Functional trait divergence and trait plasticity confer polyploid advantage in heterogeneous environments

Abstract

Polyploidy, or whole-genome duplication often with hybridization, is common in eukaryotes and is thought to drive ecological and evolutionary success, especially in plants. The mechanisms of polyploid success in ecologically relevant contexts, however, remain largely unknown. We conducted an extensive test of functional trait divergence and plasticity in conferring polyploid fitness advantage in heterogeneous environments, by growing clonal replicates of a worldwide genotype collection of six allopolyploid and five diploid wild strawberry (Fragaria) taxa in three climatically different common gardens. Among leaf functional traits, we detected divergence in trait means but not plasticities between polyploids and diploids, suggesting that increased genomic redundancy in polyploids does not necessarily translate into greater trait plasticity in response to environmental change. Across the heterogeneous garden environments, however, polyploids exhibited fitness advantage, which was conferred by both trait means and adaptive trait plasticities, supporting a 'jack-and-master' hypothesis for polyploids. Our findings elucidate essential ecological mechanisms underlying polyploid adaptation to heterogeneous environments, and provide an important insight into the prevalence and persistence of polyploid plants.

Keywords: adaptation; adaptive plasticity; common gardens; functional traits; polyploidy; wild strawberry.

Figure 1

Seventy‐two source populations of diploid (circles) and polyploid (triangles) Fragaria, and their reticulate evolutionary histories (inset). The inset dendrogram represents the known evolutionary relationships among the five diploid (2x) taxa in this study (black), as well as those not in this study (grey; F. mandshurica, and an extinct F. iinumae‐like diploid^§ with dashed branch), along with an outgroup taxon (Dasiphora). Among the six polyploids, the octoploid (8x) taxa are derived from the 2x F. vesca ssp. bracteata, F. iinumae and the extinct F. iinumae‐like diploid (each contributing, respectively, two, two and four sets of chromosomes to the 8x genomes, reflected by line width) (Tennessen et al., 2014; Wei et al., 2017a). The 10x F. cascadensis has two sets of chromosomes from F. vesca ssp. bracteata, two sets from F. iinumae and six sets from the F. iinumae‐like diploid (Wei et al., 2017b). The 6x F. moschata is derived from F. vesca ssp. vesca, F. viridis and F. mandshurica (Kamneva et al., 2017). Recurrent formation* of the same polyploid taxon in different populations has been previously identified (Dillenberger et al., 2018), whereas such information remains unclear for the remaining polyploid taxa.

Figure 2

The location, design and climate of common gardens. (a) Three common gardens were located in Oregon, USA, including the coastal garden at Newport, the valley garden at Corvallis and the montane garden at Bend. (b) We established four raised wooden beds at each garden location. Each bed (18 × 1.5 m) can host 72 × 4 plants, indicated by the dots. For each genotype, the four clones (red dots) were assigned to the four beds, and the position within each bed was chosen randomly. (c) The monthly mean temperature, rainfall and growing degree days (i.e. the cumulative heat > 10°C) were obtained (see Supporting Information Methods S1) for the three common gardens, during the course of the field experiment from October 2015 to mid‐July 2016.

Figure 3

Diploid and polyploid Fragaria differ in leaf functional traits. The least‐squares mean ±1 SE of each trait are plotted for diploids (dashed lines) and polyploids (solid lines) at each garden location, estimated from general linear mixed models where the response variables were power‐transformed if necessary (see the Materials and Methods section; sqrt, square root; log, natural logarithm). The x‐axis is arranged from the least favorable, cool/coastal garden at Newport to the most favorable, temperate/valley garden at Corvallis. SL, stomatal length; SLA, specific leaf area; N _mass, nitrogen content; SD, stomatal density; VLA, vein density; TD, trichome density; Δ¹³C, carbon isotope discrimination. VLA and TD were not available for plants at Newport. Significance levels: ***, P < 0.001; **, P < 0.01; *, P < 0.05; †, P = 0.053.

Figure 4

Polyploid Fragaria exhibit higher fitness compared with diploids. The composite fitness index, which was the product of genotypic survival rate, growth (plant size) and asexual reproduction (stolon mass), was transformed (with a power parameter λ = 0.1) in the general linear mixed model. The least‐squares means ±1 SE are plotted for diploids (dashed line) and polyploids (solid line) at each garden location. Significance levels: **, P < 0.01; *, P < 0.05; †, P = 0.072.

Figure 5

Trait means (a) and trait plasticities (b) predict average fitness of diploids and polyploids across the heterogeneous garden environments. Standardized regression coefficients (β′) of the main (grey) and ploidy‐specific (red, polyploid; blue, diploid) effects of trait means and plasticities on fitness are presented from general linear mixed models fitted separately for each functional trait. The average estimates of standardized coefficients are denoted by the symbols and the values to the right of the error bars (95% confidence intervals). Significance levels: ***, P < 0.001; **, P < 0.01; *, P < 0.05.