Journal of Research (2018) 131:633–640 https://doi.org/10.1007/s10265-018-1019-8

REGULAR PAPER

Reproductive biology of the threatened pomponium (), a species endemic to Maritime and Ligurian Alps

Gabriele Casazza1 · Angelino Carta2 · Paolo Giordani3 · Maria Guerrina4 · Lorenzo Peruzzi2 · Luigi Minuto1

Received: 5 April 2017 / Accepted: 7 February 2018 / Published online: 2 March 2018 © The Botanical Society of Japan and Springer Japan KK, part of Springer Nature 2018

Abstract Pollination ecology and breeding system of Lilium pomponium L. were studied, and their effect on the reproductive outcome was assessed. This species has high conservation interest in Europe, because it is included in Annex V of the EU Habitat Directive and it is one out of the fiveLilium species listed in IUCN Global Red List. To achieve our aim, the pollen vec- tors as well as the effect of bagging, emasculation and artificial pollination on reproductive output were studied. The most frequent visitor was the Lepidopteran Gonepteryx rhamnii. In general, reproductive outputs were close to zero for all the self-pollination treatments; however, geitonogamy and facilitated selfing seem slightly more efficient than autogamy, as also confirmed by self-compatibility and autofertility indices. Altogether, our results suggest a self-incompatible outcrossing breeding system, with a poor capacity for selfing. Nevertheless, climate change and anthropic threats might promote a shift toward self-fertilization, even maladaptive, favouring the few individuals able to produce selfed seeds.

Keywords Conservation · Pollen vectors · Reproductive strategy · Threatened species

Introduction and species survival (Gilman et al. 2012; Liu et al. 2012; Woodward 1987). Climate change may decrease popula- Studies on the reproductive systems of rare or threatened tion size of pollinators and pollination service, influencing species are useful for understanding the reasons of their the spatial and temporal distribution of genetic variation status (Evans et al. 2003; Schemske et al. 1994) and can and ultimately the processes that lead to plant speciation yield improved strategies for their conservation (Eckert and extinction (Barrett 2003; Eckert et al. 2010; Holsinger et al. 2010). Indeed, a relationship may exist between some 2000; Stebbins 1974). For these reasons, extensive studies reproductive features and rarity or threats of species (Gabri- on reproductive systems of are needed for supporting elová et al. 2013). In addition, anthropogenic changes, and their effective conservation, and in the absence of these data particularly climate change, may affect plant reproduction any conservation effort is arbitrary and may remain inef- fective (Shivanna and Tandon 2014). Lilium pomponium L. is endemic to the Maritime and Ligurian Alps. It is a spe- Electronic supplementary material The online version of this cies of high conservation interest in Europe, being listed in article (https​://doi.org/10.1007/s1026​5-018-1019-8) contains Annex V of the EU Habitat Directive (Council of European supplementary material, which is available to authorized users. Communities 1992), and it is one out of the five Lilium spe- L. ciliatum, L. jankae, L. polyphillum * Gabriele Casazza cies (together with [email protected] and L. rhodopeum) included in IUCN Global Red List (last accession 02.10.2017). Populations are declining in Italy, 1 DISTAV, University of Genoa, Corso Europa 26, although they are considered stable in France (Gargano 16132 Genoa, Italy 2015). For this reason, this species was recently assessed as 2 Department of Biology, Unit of Botany, University of Pisa, Endangered in Italy (Rossi et al. 2013) and Least Concern at Via Derna 1, 56126 Pisa, Italy worldwide level (Gargano 2015). The major threats to this 3 DIFAR, University of Genoa, Viale Cembrano 4, species are overgrazing, abandonment of pastoral systems, 16148 Genoa, Italy forest planting and the illegal collection of plants (Commis- 4 Department of Ecology and Genetics, Uppsala University, sion of the European Communities 2009). Furthermore, L. Norbyvägen 18D, 75236 Uppsala, Sweden

Vol.:(0123456789)1 3 634 Journal of Plant Research (2018) 131:633–640 pomponium appears to be prone to extinction risk because been shown to be an efficient morphological mechanism to of strong range loss induced by climate change (Casazza reduce self-fertilization and promote outcrossing in self- et al. 2014). compatible species (Webb and Lloyd 1986). Alternatively, Despite its rarity and conservation importance, nothing it can also act to reduce sexual interference between self- is known about the reproductive biology of L. pomponium, pollen and stigmas in self-incompatible species (Barrett which is, for this reason, investigated here. To accomplish 2002). this objective, we checked the pollinators, the pollen/ovule The present study was performed in a wide population ratio, the breeding system, and the germination performance extending from Monte Grai to Monte Lega (Val Nervia, of this species. Imperia, Italy), at the border between France and Italy (Fig. 1). In this area, the species occurs mainly in rocky and bushy places from 1,400 to 1,900 metres of altitude. Materials and methods This population has the highest richness of individuals within the species range. Study species and site

