Variation in heteroploid reproduction and gene flow across a polyploid complex: One size does not fit all

Abstract

Whole-genome duplication is considered an important speciation mechanism in plants. However, its effect on reproductive isolation between higher cytotypes is not well understood. We used backcrosses between different ploidy levels and surveys of mixed-ploidy contact zones to determine how reproductive barriers differed with cytotype across a polyploid complex. We backcrossed F1 hybrids derived from 2X-4X and 4X-6X crosses in the Campanula rotundifolia autopolyploid complex, measured backcross fitness, and estimated backcross DNA cytotype. We then sampled four natural mixed-ploidy contact zones (two 2X-4X and two 4X-6X), estimated ploidy, and genotyped individuals across each contact zone. Reproductive success and capacity for gene flow was markedly lower for 2X-4X than 4X-6X hybrids. In fact, 3X hybrids could not backcross; all 2X-4X backcross progeny resulted from neotetraploid F1 hybrids. Further, no 3X individuals were found in 2X-4X contact zones, and 2X and 4X individuals were genetically distinct. By contrast, backcrosses of 5X hybrids were relatively successful, particularly when crossed to 6X individuals. In 4X-6X contact zones, 5X individuals and aneuploids were common and all cytotypes were largely genetically similar and spatially intermixed. Taken together, these results provide strong evidence that reproduction is low between 2X and 4X cytotypes, primarily occurring via unreduced gamete production, but that reproduction and gene flow are ongoing between 4X and 6X cytotypes. Further, it suggests whole-genome duplication can result in speciation between diploids and polyploids, but is less likely to create reproductive barriers between different polyploid cytotypes, resulting in two fundamentally different potentials for speciation across polyploid complexes.

FIGURE 1

Campanula rotundifolia individual in alvar habitat on Manitoulin Island, Ontario, Canada

FIGURE 2

Schematic diagram of crossing design. F1 plants (middle column) were reciprocally crossed to both parental cytotypes. Two F1 cytotypes were crossed against parental 2X and 4X cytotypes (a), while only one F1 cytotype was crossed against 4X and 6X parental cytotypes (b). Crosses were repeated for both F1 hybrids per cytotype and for two replicates per arrow

FIGURE 3

Performance of backcrosses of F1 hybrids between 2X‐4X or 4X‐6X and their parental populations: relative seed set (a), relative germination proportion (b), and composite fitness (c). Seed set standardized to the maternal population; germination standardized to both parental populations. The top row denotes the backcross parent cytotype, and the bottom row denotes the F1 hybrid parent cytotype. Colors differentiate F1 cytotypes; error bars denote standard error. Differences between backcrosses of the parental cytotypes as determined by orthogonal contrasts: ***p‐value < .0001, all others p > .05

FIGURE 4

Cytotypic composition of backcross progeny. “+” denotes putatively aneuploid cytotypes. Cytotypes crossed shown on x‐axis

FIGURE 5

Cytotypic distribution of Campanula rotundifolia individuals in each contact zone. X‐axis denotes the estimated genome size. Colors denote assigned ploidy level based on genome size (see Methods for details; aneuploids between 4X and 6X were grouped with 5X). Contact zones are as follows: (a) Mittelndorf, (b) Prague, (c) Misery Bay, and (d) Cheddar Gorge

FIGURE 6

Spatial distribution of C. rotundifolia individuals in 2X‐4X contact zones in Mittelndorf, Germany, and Prague, Czechia. Due to the distance between diploids and tetraploids in Prague, Czechia, individuals are displayed in separate groups. The x‐axis denotes relative distance in meters along a latitudinal line, and the y‐axis denotes relative distance in meters along a longitudinal line

FIGURE 7

Spatial distribution of C. rotundifolia individuals in 4X‐6X contact zones in Cheddar Gorge, England, and Misery Bay Provincial Nature Reserve, Ontario, Canada. The x‐axis denotes relative distance in meters along a latitudinal line, and the y‐axis denotes relative distance in meters along a longitudinal line

FIGURE 8

First two axes from a principal coordinates analysis of pairwise genetic distances between individuals in each of the C. rotundifolia four ploidy contact zones. (a) Mittelndorf and (b) Prague are 2X‐4X contact zones, and (c) Misery Bay and (d) Cheddar Gorge are 4X‐6X contact zones. Aneuploids between 4X and 6X were grouped with 5X