Features of floral odor and nectar in the distylous Luculia pinceana (Rubiaceae) promote compatible pollination by hawkmoths

Abstract It is hypothesized that in heterostylous plant species, standardization of signals of floral attraction between different morphs is advantageous, encouraging flower visitors to switch between morphs. It remains unclear whether signals of floral attraction (floral odor and properties of nectar) are similar between morphs in distylous species pollinated by hawkmoths, and how these relate to hawkmoth behavior. We observed the behavior of visitors to distylous Luculia pinceana (Rubiaceae), collected and analyzed floral odor, and examined properties of nectar (volume, sugar concentration, and composition) of long‐styled and short‐styled morphs during the day and night. Pollinator responses to the floral scent were tested with a Y‐tube olfactometer. We conducted diurnal and nocturnal pollination treatments and six other pollination treatments to test the importance of nocturnal pollinators and to examine the self‐incompatibility system. A species of hawkmoth, Cechenena lineosa, was the effective pollinator. The floral odor was rich in methyl benzoate, and sucrose was dominant in the nectar. There were no significant differences between the two morphs in the methyl benzoate content or the properties of nectar. Flowers released more methyl benzoate and secreted larger volumes of nectar with lower sugar concentration at night than during the day. The hawkmoth had a significant preference for methyl benzoate. Luculia pinceana was partially self‐incompatible and relied on nocturnal pollinators for reproductive success. This study verifies that floral attraction signals are consistent between different morphs in this distylous species, promoting compatible pollination, and the features and the diel pattern of these signals between day and night are adapted to hawkmoth behavior.


| INTRODUC TI ON
Plant-pollinator interactions play an important role in shaping floral diversity in angiosperms (Fenster et al., 2004;Harder & Johnson, 2009). Heterostyly is a polymorphic system in which individual plants of a species have flowers of either of two (distyly) or three (tristyly) floral morphs. Distyly is usually accompanied by self-incompatibility, which prevents self-and intramorph fertilization, and promotes intermorph pollination (transmission and receipt of pollen between anthers and stigmas at the same level, also called compatible pollination; Baker, 1964;Barrett, 2002;Darwin, 1877;Lloyd & Webb, 1992). Heterostylous plants commonly have narrow tubular corollas and achieve compatible pollination via long-tongued pollinators, like bees, butterflies, and flies (Lloyd & Webb, 1992).

Standardization of floral signals between different heterostylous
morphs is advantageous, encouraging flower visitors to switch from one morph to another morph (reviewed in Lunau et al., 2017).
If these signals differ between floral morphs (long-styled or shortstyled morphs), driving floral visitors to one morph preferentially, more visitations will result in incompatible pollen transfer due to the self-incompatibility of heterostylous species.
There was no significant difference in relative amounts of volatile compounds between the two morphs of Primula elatior or P. farinosa (Gaskett et al., 2005).
In tropical and temperate parts of the globe where dusk temperatures are high, hawkmoths are crucial for the compatible pollination of distylous plants (McMullen, 2012;Wang et al., 2022).
Even in low light conditions, hawkmoths can accurately locate a flower and extend the proboscis into the base of the flower tube to extract nectar (Jürgens, 2004;Wiesenborn & Baker, 1990). In hawkmoth-pollinated flowers, floral odors are generally dominated by terpenes, aromatic esters, and/or nitrogen compounds (Kaiser & Tollsten, 1995;Raguso et al., 2003) and hawkmoths have a preference for sucrose-rich nectar (Nicolson et al., 2007). For example, the scent emission and mean number of scent compounds of African woodland orchid were highest at 19:30 h and the scent compounds differed between day and night. The nectar volume increased throughout the day and peaked just before the onset of pollinator activity, and the nectar was rich in sucrose (Balducci et al., 2020).
Qualea grandiflora secreted nectar with a high proportion of sucrose and only at night, consistent with hawkmoths' foraging period and preference (Potascheff et al., 2020). These diel patterns and features of floral scent emission and nectar production of plants are adapted to hawkmoth behavior. However, it remains unclear whether the floral attraction signals (floral odor and nectar properties) of L-and Smorphs in distylous hawkmoth-pollinated plants are similar and how these traits are related to hawkmoth pollination.