Lilium pomponium (Liliaceae, tribe Lilieae) belongs to Pollen vectors L. sect. Lilium (= L. sect. Liriotypus Asch. & Graebn.), a strongly supported phylogenetic clade of European lilies We documented insect visits and activity on L. pomponium diverging from the rest of the genus 9 Mya ago (İkinci inflorescences at the peak of the flowering season (from 2011), and including all the European and Caucasian spe- June to July). Two independent observers simultaneously cies, with the exception of L. martagon (Rešetnik et al. monitored insect presence on five individuals located 2007). The species grows in xeric stony or rocky calcare- 2–3 m apart from each other at three randomly selected ous habitats between 100 and 1,900 metres of altitude and sites (at least 50 metres apart from each other, in order to shows 1–10 flowers per inflorescence (Pignatti 1982). It avoid interference between observers and overestimation has scarlet Turk’s cap flowers with dark purple lines-spots of insect observations). Observations were monitored dur- and purplish papillae inside. Like several Lilium species, ing periods of 30 min from 9 a.m. to 11 p.m. over a total L. pomponium shows spatial separation of anthers and stig- of nine sunny days (from June 23rd to July 26th 2014), mas, the so-called ‘approach herkogamy’, with stigmas for a total observation time of 4,800 min. Insects were that stick out to be contacted first as the visitor enters the identified directly in the field or collected for laboratory blossom while pollen is picked up subsequently (Fryxell identification. 1957; Webb and Lloyd 1986). In general, herkogamy has

Fig. 1 Distributional ranges of Lilium pomponium (continuous black line). The location of the study area is reported (white square)

1 3 Journal of Plant Research (2018) 131:633–640 635

Breeding system At the end of the anthesis, bags were removed to prevent any damage to developing fruits. To evaluate the pollen/ovule ratio in the study popula- A total of 76 mature fruits from all treatments were col- tion, 15 floral buds were randomly chosen. To estimate the lected in September, preventing seed loss during dehis- number of pollen grains, anthers were put into Eppendorf cence. To estimate fruit set, we calculated the proportion tubes containing 30 µl of butanol to remove the pollenkit. of flowers producing a fruit with mature seeds. To estimate After the evaporation of butanol, we added 350 µl of 70% the seed set per flower, seeds from mature capsules (n = 55) ethanol and 50 µl of safranin, and vortexed for 10 min. were counted under a Leica M205 C stereomicroscope. We Grains were counted using Fuchs-Rosenthal cell counter calculated seed set as filled seeds/total number of ovules under an optical microscope (Leica DM2000). We made (filled seeds + aborted seeds + unfertilized ovules). We cal- three replicates per each sample and the recorded data culated the combined reproductive output for each treatment were multiplied by the dilution factor to obtain the total as the product of fruit set for the average seed set. This value number of pollen grains per flower. In the previous flow- ranged from 1 when both fruit and seed set are 1 to 0 when ers, we also manually counted the ovules per ovary using fruit set, seed set or both are 0. a Leica M205 C stereomicroscope. The mean pollen/ovule ratio per flower was calculated to define the reproductive Data analysis strategy adopted by the species. To examine the breeding system of L. pomponium, dur- Indices of self-compatibility (SCI) and autofertility (AFI) ing the years 2014–2015 a total of 246 flowers, one per were calculated according to Lloyd and Schoen (1992). The each individual, were exposed to open pollination (N = 37; SCI represents the capacity of a plant to reproduce by means N2014 = 20, N2015 = 17) as control (C, see Table 1) or allo- of self-pollination relative to cross-pollination. According to cated to one of six experimental pollination treatments Stout (2007), self-compatibility index was calculated both (T1-T6, see Table 1). The treatments were: T1, emascu- within flower(SCI ­ within = T3/T5) and between flowers of lation and free exposure, to test for allogamy (N = 30; the same inflorescence(SCI ­ between = T4/T5). The SCI ranges N2014 = 17, N2015 = 13); T2, bagged flowers with non-woven from a little below 0 to greater than 1, and values lower fabric, to test for spontaneous autogamy (N = 17; N2014 = 6, than 0.75 are interpreted as representing self-incompatibility N2015 = 11); T3, bagged flowers with non-woven fabric (Lloyd and Schoen 1992). The AFI (AFI = T2/T5) ranges and hand self-pollination, to test for facilitated autogamy from 0 to above 1, and represents the capacity of the flower (N = 25; N2014 = 12, N2015 = 13); T4, bagged flowers with to reproduce by autonomous self-pollination (Lloyd and non-woven fabric with emasculation and hand self-polli- Schoen 1992). All indices were calculated using fruit set, nation, applying pollen picked up from dehiscing anthers seed set and the combined reproductive output. from a donor flower belonging to the same inflorescence, Fruit and seed production data are of binomial type, to test for geitonogamy (N = 56; N2014 = 24, N2015 = 30); which lack the property of linearity and additivity. Hence, to T5, bagged flowers with non-woven fabric with emascula- determine the effect of breeding treatments on fruit produc- tion and hand cross-pollination (using pollen from a differ- tion, we conducted a binomial logistic regression with fruit ent plant at least 2 m away), to test for facilitate allogamy production as a binary dependent variable, and pollinator (N = 56; N2014 = 27, N2015 = 29); T6, bagged flowers with treatments as predictor variables. Pollination treatments that non-woven fabric with emasculation and no-pollination, did not set fruits were excluded by the analysis. Moreover, to test for agamospermy (N = 25; N2014 = 15, N2015 = 10). the effects of pollination treatments on plant fitness were

