Different Morphs of Heterostylous Plant (Tirpitzia Sinensis) Associated With Floral Characters Have Various Adaptation Strategies to Pollinator to Ensure Reproductive Success


 Background: Heterostylous plants are commonly associated with pollinators promoting disassortative pollination. How different morphs adapt to pollinators to ensure reproductive success and whether floral characters (such as pistil, stamen and nectar traits) are relevant to this process remain unclear. Results: Tirpitzia sinensis is distylous flowers. Its floral characters were measured. Field observation of effective pollinator and pollination efficiency to different morphs was conducted, and breeding systems were examined by hand pollination treatments. Our results showed that L-morph produced more but smaller pollen grains per anther than S-morph. T. sinensis secreted more nectar at night as a possible adaptation to hawkmoths (Macroglossum spp.) active at dusk. L-morph produced much nectar due to its large secretion volume at night. The nectar was rich in sucrose. The sucrose/hexose ratio was higher in the nectar of S-morph than in that of L-morph. S-morph had higher visit rate but lower pollination efficiency of hawkmoths compared with L-morph. T. sinensis was self-incompatible. Seed sets in nature did not differ between morphs. Conclusions: Our findings suggest that L-morph increases its pollination efficiency through outstanding stigma, many pollen grains per anther, and high amount of nectar. S-morph attracts many hawkmoths to pollinate due to its high sucrose/hexose ratio in its nectar.


Difference in oral characters between the two morphs
The leaves were relatively larger in the S-morph than in the L-morph (marginal signi cant difference, see Table 1). The S-morph has larger sepal, petal width, oral tube depth, stamen length and pollen size than the L-morph (all P < 0.05). The ower length, petal length, pistil length, anther length, pollen grain number and pollen/ovule ration of L-morph were larger than those of S-morph (all P < 0.05). Other oral traits including ower width, opening diameter, anther width, anther thick, single owering days and ovule number were not signi cantly different between the two morphs (all P > 0.05) ( Table 1). Generally, Lmorph had longer pistil, produced more pollen grains per ower and consequently had higher pollen/ovule ratios than S-morph. Meanwhile, S-morph had longer stamen and larger pollen than Lmorph. Table 1 Comparisons of vegetative and reproductive traits (mean ± SE) between long-styled morphs (hereafter L-morph) and short-styled (hereafter S-morph) morph of Tirpitzia sinensis tested by a generalized linear model (GLM)analysis. Values of one morph signi cantly larger than the other are written in bold.  (Fig. 1B).
Fructose, glucose and sucrose were detected in T. sinensis nectar through HPLC analysis. The sucrose contents (9.15 ± 0.65) in nectar (all nectar property data of L-and S-morphs combined together and analysed) were signi cantly higher (P < 0.001, Waldχ 2 = 51.145, df = 2) than the fructose (4.75 ± 0.34) and glucose contents (5.96 ± 0.37). The nectar sugar properties of L-and S-morphs had no signi cant difference (P = 0.212, Waldχ 2 = 1.556, df = 1). For the nectar of the L-morph, the sucrose contents were signi cantly higher than the fructose contents, and glucose contents had no signi cant difference with sucrose and fructose contents (P = 0.005, Waldχ 2 = 10.538, df = 2) (Fig. 1C). For the nectar of the S-morph, the sucrose contents were signi cantly higher than the fructose and glucose contents, and the fructose and glucose contents had no signi cant difference (P < 0.001, Waldχ 2 = 48.048, df = 2) (Fig. 1C). Baker and Baker (1983) conducted a sucrose to hexose ratio (r) assessment, where r equals to amounts of sucrose / (amounts of glucose + amounts of fructose). The nectar of L-morph of T. sinensis was rich in sucrose because the r equals to 0.696, 8.5 / (6.7 + 5.5), and the nectar of S-morph of T. sinensis was sucrose dominant, r = 1.1, 10 / (3.8 + 5.0). The sucrose/hexose ratio was higher for S-morph owers than for L-morph owers.