The distylous Luculia pinceana (Rubiaceae) has a long narrow corolla tube with obvious nectar, white flowers, and strong fragrance.
It is an ideal species for testing whether the floral odor and nectar properties are similar between two morphs and how these signals are adapted to hawkmoth behavior. This study aims to (1) explore the effective pollinator of the distylous L. pinceana; (2) identify the main floral odor compounds in the two morphs during the day and at night and test whether pollinators have a preference for the main odor compounds; (3) compare the nectar volume, sugar concentration, and sugar composition of the two morphs and compare the nectar volume and sugar concentration of the two morphs between day and night; (4) investigate the relative contribution to pollination of diurnal and nocturnal pollinators and examine the breeding system and test whether L. pinceana is self-incompatible.

| Study materials and study sites
Luculia pinceana (Rubiaceae) is a perennial evergreen shrub or tree, 2-10 m tall, mainly distributed in forests or thickets on mountain slopes and streamside in valleys at an elevation of 600-3000 m in Yunnan and Guangxi provinces of China and in Thailand and Vietnam. The cymose inflorescence (Figure 1) is large and graceful, with a strong floral scent, and is generally terminal at the top of a stem or branch. The corolla is usually pink, sometimes white, and the corolla tube is cylindrical and glabrous. The stamens are inserted at the throat of the corolla tube at a level that depends on the morph, and the anthers are linear oblong. The flowering period generally lasts from June to December. The seeds are numerous, nearly elliptic, with wings at both ends. Luculia pinceana is a typical distylous plant with an L-morph (with anthers low in the corolla and high stigmas) and an S-morph (with anthers high and stigmas low; Figure 1a-
In each plot, we randomly chose about 50 blooming flowers from two or three individuals per morph. When one visitor came, the visit number per foraging bout, visitor species, and foraging behavior (for the nectar or pollen) were recorded, and the total open flowers in each plot were counted. The visit frequency of one visitor was equal to the mean number of visits per flower per hour. If the visiting insect extended its long tongue into the corolla tube and touched the anthers (removing pollen grains) and stigmas (depositing pollen grains), it was considered an effective pollinator. The species and behavior of visitors were recorded using cameras (Nikon D7500).
The pollinators were considered effective not only based on their foraging behavior (extending its long tongue into the corolla tube and touching the anthers and stigmas) but also the mechanical fit between the hawkmoth tongue length and the floral tube length of L. pinceana (the data have been published in Chen et al., 2021).
Sunrise was about 06:00 h and sunset was about 18:00 h during this period.

| Floral odor collection and component analysis
To detect the floral odor composition of L. pinceana in the field, floral volatiles were collected using dynamic head-space methods (Dotterl et al., 2005). Each sample contained 10 newly opened flowers and some leaves enclosed within a disposable polyethylene oven bag The flow rate was adjusted to approximately 200 mL/min using a power supply and a flow meter and each collection lasted 30 min.
A transparent quartz glass test tube (length: 20 mm, inner diameter: 2 mm, Lianyungang Dong Hong Quartz Glass Company) was filled with Tenax-TA powder (mesh 60-80) and used as the adsorbent tube. The adsorbent powder was fixed in the tube using glass wool (HJ637-2018). To control for scent from leaves and the ambient air, three samples were simultaneously taken from oven bags with branches including just leaves from three individuals (Feng et al., 2023). After collection, both ends of the adsorbent tube were F I G U R E 1 (a) The cymose inflorescence of the long-styled morph (the long style marked with a red arrow) of Luculia pinceana; (b) the cymose inflorescence of the short-styled morph (the long anther marked with a pink arrow) of L. pinceana; (c) flowers of the short-styled morph (left) and long-styled morph (right) showing the reciprocal positioning of stigmas (red arrows) and anthers (pink arrows); (d,e,f) the longtongued Cechenena lineosa foraged for nectar; the tongue could touch the stigma and anther and pollinate the distylous L. pinceana; (g) the hawkmoth C. lineosa; the tongue length was about 6 cm which is mechanical fit with the L. pinceana floral tube; (h) a bumblebee robbing nectar from the bottom of the corolla tube; (i) a honeybee gathering pollen grains on a short-styled inflorescence. sealed with plastic film, and the samples were placed into brown injection bottles and stored at −18°C for about 3 months.