Table 1 Description of treatments used to examine reproductive biology of Lilium pomponium Treatment Treatment Effect measured Specific type of selfing Specific code type of crossing

C Free exposure (no treatment, control) Natural production Natural Natural T1 Emasculated, free exposure Spontaneous allogamy Natural geitonogamy Natural T2 Bagged Spontaneous autogamy Natural autogamy No T3 Bagged, hand self-pollination Facilitated autogamy Artificial autogamy No T4 Bagged, emasculation, hand self-pollination Geitonogamy Artificial geitonogamy No T5 Bagged, emasculation, hand cross-pollination Facilitated alloogamy (Artifi- No Artificial cial allogamy) T6 Bagged, emasculation Agamospermy (apomixis) No No

1 3 636 Journal of Plant Research (2018) 131:633–640 analysed by fitting factorial generalized linear mixed models (Mascarello et al. 2011). We used Chi square test to deter- (GLMMs, logit link function, binomial distribution) to the mine whether germination rates differed among treatments. seed set data with pollination treatments as fixed predictors, and flower as random factor (Bolker et al. 2009). Results from hand self-pollination (T3) were not included in the Results model, because a single flower set seeds. Statistical analyses were performed using the lme4 package (Bates et al. 2015) Pollination ecology implemented in R (R Development Core Team 2011). Post- hoc tests were conducted to evaluate pair-wise differences in A total of five taxa of insects were observed on L. pompo- measured traits between treatments using the glht function in nium flowers (Table 2). The most frequent visitor was the the R ‘multicomp’ package (Hothorn et al. 2008). Lepidopteran Gonepteryx rhamnii (52.94%—only female Progeny performance, measured as seed germination, was individuals were recorded), followed by Hymenoptera evaluated in temperature- and light-controlled incubators. (29.41%) and Coleoptera (11.76%), while Diptera (5.89%) After collection, the seeds were subjected to visual inspec- showed a sporadic presence. A maximum of two visits per tion and only filled seeds were used in the experiments. For flower per hour were recorded. Insect visits were most fre- each pollination treatment yielding enough seeds (C, T1, T4 quent during the warmest hours of the day (from 10 a.m. to 3 and T5), we sowed up to 50 seeds (randomly selected when p.m.), and despite our observations protracted until 11 p.m., available) in two 90 mm Petri dishes (25 seeds per dish) we did not observe any nocturnal visitors. Nevertheless, the containing 1% distilled water agar. Seeds were incubated for total number of visits recorded (N = 17) was low considering 100 days at 10 ± 1 °C under a 12 h photoperiod (white light) the time of observation (4,800 min).

Breeding system Table 2 List and frequencies of insect visitors observed on Lilium pomponium flowers In L. pomponium each flower produced an average of 47,648 Order Family Genus N N% (SD 10,806) pollen grains and an average of 147.93 (SD 28.61) ovules. Mean pollen/ovule ratio per flower resulted Lepidoptera Satyridae Gonepteryx rhamnii 9 52.94 332.76 (SD 94.18). Hymenoptera Formicidae Lasius 3 17.65 Significant differences were detected between open pol- Hymenoptera Megachilidae 2 11.76 lination test (C) and self-pollination treatments (T3 and T4 Coleoptera Cerambycidae Tetrops 2 11.76 in Table S2). In general, fruit set outputs (Fig. 2a, Table S1) Diptera Muscidae 1 5.89 were significantly higher in free pollination treatments (C Total number of insect 17 and T1) compared to bagged treatments (T2-T4 and T6). visitors Within free exposure treatments (C and T1), the fruit set Total number of taxa 5 was higher in emasculated treatment (T1), even if the differ- N number of individuals, N% percentage of total observed individuals ence is not significant. Fruit set outputs in cross-pollination

Fig. 2 Results for fruit set (a), seed set (b) and combined reproduc- T4 Bagged, emasculation, hand self-pollination; T5 Bagged, emas- tive output (c) for each treatment (treatment codes are the same than culation, hand cross-pollination; T6 Bagged, emasculation. Results in Table 1). Means and ± 95% confidence intervals are given for each of post-hoc tests for statistical differences are reported for treat- pollination test. C Free exposure (no treatment, control); T1 Emas- ments yielding fruits: different letters indicate significant differences culated, free exposure; T2 Bagged; T3 Bagged, hand self-pollination; (p < 0.05)

