May 30 April 20 April 25 April 30 May 05 May 10 May 15 May 20 May 25 May 30 April 20 April 25 April 30 May 05 May 10 May 15 May 20 May 25 May 30 (B) t d

36 <1 1–4 9–14 esent, bu 74–173Pr not determine

Study sites

Zadar Dr - Dragonja (Istria) MR - Most Raša (Istria) O - Omisalij lake (Krk) temperature (°C) Precipitation (mm) Ca - Časka (Pag) DN - Dračevak Ninski (Ravni Kotari) 0 5 10 15 20 25 5101520253035 DP - Dinjiško polje (Pag) SV DP 28 St - Stari grad (Pag) 101 SV - Mt. Sveti Vid (Pag) SV Ca 830 ć DP SV St 92 Ca Va - Vlaši i (Pag) 46 65 173 185 VL - Veli Lošinj (Losinje) Va Ca VL Ca SV Va Ca DN Va Va Ca Va 55 103 68 49 22 534 24 107 1040 270 259 467

Fieldwork period April 20 April 25 April 30 May 05 May 10 May 15 May 20 May 25 May 30 April 20 April 25 April 30 May 05 May 10 May 15 May 20 May 25 May 30 April 20 April 25 April 30 May 05 May 10 May 15 May 20 May 25 May 30 April 20 April 25 April 30 May 05 May 10 May 15 May 20 May 25 May 30 Date

Fig. 2. Pollinator collection, pollinator abundance and weather conditions during April and May 2009 to 2012. Daily minimum and maximum temperature and daily amount of precipitation are shown for the periods 20 April to 31 May every year recorded by the weather stations (A) Pula airport in Istria and (B) Zadar in Ravni Kotari. Aristolochia rotunda populations are assigned to the nearest weather station. For each collection event from top to bottom the amount of pollinators per 100 ﬂowers and dipteran family, the locality and the sample size (number of investigated female-stage ﬂowers) are given. Samples of several days’ duration (Table S1) are assigned to the latest collection date. Fieldwork periods are highlighted with pale green bars. For ease of comparison among years. the periods 1 May to 15 May are highlighted in grey.

Breeding system DISCUSSION Pollination experiments revealed significant differences in fruit Pollinators of Aristolochia rotunda set among autogamic, geitonogamic and xenogamic experi- Employing a broad sampling spanning 10 populations ments (Kruskal–Wallis v2 = 7.97, df = 2, P = 0.02). All flowers throughout Croatia, this study reveals kleptoparasitic T. rufi- that were not hand-pollinated yielded no fruits. Hand-pollina- ceps being the most important pollinator of A. rotunda in the tion experiments yielded fruit sets of 18.4% by geitonogamic study region, thus confirming findings of Oelschl€agel et al. and 12.7% by xenogamic pollination. No significant difference (2015). The data presented here however complement the was found between geitonogamic and xenogamic hand-pollina- previously identified pollinator guild by adding the species tion treatments (W = 130, P = 0.93), and both treatments Oscinimorpha koeleriae, Tricimba humeralis (Chloropidae), yielded an average of 11 seeds per fruit. Leptometopa niveipennis and Neophyllomyza acyglossa (Milichi- idae) (Table 1). Nevertheless, T. ruficeps was the only species Variability of the trapping flower stage that at least temporarily occurred in high abundance. Addi- tionally and in contrast to the other chloropid pollinators that In Vlasici in 2010 the trapping flower stage averaged 6.1 are geographically more restricted (Nartshuk 2013), T. ruficeps 1.6 days (mean SD; min = 2 days, max = 8 days, n = 34). has a wide western Mediterranean distribution range, including However, 12 flowers were observed as being in the trapping Corsica, Sardinia and Sicily (Nartshuk 2013), where A. rotunda stage for at least 9 days. Their total lifespan could not be also occurs (Nardi 1984). Therefore, we assume T. ruficeps is recorded as the observation period ended at day 9. A signifi- the most important pollinator throughout the whole distribu- cantly shorter trapping stage of 2.3 0.6 days on average tion area of the plant, even though this species was not found (min = 1, max = 3, n = 26) was observed in Most Rasa 2012 in all investigated populations. (W = 17, P < 0.001). One flower was recorded in the trapping It is most likely that the newly found pollinator species are stage for 3 days but did not enter the male stage during the also deceived by the flower’s kleptomyiophilous pollination observation period. Pollination conditions differed consider- strategy. Representatives of both milichiid genera and the ably among sites and years. In Vlasici in 2010 pollinators were chloropid genus Oscinimorpha sp. are known to be attracted by absent (534 investigated flowers), whereas in Most Rasa in hexyl butyrate and other aliphatic esters (Sugawara & Muto 2012 a high pollinator frequency was observed (74 pollinators 1974; Zhang & Aldrich 2004; Heiduk et al. 2010), the key per 100 flowers, 86 investigated flowers; Fig. 2, Table S1).

