Repeated polyploidization shapes divergence in floral morphology in Lithophragma bolanderi (Saxifragaceae)

Polyploidization is an important driver of evolution and diversification in flowering plants. Here, we assess how repeated polyploidization may have shaped diversification of floral morphology in Lithophragma bolanderi (Saxifragaceae). This species comprises multiple cytotypes and varies geographically in its interactions with specialized pollinating moths in the genus Greya (Prodoxidae). Past studies have shown that coevolution with these moths has favored particular suites of floral characters but does not fully explain local and regional floral diversification. We combined phenotypic and genomic data from more than 1800 individuals from 40 L. bolanderi populations spread across its entire range. Flow-cytometric analyses revealed a geographic mosaic of populations comprising one to four of three dominant (diploid, tetraploid, hexaploid) and three rare (triploid, pentaploid, octoploid) cytotypes. Whole-genome resequencing of a subset of populations suggested that polyploids arose from multiple autopolyploidization events, rather than a single event and/or through hybridization, albeit with some signals consistent with low levels introgression from the congener Lithophragma glabrum. Quantification of flower traits from plants grown in a common garden showed that cytotype explained more than 15% of the variation in floral morphology, with polyploids showing more variability than diploids. Experimental induction of neopolyploids directly induced phenotypic changes but also indicated that local selection may have favored subsequent convergence in floral morphology among cytotypes in natural populations. Collectively, this comprehensive and integrative approach provides novel insights into how variability generating processes, such as polyploidization integrates with selection from species interactions to shape local floral diversification.

Methods
In 29 natural populations of the plant Lithophragma bolanderi in the Sierra Nevada, California, USA, seeds were collected. A subset of these seeds was grown in a greenhouse common-garden. When the plants were well established, leaf material was collected and analyzed using flow cytometry (conducted by one of the authors with Brassica napus as internal standard) to assess plant ploidy level, flower diameter and corolla-opening diameter were measured of fresh flowers using digital calipers, and flowers were collected, stored in 70% ethanol, dissected, and photographed to quantify additional floral morphological traits based on the photos and to assess the shape of the petal edge. As the flowers were apportioned among three persons to conduct the measurements, one of these persons remeasured a subset of the flowers measured by the other two persons. In cases where the measurements were not highly correlated, this one person remeasured these traits for all samples. This one person also visually checked for “outliers” separately for each population and cytotype, double-checked these measures, and remeasured them if necessary. The morphological floral traits with all pairwise correlations < 0.7 were included in a principal coordinate analysis (PCA) and the principal components (PCs) used for further analyses were also included in the dataset.

In a subset of the natural populations of the plant L. bolanderi, in which seeds were collected, flower diameter and corolla-opening diameter were measured of fresh flowers using digital calipers and compared to the flower diameter and corolla-opening diameter measured on a randomly selected subset of the plants from the same populations grown in the greenhouse common-garden (see above).

In an addition 11 natural populations of the plant L. bolanderi, leaf material was collected and analyzed using flow cytometry (conducted by Plant Cytometry Service [https://www.plantcytometry.nl/] with Allium schoenoprasum as internal standard) to assess plant ploidy level.

Some L. bolanderi plants were grown from seeds (derived from crosses between individuals of the same cytotype and from the same population but from different seed families of plants grown from root bulbils of the greenhouse common-garden plants) in a greenhouse common-garden. When the plants were well established, leaf material was collected and analyzed using flow cytometry (conducted by Plant Cytomerty Service once with B. napus and once with A. schoenoprasum as internal standard). The sample to internal standard rations were compared between the two internal standards and with those of the parental seed families quantified by one of the authors with B. napus as internal standard.

An additional subset of the seeds collected in one natural L. bolanderi population, in which both diploids and tetraploids grow, was grown in a greenhouse common-garden. A subset of the diploid seedlings was treated with colchicine to induce polyploidization. When plants were well established, leaf material was collected and analyzed using flow cytometry to verify plant ploidy level. When these plants were flowering, hand-pollination crossings were conducted among colchicine treated plants that were at least partially polyploid, among diploid control plants, and among tetraploid control plants. These crosses were done to mitigate side effects of colchicine. The seeds resulting from these crossings were used to grow F1 plants. When the F1 plants were well established, leaf material of all colchicine-treated plants and a subset of the diploid and tetraploid control plants was collected and analyzed using flow cytometry (conducted by Plant Cytomerty Service with A. schoenoprasum as internal standard) to verify plant ploidy level, flower diameter and corolla-opening diameter were measured of fresh flowers using digital calipers, and flowers were collected, stored in 70% ethanol, dissected, and photographed to quantify the additional floral morphological traits based on the photos (see above). The same morphological floral traits as for the data of the L. bolanderi plants from the multi-population greenhouse common-garden were included in a PCA and the PCs used for further analyses were also included in the dataset.

For details, see:

Karin Grossa, Homa Papoli Yazdi, Elisabeth Schlager, Jodie Lilley, Andrés Romero-Bravo, Anna Runemark, John N Thompson, and Magne Friberg. Repeated polyploidization shapes divergence in floral morphology in Lithophragma bolanderi (Saxifragaceae). Proceedings of the National Academy of Sciences.