All experiments were carried out at 24°C. Both species were cultivated in deionized water as a medium and fed with 2–3 mg/ml suspension of powdered fish food (Friskies®) [15,19,27].
Desiccation was performed every 30th day for 15 months and the experiment was replicated in two different years. The whole population was desiccated by transferring the animals on filter paper substrates and by removing excess water by gentle filtration. During each desiccation event, 3–5 batches with about 250 rotifers each were used to test their recovery rates. Anhydrobiosis was induced in a humido-thermostatic chamber according to a previously established desiccation protocol ("C" in ). The anhydrobiotic animals were re-hydrated after 7 days of desiccation. Only active bdelloids were considered alive and recovery percentage was recorded about 24 h after water addition. Differences in recovery rates have been tested by ANOVA on original data, as they were normally distributed.
Life table experiments
Life-table experiments were run at 24°C under continuous dark condition. For each species a first life-table experiment was run before starting the desiccation series and represented the M0 reference. Paired life table experiments of D and H lines were performed after 4, 8, and 12 months, that is after 4, 8, 12 desiccations. At prefixed times, life-table cohort experiments were set up by isolating about 30 newly laid eggs from each D and H population of both species. Along the D line, the eggs isolated were laid soon after recovery from desiccation. The hatched bdelloids represented the experimental cohort and each rotifer received about 5 μg of food per day for the duration of its life. Daily, culture medium and food were renewed, laid eggs were counted and removed and deaths were recorded. From each experiment, a life table was compiled . From the life table data, fecundity (number of eggs per rotifer per lifetime), number of reproductive days, reproductive effort (number of eggs per reproductive days), eggs produced till 10-d-old, age at the first reproduction, and longevity were calculated.
For statistical analysis, life-history parameters have been transformed when data were not normally distributed or resulted in non-normally distributed residuals. Correlation analysis between parameters obtained from life table has been performed with Pearson test, in order to reduce the number of parameters to analyse, avoiding redundancy. This factor reduction has been performed with Principal Component analysis on a covariance matrix.
Uni- or multivariate GLM (Generalised Linear Model ) was performed to test the effect of (1) treatment (H or D), (2) months (number of subsequent desiccations in the D lines can reveal a cumulative effect of the desiccation), and (3) interaction between treatment and months.