Plant Biology 18 (2016) 928–937 © 2016 German Botanical Society and The Royal Botanical Society of the Netherlands 933 Spatio-temporal patterns in pollination Oelschlagel,€ Tschirnhaus, Nuss, Nikolic, Wanke, Dotterl€ & Neinhuis

(A) Coll. day One day Days 1 to 5 Days 5 to 15 Days 15 to 30 before coll. before coll. before coll. before coll.

15 15 75 10 10 20 40 50 5 10 20 5 25 0 0 0 0 0 Cum. prec. (mm) Cum. prec. (mm) Cum. prec. (mm) Precipitation (mm) Precipitation (mm) 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 Pollinators Pollinators Pollinators Pollinators Pollinators 15 125 160 60 100 140 10 10 50 120 40 75 5 100 5 30 50 80 0 20 60 Sunshine hours (h) Sunshine hours (h) Cum. sun. hours (h) Cum. sun. hours (h) 0 50 100 150 0 50 100 150 Cum. sun. hours (h) 0 50 100 150 0 50 100 150 0 50 100 150 Pollinators Pollinators Pollinators Pollinators Pollinators 25.0 24 17 17 16 22.5 21 20.0 15 20.0 16 18 17.5 14 17.5 15 13 15.0 15 15.0

Tmean (°C) Tmean (°C) 12 v. Tmean (°C) 14 12.5 A Tmean (°C) Av. Tmean (°C) Av. 11 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 Pollinators Pollinators Pollinators Pollinators Pollinators 23

27 25.0 21 25 22 24 22.5 19 20 21 20 20.0

Tmax (°C) Tmax (°C) 17 v. Tmax (°C) v. Tmax (°C) Av. Tmax (°C) Av. 18 A 17.5 A 18 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 Pollinators Pollinators Pollinators Pollinators Pollinators 17.5 17.5 14 15.0 13 12 15.0 12 12.5 12.5 11 10 10.0 10 10.0 8 Tmin (°C) 7.5 Tmin (°C) 9 Av. Tmin (°C) Av. Av. Tmin (°C) Av. 8 Tmin (°C) Av. 5.0 7.5 6 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 ) ) Pollinators Pollinators ) Pollinators Pollinators Pollinators –1 –1 –1 4.5 5 6 4.0 4 5 3.5 Legend vel. (m s 3.0 3 4 elocity (m s Pollinators absent 2.5 2 3 2.0 Pollinators present Av. wind Av. Wind velocity (m s Wind velocity Wind v 020406080 0 50 100 150 0 50 100 150 Pollinators Pollinators Pollinators

(B) Coll. day One day Days 1 to 5 Days 5 to 15 Days 15 to 30 before coll. before coll. before coll. before coll.

4 4 2 2 4 W SumSh P 2 SumSh 2 Tmin Sh Tmax 0 Tmean 0 2 SumSh Tmax Tmax 0 PC2 PC2 PC2 PC2 Tmax Tmax SumP PC2 SumP Tmean SumP 0 –2 Tmean W 0 Tmean –2 Tmean Sh P W –2 Tmin Tmin Tmin Tmin –2 –4 –2 –4 –4 –4 –2 0 2 4 –4 –2 0 2 4 –4 –2 0 2 4 6 –4 –2 0 2 4 –4 –2 0 2 4 PC1 PC1 PC1 PC1 PC1 PC1: 44.8 % PC1: 48.6 % PC1: 55.3 % PC1: 45.2 % PC1: 60.4 % PC2: 26.1 % PC2: 22.2 % PC2: 28.8 % PC2: 31.5 % PC2: 24.8 %

