The two-fold cost of sex: experimental evidence from a natural system

Abstract

Over four decades ago, John Maynard Smith showed that a mutation causing asexual reproduction should rapidly spread in a dioecious sexual population. His reasoning was that the per-capita birth rate of an asexual population would exceed that of a sexual population, because asexual females do not invest in sons. Hence, there is a cost of sexual reproduction that Maynard Smith called the "cost of males." Assuming all else is otherwise equal among sexual and asexual females, the cost is expected to be two-fold in outcrossing populations with separate sexes and equal sex ratios. Maynard Smith's model led to one of the most interesting questions in evolutionary biology: why is there sex? There are, however, no direct estimates of the proposed cost of sex. Here, we measured the increase in frequency of asexual snails in natural, mixed population of sexual and asexual snails in large outdoor mesocosms. We then extended Maynard Smith's model to predict the change in frequency of asexuals for any cost of sex and for any initial frequency of asexuals. Consistent with the "all-else equal" assumption, we found that the increase in frequency of asexual snails closely matched that predicted under a two-fold cost.

Keywords: Potamopyrgus antipodarum; all-else-equal assumption; asexual reproduction; evolution of sex; experimental evolution; paradox of sex; sexual reproduction; two-fold cost of males.

Figure 1

Increase in asexual frequency in experimental mesocosms. (A) Mesocosms were initiated with 800 field‐collected juveniles (gray), which matured to adulthood and produced offspring (black) over the course of one year. Parents (originally juveniles) and offspring were separated by size and split into discrete generations (t and t+1, respectively). We then estimated the frequency of asexual individuals in parent (q_t) and offspring (q_t+1) generations. (B) The frequency of asexuals increased from the parent (t) to offspring (t+1) generation. Box plot shows median (black bar), upper, and lower quartiles (limits of box), minimum and maximum (whiskers, excluding outliers), and outliers (dots). The measure of significance is derived from the logistic model reported in the text. Each generation is represented by 24 mesocosms. The numbers of triploid females represented by each mesocosm are: 28.33 ± 1.50 SEM for parents and 23.67 ± 3.60 for offspring for the six mesocosms in 2012; 21.00 ± 1.97 for parents and 37.00 ± 2.29 for offspring in 2013 mesocosms; 16.67 ± 2.75 for parents and 34.33 ± 2.03 for offspring in 2014 mesocosms; and 16.67 ± 1.52 for parents and 27.83 ± 2.82 for offspring in 2015 mesocosms.

Figure 2

Theoretical predictions for the cost of sex. (A) Under a two‐fold cost of sex (c = 2), asexual females can produce twice as many childbearing offspring (females) as sexual females. The net cost c is the product of the female fecundity‐survival ratio r and the cost of males. Here, sexual and asexual females produce an equivalent number (n = 2) of surviving offspring (fecundity‐survival ratio r = 1), consistent with the all‐else‐equal assumption. Sexual females make 50% daughters (s = 0.5), so the cost of males is two (1/s = 2). The total cost of sex is then two (c = r * 1/s). (B) Equation (3) shows the fold‐increase in asexual reproduction: under a two‐fold cost (c = 2, black solid line), doubling is observed only at very low starting frequencies of asexual individuals. The proportional increase in asexual frequency declines from two to one as the initial frequency of asexuals (q_t) increases from rarity to fixation. Equation (2)’s corresponding prediction for the frequency of asexual individuals in the offspring generation (q_t+1) is shown in (C). We use equation (2) when fitting models to experimental data. When there is no net cost to sexual reproduction (c = 1, gray dashed line), asexuals have no intrinsic birth rate advantage and will not change in frequency.

Figure 3

Experimental data are consistent with model predictions of a two‐fold cost of sex. We fit our simple model (Fig. 2C; eq. (2)) to experimental data (Fig. 1B) on the frequency of asexuals q in generations t and t+1 in 24 seminatural mesocosms (purple points). We used standard maximum likelihood techniques and Akaike's information criterion to compete different estimates of the cost of sex c in P. antipodarum. The predicted frequency of asexual offspring (q_t+1) for a given frequency of asexual parents (q_t) is shown for three values of the cost of sex: no cost (c = 1, gray dashed line), a twofold cost (c = 2, black solid line), and the maximum likelihood estimate (c = 2.14, solid orange line). The 95% confidence intervals of the maximum likelihood estimate include two ([1.81, 2.55], dotted orange lines). Each point represents one mesocosm. For each mesocosm, the average number of triploid parents was 20.67 ± 1.57 SEM and the average number of triploid offspring was 30.71 ± 1.95.