Genetic slippage after sex maintains diversity for parasite resistance in a natural host population

Abstract

Although parasite-mediated selection is a major driver of host evolution, its influence on genetic variation for parasite resistance is not yet well understood. We monitored resistance in a large population of the planktonic crustacean Daphnia magna over 8 years, as it underwent yearly epidemics of the bacterial pathogen Pasteuria ramosa. We observed cyclic dynamics of resistance: Resistance increased throughout the epidemics, but susceptibility was restored each spring when hosts hatched from sexual resting stages. Host resting stages collected across the year showed that largely resistant host populations can produce susceptible sexual offspring. A genetic model of resistance developed for this host-parasite system, based on multiple loci and strong epistasis, is in partial agreement with our findings. Our results reveal that, despite strong selection for resistance in a natural host population, genetic slippage after sexual reproduction can be a strong factor for the maintenance of genetic diversity of host resistance.

Fig. 1.. Cyclic resistotype dynamics across 8 years in the Aegelsee.

From 2010 to 2018, samples of D. magna were collected from early April to early October every 2 to 4 weeks. Parasite prevalence was recorded, and about 60 to 100 animals were cloned and their resistotypes (resistance phenotypes) were assessed. (A) P. ramosa prevalence (=proportion of infected females) in the D. magna population. (B) Resistotype frequency in the D. magna population. Resistance and susceptibility to individual P. ramosa isolates are denoted as R and S, respectively. The combined resistotype shows resistance for up to five P. ramosa isolates: C1, C19, P15, P20, and P21. Until 2013, only C1 and C19 were tested; in 2014 and 2015, isolates C1, C19, P15, and P20 were tested; and all five isolates were tested after 2015. We use the placeholder ⎵ when an isolate was not tested. Resistance to P20 is pinpointed because of its importance in the evolution of the host population (50). n denotes the total number of genotypes tested in a given year. (C) Resistotype frequency to each of the five P. ramosa isolates. Note the strong increase in resistance to P20 every year.

Fig. 2.. Genetic slippage resulting from sexual reproduction in the Aegelsee D. magna population.

(A) Observed resistotype (resistance phenotype) frequencies in the D. magna population from 2014 to 2018 (same as Fig. 1B for 2014–2018; repeated here for better comparison). (B) Mean resistance to P. ramosa across time. Mean resistance increases across every summer planktonic phase. We attributed to each resistotype a resistance score ranging from zero to the number of isolates tested, and weighted the mean per sampling point by the number of tested isolates, resulting in a score between zero and one (e.g., RRRRR would have an overall resistance score of 1 and SSSSS would be 0). The dashed lines span the time windows during which sexual offspring overwinter and hatch the following spring. (C) Variance of resistance across time, calculated along with the mean in (B). Note that in 2014 and 2015, four bacterial isolates were tested, while we used five from 2016 to 2018; hence, we do not represent the dashed line between 2015 and 2016. Therefore, mean and variance cannot be directly compared between years when different numbers of parasite isolates are used.

Fig. 3.. Longitudinal resting stage hatching of D. magna from the Aegelsee.

(A) Observed resistotype (resistance phenotype) frequencies in the D. magna population from 2014 to 2018 (same as Fig. 1B; repeated here for better comparison). (B) Observed relative number of D. magna resting stage cases (ephippia) produced in the pond and recovered from five to nine sediment traps, in 2- to 4-week intervals from early April to early October in 2014, 2015, 2017, and 2018. Time on the x axis represents the midpoint between two consecutive emptying of the traps. n indicates the total number of ephippia for a given year. (C) Resistotype frequencies of the hatchlings from the sediment traps plotted against the collection time (only for 2014, 2015, and 2017). Resting stages from 2018 were collected but not hatched. Note that in 2014, the first resting stage sample was lost. In 2015, no hatchlings emerged from the last sample. We represent the four-letter resistotype (C1, C19, P15, and P20) to be comparable with (E). (D) As in (C) but for each of the five P. ramosa isolates separately. (E) Expected resistotype frequencies of hatchlings from sexually produced eggs (resting stages) by the planktonic population across the entire planktonic season (also for parts of the season where no resting stages were produced). Expected resistotype frequencies were calculated using the genetic model of resistance in the D. magna–P. ramosa system, assuming random mating of the parent population at the time of resting stage production. Detailed methods and results of these calculations are given in the text and in figs. S5 and S6 and tables S1 and S2.

Fig. 4.. Scatterplot of resistotype frequencies of the hatchlings from the overwintering resting stages (collected in the sediment traps) against those of the D. magna collected the following spring in the Aegelsee.

The x axis represents cumulated resistotype frequencies in the hatchlings from the sediment traps. These frequencies were calculated by weighing resistotype frequencies in the hatchling population by the relative number of resting stages produced at each sampling point. The y axis represents cumulated resistotype frequencies in the first sample collected the following spring after the resting stages. Dots are plotted using jitter to reduce overlap. The gray line represents the y = x function and depicts an expected perfect match between both resistotype frequencies. The black line represents the fitted linear regression, with 95% error as the gray area (not visible in the 2017–2018 panel because it is too small).