Fig. 3. Variability of meteorological parameters during and up to 30 days before pollinator collection from 2009 to 2012. Absolute values of meteorological parameters are drawn on the collection day and the day before the collection day, whereas for the remaining time spans cumulative values (sunshine hours and precipitation) and averages (wind velocity and temperature) of parameters were used. Samples are discriminated by colour according to presence and absence of Trachysiphonella ruficeps pollinators. A: Scatter plots of meteorological parameters in relation to pollen-carrying T. ruficeps individuals per 100 investigated female stage flowers. B: PCA conducted on normalised meteorological parameters of collection samples. Variation explained by first two principal components (PC) is given and the contribution of individual parameters to ordination is plotted in the diagrams. P - daily precipitation, Sh - daily sunshine hours, SumP - cumulative precipitation, SumSh-cumulative sunshine hours, Tmax - maximum daily temperature, Tmean - average daily temperature, Tmin - minimum daily temperature, W - wind velocity.

934 Plant Biology 18 (2016) 928–937 © 2016 German Botanical Society and The Royal Botanical Society of the Netherlands Oelschlagel,€ Tschirnhaus, Nuss, Nikolic, Wanke, Dotterl€ & Neinhuis Spatio-temporal patterns in pollination compounds of the kleptomyiophilous pollination system of specialised pollination systems even small changes in qualita- A. rotunda (Oelschl€agel et al. 2015). Additionally, kleptopara- tive and/or quantitative floral traits (i.e. scent chemistry) may sitism is a common phenomenon in Milichiidae and was provoke pollinator-induced reproductive isolation (Schiestl & reported for female Neophyllomyza sp. on heteropteran prey Ayasse 2002; Peakall et al. 2010). This may be the case in Din- (Coreidae and Pentatomidae; Eisner et al. 1991) as well as for jisko polje where A. rotunda is almost exclusively pollinated by N. acyglossa on beetle prey (Brake 2015). Knowledge about the Tricimba humeralis (Table 1). However, we do not know, habit of the chloropid genus Tricimba is currently unavailable. whether the absence of a specific pollinator species in a popula- Although A. rotunda flowers show low ecological specialisation is due to the general absence of this species in the respec- tion (several pollinator families and species), these findings tive locality or is due to different scent traits. This still needs to substantiate the high functional specialisation of the flowers be tested and is the subject to on-going studies. attracting kleptoparasites through olfactory mimicry of their heteropteran prey. In scent-based pollination systems, com- Temporal pollinator variability and dependence on climatic plex composition of the floral scent may promote the attrac- factors tion of several pollinator species, as different scent compounds or differences in their quantity may attract differ- During the study period considerable inter-annual variations ent insect taxa. The flower scent of A. rotunda is composed of of meteorological conditions were found along with differences numerous aliphatic esters and hydrocarbons, many more in pollinator availability and abundance. Similar variations in compounds than were shown to be crucial for attraction of pollinator availability were reported in other Mediterranean the main pollinator T. ruficeps (Oelschl€agel et al. 2015). It Aristolochia species (Berjano et al. 2009) and are likely also needs to be shown in future studies whether these compounds common outside the Mediterranean region. are involved in attraction of pollinators other than T. ruficeps. Contrary to our expectations, no statistically significant cor- Complex blends of floral scent are also known from other relation between meteorological data and pollinator availability deceptive pollination systems (Brodmann et al. 2008; Stokl€ was found during the flowering period. However, the collected et al. 2010). It has been shown that different pollinator species data are biased because observations were mainly performed detect different compounds of complex scent blends (Stokl€ under dry weather conditions, and potentially differing micro- et al. 2010). Consequently, it is likely that different pollinator climatic conditions of individual populations could not be taxa of the same plant species are attracted to different com- considered. However, a significant correlation was found pounds of the floral scent. between the availability of the main pollinator T. ruficeps and the weather conditions 15–30 days before pollinator collection (Fig. 3B). Unfavourable weather conditions in this period, i.e. High pollinator variability among populations low values for temperature and sunshine hours, lead to strong In contrast to our hypothesis that A. rotunda flowers trap the pollinator limitation and failure of reproductive success of the same pollinator species among different populations, we flowers later on. This period is assumed to include at least detected considerable variability of its pollinator composition some developmental stages of the flies, and most probably on a spatial scale. The occurrence of individual pollinator spe- occurs ahead of the flowering period of the plants. It is well cies and their relative abundance varied strongly among popu- known that insects strongly depend on weather conditions dur- lations (Table 1). This might give rise to different selection ing development (e.g. Coulson et al. 1976; Tolley & Niemczyk pressures on floral traits among populations (Fenster et al. 1988) and our results show that a high number of sunshine 2004; Gomez et al. 2009). The reason for the differing pollina- hours, and high daily maximum and average temperatures tor guilds remains unknown but might be caused by differences favour the appearance of T. ruficeps 2 or 3 weeks later. in habitat-based biotic and/or abiotic conditions (i.e. humidity, salinity, vegetation) or floral traits (i.e. scent). Breeding system The existence of geography-related, community-based differences in the pollinators has also been shown in other decep- We reject our hypothesis that autonomous selfing is a strategy tive plants, such as Arum maculatum (Espındola et al. 2010). to ensure reproduction in A. rotunda flowers, as autonomous A. rotunda plants tolerate highly anthropogenic environments selfing was not observed. A. rotunda flowers consequently rely such as field margins and dykes. We assume that the original on insects as pollen vectors for successful sexual reproduction. habitat of the species might have been grassy border areas of The inability for autonomous selfing, together with the the Mediterranean sclerophyll forests, i.e. treefall gaps, clearings observed periods of pollinator absence during flowering, might derived from forest fires, open areas along rivers or the coast. strongly decrease the reproductive success of the plants. Artifi- Such habitats typically underlie constant modifications that cial pollination of the flowers yielded relatively low numbers of might lead to varying pollinator climates. We hypothesise that ripe fruits (12–18%) independent of whether the pollen origin a pollination system that is able to attract a functional group of was from a different flower of the same or another plant. We pollinators, i.e. carnivorous kleptoparasites, instead of a single assume that inefficient transfer of pollen to the stigmatic lobes pollinator species is more likely able to adapt to altering and using a toothpick or damage to the stigmatic lobes might be unpredictable environmental conditions. the reason for the low fruit set (Bliss et al. 2013), as well as pos- Another reason for the different pollinator guilds could be a sible resource limitation in the greenhouse. Quantitative data hitherto not studied variability of floral scent among popula- under natural conditions are not available for A. rotunda. tions. A scent-driven pollinator variability among conspecific However, in most studied Aristolochia species artificial pollina- populations has been shown for deceptive Ophrys spp. in tion resulted in considerably higher fruit set than natural polli- southern Italy (Mant et al. 2005). Especially in highly nation (Sakai 2002; Berjano et al. 2006; Murugan et al. 2006;