1 3 Journal of Plant Research (2018) 131:633–640 637 treatments (C, T1 and T5) were always significantly higher visual stimuli (Andersson 2003) and it shows a strong pref- compared to the self-pollination ones (T2, T3 and T4) that erence for blue and purple flowers, though it also visits red frequently resulted close to zero. Among these, geitonogamy ones (Ilse 1928) like those of L. pomponium. The prevalence treatment (T4) was slightly more efficient than spontane- of female individuals of Gonepteryx rhamnii in late spring ous and facilitated autogamy (T3 and T2, see in Fig. 2a, (when Lilium mating occurs) may be explained by differ- Table S1). No fruits were produced in T6, suggesting that ences in foraging tasks of females and males (Andersson apomixis may not occur. Seed-set outputs (Fig. 2b, Table S1) 2003). Males are more active in seeking mates than females in natural and hand cross-pollination treatments (C, T1 e T5) (Wickman 1996), while the latter are often busy searching were significantly higher compared to hand self-pollination for acceptable host plants and for food (Andersson 2003). treatments (T3-T4). Contrary to expectation, seed set in the Our results are in line with previous studies on pollinators hand cross-pollination treatments (T5) was slightly lower in Lilium. Indeed, according to literature, lilies are mainly compared to the natural insect cross-pollination ones (C, pollinated by Lepidoptera (Skinner 1988; Yokota and Yahara T1), even if the difference is not significant. The same trend 2012). Specifically, Turk’s-cap lilies, like L. pomponium, are recorded in fruit set and seed set was detected in the com- usually pollinated by butterflies or hummingbirds (Skinner bined reproductive output (Fig. 2c, Table S1). 2002; Skinner and Sorrie 2002). Despite their body size, All AFI values resulted in zero. The SCI values calcu- Hymenoptera, Coleoptera and Diptera have been found to lated on fruit set, seed set and combined reproductive output be effective pollinators in L. japonicum (Yokota and Yahara were 0.07, 0.33 and 0.03 ­(SCIwithin) and 0.16, 0.30 and 0.06 2012). Due to the similarity in flower shape and arrangement ­(SCIbetween), respectively. between L. japonicum and L. pomponium, we can expect All treatments yielding fruits produced viable seeds. The the same classes of insects to be occasional pollinators in proportion of germinated seeds was high (> 80%) in all treat- both species. The total amount of insects we recorded is ments (Table 3), and there were no significant differences low in relation to the total time of observation, and this is among them (p > 0.05). in accordance with what is reported for other stenochorous species (Guerrina et al. 2016; Lehnebach and Riveros 2003; Philipp et al. 2004). Nevertheless, fruit set in L. pomponium Discussion does not seem limited by pollinator visits, since we found no significant differences between open pollinated (C and Reproductive biology T1) and hand cross-pollinated (T5) flowers. This result sug- gests that pollinators are effective at pollen transfer despite A detailed knowledge of the reproductive biology of rare being scarce. Similarly, low visitation rate without strong and threatened species is necessary for conservation, and in reduction in reproductive output was recorded in other Lili- the absence of these data any conservation effort is arbitrary aceae like Fritillaria meleagris L. (Zych and Stpiczyńska and may remain ineffective. Despite their use in traditional 2012) and Erythronium grandiflorum Pursh (Thomson et al. medicine and their importance as ornamental plants, several 2000). The low values of fruit and seed set recorded in our Lilium species are threatened because of their rarity, and yet hand-pollinated treatment (T5) may also depend on the sin- very few data are available on their reproductive biology. gle pollen donors we used in the treatment. In fact, there In L. pomponium, the observed insects are common plant is a positive relationship between number of pollen donors diurnal pollinators (i.e. Lepidoptera, Hymenoptera and and reproductive success (Paschke et al. 2002). Because low Coleoptera) and the most frequent visitor is Gonepteryx fruit may result from both pollen limitation and limitation rhamnii (ca. 50% - particularly female individuals) suggest- in available resources (Ashman et al. 2004; Bateman 1948; ing a certain degree of specificity in pollination system. The Burd 1994), the low fruit set detected in L. pomponium may low number of observations of G. rhamnii recorded in our also be explained by a limitation in resource availability. In study (even if each individual visits several plants, pers. obs.) multi-flowered species, like L. pomponium, both fruit and is probably due to the fact that this butterfly mostly reacts to seed set are limited by competition for resources among flowers of the same inflorescence (Kliber and Eckert 2004), Table 3 Percentages of germinating seeds (mean ± 95% binomial resulting in a strongly reduced combined output. The influ- confidence intervals) resulting from pollination treatments (as indi- ence of resource limitation on reproductive output was also cated in Table 1) detected in other endemic species growing in soils poor in C T3 T4 T5 mineral nutrients and in dry habitat like montane dry forest, sandy soil, cliff and rocky habitats (Guitián et al. 1994; Rod- 98 (97–99) 92 (80–97) 92 (80–97) 98 (97–99) ríguez-Riaño et al. 1999; Shi et al. 2010). Similar low repro- Seeds were incubated at constant 10 °C and exposed to a 12 h daily ductive output was previously recorded in other SW Alps photoperiod endemics growing on cliffs or screes likePrimula allionii