Pollinator species and abundance
Hawkmoth (Macroglossum), bumblebees (Bombus) and honeybees (Apis) were the major oral visitors of T. sinensis through the eld observation in 2018 and 2019. The long-tongued hawkmoths probed the nectar at the base through the narrow oral tube. In this process, the pollen was removed and deposited onto the tongue, which was then stained with pollen and could touch the stigma to achieve effective pollination ( Fig. 2A, B). The bumblebees always only robbed the nectar (Fig. 2C), and the honeybees mainly groomed the T. sinensis pollen into their corbiculae (Fig. 2D). We also found that Macroglossum frequently pollinated the T. sinensis owers at 18:00-20:30 but seldom during the rest of the day.

Discussion
Hawkmoth owers produce copious nectar to support the large energy requirement of the visitor. The nectar is usually diluted to ensure that it ows easily through the thin proboscis [20][21][22]. The Petunia axillaris were pollinated by hawkmoths, and the nectar volume was high (approximately 35 µl), but the sugar concentration (approximately 16%) was low [23]. The Agrius convolvuli pollinated by long-tongued hawkmoths produced higher volume (approximately 15 µl) and dilute (sugar concentration approximately 27%) nectar [24]. Similarly, the nectar sugar concentration of T. sinensis was low. Although one ower secreted a small amount of nectar, the plant still has numerous owers.
Flowers may control their nectar and thus affect pollinator's behaviours [21]. By manipulating the nectar volume of Mirabilis multi ora, Hodges [25] showed that hawkmoths always visit e owers on plants with a large amount of nectar. The amount of nectar produced by mutational Petunia integrifolia individuals per ower was only 1/3 of that produced by wild-type individuals; hawkmoth pollinators reduced the probing duration on low-nectar plants when they are exposed simultaneously to low-nectar and wild-type P. integrifolia [23]. The nectar volume of T. sinensis secreted during the night was higher than that secreted during the day, a possible adaptation strategy to Macroglossum pollinator ying at dusk. Lmorph secreted more nectar than S-morph and could increase the probing duration of hawkmoths and consequently increase the pollination e ciency. The nectar composition could potentially act as an important agent for visitor activities. Hummingbirds, hawkmoths and long-tongued bees prefer sucroserich nectar, and short-tongued bees and ies prefer hexose (glucose and fructose)-rich nectar [15,21]. The nectar of hawkmoth-pollinated Petunia axillaris was also rich in sucrose (approximately 57% in nectar sugar proportion) [23]. The nectar of buckwheat pollinated by honeybees is rich in glucose and fructose [15]. The nectar of T. sinensis is rich in sucrose as an adaptation to hawkmoth pollination.
The natural seed production between the different morphs of distylous plants always has no signi cant difference to maintain the balance between the numbers of L-and S-morphs in one population, such as in Pentanisia species (Rubiaceae) and Fagopyrum esculentum (Polygonaceae) [15,26,27]. The seed set of the L-and S-morphs of T. sinensis did not differ signi cantly. The seed set with the control treatment of Lmorph was signi cantly lower than that with intermorph pollination treatment. The seed set with control treatment and intermorph pollination treatment of S-morph had no signi cant difference. These results indicated that the L-morph of T. sinensis exhibited a legal pollen (intermorph pollen) limitation may due to the low visit rate of hawkmoth to this morph type.
Heterostyly promotes compatible pollination between oral morphs within species since Darwin [1]. Therefore, the legal and illegal pollens on the stigma of L-and S-morphs must be detected to test the above hypothesis. Most heterostylous species are tubular owers (such as Tirpitzia sinensis), and some are bowl-shaped owers. However, the reproductive strategies of the two types of corolla formations in heterostylous plants remain unknown. The tubular ower T. sinensis is pollinated by hawkmoths (Macroglossum), and the bowl-shaped ower Linum suffruticosum is pollinated by bee-ies (Bombyliidae) [28]. We plan to explore whether and how distylous plants with different corolla shapes promote compatible pollination by using two species in Linaceae disturbed in China including Tirpitzia sinensis (tubular ower) and Linum usitatissimum (open bowl-shaped ower).

