Floral biology and reproductive effort of andromonoecious Aesculus indica Wall. ex Camb. Hook (Hippocastanaceae)

Background Andromonoecious is an unusual sex-expression in trees; in which an individual plant bears both functionally staminate and hermaphrodite owers on the inorescences. The signicant effects of crown layers on reproductive success have not been studied so far. This study aims to investigate the effect of crown layers on oral biology and reproductive effort of Aesculus indica. Results The results revealed that the peak period of peak period of anthesis was between 06:00-08:00 h of the day. Male ower production was predominantly higher as compared to the perfect ower on the inorescence. There was no signicant variation between total pollen production in staminate and perfect owers. The features like protogyny and inter-level asynchrony promote xenogamy, however, the intra-level asynchrony resulted geitonogamy. Controlled pollination treatment revealed the existence of self-incompatibility in owers. Pollination syndromes of owers are supporting ambophily. A trend of consistent improvement in reproductive success from lower canopy layers to upper canopy layers in trees was recorded. The crown layers have a signicant impact on ower production, fruit, and seed set. Conclusions Increment in male ower production due to the increment in crown is a mechanism of reproductive assurance as a pollen donor and pollinator recipient and also due to the differential cost of expenditure of reproduction in crown layers.


Background
The genus Aesculus (family: Hippocastanaceae) comprises of thirteen species, numerous botanic varieties, cultivars and natural hybrids spread across the Northern hemisphere, mainly in temperate climates. The species are organized into 5 sections Aesculus, Calothyrsus, Macrothyrsus, Parryana, and Pavia (Hardin 1956). Aesculus indica Wall. ex Camb. Hook. (Indian horse chestnut) is a large, rounded tree and falls under the section Calothyrsus. A. indica is common along the Himalayan lowlands in the elevations of 1000-3000m, in the forests and shady valleys across Northern Afganistan, Pakistan, Kashmir, North India and Nepal (Troup 1921). A. indica is the most climatically sensitive species of all Aesculus. It has the widest acceptable pH range of 6.1 to 7.5 (Sternberg and Wilson 1995). It is widely grown as a shade tree for parks, arboreta, college campuses and home landscapes (Dirr 1990) and also recommended for buffer strips. The bark of A. indica has been used as a tonic because of its astringent properties (Hooker 1859). The wood is used to make utensils and pots and is also used for making, packing-cases, tea boxes and decoration articles (Rajasekaran and Singh 2009).
The genus Aesculus has an unusual reproductive strategy including a large portion of functionally staminate owers, and a high microspore to mature seed ratio. The owering biology of the genus characterizes andromonecious (having perfect and staminate owers on the same plant). Hardin (1956) stated that functionally staminate owers have a vestigial gynoecium with an undeveloped ovary and stunted style. Complete owers have fully developed androecium and gynoecium. Pistillate in orescences were found infrequently in A. hippocastanum and in A. pavia, where two plants out of many hundreds had only functionally female owers (Bertin 1980). Proportions of female-fertile owers are as low as 0.1% in Aesculus (Sedgley and Gri n 1989). The sex ratio of different Aesculus species was reported as 1.3 to 9.3% for A. turbinate (Suo et al. 1995); 1.1% for A. pavia, (Bertin 1980); <5% for A. californica (Benseler 1968); and 7.0% for A. sylvatica (Coker and Totten 1937). The empirical studies on the reproductive biology of A. indica are required in order to understand the variability and relationships among oral morphology, mating system and oral visitors under different ecological contexts.
Yet, there is no study regarding the reproductive effort of tree species under various layers of the crown. Reproductive success may be varied among crown layers of the tree. The possible assumptions are the differential allocation of photosynthates in crown layers or in uence of sunlight on fruit and seed set or variation in effective pollination and fertilization success in crown layers would be expected. Suo et al. (1995), using A. turbinata, reported that panicles from the upper crown (presumably receiving optimum photosynthetically active radiation) were longer and had a greater number of owers per panicle and a higher sex ratio than those from lower portions of the crown. Furthermore, fruit development depends strongly on photosynthesis and light availability (Brady 1987;Raven et al. 1992;Taiz and Zeiger 1998). Leaves in the upper part of trees shade those below; conceiving a vertical strati cation of light within the tree. Consequently, upper leaves are expected to produce more photosynthates than lower, shaded leaves.
We investigated the oral morphology and biology, pollination and breeding systems of A. indica for addressing the following questions: (1) Do any difference in reproductive success and reproductive potential in terms of fruit and seed set percentage in different crown layers? (2) Is there any signi cant difference between the production of staminate and bisexual owers in the sample trees? (3) What is the functional role of staminate ower? (4) Whether pollen production in staminate ower and perfect ower are equal or not? (5) What is the main pollination system of study species? (6) What are the major oral visitors and how they take part in reproduction?

