Two reciprocal crosses (i.e., four crosses in total) were performed using four cultivars of water lotus, i.e. N. nucifera ‘Guili’ (pollen receptor) × N. nucifera ‘Aijiangnan’ (pollen donor) (hereafter referred to as GA), N. nucifera ‘Aijiangnan’ (pollen receptor) × N. nucifera ‘Guili’ (pollen donor) (AG), N. nucifera ‘Molingqiuse’ (pollen receptor) × N. nucifera ‘Qinhuaiyanzhi’ (pollen donor) (MQ), and N. nucifera ‘Qinhuaiyanzhi’ (pollen receptor) × N. nucifera ‘Molingqiuse’ (pollen donor) (QM). About 100 plants of each cultivar were grown in pools in Nanjing Yileen, Zhujiang town, Pukou district, Nanjing, China (32°07' N, 118°62' E). The four cultivars are diploid and have a chromosome number of 16 (2 n = 2 x = 16). These specific cultivars were selected for two main reasons. First, the ornamental values of ‘Aijiangnan’ and ‘Guili’ are complementary, for example the flower color of ‘Aijiangnan’ is butter-yellow and that of ‘Guili’ is pink. If these two cultivars can be crossed successfully, it should be possible to obtain genotypes with novel traits. This is also the case for ‘Molingqiuse’ and ‘Qinhuaiyanzhi’. Second, our preliminary studies indicated that seed set differed among these crosses, which motivated us to examine the factors that influence the fecundity of these crosses. In the present study, flowers from multiple plants of each cultivar were used, because one plant was unable to produce a sufficient number of flowers for the present experiment. However, within each cultivar the plants had an identical genetic background, because they were propagated vegetatively from rhizomes. In addition, before we formally carried out the present study, we performed preliminary tests to examine if there was a significant difference in behavior between flowers from the same plant and between flowers those from different plants within a given cultivar. No significant differences were observed.
Determination of pollen viability
Previous studies have shown that lotus pollen is only viable for a few hours after anther dehiscence and that it is difficult to determine lotus pollen viability . In the present study, we systematically estimated lotus pollen viability using a variety of methods, such as fluorescein diacetate (FDA), triphenyltetrazolium chloride (TTC), germination in vitro, and the peroxidase reaction. Only the peroxidase test was successful, and thus was used to detect pollen viability, and the other methods (FDA, TTC, and germination in vitro) were not used further. Pollen grains from freshly dehisced anthers were collected at 05:00–06:00, 06:00–7:00 and 07:00–08:00, respectively, on a sunny day in July (sunrise occurred at about 05:00 at the study site in July). All plants were growing in the field. The soultion for determination of pollen viability comprised two reagents, i.e. reagent I (1:1:1 0.5% benzidine: 0.5% α-naphthol: 0.25% sodium carbonate) and reagent II (0.3% hydrogen peroxide solution). The pollen grains were placed onto a glass slide that bore a culture medium comprising one drop of reagent I and one drop of reagent II, and then incubated at 30°C for 30 min. The pollen grains turned red if there were viable. The number of viable pollen grains in 10 optical fields was counted under an Olympus BX41 microscope and at least 50 pollen grains were found in each field. The pollen grains were collected from 10 flowers and five plants (two flowers per plant) and mixed together. Each cross was repeated three times, and approximately 2000 pollen grains were tested in each experiment.
Lotus plants are insect-pollinated and the pistil matures one or two days before stamen maturation. In order to prevent insect pollination in the present study, we bagged the flowers of the female parents before the pistil matured. The optimal time for pollination is 05:00 to 08:00 on a sunny day. The pollen grains from 20 bagged male flowers from five plants (four flowers per plant) were collected between 05:00 and 06:00, and used for artificial pollination once the female flowers opened and their stigma surface was coated with bright yellow mucus. We used 150–200 flowers from about 50 plants as the pollen receptor for each cross. The immature stamens of the pollen-receptor flower were first removed, and then the stigmas were pollinated with the freshly collected pollen grains. After artificial pollination, the pollinated flowers were bagged again.
Pollen behavior on the stigma after pollination
Pollen germination on the stigma in vivo was investigated as described by Sun et al.  with minor modifications. Twenty pistils per cross were fixed in FAA solution (5:5:90 formalin: acetic acid: 70% ethanol) at 0.5, 1, 2, 4, 6, 8, 10 and 12 h after pollination, respectively, then stored at 4°C until use. Each flower usually contained 20–35 pistils from which we randomly collected 20 pistils from three flowers at each time-point per cross. Pistils were collected from a total of 24 flowers per cross. The ovaries were removed and the stigmas softened overnight in 1 mol l-1 NaOH, rinsed in water and mounted on a microscope slide with a drop of 0.1% aniline blue (0.1 mol l-1 K3PO4 supplemented with 18% glycerol), and then observed under a fluorescence microscope (Zeiss Axioskop 40) with a BP 395-440 excitation filter BP 395-440, FT 460 chromatic beam splitter, and LP 470 barrier filter. Digital images were captured with an Axiocam MRC camera . In addition, some pistils were fixed in 2.5% glutaraldehyde (0.1 M phosphate buffer, pH 7.2), dehydrated in an ethanol series (40, 70, 90 and 100%, for 15 min at each concentration), subjected to critical point drying and coated with gold for scanning electron microscopy (Hitachi S-3000 N), followed by digital processing of the images .
Examination of embryo development and seed set
About 80 ovaries (or seeds) were collected from 10 flowers on five plants (two flowers per plant) for each pollination treatment at both 1 and 2 days after pollination and immediately immersed in FAA until use for examination of embryo development. The ovules were dissected from the ovaries and dehydrated through an alcohol series (70, 85, 95 and 100%, for 5 min at each concentration), then infiltrated with xylene, and embedded in paraffin wax . Sections were cut to a thickness of 8-10 μm, stained in Heidenhain's haematoxylin, then observed and photographed under an Olympus BX41 microscope. In addition, we collected about 80 ovaries at 4, 6, 8 and 11 days after pollination for determination of the ratio of plump seeds to shriveled seeds over time. These samples were observed and photographed under a stereo microscope equipped with a digital cameral.
Among the pollinated flowers, we chose 30 flowers from 10 plants to determine seed set in each cross at 1 month after pollination. Seed set was calculated using the formula: seed set = (the number of plump seeds/the total number of pollinated stigmas) × 100% . The crosses were performed in July and August, 2009, during which the average temperature was approximately 28°C (range 26–38°C). In addition, in order to investigate the percentage of normal pistils just before pollination, the seed set in emasculated flowers of female parents were determined under open pollination conditions. We also performed self-pollinations to determine self-compatibility of each cultivar. About 30 flowers from 10 plants per cultivar were bagged one day before the flowers opened, and the bags were immediately removed after the flowers withered. Furthermore, we emasculated 30 flowers from 10 plants, and these flowers were later pollinated with fresh pollen from other plants of the same cultivar.
The data were subjected to a one-way analysis of variance using SPSS 16.0 (SPSS Inc, Chicago, IL, USA). The means were compared using the Bonferroni t-test with α = 0.05 (the type I experimentwise error rate). The data are presented as the mean value ± standard deviation.