In the laboratory, each sample was washed with a mixture of nhexane (100 μL) and ether (100 μL), extracted ultrasonically for 30 min, and centrifuged at 8609 g for 3 min. The volume of each sample solution (mL) was recorded. The floral scent of L. pinceana was analyzed quantitatively on a Varian 450 gas chromatography (GC)-Varian 320 TQ mass spectrometer (MS) on a Varian Saturn 2000 System. The GC was performed with Agilent 7694 E headspace samplers. The conditions of GC were as follows: Agilent 19,091 J -413 column (30.0 m long; inner diameter, 320 μm; film thickness, 0.25 μm; nonpolar); the starting temperature of 60°C was maintained for 3 min, increased from 20°C/min to 290°C, and kept for 1 min. The injection volume was 2 μL. Nondiversion mode was adopted. The injection inlet temperature was 280°C. The carrier gas was helium. The conditions of MS were as follows: the ion source temperature was 300°C, electron energy was 10 V at full scanning mode, the transmission line tem-

| Behavioral preference of pollinating hawkmoths
To examine whether the effective adult pollinators of L. pinceana have a preference for the main floral scent component, 60 hawkmoths were captured with high-pressure mercury vapor lamps (450watt, Shanghai Yaming Factory) at night in Da Xi Factory, Yunling Village, Djinchang Town, Ma Lipo County (104°84′60″ E, 23°17′9″ N) in August 2020. After capture, the hawkmoths were tested immediately on the same night. The preliminary experiment showed that hawkmoths that were kept in a cage did not make a preference choice, perhaps because they were frightened in the breeding cage.
When different odor sources were connected to each end of the Y-tube olfactometer, the insects crawled toward the favorite odor.
Considering all the features of the equipment and the purpose of this research, we designed a bespoke Y-tube glass olfactometer with one vacuum pump, two drying towers, two flowmeters, and two gas washers (produced by Nanjing Shelly company). The Y-tube had three channels each 100 cm long (Jarriault et al., 2009), with an inner diameter of 25 cm, schematically shown in Figure 2c. The wingspan of the hawkmoth was 73.13 ± 11.74 (mm), the body length was 49.12 ± 4.65 (mm; Chen et al., 2021). The channel was big enough for the hawkmoth to crawl and fly.
The methyl benzoate standard was diluted with n-hexane to 0.2 mg/mL to correspond with the detected average content of methyl benzoate (the main chemical substance) in L. pinceana floral odor. The methyl benzoate and n-hexane blend were randomly placed at one end of the Y-tube olfactometer, and n-hexane was placed at the other end as a control. Odors were passed from both arms to the stem in equal flow rates of cleaned and humidified air flow created by an air pump system (super silent positive pressure vacuum pump SCI-HS25ZF: voltage 220 V, power 1.5 W, average flow 3 L/min, air pressure 0.016 MPA, pressure adjustable) via an activated charcoal filter and distilled water. To improve experimental operation without interfering with the hawkmoth's behavior, a small flashlight (KM-8911 lamp, Kangming Company) covered with thick red plastic film (radiating faint red light) was placed more than 3 m away from the olfactometer and not directly exposed to it. The olfactometer was surrounded with a dark environment. Liu and Huang (2013) showed that a small flashlight covered with thick red plastic film could be used for nocturnal moth observations and would not disturb the moth's foraging behavior. Each hawkmoth was tested only once, and its behavior was assigned to one of the three choices: (1) flying completely to the floral scent source and staying at the end for more than 10 s; (2) flying to the control end; (3) if the hawkmoth flew irregularly for 10 min, it was deemed that no choice had been made (Wan et al., 2015). We repeated this experiment 60 times (i.e., 60 hawkmoths) and compared the number of different choices using a G-test of goodness of fit to see whether the hawkmoths preferred one end to the other.