Study area
This study was conducted in the natural moist temperate forests of Ranichauri area situated adjacent to the College of Forestry, Ranichauri, Tehri Garhwal (latitude 30º 18' N, longitude 78 º 24' E, altitude between 1750-2100 m asl) in the state of Uttarakhand, North West Himalaya, India. The nearby vegetation is comprises of the temperate broad leaved and coniferous forest, viz. Myrica esculenta, Pyrus pashia, Plantanus orientalis, Quercus leucotrichophora, Rhododendron arboreum, Cedrus deodara and Pinus roxburghii. The study was carried out from April 2016 to November 2016. The study area typically experiences a moist temperate type of climate with chilled winters. The mean monthly maximum and minimum temperature during the study period was oscillated between 7.0 0 C and 29.9 0 C. The trees of A. indica were selected under varying microclimatic conditions. The trees were chosen on the basis of accessibility, convenience, and free from biotic pressure.

Floral morphology and anthesis
Twenty in orescences at ower bud stage were selected and marked on the sample trees for observation. Basic stages of ower development were de ned by a detailed examination of the owers. To examine anthesis, owers were observed frequently at 2h interval within a day between 0600 and 1600 h. during the whole period of study and noted the number of owers opened. Opened owers were marked with a permanent marker to avoid error due to duplication and overcounting at every two hours intervals. To elucidate the anther structure and morphological changes of the stigma, owers were collected and observed under a microscope. Regular phenological observations were made for individual sample trees.

Pollen production
Pollen estimation was done separately for both male ower and perfect ower. The pollen production per ower in trees was assessed by using the noon loop method (Khanduri 2012). Anthers were obtained from closed owers just prior to anthesis and were laid into a test tube containing 5 drops of 5% glycerol, crushed with a glass rod and pollen grains were suspended in the test tube. The number of pollen grains was counted under a compound microscope.
Pollen production per anther was determined with the formula P = (Tp/N)×n where, Tp-is the summation of the total number of counted pollen grains in all the ve drops, N-is the total number of samples (slides) used for counting, and n-is the total number of drops. Production of pollen grains per ower was estimated by multiplying the number of pollen grains per anther with the number of anthers per ower. Mean pollen production of a tree was determined by using the equation TP= ∑N× F×A×P where TP is the total pollen grains per tree, ∑N is number of in orescence per tree, F is the average number of owers per in orescence, A is the average number of anthers per ower, P is the average number of pollen grains per anther.
Stigma receptivity and pollen ovule ratio Stigma receptivity was determined by immersing stigmas of owers in 3% aqua's hydrogen peroxide; bubbling indicating peroxide activity and stigma receptivity (Carrington et al. 2003). Flowers from different in orescences opened at the almost same time were immersed at two hours intervals up to thirty-eight hours to determine the average duration of stigma receptivity. Pollen-ovule (P/O) ratios were determined by dividing the estimate of the number of pollen grains per ower by the number of ovules per ower (Cruden, 1977). To determine ovule number cross section of the ovary was taken and counted the ovule number.

Pollen viability
The viability of the pollen grains was tested under in-vitro by using the acetocarmine staining method (Alexander 1969). We used anthers from fresh owers prior to the anthesis. Pollen grains were dusted onto a clean microscopic slide to which equal amount of acetocarmine (0.5%) and glycerol was added and warmed gently. The slides were incubated for 10-15 minutes. Stained pollen grains within the microscopic slide were counted as viable pollen while shrivelled, empty and weakly stained were recorded as non-viable.

