Ecological correlates of population genetics in Linum suffruticosum, an heterostylous polyploid and taxonomic complex endemic to the Western Mediterranean Basin

Abstract

Linum suffruticosum s.l. is a taxonomic complex widespread in the Western Mediterranean basin. The complex is characterized by a high phenotypic and cytogenetic diversity, and by a unique three-dimensional heterostyly system that makes it an obligate outcrosser. We studied the patterns of genetic diversity and structure of populations throughout the entire distribution of L. suffruticosum s.l. with microsatellite markers. We analysed their relationships with various biological and ecological variables, including the morph ratio and sex organ reciprocity of populations measured with a novel multi-dimensional method. Populations consistently showed an approximate 1:1 morph ratio with high sex organ reciprocity and high genetic diversity. We found high genetic differentiation of populations, showing a pattern of isolation by distance. The Rif mountains in NW Africa were the most important genetic barrier. The taxonomic treatment within the group was not related to the genetic differentiation of populations, but to their environmental differentiation. Genetic diversity was unrelated to latitude, elevation, population size, niche suitability or breeding system. However, there was a clear influence of ploidy level on the genetic diversity of populations, and a seeming centre-periphery pattern in its distribution. Our results suggest that polyploidization events, high outcrossing rates, isolation by distance and important geographical barriers to gene flow have played major roles in the microevolutionary history of this species complex.

Keywords: Linum; Western Mediterranean; cytotype; heterostyly; microsatellites; polyploidy; reciprocity.

Figure 1.

Linum suffruticosum flower (A), details for the long-styled (B) and short-styled (C) morphs, habit (D) and natural population (E).

Figure 2.

Genetic structure of Linum suffruticosum populations. Results of the Bayesian clustering of populations and geographic distribution of clusters, represented by different colours. Results for the most likely numbers of genetic clusters are shown: two genetic clusters when analysing the whole dataset (K = 2; A, D); and two (K = 2; B, E) and seven (K = 7; C, F) genetic clusters after excluding Riffan cluster from the analyses. The cytotype of each population is indicated above plot (A); population labelled as 3x had a mixture of diploid and triploid individuals.

Figure 3.

Results of the PCA analyses with populations grouped by STRUCTURE genetic cluster (A); ploidy level (B) and taxonomic entity (C). Plot (A) represents the three clusters resulting from K = 2 in both full datasets and excluding Riffan cluster in shadows; and the eight clusters resulting from K = 2 in the full dataset and K = 7 excluding Riffan cluster in empty circles. A Minimum Spanning Tree has been superimposed on plots (B) and (C). LS: Linum suffruticosum s.s.; LAS: L. appressum-salsoloides, LSM: North African Linum suffruticosum.

Figure 4.

Genetic diversity (Na) of populations grouped by STRUCTURE genetic cluster (A), ploidy level (B) and taxonomic entity (C). Significant differences between groups according to the Tukey test are represented by different letters over each box in each plot. LS: Linum suffruticosum s.s.; LAS: L. appressum-salsoloides, LSM: North African Linum suffruticosum.

Figure 5.

Scatter plots representing the linear regressions between the expected heterozygosity (H_S) and latitude (A), elevation (B), population size (C), percentage of long-styled (L-) morph (D), reciprocity (R2) (E), and environmental suitability (F). P-values from Dutilleul’s modified t-tests are presented in each plot.

Figure 6.

Scatter plots representing the linear regressions between the percentage of long-styled (L-) morph and reciprocity (R2) (A), between the percentage of long-styled (L-) morph and environmental suitability (B), and between reciprocity (R2) and environmental suitability (C). P-values from Dutilleul’s modified t-tests are presented in each plot.

Figure 7.

Environmental PCAs of populations of Linum suffruticosum grouped by STRUCTURE genetic cluster (A), ploidy level (B) and taxonomic entity (C). LS: Linum suffruticosum s.s.; LAS: L. appressum-salsoloides, LSM: North African Linum suffruticosum.