Plant Biology 18 (2016) 928–937 © 2016 German Botanical Society and The Royal Botanical Society of the Netherlands 935 Spatio-temporal patterns in pollination Oelschlagel,€ Tschirnhaus, Nuss, Nikolic, Wanke, Dotterl€ & Neinhuis

Trujillo & Sersic 2006). Fruit set of artificial-pollinated flowers Aristolochia species, that fruit set is possible in flowers polli- ranges from 7–19% in A. baetica (Berjano et al. 2010) up to nated on day 2 or 3 of anthesis. In contrast, it has been shown 94% in A. tagala (Murugan et al. 2006). Fruit set of open-polli- for at least one Aristolochia species that stigmata receptivity nated flowers, in contrast, ranges from 2% in A. maxima (Sakai decreases rapidly during the first day of anthesis (Murugan 2002) to about 40% in A. chilensis and A. gigantea (Hipolito et al. 2006). In this case, prolongation of the trapping stage of a et al. 2012; Stotz & Gianoli 2013), but in most investigated spe- flower would increase its male rather than its female fitness. cies a fruit set <20% was found (Sakai 2002; Berjano et al. Additionally, by contributing to a flower surplus in the popula- 2006, 2010; Murugan et al. 2006; Trujillo & Sersic 2006). Sev- tion, an expanded lifespan of a flower may increase the female eral factors were suggested as promoting pollen limitation and fitness of neighbouring, younger flowers, as flower surplus in a consequently low fruit set in the genus: scarcity of pollinators, population is assumed to attract pollinators more efficiently limited ability of small dipterans to cover larger distances, (Podolsky 1992; Berjano et al. 2010). insufficient amount of pollen grains transported by single pollinators, the need for several pollen grains to fertilise one ovule ACKNOWLEDGEMENTS (e.g. A. baetica – 2.75 pollen grains per seed, A. paucinervis – 4.5 per seed), short pollen viability, and a rapid decline in We dedicate this publication to the late Professor Dr Stefan stigma receptivity after anthesis (Sakai 2002; Berjano et al. Vogel. We thank Monika Skegro (University Zagreb), Anna– 2006, 2010; Murugan et al. 2006; Trujillo & Sersic 2006). Magdalena Barniske (University Kassel), Markus Gunther,€ Cindy Thomas, Claudia P€atzold and Sarah Wagner for their valuable assistance during fieldwork, and Katharina Doll for Variability of the trapping flower stage assistance in pollination experiments. Dirk Pavlik and Thomas Flowers of A. rotunda show a significantly longer trapping Pluntke (all TU Dresden) gave valuable advice for analysis of flower stage compared to short-lived tropical species (e.g. Bur- meteorological data. The identification of flower visitors by gess et al. 2004; Trujillo & Sersic 2006) and similar floral long- Frauke Nielsen and Andre Reimann (Senckenberg Museum fur€ evity as other Mediterranean species (Berjano et al. 2009). Tierkunde, Dresden) and of milichiid pollinators by Irina These findings substantiate the hypothesis of Stotz & Gianoli Brake (Burgdorf) is acknowledged. Aristolochia plants used for (2013). Additionally, high plasticity of the duration of the trap- pollination experiments are cultivated in the Botanical Garden ping flower stage was observed. The trapping phase of flowers Dresden. Meteorological data were provided by the Meteoro- in the Vlasici population in a period with virtually no pollina- logical and Hydrological Service Croatia. The Croatian author- tors lasted significantly longer compared to the trapping phase ities issued the collection permit (UP/I-612-07/12-33/0292, of flowers in the Most Rasa population experiencing a period 517-12-02). with high pollinator frequency. A possible influence of weather conditions on the duration of the flower trapping stage in SUPPORTING INFORMATION A. rotunda also cannot be excluded. However, we assume that successful pollination, instead of weather, might be the major Additional Supporting Information may be found online in the influencing factor on the floral life span (Berjano et al. 2009). supporting information tab for this article: A prolonged trapping flower stage enhances both, the female Table S1. Arthropods entrapped in female stage A. rotunda and male fitness of a flower, as the chance to trap scarce polli- flowers in Croatia. nators and, hence, to receive foreign and distribute own pollen Table S2. Analysis of meteorological data for the months increases. Bliss et al. (2013) showed, for two tropical April and May 2009 to 2012.