1 3 638 Journal of Plant Research (2018) 131:633–640

(fruit set = 0.6, seed set = 0.5; Minuto et al. 2014) and Berar- (Ramsey and Vaughton 2000; Vogler and Stephenson 2001) dia subacaulis (fruit set = 0.5; Guerrina et al. 2016). or by promoting fruit abortion (Webb and Lloyd 1986). The The index of autofertility (AFI) was null in our experi- differences between emasculated treatments (T1 and T5) ment, indicating the inability of flowers to self spontane- may result from the number of pollen donors used in the ously. The index of self-compatibility (SCI) calculated treatments that in turn may affect reproductive output. In between flowers on the same plant ­(SCIbetween) was higher fact, in T5 only one pollen donor was used because of the than that calculated within flower(SCI ­ within), suggesting scarcity of flowers, while in free exposed T1 pollination may that geitonogamy is more efficient than autogamy. These occurs from multiparental donors. results are in line with the mean value of pollen/ovule ratio The lack of differences in seed germination between (332.76), that falls in the range established by Cruden (1977) self-pollination and cross-pollination treatments is not for facultative allogamy (244.7–2,588) and they are congru- uncommon (Carta et al. 2015; Husband and Schemske ent with the evidence that Lilium species are usually cross- 1996). The high final germination proportions in all treat- pollinated and self-incompatible (Austin-McRae 1998; ments together with the low reproductive output in self- Pelkonen et al. 2007). The differences detected in self-pol- pollinated individuals is consistent with the theory of lination treatments may be explained both by spatial separa- inbreeding depression, according to which in outcrossing tion of stigmas and anthers (approach herkogamy) and by species inbreeding depression acts mainly at the stage of proterandry within flowers, as observed in other lily species seed production (Husband and Schemske 1996). (Sun and Yao 2013). In fact, the lower efficiency of spon- taneous autogamy (T2) compared to facilitated autogamy (T3) in L. pomponium may be because the spatial separa- Implication for conservation tion of stigmas and anthers restricts pollen deposition on stigma of the same flower. Furthermore, a lower efficiency Taken together, our results suggest that L. pomponium is of spontaneous autogamy (T2) than geitonogamy (T4) may essentially an insect pollinated and self-incompatible spe- derive from the supposed proterandry of L. pomponium cies, despite few individuals showing leakiness in their together with the sequential blooming in the inflorescence. self-incompatibility system. Nevertheless, mating system On the one hand, proterandry may decrease probability of may vary among populations of a species. Species may autogamy. On the other hand, proterandry combined with exhibit both variation in mating system (self-incompati- sequential blooming may allow the mature pollen of the bility vs self-compatibility) and variation in selfing rate. upper flower to fertilize the receptive stigma of any below An example of the first case is Leavenworthia alabamica one. This behaviour has been observed also in several other that exhibits variation among peripheral and central popu- species of the genus Lilium where both proterandry and/ lations in the presence or absence of self-incompatibility or approach herkogamy may limit self-fertilization (Fryxell (Busch 2005). An example of the latter case is Clarkia 1957; Sun and Yao 2013; Webb and Lloyd 1986), though xantiana that has a higher rate of selfing and reduced the asynchrony of blooming may favour the self-fertilization herkogamy in geographically peripheral populations via geitonogamy, as observed in other species (Montaner compared to central populations (Moeller 2006). In Picea et al. 2001). Nevertheless, some species of Lilium are self- sitchensis mating system is influenced by both population compatible (Rodger et al. 2013) and others show substantial size and isolation at range peripheries (Mimura and Aitken quantitative variation in self-incompatibility degree among 2007). Similarly, in Lilium longiflorum self-compatible individuals (Sakazono et al. 2012). The higher fruit set in individuals are more frequent at the periphery of distribu- emasculated compare to non-emasculated flowers when tion range (Sakazono et al. 2012). In summary, climate natural pollination occurs (T1 vs C) may be explained by change may increase the self-fertilization rate in periph- proterandry together with self-incompatibility system (Lun- eral populations as a result of a chronic reduction in mate dqvist 1991; Tezuka et al. 2007) in individuals where the availability induced by a decline in population size (Eckert low degree of herkogamy does not prevent deposition of et al. 2010; Levin 2012). In low altitude populations of L. self-pollen on stigma. On the one hand, the non-viable pol- pomponium, future climate change and anthropic threat len may occupy the surface area of the stigma of the same will probably induce a decline in population size (Noble flower that would otherwise be available for pollen of other and Diadema 2011). This change may promote an evolu- flowers, the so-called sexual interference. This is in line with tionary shift towards reduced herkogamy and increased the idea that proterandry combined to self-incompatibility autogamy, in order to assure reproduction. However, self- may induce within-flower sexual interference (Harder et al. fertilization on the one hand may assure reproduction in 2000; Routley and Husband 2003; Routley et al. 2004). On front of low pollinator service and, on the other hand, may the other hand, self-pollen may interfere with the outcross be maladaptive because of strong inbreeding depression one by reducing germination rates or pollen tube growth and pollen discounting (Barrett 2003).