An inflorescence (with 10 flowers) and a branch (with 10 leaves) from the same plant of L. pinceana were picked and the inflorescence was randomly placed at one end of the Y-tube olfactometer, and the branch was placed at the other end as control. In total, we got 30 groups of inflorescences and branches from different individuals. A total of 30 hawkmoths were captured and we tested the preference choice response to the inflorescence or branch with the methods described above. Each test used a new inflorescence and branch.
During the experiment, the methyl benzoate or inflorescence (treatment) and the hexane or branch (control) were randomly placed on either side of the Y-tube olfactometer. Each hawkmoth was tested only once. After testing and before testing the next hawkmoth, the olfactometer was cleaned firstly with 75% ethanol and then with distilled water, and dried with a clean dry cloth, to prevent the accumulation of odor. After each measurement, the tested hawkmoth was temporarily placed in a cage and then released back to the field.

| Nectar volume, sugar concentration, and sugar composition of Luculia pinceana
To compare the nectar volume and sugar concentration from the same individual L. pinceana flower between day and night during anthesis, we bagged and labeled 50 buds with nylon mesh bags per morph from 50 L-morph individuals and 50 S-morph individuals.
During blooming, the nectar in the bagged flower (on the first day of anthesis) was removed using glass microcapillary tubes (0.3 mm in diameter), which were inserted into the base of the corolla. The nectar volume secreted during the day (from 06:00 to 18:00 h) was measured at 18:00 h. The same flower was bagged again after the measurement. The nectar volume secreted from the same flower during the night (from 18:00 to 06:00 h the next day) was measured at 06:00 h the next day. Then, the length (L) of the microcapillary tube occupied by nectar was measured using a caliper micrometer (accurate to 0.01 mm). The nectar volume was equivalent to π * 0.15 2 * L. The sugar concentration of each nectar sample was measured using a hand-held refractometer (Eclipse 0%-50%; Bellingham and Stanley Ltd.) and the units of the nectar sugar concentration were g sucrose per 100 g solution, known as % Brix (Corbet, 2003). Three buds in L-morph and four buds in S-morph were not fully developed and we only get 47 samples for L-morph and 46 samples for S-morph.
To examine the composition of the nectar, we used 30 blooming flowers per morph from 30 individuals. The nectar of each flower was extracted with a capillary tube, and the nectar volume was measured.
Then, the nectar was blown onto a filter paper and air-dried at room temperature. The filter papers with nectar were placed in a 2-mL centrifuge tube and stored in a −18°C refrigerator. Nectar samples were dissolved in 100 μL of deionized water for 24 h at room temperature before detection. The sugar contents of nectar (mainly glucose, fructose, sucrose, and maltose) were analyzed by high-performance liquid chromatography (HPLC) with a refractive index detector and an Agilent Zorbax carbohydrate analysis column 843300-908 (Agilent Technologies). The mobile phase was an acetonitrile: water system (80:20 by volume) at a flow rate of 1 mL/min. The column temperature was 35°C, and the injection volume was 20 μL. The quantities of each sugar in nectar samples were calculated by comparison with standards (fructose, glucose, sucrose, and maltose) using the regression equation (based on response peak to standard sugar) and were expressed as relative percentages by mass . plants per morph and selected eight buds in each plant. Three of the eight buds were used for the pollination treatments: (1) Open pollinated treatment as control: flowers were always exposed; (2) nocturnal pollination: the flowers were bagged with nylon mesh bags from 06:00 to 18:00 h and exposed from 18:00 to 06:00 h the next day; (3) diurnal pollination: the buds were exposed from 06:00 to 18:00 h and bagged with nylon mesh bags from 18:00 to 06:00 h the next day. The pollination treatments continued until the flowers faded. The other five buds on each plant were labeled with cotton threads of different colors. Four of the five buds were emasculated and bagged with nylon mesh bags until they developed into the female phase and then received the following pollination treatments:

| Pollination treatments
(4) self-pollination (using pollen from the flowers of the same individual); (5) test for apomixis (after emasculation, the flowers were always bagged); (6) intramorph pollination (pollen from flowers of the same morph); (7) intermorph pollination (pollen from flowers of the other morph). The last one bud was used as (8) autonomous

| Data analysis
We compare the nectar volume/nectar sugar concentration between day and night (time factor), between L-and S-morph (morph factor) in generalized linear model (GLM) with normal distribution and identity link function in SPSS 20.0, with nectar volume (μL)/nectar sugar concentration (%) as dependent variable, day and night as factor 1, Land S-morph as factor 2, and the interaction between factor 1 and 2 was also analyzed. To detect the difference of nectar volume/nectar sugar concentration between day and night in L-morph/S-morph, we compare the nectar volume/nectar sugar concentration between day and night in L-morph/S-morph in GLM with normal distribution and identity link function (nectar volume (μL)/nectar sugar concentration (%) in L-morph/S-morph as dependent variable, day and night as factor).The sugar composition was compared between the two morphs in GLM with normal distribution and identity link function (with glucose, fructose, and sucrose components (μg/μL) as dependent variables, and L-and S-morphs as factors). The methyl benzoate amount between L-and S-morphs and between day and night was compared in GLM with normal distribution and identity link function (with methyl benzoate amounts (mg/mL) as the dependent variable, and day and night as factor 1, L-and S-morphs as factor 2, the interaction between factor 1 and 2 was also analyzed). A G-test of goodness of fit was used to see whether hawkmoth pollinators prefer one end over the other during the Y-tube olfactometer test.
The seed sets of different pollination treatments in L-and S-morph were compared using binary logistic analysis in GLM (with full seed number as the dependent variable, total seed number as trials variable, and different treatments as factor 1, two morphs as factor 2, the interaction between factor 1 and 2 was also analyzed). All data were analyzed in SPSS 20.0 (IBM Inc.) software.  and terpenes (about 1.55%). Some iron peaks do not match chemicals (unknown compounds) during our analysis and the summary of these relative amount for these was 7.55 ± 0.35 (%). Methyl benzoate, 2,3-butanediol, palmitic acid, acetic acid, hydrazide, (e)isoeugenol occurred in relatively large amounts (>2%) and were consistently abundant across samples in 2019 and 2020. The scent was dominated by methyl benzoate (C 8 H 8 O 2 ; an ester; the relative amount was about 32%; Figure 2a). Methyl benzoate was not detected in the control floral odor samples and the control samples mainly matched 2,3-butanediol (the relative amount was about 77.54%) and several alkane chemicals (the relative amount was about 10.65%). The summary of the relative amount of these unknown compounds in control samples was 9.55 ± 0.29 (%; Table 1).

| Floral odor composition
Generally, there was no significant difference in the methyl benzoate content between L-morphs (0.22 ± 0.02 mg/mL, n = 23) and S-morphs (0.21 ± 0.01 mg/mL, n = 26) flowers (the methyl benzoate content data of day and night of each morph combined together and analyzed), but significantly more methyl benzoate was released at night (0.25 ± 0.02 mg/mL, n = 22) than by day (0.19 ± 0.01 mg/mL, n = 27; the methyl benzoate content data of Land S-morphs during the day or at night were combined together and analyzed; Table 2). There was no interaction between floral morphs and emission times with respect to the methyl benzoate content ( Table 2). The L-morph inflorescences released significantly more methyl benzoate at night than during the day (1.59 times as much; the mean methyl benzoate content at night divided by that during the day, Wald χ 2 = 5.290, df = 1, p = .021), whereas in the S-morph, there was no significant difference between night and day in the amounts of methyl benzoate released (Wald χ 2 = 1.519, df = 1, p = .218). During the day, there was no significant difference between the two morphs in the methyl benzoate content (Wald χ 2 = 0.437, df = 1, p = .508). The same was true at night (Wald χ 2 = 1.581, df = 1, p = .209; Figure 2b).