REFERENCES in two Mediterranean Aristolochia species, with a volatiles to attract prey-hunting wasps for pollina- review of flower visitor records for the genus. Plant tion. Current Biology, 18, 740–744. Armbruster W.S., Fenster C., Dudash M. (2000) Polli- Biology, 11,6–16. Burgess K., Singfield J., Melendez V., Kevan P. nation “principles” revisited: specialization, pollina- Berjano R., Arista M., Ortiz P.L., Talavera S. (2010) (2004) Pollination biology of Aristolochia grandi- tion syndromes, and the evolution of flowers. The Persistently low fruiting success in the Mediterranean flora (Aristolochiaceae) in Veracruz, Mexico. Scandinavian Association for Pollination Ecology Hon- pipevine Aristolochia baetica (Aristolochiaceae): a Annals of the Missouri Botanical Garden, 91, 346– ours Knut Faegri, 39, 179–200. multi-year study. Plant Biology, 13, 109–117. 356. Ashmann T.-L., Schoen D.J. (1994) How long should Beshovski V.L. (1985) Zhitni mukhi (Diptera, Chloropi- Cammerloher H. (1923) Zur Biologie der Blute€ von flowers live? Nature, 371, 788–791. dae). Izdatelstvo na Balgarskata Akademija na Nau- Aristolochia grandiflora Swartz. Osterreichische€ Ashmann T.-L., Schoen D.J. (1996) Floral longevity: kite, Sofija, Bulgaria. Botanische Zeitung, 72, 180–198. fitness consequences and resource costs. In: Lloyd Bliss B.J., Wanke S., Barakat A., Ayyampalayam S., Castro S., Silveira P., Navarro L. (2008) Effect of D.G., Barrett S.C.H. (Eds), Floral Biology. Studies on Wickett N., Wall P.K., Jiao Y., Landherr L., Ralph pollination on floral longevity and costs of delaying Floral Evolution in Animal-Pollinated Plants. Chap- P.E., Hu Y., Neinhuis C., Leebens-Mack J., Arumu- fertilization in the out-crossing Polygala vayredae man & Hall, London, UK, pp 112–139. ganathan K., Clifton S.W., Maximova S.N., Ma H., Costa (Polygalaceae). Annals of Botany, 102, 1043– Baker A.M., Barrett S.C.H., Thompson J.D. (2000) Depamphilis C.W. (2013) Characterization of the 1048. Variation of pollen limitation in the early flowering basal angiosperm Aristolochia fimbriata: a potential Clarke K.R. (1993) Non-parametric multivariate analy- Mediterranean geophyte Narcissus assoanus experimental system for genetic studies. BMC Plant ses of changes in community structure. Australian (Amaryllidaceae). Oecologia, 124, 529–535. Biology, 13, 13. Journal of Ecology, 18, 117–143. Berjano R., de Vega C., Arista M., Orti P., Talavera S. Brake I. (2015) [Neophyllomyza acyglossa (Villeneuve, Clarke K.R., Gorley R.N. (2006) PRIMER v6: user man- (2006) A multi-year study of factors affecting fruit 1920). Descriptions.]. Milichiidae online. Available ual/tutorial. PRIMER-E, Plymouth, UK. production in Aristolochia paucinervis (Aristolochi- from http://milichiidaeinfo/taxonomy/term/557/de- Collin J.E. (1946) The British genera and species of aeae). American Journal of Botany, 96, 599–606. scriptions (accessed 17 July 2015). Oscinellinae (Diptera, Chloropidae). Transactions of Berjano R., Ortiz P.L., Arista M., Talavera S. (2009) Brodmann J., Twele R., Francke W., Holzler€ G., Zhang the Royal Entomological Society of London, 97, 117– Pollinators, flowering phenology and floral longevity Q.H., Ayasse M. (2008) Orchids mimic green-leaf 148.

936 Plant Biology 18 (2016) 928–937 © 2016 German Botanical Society and The Royal Botanical Society of the Netherlands Oelschlagel,€ Tschirnhaus, Nuss, Nikolic, Wanke, Dotterl€ & Neinhuis Spatio-temporal patterns in pollination