1 3 Journal of Plant Research (2018) 131:633–640 639

Acknowledgements We are grateful to Simona Bonelli (Department Gabrielová J, Münzbergová Z, Tackenberg O, Chrtek J (2013) Can of Life Sciences and Systems Biology, University of Turin) and Loris we distinguish plant species that are rare and endangered from Galli (DISTAV, University of Genoa) for insect identification. We other plants using their biological traits? Folia Geobot 48:449– thank Stefano Iardella, Carmelo Nicodemo Macrì, Elio Guerra and 466. https​://doi.org/10.1007/s1222​4-012-9145-x Luca Ulzi for their support in field and laboratory work. Gargano D (2015) Lilium pomponium. The IUCN red list of threatened species 2015: e.T190872A70290522. https​://doi. org/10.2305/IUCN.UK.2015-1.RLTS.T1908​72A70​29052​2.en Gilman RT, Fabina NS, Abbott KC, Rafferty NE (2012) Evolution of References plant–pollinator mutualisms in response to climate change. Evol Appl 5:2–16. https://doi.org/10.1111/j.1752-4571.2011.00202​ .x​ Andersson S (2003) Foraging responses in the butterflies Inachis io, Guerrina M, Casazza G, Conti E et al (2016) Reproductive biology Aglais urticae (Nymphalidae), and Gonepteryx rhamni (Pieridae) of an Alpic paleo-endemic in a changing climate. J Plant Res to floral scents. Chemoecology 13:1–11. https​://doi.org/10.1007/ 129:477–485. https​://doi.org/10.1007/s1026​5-016-0796-1 s0004​90300​000 Guitián J, Sánchez JM, Guitián P (1994) Pollination ecology of Ashman T-L, Knight TM, Steets JA et al (2004) Pollen limitation of Petrocoptis grandiflora Rothm. (Caryophyllaceae); a species plant reproduction: ecological and evolutionary causes and conse- endemic to the north-west part of the Iberian Peninsula. Bot J quences. Ecology 85:2408–2421. https://doi.org/10.1890/03-8024​ Linn Soc 115:19–27. https://doi.org/10.1111/j.1095-8339.1994.​ Austin-McRae E (1998) Lilies: a guide for growers and collectors. tb017​64.x Timber Press, Incorporated Harder LD, Barrett SCH, Cole WW (2000) The mating consequences Barrett SCH (2002) The evolution of plant sexual diversity. Nat Rev of sexual segregation within inflorescences of flowering plants. Genet 3:274–284. https​://doi.org/10.1038/nrg77​6 Proc R Soc Lond B Biol Sci 267:315–320. https://doi.org/10.1098/​ Barrett SCH (2003) Mating strategies in flowering plants: the outcross- rspb.2000.1002 ing-selfing paradigm and beyond. Philos Trans R Soc B Biol Sci Holsinger KE (2000) Reproductive systems and evolution in vascular 358:991–1004. https​://doi.org/10.1098/rstb.2003.1301 plants. Proc Natl Acad Sci 97:7037–7042. https://doi.org/10.1073/​ Bateman AJ (1948) Intra-sexual selection in Drosophila. Heredity pnas.97.13.7037 2:349–368 Hothorn T, Bretz F, Westfall P (2008) Simultaneous inference in gen- Bates D, Mächler M, Bolker B, Walker S (2015) Fitting linear mixed- eral parametric models. Biom J 50:346–363 effects models using lme4. J Stat Softw. https​://doi.org/10.18637​ Husband BC, Schemske DW (1996) Evolution of the magnitude and /jss.v067.i01 timing of inbreeding depression in plants. Evolution 50:54–70. Bolker BM, Brooks ME, Clark CJ et al (2009) Generalized linear https​://doi.org/10.2307/24107​80 mixed models: a practical guide for ecology and evolution. Trends İkinci N (2011) Molecular phylogeny and divergence times estimates of Ecol Evol 24:127–135. https://doi.org/10.1016/j.tree.2008.10.008​ Lilium section Liriotypus (Liliaceae) based on plastid and nuclear Burd M (1994) Bateman’s principle and plant reproduction: The role ribosomal ITS DNA sequence data. Turk J Bot 35:319–330 of pollen limitation in fruit and seed set. Bot Rev 60:83–139. https​ Ilse D (1928) Über den farbensinn der tagfalter. Z Für Vgl Physiol ://doi.org/10.1007/BF028​56594​ 8:658–692. https​://doi.org/10.1007/BF003​38976​ Busch JW (2005) The evolution of self-compatibility in geographi- Kliber A, Eckert CG (2004) Temporal decline in reproductive invest- cally peripheral populations of Leavenworthia alabamica (Bras- ment among flowers in a sequentially blooming plant: proximate sicaceae). Am J Bot 92:1503–1512. https​://doi.org/10.3732/ mechanisms and adaptive significance. Ecology 85:1675–1687 ajb.92.9.1503 Lehnebach C, Riveros M (2003) Pollination biology of the Chilean Carta A, Bedini G, Giannotti A et al (2015) Mating system modulates endemic orchid Chloraea lamellata. Biodivers Conserv 12:1741– degree of seed dormancy in Hypericum elodes L. (Hypericaceae). 1751. https​://doi.org/10.1023/A:10236​66800​948 Seed Sci Res 25:299–305. https​://doi.org/10.1017/S0960​25851​ Levin DA (2012) Mating system shifts on the trailing edge. Ann Bot 50002​52 109:613–620. https​://doi.org/10.1093/aob/mcr15​9 Casazza G, Giordani P, Benesperi R et al (2014) Climate change has- Liu Y, Mu J, Niklas KJ et al (2012) Global warming reduces plant tens the urgency of conservation for range-restricted plant spe- reproductive output for temperate multi-inflorescence species on cies in the central-northern Mediterranean region. Biol Conserv the Tibetan plateau. New Phytol 195:427–436. https​://doi.org/10 179:129–138. https​://doi.org/10.1016/j.bioco​n.2014.09.015 .1111/j.1469-8137.2012.04178​.x Commission of the European Communities (2009) Report on the con- Lloyd DG, Schoen DJ (1992) Self- and cross-fertilization in plants. I. servation status of habitat types and species as required under Functional dimensions. Int J Plant Sci 153:358–369 article 17 of the Habitats Directive. Commission of the European Lundqvist A (1991) Four-locus S-gene control of self-incompatibility Communities, Brussels made probable in Lilium martagon (Liliaceae). Hereditas 114:57– Council of European Communities (1992) Council Directive 92/43/ 63. https​://doi.org/10.1111/j.1601-5223.1991.tb005​53.x EEC of 21 May on the conservation of natural habitats and of wild Mascarello C, Sacco E, Carasso V et al (2011) Evaluation of the fauna and flora. Off J Eur Commun 35:7–50 seed germination of two protected species: Lilium pomponium Cruden RW (1977) Pollen-ovule ratios: A conservative indicator of L. and Lilium martagon L. Acta Hortic 385–391. https​://doi. breeding systems in flowering plants. Evolution 31:32–46. https​ org/10.17660​/ActaH​ortic​.2011.900.48 ://doi.org/10.2307/24075​42 Mimura M, Aitken SN (2007) Increased selfing and decreased effective Eckert CG, Kalisz S, Geber MA et al (2010) Plant mating systems pollen donor number in peripheral relative to central populations in a changing world. Trends Ecol Evol 25:35–43. https​://doi. in Picea sitchensis (Pinaceae). Am J Bot 94:991–998. https://doi.​ org/10.1016/j.tree.2009.06.013 org/10.3732/ajb.94.6.991 Evans MEK, Menges ES, Gordon DR (2003) Reproductive biology Minuto L, Guerrina M, Roccotiello E et al (2014) Pollination ecology of three sympatric endangered plants endemic to Florida scrub. in the narrow endemic winter-flowering Primula allionii (Primu- Biol Conserv 111:235–246. https​://doi.org/10.1016/S0006​ laceae). J Plant Res 127:141–150. https​://doi.org/10.1007/s1026​ -3207(02)00293​-8 5-013-0588-9 Fryxell PA (1957) Mode of reproduction of higher plants. Bot Rev Moeller DA (2006) Geographic structure of pollinator communities, 23:135–233 reproductive assurance, and the evolution of self-pollination.

