Cryptic fitness advantage: diploids invade haploid populations despite lacking any apparent advantage as measured by standard fitness assays

Abstract

Ploidy varies tremendously within and between species, yet the factors that influence when or why ploidy variants are adaptive remains poorly understood. Our previous work found that diploid individuals repeatedly arose within ten replicate haploid populations of Saccharomyces cerevisiae, and in each case we witnessed diploid takeover within ~1800 asexual generations of batch culture evolution in the lab. The character that allowed diploids to rise in frequency within haploid populations remains unknown. Here we present a number of experiments conducted with the goal to determine what this trait (or traits) might have been. Experiments were conducted both by sampling a small number of colonies from the stocks frozen every two weeks (~ 93 generations) during the original experiment, as well through sampling a larger number of colonies at the two time points where polymorphism for ploidy was most prevalent. Surprisingly, none of our fitness component measures (lag phase, growth rate, biomass production) indicated an advantage to diploidy. Similarly, competition assays against a common competitor and direct competition between haploid and diploid colonies isolated from the same time point failed to indicate a diploid advantage. Furthermore, we uncovered a tremendous amount of trait variation among colonies of the same ploidy level. Only late-appearing diploids showed a competitive advantage over haploids, indicating that the fitness advantage that allowed eventual takeover was not diploidy per se but an attribute of a subset of diploid lineages. Nevertheless, the initial rise in diploids to intermediate frequency cannot be explained by any of the fitness measures used; we suggest that the resolution to this mystery is negative frequency-dependent selection, which is ignored in the standard fitness measures used.

Figure 1. Polymorphism for genome size across the time series.

30 000 cells from each of 24 colonies were measured on a FACSCalibur at each time point, with haploid and diploid assignment determined by the kmeans function in the R programming language . Points are plotted with slight jitter on the x-axis for viewing purposes.

Figure 2. Cell size and shape at 1023 and 1302 generations.

Cell size (A: volume, B: surface area) and shape (C: eccentricity, D: surface area to volume ratio) of ancestral and evolved populations, as well as five haploid and five diploid colonies isolated at 1023 and 1302 generations. Numbers assigned to a colony are used consistently throughout all assays (and in all figures).

Figure 3. Cell size and shape across the time series.

Cell size (A: volume, B: surface area) and shape (C: eccentricity, D: surface area to volume ratio) measures for 15 cells from two to four colonies isolated at each time point. Haploid and diploid colonies were analyzed separately to test the relationship between cell size/shape and generation of colony isolation. Solid lines indicate a a significant linear regression (p 0.05) while dashed lines indicate a nonsignificant trend (p 0.5). Here and in later figures, error bars represent 1 SE.

Figure 4. Lag phase fitness proxy.

Glucose % (w/v) measured by HPLC post transfer into new medium. Haploids (isolated at the first time point and after 1302 generations of evolution) and diploids (1302 generation and 1767 generation) do not differ significantly in the amount of glucose present in the medium at the early time points, suggesting that their growth lags do not differ substatially.

Figure 5. Growth rate and biomass production across the time series.

Growth rate and biomass production were measured for 20 haploid and 17 diploid colonies isolated throughout the original experiment. Points are plotted with slight jitter on the x-axis for viewing purposes.

Figure 6. Growth rate, biomass production and the number of cells at 24 hours.

Growth rates (top), biomass production (middle) and the density of cells (bottom) measured for ancestral haploid and diploid lines, five colonies of each ploidy after 1023 generations and 1302 generations of evolution, and the diploid population after 1767 generations.

Figure 7. Competition against a common competitor.

Haploid and diploid colonies were competed directly again a common marked competitor for 72 hours. The y-axis (m) is the difference in malthusian growth rate between the given strain and the common competitor.

Figure 8. Competition against the haploid population from 1302 generations.

Thirteen diploid colonies isolated at 1302 generations, a pool of 12 1302 generation diploids (“Gen1302 2N population”) and a pool of 24 diploid colonies isolated at 1488 generations (“Gen1488 2N population”) were competed against a population of the 6 haploid colonies isolated at 1302 generations. Contamination arose in one of the diploid colony competitions, and we were unable to measure the results of this competition (the blank space in the first panel). Only the population of diploids from 1488 generations (the first time point after diploidy swept in the original experiment, rightmost panel) was consistently able to outcompete the haploid population. All competitions were started at 50:50 (v/v) with transfers into fresh medium every 24 hours for 14 days. Standard error bars based on two replicate competitions.

Figure 9. Replicate evolution experiment.

The experimental evolution study was restarted from A: culture from 1302 generations and B: culture from 1302 generations spiked with 5% diploid culture from 1488 generations. Replicates were evolved through batch culture in exactly the same way as the original experiment [?]. The 1302 generation culture was evolved for 41 days (20 replicates) and 14 days (10 replicates - shaded bars in top panel), mixed culture was evolved for 14 days (20 replicates).