Pollen germination
Pollen germination of A. indica was tested separately for freshly collected pollen grains from both male and perfect owers under in-vitro condition in sucrose solutions (germination media) with the concentrations of 10%, 20%, and 30 %. Pollen grains were placed in Petri dishes containing germination media (20ml) and maintained at 20.4 °C (room temperature) for 48 hours. Germinated pollen grains were observed under the binocular microscope in each solution.
The pollen grains were considered germinated when the pollen tube length was greater than the diameter of the pollen grain (Tuinstra and Wedel 2000). Germination rate was quanti ed as the percentage of germinated pollen grains per 100 evaluated grains.

Breeding system and oral visitors
A total of 100 owers from each sample trees were subjected to four types of treatments (25 owers for each treatment), 1) Open pollination-the owers were left without any intervention (no bag, no arti cial pollination) and observed up to fruit set, 2) obligate self-pollination -ower buds were bagged throughout the owering period and observed up to fruit set, 3) Open cross-pollination-removing the anthers from the owers prior to anthesis and was observed for fruit set, 4) Apomixes-removing both the anther and the stigma of the owers in bud stage and observed for fruit set.
Pollinator observations were carried out throughout the day between 06:00-16:00 hours on each experimental tree continuously on 8 days in the month of May. Insects visiting the in orescences were observed for their foraging behaviour, the number of individuals and duration of the visit. The insects were captured using insect trapping net (Sweep net) and polythene bags, every hour during the foraging period of ower visitors (06:00h to 18:00h). Unidenti ed insects' specimens were dried and placed into separate rectangular papers used for further identi cation. The double mounting method has opted for small insects.

Reproductive Success in different crown layers
The crown length of the selected sample trees was measured using Haga altimeter. The total crown length was divided into three equal proportions named upper canopy layer (UC), middle canopy layer (MC), and lower canopy layer (LC) (Fig. 1). The total number of in orescences in each canopy layer was counted manually and was multiplied by the average number of owers per in orescence to get the total number of owers. Fruit set and seed set were calculated separately for each canopy layer. Here fruit set is de ned as the proportion of owers that developed into fruits in each in orescence. Seed set is de ned as the proportion of ovules that developed into seeds in all the mature fruits within an in orescence. Twenty in orescences each in upper, middle and lower canopy were tagged in each sample tree and observed for the fruit formation. The average fruit set in each canopy layer was calculated by multiplying the number of fruit set per in orescence by the total number of in orescence in each canopy layer and expressed in percentage. Similarly, seed set per tree was calculated as, the average number of seed set per in orescence multiplied by the total number of in orescence within the canopy. The female reproductive success of tree was calculated by multiplying the mean number of fruits per plant by the mean number of seeds per fruit (Dafni 1992).

Data analysis
Mean ± standard deviations of the mean were calculated for all the measurements. The effect of time and tree to tree variation on ower opening was examined by two-factor analysis of variance (ANOVA) with time and tree to tree difference as xed effects. Student's t-test (test for difference in mean) was performed for assessing the difference in pollen production in staminate and perfect owers, number of staminate and perfect owers within the in orescence. The effect of time and sucrose concentration on pollen germination was also examined with two-way analysis of variance (ANOVA) with time and concentration of the solution as xed effects. Among crown layers variation in the production of in orescence, number of owers, number of fruit per in orescence, and reproductive success was assessed by single-factor ANOVA with crown layers as xed effects. The effect of pollination treatments on fruit set was also veri ed by single-factor ANOVA with treatment as xed effects.

Floral morphology and anthesis
The sex expression in A. indica is andromonoecious in which the in orescences bear both staminate and perfect owers. There was a marked difference in the number of staminate owers and perfect owers within the in orescence (t= 10.60, P 0.05, d.f.=18). Majority of the owers are staminate, with the pistil absent or rudimentary. The in orescence is compound raceme (Tyrsus) bearing terminal panicles, broad at the base forming a pyramidal shape . The owers are pedicellate, zygomorphic, the number varied from 341 ± 65 to 441± 74 per in orescence, the number of perfect owers per in orescence is 13 ± 3 to 20 ± 6. Flower composed of the polypetalous corolla and has four petals clawed which are narrow at the base and broad at the top (unequal in breadth) the place of 5 th usually vacant. The sepals are fused (gamosepalous), tubular, tomentose and have ve lobes. The androecium includes seven stamens (occasionally 6, 8 and 9) which are liform, tomentose, narrow and oblong. It is curved upward and longer than the petals. The gynoecium is hypogynous, simple and slender and sessile ovary having 6 ovules arranged in three cells with two ovules each.
Reproductive phenological observations showed that the ower bud initiation takes place in the month of March and the fruit set appeared in the last week of April and fully ripened during October to November thus took 7.5 months for the formation of zygote. In orescence was commenced in March and took 27 ± 2 d for enlargement of buds, individual bud took 8-12 d for complete development (Fig. 2), the mean time taken for the complete blooming of one in orescence was 56 ± 2 d, and mean number of days for fruit development from bud initiation was 142 ± 5 d. Flowers are opened in acropetal pattern to be noted from the second week of April to till June. The peak period of anthesis was recorded between morning hours 06:00 to 08:00 h of the day (Table 1). The effect of tree to tree difference on anthesis was not signi cant (Table 2, F=3.26, P 0.05), however, there was a signi cant effect of time on anthesis (Table 2, F=968.55, P 0.05). Anther dehiscence begins after 5-6 h of the anthesis.