Conclusion
The distylous T. sinensis was effectively pollinated by hawkmoths (Macroglossum). The S-morph had higher visit rate of hawkmoths but lower pollination e ciency of hawkmoths compared with the L-morph. The visitor Bombus always robbed the nectar, and Apis gathered the pollen as pollen thieves. T. sinensis produced nearly no seeds with intramorph pollination, and the natural seed production of both morphs had no signi cant difference. The long oral tube and nectar traits (large amount of nectar secreted during the night, low nectar sugar concentration and high sucrose contents in the nectar) of T. sinensis may be used to adapt to hawkmoth pollination. Moreover, the outstanding stigma, high pollen amount per anther and large amount of nectar of L-morph could be employed to adapt to the pollinators and increase the pollination effectiveness of hawkmoths to L-morph. L-morph stigmas are usually accessible to pollinator [7,13,27]. The high sucrose/hexose ratio in the nectar of S-morph might attract many hawkmoths to visit the in orescence. Flowers are homogamy and usually last 3-4 days. Plants usually ower from May to August. Capsules contain three to eight seeds and mature 3 months after fertilisation [29]. Our eld investigation of pollination ecology revealed that T. sinensis is a typical distylous plant with L-morph (anthers are placed low in the corolla, and stigmas are located high) and S-morph (anthers are placed high, and stigmas are located low) ( Figure 1A

Difference in traits between the L-and S-morphs of T. sinensis
To compare plant performance between the two morphs, we randomly selected 50 plants (each plant selected one ower) per morph and measured two vegetative and thirteen reproductive traits, including leaf length and width; sepal length and width; ower length, width and opening diameter; tube depth; petal length and width; stamen length; pistil length; and anther length, width and thickness to 0.01 mm using a caliper micrometre.
To compare pollen, ovule production and pollen size, we selected 30 ower buds from 30 L-morph individuals and 30 S-morph individuals respectively and stored them in a 1.5 ml centrifuge tube lled with 75% alcohol for xation and preservation. The anther and ovary from one ower were separated using forceps in the laboratory, and the anthers were suspended in 500 ml of water. Three drops (each drop of 50 ml) of every pollen solution sample were counted under the Nikon E100 optical microscope. The mean of the three pollen drops was multiplied by 10 to estimate the pollen production of one ower. The ovules were counted under a stereomicroscope. The P/O ratio was equal to the number of pollen grains divided by the corresponding ovule number. For pollen size estimation, three pollen grains per ower were rst photographed, and the length and width were then measured using Digimizer Version 4.6.0. To compare the single ower period between the two morphs, we marked one bud of the above selected 30 individuals each for L-and S-morph and recorded the rst day of the opening state. Every 2 days, we recorded the ower opening state until the anthers and pistil lost function. These days were denoted as the single ower period.

Measurements of nectar volume, sugar concentration and properties of T. sinensis
To compare the nectar volume and concentration in T. sinensis during anthesis between day and night, we bagged and labelled 30 owers before anthesis from 30 plants each from L-and S-morphs. During the male owering phases, the nectar in the bagged ower was removed using glass microcapillary tube (0.3 mm in diameter) on the day before the measurement. Nectar was extracted from the owers bagged from 18:30 to 06:30 (secreted during the night). After the treatments, the same ower was bagged again, and the nectar was extracted from 06:30 to 18:30 (secreted during the day) the next day. The length (L) of the microcapillary tube occupied by nectar was measured using a caliper micrometre. The volume (V total) and length (L total ) of one standard microcapillary were calculated, and the volume of nectar (V) is equal to L/L total * V total . And the concentration of nectar was measured with a hand-held refractometer (Eclipse 0%-50%; Bellingham and Stanley Ltd., Basingstoke, United Kingdom; see [30]. To measure sugar components, we collected nectar from control-bagged owers of L-morph (30 owers from 30 individuals) and S-morph (27 owers from 27 individuals) of T. sinensis by using microcapillary tubes. After the nectar length was measured using a caliper micrometre, the nectar was spotted onto lter paper and was air-dried at room temperature [31]. The spotted lter papers were placed in a 1.5 ml centrifuge tube and stored in the refrigerator at −20 °C. The sugars were removed by elution with 100 µl of deionised water at room temperature for 24 h. Sugar type (glucose, fructose, sucrose and maltose) was identi ed, and the relative mass was quanti ed by High Performance Liquid Chromatography (HPLC, Waters Corporation, Milford, Massachusetts) with a refractive index detector and an Agilent Zorbax carbohydrate analysis column 843300-908 (Agilent Technologies, Santa Clara, California) under the column temperature of 35 °C. The mobile phase was 80% acetonitrile, the ow rate was 1 ml/min and the injection volume was 20 µl. Quantities of each sugar in nectar samples were determined by the standards (glucose, fructose, sucrose and maltose) using the regression equations (based on response peak areas to standard sugar mass) and were expressed as relative percentage by mass [32].

