Biological material
We used 10 different genotypes (clones) of D. magna isolated from different pools in a metapopulation in south-western Finland, where P. ramosa occurs naturally [31]. D. magna clones from this region are known to produce relatively large numbers of males (laboratory and field observations) but are still strongly female-biased. This allowed us to have male and female host individuals from mothers raised in the same laboratory conditions. Host clones were kept in the laboratory in standardized medium (ADaM [32]) at 20°C, and fed daily with chemostat cultured unicellular green algae Scenedesmus obliquus. Per day and per individual host, we provided 2.5 million algae cells for the first three days, 3 million for the next four days, and 5 million afterwards. During the experiments, individual Daphnia were kept in 100-mL jars with 80 mL ADaM medium, which was changed weekly. The male-specific long antennules that are vestigial in females allowed us to sex D. magna individuals shortly after birth (one-day-old host individuals, as used in some experiments), before the differentiation of major sexual dimorphic traits. Other experiments used D. magna individuals that were three-days old, an age at which sexual dimorphism (also in body size) starts to be obvious. When applicable, body length of adult Daphnia individuals was measured as the distance from the top of the head to the base of the apical spine under a dissecting microscope.
For the bacterial parasite P. ramosa, we used clone C19, which was originally sampled from infected D. magna females in a population in Gaarzerfeld, Germany [7]. This parasite genotype is not qualitatively different from other known genotypes in terms of induced host symptoms. Parasite spore suspensions were obtained by homogenizing infected D. magna in 500 μL of water. Spores were then counted under phase contrast microscopy (Leica microsystems DM 2500, magnification 400×) with a hemocytometer (Neubauer improved) and diluted to the desired concentration for host exposure (see below). As control, we used placebo suspensions obtained by homogenizing uninfected Daphnia. Particular host and parasite clones used in the experiments were not coevolving, which allowed us to specifically test for the factor "sex".
Infections were performed by exposing single host individuals to suspensions of parasite spores. For the larger and sexually dimorphic three-day-old D. magna individuals, exposure took place in 100-mL jars filled with ADaM and lasted 11 days (4 days in 20 mL followed by 7 days in 80 mL medium) before individuals were transferred to 80 mL of clean medium. For the smaller, one-day-old individuals, exposure took place in wells of 24-well plates containing 1 mL of ADaM and lasted two days before transfer to jars with fresh medium. The infection status of D. magna at the end of the experiments was assessed by checking, with phase-contrast microscopy, single individuals homogenized in 500 μL of medium. Individuals that died before Day 14 of the experiments, largely due to handling during sorting, were excluded because detection of P. ramosa infection is less reliable during early stages of infection. An overview of the experiments carried out is given in Table 1.
Likelihood of infection upon exposure
We tested for a difference in infection likelihood between female and male Daphnia hosts in two experiments (Experiments 1 and 2, Table 1). For Experiment 1, we used 30 females and 30 males of D. magna clone SP1-2-3 for each of five treatments corresponding to exposure to different doses (on a log-linear scale) of parasite spores: control (placebo obtained as described above), 5,000, 12,500, 31,300 or 78,100 parasite spores per jar. At exposure, hosts were three-days old. Eleven days after exposure, Daphnia were transferred to fresh medium and 21 days after that, we inspected all individuals (n = 264, excluding 36 that died before Day 14 of the experiment) for the presence of infection with the naked eye. P. ramosa infections produce very clear symptoms visible by eye. By Day 20 post infection 100% of the infected hosts show these symptoms. For Experiment 2, we exposed very young (one-day-old) animals, which do not yet show sex differences in traits, such as body size. We used 20 males and 20 females of each of 7 D. magna clones (Kela 08-10, Kela 10-01, Kela 12-06, Kela 18-11, Kela 20-13, Kela 28-08 and Kela 39-01) for exposure to each of 2 doses of parasite spores: 5,000 or 20,000 spores per well. As control, we used 14 control animals per clone and sex exposed to a placebo parasite suspension. Individuals dying during the experiment were recorded daily and stored for later analysis. We stopped the experiment 120 days after exposure (when all infected and most control individuals had died) and checked infection status of every individual (n = 582, excluding 174 that died before Day 14 of the experiment).