| Preference of pollinators on floral scent
During the Y-tube olfactometer test, the number of C. lineosa individuals flying to the methyl benzoate end (37) was significantly higher (p = .022, G = 5.243) than the number flying to the control end (19), and four C. lineosa did not make a choice. The number of C. lineosa flying to the inflorescence end (20) was significantly higher (p = .019, G = 5.524) than the number flying to the branch (as control) end (7), and three C. lineosa did not make a choice (Figure 2d).
These results indicated that the effective pollinator C. lineosa had a significant preference for the methyl benzoate and the L. pinceana inflorescence.
Moreover, the two morphs differed in the pattern of nectar volume and sugar concentration secreted during the day and at night. The L-morph secreted a significantly larger volume at night than by day (1.40 times as much; the mean nectar volume at night divided by that during the day, Wald χ 2 = 3.828, df = 1, p = .01) but at a lower sugar concentration (0.88 times; the mean of sugar concentration at night divided by that during the day, Wald χ 2 = 5.878, df = 1, p = .015). For the S-morph, there was no significant difference between day and night in volume (Wald χ 2 = 1.962, df = 1, p = .161) or sugar concentration (Wald χ 2 = 1.045, df = 1, p = .307; Figure 3).

| Pollination treatments
Seed sets of the L-(43.32 ± 3.03%) as pollen recipient were 1.41 times higher than seed set of S-morph as pollen recipient (30.68 ± 2.72%). The seed sets of different pollination treatments differed significantly and there was no interaction between pollination treatments and pollen recipient morph with respect to seed set (Table 2).
Generally, there was no significant difference in seed set between nocturnal pollination treatment (72.58 ± 5.24%) and openpollinated control treatment (79.13 ± 4.85%; Wald χ 2 = 0.832, df = 1, p = .362), and all were significantly higher than diurnal pollination treatments (19.91 ± 3.72%; all p < .05; seed set data of L-morphs and S-morphs as pollen recipients were combined and analyzed; Figure 5). For L-morphs as pollen recipients, there was no significant difference in the seed set between nocturnal pollination and the open-pollinated treatment (Wald χ 2 = 0.182, df = 1, p = .669), and both were significantly higher (Wald χ 2 = 55.511, df = 2, p < .001) than those of diurnal pollination treatment. The same was true for the S-morph (Wald χ 2 = 63.377, df = 2, p < .001). TA B L E 1 Volatile compounds identified of Luculia pinceana flower using GC-MS and the relative amount (%) and number of samples (n) that each compound found in floral odor (10 in total) and control samples (3 in total). The compounds are arranged by compound class and retention time. The seed sets of nocturnal pollination in S-morph (62.65 ± 7.89%)
Linalool is the most common attractant for nocturnal hawkmoths in the floral scent of Lonicera japonica, and there is a significant correlation between the visitation of hawkmoths and specific floral odor . The scent of Dianthus inoxianus is dominated by aliphatic 2-ketones that contribute to attracting hawkmoth pollinators (Balao et al., 2011) and the nocturnal hawkmoth Deilephila elpenor responds to the odor of available extract of lavender positively (Balkenius et al., 2006). The hawkmoth Manduca sexta learns to feed from Agave palmeri flowers through olfactory conditioning but prefers Datura wrightii flowers based on an innate odor preference (Riffell et al., 2008). We found that the methyl benzoate content of L. pinceana floral odor was higher during the night than by day, and the hawkmoth (C. lineosa) was attracted to the methyl benzoate and made a behavioral preference response. Methyl benzoate is an important constituent in the floral odor of hawkmothpollinated plants (Manning & Snijman, 2002;Raguso et al., 2003;Schlumpberger & Raguso, 2008) and has been shown to elicit strong antennal responses in two hawkmoth species Manduca sexta (Hoballah et al., 2005) and Hyles lineata (Raguso 1996).