Correns C. (1891) Beitr€age zur biologischen Anatomie weather conditions on fruit set in two spring wood- generalization. University of Chicago Press, Chicago, der Aristolochia-Bluthe.€ Jahrbucher€ fur€ Wis- land orchids. Biology Letters, 5, 495–498. IL, pp 123–144. senschaftliche Botanik, 22, 161–189. Mant J., Peakall R., Schiestl F.P. (2005) Does selection RStudio (2014) Integrated development environment Coulson J.C., Horobin J.C., Butterfield J. (1976) The on floral odor promote differentiation among popu- for R. Version 0.98.978. Available from: http:// maintenance of annual life-cycles in two species of lations and species of the sexually deceptive orchid www.rstudio.org (accessed 11 July 2015). Tipulidae (Diptera): a field study relating develop- genus Ophrys? Evolution, 59, 1449–1463. Rulik B., Wanke S., Nuss M., Neinhuis C. (2008) ment, temperature and altitude. Journal of Animal Martin-Vide J., Lopez-Bustins J.A. (2006) The Western Pollination of Aristolochia pallida Willd. (Aristolochi- Ecology, 45, 215–233. Mediterranean Oscillation and rainfall in the Iberian aceae) in the Mediterranean. Flora, 203, 175–184. Dafni A. (1984) Mimicry and deception in pollination. Peninsula. International Journal of Climatology, 26, Sakai S. (2002) Aristolochia spp. (Aristolochiaceae) Annual Review of Ecology and Systematics, 15, 259– 1455–1475. pollinated by flies breeding on decomposing flowers 278. Murugan R., Shivanna K.R., Rao R.R. (2006) Pollina- in Panama. American Journal of Botany, 89, 527– Daumann E. (1971) Zur Best€aubungsokologie€ von tion biology of Aristolochia tagala, a rare species of 534. Aristolochia clematitis L. Preslia(Praha), 43, 105– medicinal importance. Current Science, 91, 795–798. Schiestl F.P., Ayasse M. (2002) Do changes in floral 111. Narchuk E.P., Smirnov E.S., Fedoseeva L.I. (1989) 30. odor cause speciation in sexually deceptive orchids? Dely-Draskovits A. (1981) Revision der palaearktischen Order Diptera. 99. Family Chloropidae. In: Bei- Plant Systematics and Evolution, 234, 111–119. Arten der Gattung Aphanotrigonum Duda, 1932, Bienko G., Steyskal G. (Eds), Keys to the insects of the Scopece G., Cozzolino S., Johnson S.D., Schiestl F.P. und Aphanotrigonella Nartshuk, 1964 (Diptera: European part of the USSR. Volume V, Diptera and (2010) Pollination efficiency and the evolution of Chloropidae). Acta zoologica Academiae Scientiarum Siphonaptera, Part II. Amerind Publishing, New specialized deceptive pollination systems. The Amer- Hungaricae, 27, 115–138. Delhi, Indiaand Brill Publishing, Leiden and Kinder- ican Naturalist, 175,98–105. Dunkeloh€ A., Jacobeit J. (2003) Circulation dynamics hook, pp 667–731. Sivinski J., Marshall S., Petersson E. (1999) Kleptopara- of Mediterranean precipitation variability 1948–98. Nardi E. (1984) The genus Aristolochia L. (Aristolochi- sitism and phoresy in the Diptera. Florida Entomolo- International Journal of Climatology, 23, 1843–1866. aceae) in Italy. Webbia, 38, 211–300. gist, 82, 179–197. Eisner T., Eisner M., Deyrup M. (1991) Chemical Nardi E. (1991) The genus Aristolochia L. (Aristolochi- Stokl€ J., Strutz A., Dafni A., Svatos A., Doubsky J., attraction of kleptoparasitic flies to heteropteran aceae) in Greece. Webbia, 45,31–69. Knaden M., Sachse S., Hansson B.S., Stensmyr M.C. insects caught by orb-weaving spiders. Proceedings of Nartshuk E.P. (2013) Grassflies of the family Chloropi- (2010) A deceptive pollination system targeting dro- the National Academy of Sciences USA, 88, 8194– dae (Diptera) on Mediterranean islands. Zoosystem- sophilids through olfactory mimicry of yeast. Cur- 8197. atica Rossica, 22,129–140. rent Biology, 20, 1846–1852. Espındola A., Pellissier L., Alvarez N. (2010) Variation Nartshuk E.P., Andersson H. (2013) The frit flies Stotz G.C., Gianoli E. (2013) Pollination biology in the proportion of flower visitors of Arum macula- (Chloropidae, Diptera) of Fennoscandia and Den- and floral longevity of Aristolochia chilensis in an tum along its distributional range in relation with mark. Brill Academic, Leiden, the Netherlands. arid ecosystem. Plant Ecology & Diversity, 6, 181– community-based climatic niche analyses. Oikos, Oelschl€agel B., Gorb S., Wanke S., Neinhuis C. (2009) 186. 