1 3 640 Journal of Plant Research (2018) 131:633–640

Ecology 87:1510–1522. 10.1890/0012-9658(2006)87[1510:GSO Shi X, Wang J-C, Zhang D-Y et al (2010) Pollen source and resource PCR]2.0.CO;2 limitation to fruit production in the rare species Eremosparton Montaner C, Floris E, Alvarez JM (2001) Geitonogamy: a mechanism songoricum (Fabaceae). Nord J Bot 28:438–444. https​://doi.org/ responsible for high selfing rates in borage (Borago officinalis 10.1111/j.1756-1051.2010.00658​.x L.). Theor Appl Genet 102:375–378. https​://doi.org/10.1007/ Shivanna KR, Tandon R (2014) Reproductive ecology of flowering s0012​20051​656 plants: A manual. Springer India Noble V, Diadema K (2011) La flore des Alpes-Maritimes et de la Skinner MW (1988) Comparative pollination ecology and floral evo- principauté de Monaco. Originalité et diversité. Naturalia Pub- lution in Pacific CoastLilium . Disertation, Harvard University lications, Turriers Skinner MW (2002) Nomenclatural changes in North American Lilium Paschke M, Abs C, Schmid B (2002) Effects of population size and (Liliaceae). Novon 12:253–261. https://doi.org/10.2307/33929​ 64​ pollen diversity on reproductive success and offspring size in the Skinner MW, Sorrie BA (2002) Conservation and ecology of Lilium narrow endemic Cochlearia bavarica (Brassicaceae). Am J Bot pyrophilum, a new species of Liliaceae from the Sandhills Region 89:1250–1259. https​://doi.org/10.3732/ajb.89.8.1250 of the Carolinas and Virginia. USA Novon 12:94–105. https://doi.​ Pelkonen V-P, Niittyvuopio A, Pirttilä AM et al (2007) Phylogenetic org/10.2307/33932​47 background of orange lily (Lilium bulbiferum s.l.) cultivars from Stebbins GL (1974) Flowering plants: evolution above the species a genetically isolated environment. Plant Biol 9:534–540. https://​ level. Belknap Press of Harvard University Press, Cambridge doi.org/10.1055/s-2007-96504​2 Stout JC (2007) Reproductive biology of the invasive exotic shrub, Philipp M, Hansen LB, Adsersen H, Siegismund HR (2004) Reproduc- Rhododendron ponticum L. (Ericaceae). Bot J Linn Soc 155:373– tive ecology of the endemic Lecocarpus pinnatifidus (Asteraceae) 381. https​://doi.org/10.1111/j.1095-8339.2007.00719​.x in an isolated population in the Galápagos Islands. Bot J Linn Soc Sun S-G, Yao C-Y (2013) Increased seed set in down slope-facing 146:171–180. https://doi.org/10.1111/j.1095-8339.2004.00323​ .x​ flowers of Lilium duchartrei. J Syst Evol 51:405–412. https://doi.​ Pignatti S (1982) Flora d’Italia. Edagricole org/10.1111/jse.12002​ R Development Core Team (2011) R: A language and environment Tezuka T, Akita I, Yoshino N, Suzuki Y (2007) Regulation of self- for statistical computing. R foundation for statistical computing, incompatibility by acetylcholine and cAMP in Lilium longiflo- Vienna rum. J Plant Physiol 164:878–885. https://doi.org/10.1016/j.jplph​ ​ Ramsey M, Vaughton G (2000) Pollen quality limits seed set in Bur- .2006.05.013 chardia umbellata (Colchicaceae). Am J Bot 87:845–852 Thomson JD, Wilson P, Valenzuela M, Malzone M (2000) Pollen Rešetnik I, Liber Z, Satovic Z et al (2007) Molecular phylogeny and presentation and pollination syndromes, with special reference systematics of the group (Liliaceae) based on to Penstemon. Plant Species Biol 15:11–29. https​://doi.org/10.10 nuclear ITS sequences. Plant Syst Evol 265:45–58. https​://doi. 46/j.1442-1984.2000.00026​.x org/10.1007/s0060​6-006-0513-y Vogler DW, Stephenson AG (2001) The Potential for mixed mating in Rodger JG, van Kleunen M, Johnson SD (2013) Pollinators, mates and a self-incompatible plant. Int J Plant Sci 162:801–805. https://doi.​ Allee effects: the importance of self-pollination for fecundity in org/10.1086/32078​7 an invasive lily. Funct Ecol 27:1023–1033 Webb CJ, Lloyd DG (1986) The avoidance of interference between the Rodríguez-Riaño T, Ortega-Olivencia A, Devesa JA (1999) Repro- presentation of pollen and stigmas in angiosperms II. Herkog- ductive biology in two Genisteae (Papilionoideae) endemic of amy. N Z J Bot 24:163–178. https​://doi.org/10.1080/00288​ the western Mediterranean region: Cytisus striatus and Retama 25X.1986.10409​726 sphaerocarpa. Can J Bot 77:809–820 Wickman P-O (1996) Butterfly leks. Entomol Tidskr 117:73 Rossi G, Montagnani C, Gargano D et al (2013) Lista rossa della flora Woodward FI (1987) Climate and plant distribution. Cambridge Uni- italiana. 1. Policy species e altre specie minacciate. pp 1–58 versity Press, Cambridge Routley MB, Husband BC (2003) The effect of protandry on siring Yokota S, Yahara T (2012) Pollination biology of Lilium japonicum var. success in Chamerion angustifolium (Onagraceae) with different abeanum and var. japonicum: evidence of adaptation to the differ- inflorescence sizes. Evol Int J Org Evol 57:240–248 ent availability of diurnal and nocturnal pollinators. Plant Species Routley MB, Bertin RI, Husband BC (2004) Correlated evolution of Biol 27:96–105. https://doi.org/10.1111/j.1442-1984.2011.00336​ ​ dichogamy and self-incompatibility: a phylogenetic perspective. .x Int J Plant Sci 165:983–993. https​://doi.org/10.1086/42388​1 Zych M, Stpiczyńska M (2012) Neither protogynous nor obligatory Sakazono S, Hiramatsu M, Huang K-L et al (2012) Phylogenetic rela- out-crossed: pollination biology and breeding system of the tionship between degree of self-compatibility and floral traits European Red List Fritillaria meleagris L. (Liliaceae). Plant Biol in Lilium longiflorum Thunb. (Liliaceae). J Jpn Soc Hortic Sci 14:285–294. https​://doi.org/10.1111/j.1438-8677.2011.00510​.x 81:80–90. https​://doi.org/10.2503/jjshs​1.81.80 Schemske DW, Husband BC, Ruckelshaus MH et al (1994) Evaluating approaches to the conservation of rare and endangered plants. Ecology 75:585–606. https​://doi.org/10.2307/19417​18

1 3 Journal of Plant Research is a copyright of Springer, 2018. All Rights Reserved.