Recovery percentages recorded after each of the 15 desiccation events varied between 88% and 97% in the first experiment, and between 86% and 94% in the second one (Additional file 1). A two-way ANOVA revealed that neither the number of desiccations (F15,82 = 1.26, P = 0.243) nor the replicate experiments (ANOVA test: F1,81 = 0.04, P = 0.84) affected the recovery of the D population.
During the whole lifespan each bdelloid of the D line produced 26–28 eggs in a mean 9.5 reproductive days (Table 1). The fecundity of their ancestors that represented the reference cohort (M0) was not different, while the eggs produced by the H line were a mean of 17 eggs after 8–12 months of stable hydration. The reproductive effort of the D line ranged from 2.9 to 3.2 eggs/female x day after 12 desiccation events (M12), while in H line the reproductive effort was 1.3 after 12 months of continuous hydration.
All fecundity-related parameters (number of eggs, number of reproductive days, reproductive effort, eggs produced till 10-d old, and age at the first reproduction) were highly related to each other (all correlation values 1 and Additional file 2). Thus, a new parameter capable to summarise all parameters associated to fecundity was computed through a Principal Component analysis. The first PC axis explained 92.01% of the total variance. It was directly related to number of eggs, reproductive effort, and eggs produced till 10-d-old, and inversely related to the number of reproductive days and to the age at the first reproduction. This new synthetic parameter was used as a proxy for reproductive fitness. The effect of (1) treatment (D or H), (2) time from start of the experiment (expressed in months and equivalent to the number of desiccations in the D line, called months, M, from now on), and (3) their interaction was tested on fitness components and longevity. A multivariate GLM (Generalised Linear Model) gave significant effects (MANOVA test: all Wilks' Lambda 1,155 = 179.52, P 1,155 = 0.08, P = 0.779). The absence of association between fecundity and longevity may be due to absence of effect, but could be biased by the poor resolution of the statistical test (observed power = 0.059). Months, which represent the time from the start of the experiment, significantly affected longevity (F2,155 = 19.89, P 2,155 = 2.95, P = 0.055). The interaction between treatment (D or H) and months affected both fitness and longevity. Fitness components decreased with time progressively in the H line but not in the D line (Figure 1).
Recovery percentages along the 15 desiccation events were 82% – 96% in the first experiment and 88% – 97% in the second one (Additional file 1). Recovery rates did not differ between years (Two-way ANOVA: F1,78 = 2.96, P = 0.089), but for each experiment differed between months significantly (F14,78 = 2.23, P = 0.014). Apart for the variability, no trend was found in the recovery values.
In the life table experiments, each bdelloid of the D line produced 25 – 27 eggs during the duration of its life, not differently from the initial reference cohort (M0). In the H line the number of eggs produced by each bdelloid was about 20, with a minimum of 15 eggs per lifetime in M8. Reproductive effort remained almost stable through the D line, from M0 (reference) with 1.05 eggs/female x days to M12 with 1.34 eggs/female x days, while it decreased to 0.76 eggs/female x days after 12 months of continuous hydration (Table 1).
All parameters associated to fecundity (number of eggs, number of reproductive days, reproductive effort, eggs till 10-d-old, and age at the first reproduction) and longevity were closely related (Table 1 and Additional file 3). Therefore, through Principal Component analysis we found a new parameter capable of expressing fecundity, fecundity-related and longevity parameters. The first PC axis explained 99.61% of the total variance and was directly related to number of eggs, number of reproductive days, reproductive effort, eggs produced till 10-d-old and longevity and was inversely related to the age at first reproduction. The effect of (1) treatment (D or H), (2) time from start of the experiment (that is months, M), and (3) their interaction was tested on this single parameter used as a proxy for fitness. A GLM test revealed that fitness was affected by the treatment significantly (F1,160 = 105.32, P 2,160 = 9.63, P = 0.055), while the interaction between treatment and months was ineffective (F2,160 = 2.04, P = 0.133) (Figure 2).