Parasite virulence, fitness and proliferation
To measure the parasite's effect on lifespan of male and female hosts, we used longevity data collected in Experiment 2 (details above and in Table 1). Specifically, the survival analysis was done on lifespan data collected daily for infected (from both dose exposures) and healthy individuals from seven D. magna clones. All individuals dead before Day 14 of the experiment were removed from the analysis, and the six control females that were still alive at Day 120 of the experiment were censored. To estimate parasite fitness, we counted the number of spores at death for two of the seven host clones (n = 49 for Kela 08-10 and n = 46 for Kela 20-13 in Experiment 2).
To test for host sex differences in the rate of within-host proliferation, we counted spores in two groups at two different times after exposure (Experiment 3, Table 1). Individual Daphnia (clone SP1-2-3) exposed to 20,000 spores when one-day-old were killed, measured and homogenized for counting parasite spores (as described above) at Day 20 (37 females and 29 males) or at Day 27 (40 females and 36 males) of the experiment. We stopped the experiment at Day 27 because approximately 50% of the males were dead after that period (Figure 3). The number of parasite spores was estimated by homogenizing individual hosts in 0.5 mL of medium, and counting a subsample of this suspension using a hemocytometer (Neubauer improved). For each individual, we also calculated the density of spores by dividing the number of spores by the host body volume (body volume = 0.2418 × body length2.593 [33]). Note that because the formula to calculate host body volume was established for females which have a brood pouch, it is possible that male body volume was underestimated and, consequently, that parasite density in male hosts was overestimated. If this was the case, the differences in densities we found would be even higher. For the analysis of parasite proliferation, we used the difference in parasite number and in parasite density between Days 20 and 27.
Host castration and gigantism
To test for parasite-induced gigantism, we measured body length of 21-day-old live infected and non-infected individuals from Experiment 2 (clones Kela 08-10 and Kela 20-13) and from an extra dedicated experiment (Experiment 4, Table 1). Here, three-day-old males (n = 25) and females (n = 25) from each of three D. magna clones (Xinb3, SP1-2-3, XFa6) were exposed to 30,000 P. ramosa spores for 11 days. As controls, we used 13 males and 13 females per clone exposed to a placebo suspension. Twenty-one days after exposure, we measured the body length of all individuals still alive (n = 184) and recorded their infection status.
To test for the effect of parasite infection on spermatozoa production in D. magna males (Experiment 5, Table 1), one-day-old males (clone SP1-2-3) were exposed individually (n = 30 per group of the same age) in 20 mL of ADaM medium to 100,000 P. ramosa spores (expected to result in 100% infection rates) or to a placebo (control) suspension (n = 25 per group of the same age). The number of spermatozoa was estimated by homogenizing individuals in 50 μL of medium, and counting a subsample of this suspension using a hemocytometer (Neubauer improved). We estimated the number of spermatozoa in control and infected males at ages 13 (is the approximate age for sexual maturity), 16, 19, 22, 24 and 26 days. Individual males were exposed to 50 μL of 2.5% nicotine (15 minutes in the dark), which stimulates muscle contractions and results in the release of mature spermatozoa. Spermatozoa counts were performed in a total of 120 infected and 110 uninfected hosts (see details in Figure 5).
Statistical analysis
All analyses were performed with R [34]. To compare the proportion of P. ramosa infected individuals between host sexes, we used a generalized linear model (GLM) with a binomial error distribution, and logit link (Experiment 1, n = 211, one host clone, one parasite clone; and Experiment 2, n = 448, seven host clones, one parasite clone; see Table 1). Assumptions on the error distribution were checked by estimating dispersion parameters in GLM; no significant over-dispersion was detected. To study the impact of Pasteuria on female and male Daphnia survival (Experiment 2 in Table 1), we chose to use the non-parametric log-rank test for its robustness (package "Survival" R [34]). The impact of the parasite on host lifespan was assessed by the interaction between the factors "Infection status" and "Sex" in a Cox proportional hazards model. To test for the difference of parasite spore production in male and female hosts, we used non-parametric tests for their robustness (Experiments 2 and 4 in Table 1). For the other tests (specified in the results), we considered parametric assumptions, checked normality and homoscedasticity of residuals, and transformed data when appropriate (the specific data transformation in each case is reported on when the corresponding results are presented). When comparing the body size of hosts infected versus uninfected, we pooled exposed but uninfected and non-exposed individuals as they did not differ in size (linear model with data from Experiment 2, P > 0.5; and with data from Experiment 4, P > 0.05).