Stigma receptivity and pollen ovule ratio
The in orescence of A. indica was protogynous female receptivity lasts about 5-6 d and shows distinct morphological changes. The stigma looked fresh and pink in colour throughout its receptive phase. Eventually, stigma becomes dry, shrink and brown in colour which made clear the cease of receptivity.
There are six ovules per ower; hence the P/O ratio varied from 8855 ± 65 to 9078 ± 198 pollen grains per ower for perfect ower and 8863 ± 107 to 9029 ± 198 pollen grains per ower for the male. In addition, according to Cruden (1977) classi cation, the P/O ratio of A. indica falls within the reported range of Xenogamy.

Pollen viability and germination
The percentage of viable pollen per slide was ranged between 87.54 ± 4 % to 90 ± 0.8 % for perfect and 85.3 ± 3 % to 89.6 ± 4% for male owers. The effect of time and concentration of solution on pollen germination was found to be signi cant for perfect owers (F=48.73, F=18.10) and male owers (P 0.05, F=67.84, F=31.13, P 0.05) ( Table 2). The maximum germination of 88.17% for perfect owers and 88.21 % for male owers were observed in 20% sucrose concentration, followed by (83.33% for perfect owers and 84.19% for male owers) in 10% sucrose.

Breeding systems and oral visitors
There was a signi cant effect of treatment on fruit set (Table 2,  A diverse array of insects belongs to the orders Hymenoptera, Thysanoptera and Lepidoptera were observed visiting the owers of A. indica during the 8 d observation period (Table 3). Most visiting were observed in morning hours between 08:00-10:00 h. The most frequent visitors in owers were Bombus spp. and Thrips. There was little difference in the visitation rates of the insects to staminate and perfect owers. The in orescence received most Bombus spp. in the morning (06:00-08:00 h) and rarely in afternoon. Bumblebees tended to visit the outward facing owers and the owers at the base of the panicle rst. Tagiades menaka was found more frequently in the evening (16:00 -18:00 h). The butter ies were observed feeding at the top of the in orescence and visiting both buds ready to open and open owers, could be considered as a marginal pollinator. Hummingbird moth was also observed visiting the owers; they tended to visit the panicle apex rst. In aggregate, the activity of honey bees, butter ies, and bumblebees have coincided with the peak period of the ower opening, which strongly supporting the role of insect in pollination.

Reproductive success in different crown layer
The production of in orescence and production of owers did not differ signi cantly on crown layers (Table 2, F= 1.323, P=0.33341, F=1.272, P=0.3462), fruit set within in orescence showed highly signi cant difference between crown layers (Table 2, F=12.24, P=0.0076) in trees. We also have noticed the trend of increasing the number of in orescence and owers production from the lower canopy (LC) to upper canopy layer (UP) and consequently enhanced the fruit set in middle and upper canopy layers (Table 4). Surprisingly, there was a signi cant difference in reproductive success (proportion of the number of owers converted into fruit) between crown layers in trees (Table 2, F=14.237, P=0.0052). Although there are 6 ovules in each ovary, successful pollination invariably leads to the formation of only one seed in a fruit. Only the ovule located at either the rst or the second position towards the stylar end matures into a seed. The remaining ovules abort results very low reproductive potential in trees (0.167).