120, 728–734. Structure and biomechanics of trapping flower tri- Sugawara R., Muto T. (1974) Attraction of several Fenster C.B., Marten-Rodrıguez S. (2007) Reproduc- chomes and their role in the pollination biology of dipterous insects to aliphtic esters (Diptera: Milichi- tive assurance and the evolution of pollination spe- Aristolochia plants (Aristolochiaceae). New Phytolo- idae, Chloropidae and Ceratopogonidae). Applied cialization. International Journal of Plant Sciences, gist, 184, 988–1002. Entomology and Zoology, 9,11–18. 168, 215–228. Oelschl€agel B., Nuss M., von Tschirnhaus M., P€atzold Tolley M.P., Niemczyk H.D. (1988) Upper and lower Fenster C.B., Armbruster W.S., Wilson P., Dudash C., Neinhuis C., Dotterl€ S., Wanke S. (2015) The threshold temperatures and degree-day estimates for M.R., Thomson J.D. (2004) Pollination syndromes betrayed thief – the extraordinary strategy of Aris- development of the frit fly (Diptera: Chloropidae) at and floral specialization. Annual Review of Ecology, tolochia rotunda to deceive its pollinators. New Phy- eight constant temperatures. Journal of Economic Evolution, and Systematics, 35, 375–403. tologist, 206, 342–351. Entomology, 81, 1346–1351. Gomez J.M., Perfectti F., Bosch J., Camacho J.P.M. Oosterbroek P. (2006) The European families of the Trujillo C., Sersic A. (2006) Floral biology of Aris- (2009) A geographic selection mosaic in a general- Diptera. KNNV, Utrecht, the Netherlands. tolochia argentina (Aristolochiaceae). Flora, 201, ized plant–pollinator–herbivore system. Ecological Peakall R., Ebert D., Poldy J., Barrow R.A., Francke W., 374–382. Monographs, 79, 245–263. Bower C.C., Schiestl F.P. (2010) Pollinator speci- Vogel S. (1978) Pilzmuckenblumen€ als Pilzmimeten. Gonzalez F., Stevenson D. (2000) Perianth develop- ficity, floral odour chemistry and the phylogeny of Flora, 167, 329–398. ment and systematics of Aristolochia. Flora, 195, Australian sexually deceptive Chiloglottis orchids: Wagner S.T., Hesse L., Isnard S., Samain M.S., Bolin J., 370–391. implications for pollinator-driven speciation. New Maass E., Neinhuis C., Rowe N.P., Wanke S. (2014) Hall D., Brown B. (1993) Pollination of Aristolochia lit- Phytologist, 188, 437–450. Major trends in stem anatomy and growth forms in toralis (Aristolochiales: Aristolochiaceae) by males of Podolsky R.D. (1992) Strange floral attractors: pollina- the perianth-bearing Piperales, with special focus on Megaselia spp. (Diptera: Phoridae). Annals of the tor attraction and the evolution of plant sexual sys- Aristolochia. Annals of Botany, 113, 1139–1154. Entomological Society of America, 86, 609–613. tems. Science, 258,791–793. Wanke S. (2007) Evolution of the genus Aristolochia – Heiduk A., Brake I., Tolasch T., Frank J., Jurgens€ A., Proctor M., Yeo P., Lack A. (1996) The natural history Systematics, Molecular Evolution and Ecology. Disser- Meve U., Dotterl€ S. (2010) Scent chemistry and pol- of pollination. Harper Collins, London, UK. tation, TU Dresden, Germany. linator attraction in the deceptive trap flowers of R Core Team (2014) A language and environment for Waser N.M., Chittka L., Price M.V., Williams Ceropegia dolichophylla. South African Journal of Bot- statistical computing. Available from http://www.R- N.M., Ollerton J. (1996) Generalization in polli- any, 76, 762–769. project.org (accessed 11 July 2015) . nation systems, and why it matters. Ecology, 77, Hipolito J., Viana B.F., Selbach-Schnadelbach A., Razzak M., Ali T., Ali S. (1992) The pollination biology 1043–1060. Galetto L., Kevan P.G. (2012) Pollination biology of Aristolochia bracteolata Lamk. (Aristolochiaceae). Zhang Q., Aldrich J. (2004) Attraction of scavenging and genetic variability of a giant perfumed flower Pakistan Journal of Botany, 24,79–87. chloropid and milichiid flies (Diptera) to metatho- (Aristolochia gigantea Mart. and Zucc., Aristolochi- Renner S.S. (2006) Rewardless flowers in the Angios- racic scent gland compounds of plant bugs (Hetero- aceae) visited mainly by small Diptera. Botany-Bota- perms and the role of insect cognition in their evolu- ptera: Miridae). Environmental Entomology, 33,12– nique, 90, 815–829. tion. In: Waser N.M., Ollerton J. (Eds), Plant- 20. Jacquemyn H., Brys R., Honnay O. (2009) Large popu- pollinator interactions: from specialization to lation sizes mitigate negative effects of variable