Pollinator species and abundance
To Pollen transfer e ciency of hawkmoths To compare the pollination e ciency of hawkmoth between the L-and S-morphs, we estimated the pollen removal and receipt per morph. Male-phase in orescence (previously unvisited) were bagged until anther dehiscence, and each in orescence was allowed a single visit by a hawkmoth. To estimate pollen removal, we collected 48 visited owers for L-morph and 46 visited owers for S-morph from different plants with another 48 L-morph buds and 46 S-morph buds as the control. Each ower was stored in a 2 ml centrifuge tube with 75% alcohol. Pollen removal per ower was calculated from the mean number of pollen grains in unvisited owers minus the mean number of pollen grains remaining after one visit. To estimate pollen receipt per visit, we removed undehisced anthers from the 48 male-phase owers for L-morph and 46 male-phase owers for S-morph and bagged these owers with cotton mesh until they developed into the female phase. These female-phase in orescences were then removed from the bag and allowed one visit by the hawkmoth. Stigmas of these visited emasculated owers were collected and stored in a 1.5 ml centrifuge tube with alcohol. Pollen grains from the anthers and on the stigmas were counted under a light microscope (Nikon E100). The anthers were fully mashed with tweezers to form 0.5 ml of pollen suspension. Three drops (each drop of 50 μl) of every pollen solution sample were counted, and the mean was multiplied by 10 to estimate pollen production (for undehisced anthers) or pollen remaining per ower (one single visited anther) [33].

Breeding system
To determine whether T. sinensis is self-and intramorph incompatible, we conducted arti cial pollination experiments as follows: (1)  The remaining two owers were used as the control and autogamy pollination treatments. Three months after pollination, seeds per ower of six pollination treatments were collected and counted.

Data analysis
To assess the differences in plant performance between L-and S-morphs, we compared 15 plant vegetative and reproductive traits, single owering days and P/O (pollen number/ ovule number) using a generalized linear model (GLM) with normal distribution and identity-link function. The pollen number and ovule number between L-and S-morphs were compared using Poisson distribution with loglinear-link function in GLM (all plant characters as dependent variable, and L-and S-morphs as factors). Nectar volume and sugar concentration were analysed using GLM with normal distribution and identity-link function (nectar volume and sugar concentration as dependent variables, and L-and S-morphs and day and night as factors) to compare the nectar traits of the two morphs between day and night. Glucose, fructose, sucrose and maltose contents in nectar were examined using GLM with normal distribution and identity-link function (sugar components as dependent variables, and L-and S-morphs as factors) to compare the sugar components between the two morphs. Data of visits were analysed using GLM with normal distribution and identity-link function (visitation rates as dependent variables, and ower morphs and visitor types as factors) to compare the visiting rates (visits/ ower/hour) of all visitors between the two morphs. Pollen removal and receipt between the two morphs were compared using GLM with Poisson distribution with loglinear-link function (pollen number as dependent variable, and L-and S-morph as factors). Seed sets of all treatments were examined with binary logistic analysis in GLM (full seed number as event variable, total ovule number as trait variable, and pollination treatments and ower morph as factors) to compare the reproductive success of six pollination treatments between the two morphs.
All data were analysed in SPSS 20.0 (IBM Inc., New York, NY) software.   Comparison of visits/ ower/hour between honeybee, bumblebee and hawkmoth in S-and L-morphs. Bars sharing the same letters are not signi cantly different in visit rates among three visitor groups in both morphs. Arrows indicated that the visit rates of hawkmoth to S-morph are signi cantly higher than to Lmorph.