Discussion
This paper reports the rst comprehensive study of oral biology and reproductive effort of A. indica. Many of the workers have investigated the various reproductive aspect of the genus Aesculus including the species A. indica, whilst many parts being still unknown. The andromonoecious sex-expression in A. indica is depicting signi cant difference in the number of male owers and perfect owers in an in orescence. The surplus production of male ower maximises pollinator attraction and function as a pollen donor to fertilise another hermaphrodite ower which enhance the female and male tness within the population. The study conducted by past workers stated that staminate owers may have two selective advantages. First, they provide a pollen surplus that enhances male tness (Bertin, 1982;Anderson and Symon 1989;Podolsky 1993;Connolly and Anderson 2003). Secondly, staminate owers increase oral display that improves pollinator attendance, which in turn bene ts female tness (Bertin 1982;Anderson and Symon 1989;Spalik 1991). Staminate owers and perfect owers were morphologically alike except incomplete development of pistil in staminate owers, the more advanced study is required to understand the oral sexual dimorphism in this species.
In A. indica diurnal pattern of anthesis has been witnessed; the peak period of anthesis coincides with the maximum activity of the oral visitors. Clearly, owering time of each species is genetically xed, whilst, which is highly variable with the environmental factors, mainly precipitation, temperature and relative humidity (Hegazy et al. 2017;Amano et al. 2010;Khanduri et al. 2008). In addition, the owering pattern of A. indica was found to be asynchronous i.e. buds and owers are at different stages of development even on the same tree. Similar results were also reported by Bangarwa (1993) in Dalbergia sissoo, Khosla et al. (1982) in Bombax ceiba. Individuals with asynchronous owering decrease reproductive output, the amount of pollen, the number of pollen donors and the levels of outcrossing compared to individuals blooming during the same period (Schwartz, 2003).
There was no signi cant variation in pollen production between perfect and staminate owers. So it can be inferred that there would be differential cost of expenditure for production of staminate and perfect owers, probably the production of owers serves only pollen source would be less expensive. Therefore, the trees inevitably produce more staminate owers to increase the pollen output and female tness. Similar ndings were also reported by Reale et al. (2006). The hermaphrodite (perfect) owers possess protogynous dichogamy (female phase preceding the male phase). Stigmas were receptive prior to the anthesis and anther dehisced 5-6 h after the anthesis indicating that the hermaphrodite owers were well adapted to outbreeding and declined the chances of auto-pollination in A. indica. Protogynous dichogamy is one of the mechanisms promoting outcrossing in Acacia (Sedgley 1987). At anthesis, anthers were not dehisced results greater chances of viable pollen from outside may deposit over the stigma and get fertilized. The features like marked protogyny and inter-level asynchrony promotes xenogamy, however, the intra-level asynchrony perhaps results in gietonogamy.
A high percentage of pollen viability from both hemaphrodite and staminate owers exert a direct control on pollenstigma interactions (Heslop-Harrison et al. 1984;Heslop-Harrison and Heslop-Harrison 1992), fruit set and gene ow (Dafni and Firmage 2000). Pollen germination was signi cant with time and concentration of germination media. The pollen germination was increasing with increase in time till 5 h, later there was a steady decline in the germination rate. 20% sucrose concentration exhibited maximum pollen germination, as sucrose is the carbohydrate source for pollen germination and tube growth (Tupy 1960). The results on controlled pollination reveal predominantly selfincompatibility in A. indica as no fruit set was observed from autogamous (self-pollination) pollination. Hence pollination is completely dependent on vectors for its reproductive assurance. Interestingly, fruit set under open pollination was almost 10% less than that of open-cross pollination which would be due to the absence of out-crossed pollen. Kalinganire et al. (1996) mentioned that the poor fruit set following natural open-pollination is mainly due to the absence of outcrossed pollen. This is probably due to low visitation rates by effective pollinators. Since there was no spontaneous autogamy, as indicated by lack of fruit set in completely bagged owers.
