Heterostylous syndrome and clonal growth

During the flowering season, we investigated all genet in the five subpopulations for floral morphology and floral morph frequency analysis by randomly selected ten flowers from each flowering genet. Floral morph frequency is expressed by the ratio of the number of certain floral morph to the total number of counted flowers in subpopulations. As there were two distinct pollen/stigma morphologies between L- and S-morph flowers, H-morphs with the same stigma morphology and pollen ornamentation as the L-morphs were named H_L-morph flowers, whereas those with the same stigma epidermal cell morphology and pollen ornamentation as the S-morphs were named H_S-morph flowers. Plants exclusively bearing H_L-morph flowers were categorized as H_L-morph individuals, while those with only H_S-morph flowers were classified as H_S-morph individuals. For H-morph identification, we used a microscope (Nikon, ECLIPSE E200, Nikon Corporation, Japan) to observe pollen or stigma morphology.

During the flowering season of L. otolepis, we surveyed five subpopulations (S1–S5) to record genet/clonal fragment numbers, ramet counts per genet, and floral morph distributions. We also conducted genet localization to map their distribution and clonal growth patterns. Given the scattered and sporadic distribution of most genets, the consistent floral morphology and linear arrangement of ramets within the same genet, and the distinct pistil-stamen positions, pollen, and stigma morphologies, it was relatively straightforward to identify the genets. In clonal patches featuring H-morphs, we confirmed genets and their ramets by examining stigma morphology using a light microscope (Nikon, ECLIPSE E200, Nikon corporation, Japan).

To examine pollen and stigma morphology, we randomly chose 10 individuals per floral morph in each subpopulation, collecting 1–2 nearly open floral buds from each ramet. Anthers were air-dried in EP tubes, while stigmas were fixed in 2 % glutaraldehyde fixative (0.1 mol L^–1 phosphate buffer). A scanning electron microscope (LEO 1430 VP, Carl Zeiss, Oberkochen, Germany) was used for studying pollen and stigma morphology.

For assessing size parameters and floral morph variation in the L. otolepis population, we randomly marked 15 × 4 individuals, each representing L-, S-, H_L- and H_S- morphs, from the S4 subpopulation. From each marked individual, 2–4 flowers were randomly sampled. We measured various floral characteristics, such as total flower length, corolla length, corolla tube length, corolla opening diameter, corolla tube diameter and stigma and anther heights. Because the five stamens within individual flowers often differed in length, we measured the tallest and shortest stamens in each flower and calculated the average anther height. A principal component analysis was then performed on these floral parameters.

To capture variations in stigma and anther heights, we selected 30 individuals from mixed clonal patches with two mating types exhibiting different pollen/stigma morphology (including L- and H_L-morphs and S- and H_S- morphs). From each of these individuals, 3–5 flowers were randomly chosen for measuring stigma and anther heights. Stigma and anther height distributions were subsequently analysed using a digital calliper with 0.02 mm accuracy.

Owing to the occurrence of H-morph with different pollen/stigma morphology, there were four floral morphs (L-, S-, H_L- and H_S-morphs) in the population. To test the compatibility between floral morphs, 15 L-, S-, H_L- and H_S-morphs plants were labelled randomly. Before anther dehiscence, the pollen of various floral morphs were collected in the labelled plants and placed in different EP tubes for pollination treatments. Meanwhile, about 20 flowers were selected randomly from each plant for the following treatments (2–3 flowers per treatment): (i) apomixis (emasculated and netted); (ii) intramorph pollination and netted (L × L, S × S, H_L × H_L, and H_S × H_S); (iii) ‘intermorph pollination’ and netted (different from the traditional meaning of intermorph pollination) (L × S, S × L, L × H_L, L × H_S, S × H_S, S × H_L, H_S × L, H_S × S, H_S × H_L, H_L × S, H_L × L, H_L × H_S); (iv) artificial self-pollination and netted and (v) control (flowers were naturally pollinated). We assessed fruit sets in the various pollination treatments after fruit ripening. Additionally, we tested compatibility in a monomorphic S1 subpopulation of H-morph (H_S-morph). We randomly labelled 15 individuals, each with 16 flowers, and subjected them to four treatments (four flowers per treatment): (i) intramorph pollination and netted, (ii) artificial self-pollination and netted, (iii) apomixis (emasculated and netted) and (iv) control (flowers were naturally pollinated). Subsequently, we counted the fruit set for each treatment upon ripening.