Some previous research indicates that L-morph flowers are significantly more strongly scented than S-morph flowers which is consistent with the hypothesis that small flowers (L-morph flowers) produce a stronger scent to defend against herbivory or attract pollinators (Leege & Wolfe, 2002). There was no significant difference in the floral odor between the two morphs in Primula elatior and P. farinosa (Gaskett et al., 2005). It is reasoned that heterostylous flowers are self-incompatible and rely on pollinators visiting both morphs equally. The content of methyl benzoate, the main odor compound, did not differ significantly between two morphs of L. pinceana and this may contribute to attracting the hawkmoths to visit the two morphs equally, promoting compatible pollination.
Nectar properties and the different strategies of nectar presentation could influence the visiting behavior of floral visitors, thus affecting the reproductive success of plants (Hodges, 1995;Pacini et al., 2003). Nectar volume and sugar concentration may vary significantly within a day, depending on the physiological condition of the plant, the evaporation of water and collection by visiting insects (Angela et al., 2010;Corbet, 1978;Hagler & Buchmann, 1993).
Nectar volume plays a major role in regulating the behavior of floral visitors (Nicolson et al., 2007). The hawkmoth pollinators visit significantly more Mirabilis multiflora flowers with larger amounts of nectar by manipulating the nectar volume (Hodges, 1995). A South American Andean cactus (Echinopsis ancistrophora) with high nectar volume is pollinated by hawkmoths (Schlumpberger et al., 2009). The hawkmoth pollinators reduce the probing duration on low-nectar mutational Petunia axillaris individuals when they are exposed simultaneously to mutational and wild-type P. axillaris (Anna et al. 2012).
Both L-and S-morph of Tirpitzia sinensis secrete larger nectar volume at night than by day which is potentially adaptive to hawkmoth pollination (Wang et al., 2022). Luculia pinceana secreted a larger volume of nectar with a lower sugar concentration during the night than by day which might be a direct dilution effect of nectar and the dilution might result from the higher relative humidity at night and are related to hawkmoth behaviors. The high nectar volume reflects the plant's resource investment in pollinator attraction.
Pollinators have different preferences for nectar sugar composition (Rio et al., 1992;Schondube & Rio, 2004). Nectar is mostly composed of fructose, glucose, and sucrose (Wykes, 1952). Moreover, nectar components vary significantly among plant species, probably to attract different pollinator species (Cruden et al., 1983). A recent study investigating the pollinator spectra of 57 Balsaminaceae species showed that the combination of nectar features reflects the pollination syndromes of flies, bird, bee, and butterfly pollination (Vandelook et al., 2019). Long-tongued bees, butterflies, hawkmoths, and hummingbirds have a preference for sucrose-rich nectar, while short-tongued bees and flies prefer hexose nectar (Goodwin et al., 2011;Nicolson et al., 2007). According to Baker and Baker (1983), the ratio (equal to [amounts of sucrose/(amounts of fructose + amounts of glucose)]) of three common sugars in nectar could be divided into four types: sucrose dominant, sucrose rich, hexose rich, and hexose dominant. The nectar of the L-morph (r = 5.75) and the S-morph (r = 6.75) of L. pinceana is sucrose dominant. Plants with a large nectar volume, low sugar concentration, and high sucrose-hexose ratio are mainly adapted to hawkmoth pollination (Kristina et al., 2015).
Heterostylous plants are usually self-incompatible and withinmorph cross-incompatible (Armbruster et al., 2006;Barrett, 2019;Keller et al., 2014;Wolfe et al., 2009;Wu et al., 2018). For example, the distylous Sebaea grandis did not produce seeds under intramorph pollination treatment, while the seed set with intermorph pollination treatment was about 78% (Wolfe et al., 2009). Moreover, the incompatibility system of Erythroxylum species changed with deviations from a 1:1 ratio of L-and S-morphs (Matias et al., 2020). The distylous species L. pinceana is partially self-incompatible; the seed set of the intermorph pollination treatment in L. pinceana (about 77%) was significantly higher than that of the intramorph pollination treatment (about 17%) and the numbers of L-and S-morph individuals of L. pinceana in five field populations did not deviate from 1:1 (Chen et al., 2021).