A. indica was found to possess all the characters of anemophily syndrome such as long-exerted terminal in orescence, small nectar less owers, high P/O ratio, small and smooth pollen produced in large numbers (Stanley and Kirby, 1973;Faegri and van der Pijl 1979;Sharma and Khanduri 2007). The existence of protogyny, asynchronous owering and self-incompatibility rules out the possibilities of self-(intra oral or geitonogamous) pollination in the canopies. The present studies on oral visitors have revealed that the owers of A. indica were frequently being visited by ve insect species from three orders such as Hymenoptera, Thysanoptera, and Lepidoptera. It can be inferred that the pollination system in A. indica had adapted to support both entomophily and anemophily, the occurrence of which is referred to as ambophily (Yamasaki and Sakai 2013).
Most of the foragers was visiting the owers during the peak period of blooming (08:00-10:00 h) of the day. It was apparent that anther dehiscence occurred 5-6 h later the anthesis and not coinciding with the pollinator activity, clearly indicating that the foragers visiting the ower inadvertently. Moreover, there was no difference in visitation rate of insects on both staminate and perfect owers. The higher number of male owers may reduce the probability of bisexual ower visits consequently reduces the fruit set. Alternatively, pollen grains of 1 day old opened owers were contributing as pollen donor to freshly opened owers, augmenting maximum cross-fertilization. Despite, the frequent incidents of Bees, Thrips and Butter ies, owers were assumed to increase the chances of pollen theft and negatively affect the seed production (Bertin 1980;Arizmendi et al. 1996;Irwin et al. 2001;Lara and Ornelas 2001). Because of these reasons insects cannot be considered as an effective pollinator in A. indica. Furthermore, frequent visit of more number of insects to the owers increase the pollen concentration in the air, which assisting wind pollination in A. indica. The assistance of insects in anemophily was recorded in Phyllostachys nidulari (Huang et al. 2013); Plantago lanceolata (Clifford 1962); and Mallotus (Yamasaki and Sakai 2013). Therefore, the pollination syndrome of A. indica supports both wind (anemophily) and insect (entomophily) pollination.
Unlikely, we found that there was a signi cant difference in the proportion of owers set into fruit (reproductive success) in canopy layers. Reproductive success was found to be markedly high in upper canopy layers as compared to lower layers. Exactly what causes the variability in fruit production in crown layers? Clearly, a number of factors affecting the fruit production, (1) differential cost of expenditure of reproduction in crown layers results inadequate translocation of resources (2) ample amount of light availability and temperature increase the photosynthetic rate and reproductive success (3) pollen limitation arises from the preferential activity of pollinator (4) natural selection pressure. Similar ndings were reported by a number of workers (e.g. Fruit development depends strongly on photosynthesis and light availability (Brady 1987;Raven et al. 1992;Taiz and Zeiger 1998); competition by the surrounding neighborhood of plants for those resources (Lee and Bazzaz 1982; Graham et al. 2003); zones of the canopy that receive more sunlight are expected to distribute more photosynthates locally, from the leaf to the adjacent fruit: local translocation of resources hypothesis (Lynch and Gonzalez 1993;Proietti et al. 2000;Raven et al. 1992;Taiz and Zeiger 1998).
In A. indica, the fruit set and seed set rates are the same due to the production of 1-seeded fruits (Drupe). It is evident that fruit and seed production is very low when compared to the numbers of owers and ovules. Lee (1988) and Guitiá n et al. (1996) proposed several factors that may limit fruit production, including extrinsic causes (pollen limitation, herbivory, frost, etc) and intrinsic causes (genotypes, stored resource content, etc). The most common factor reported in the literature is resource limitation (e.g. Stephenson 1981;Berjano et al., 2006). Furthermore, we speculated that the low rate of reproductive potential (S/O) was due to the existence of a high degree of ovule abortion. High rates of ovule abortion in multi-ovulated species are still conjectural. It could be argued that the self-incompatibility and intra-ovule competition within the ovary would likely to be the primary reason for ovule abortion under pollen surplus condition in A. indica.

Conclusions
This study indicates that A. indica is self-incompatible and outcrossing species. Surplus male ower production is likely an ovary reserve mechanism as a pollen donor and pollinator attractant. Ambophily (anemophily as well as entomophily) is the primary pollination mechanism in A. indica. We found that crown layers have a signi cant impact on ower production, fruit, and seed set presumably due to the differential cost of expenditure of reproduction in crown layers. Floral dimorphism, ovule-abortion, and productivity of crown layers are the subject of subsequent study in A. indica.

Declarations
Tupy J (1960) Sugar absorption, callose formation and the growth rate of pollen tubes.