Wolbachiainfection
We randomly selected 10 adult springtails from second-generation populations of: (i) the F. candida control treatment (0% antibiotic); (ii) the F. candida high-dose (2.7%) tetracycline treatment; (iii) the F. candida high-dose (2.7%) rifampicin treatment; and (iv) the sibling species Folsomia fimetaria, which has males, reproduces sexually, and is reputed not to host Wolbachia. We used Wolbachia-specific primers in polymerase chain reaction (PCR) analyses of the total DNA extracted from each of these 40 individuals to assess whether the bacteria had been eliminated (see Methods for details of experimental procedures). Wolbachia-specific amplification products were not found in any springtails from the high-dose rifampicin treatment, indicating that this treatment successfully eliminated the bacteria. Similarly, none of the 10 F. fimetaria individuals tested positive for Wolbachia, and it is thus indeed likely that the two are not symbiotic. Perhaps surprisingly, all F. candida individuals from the high-dose tetracycline treatment tested positive for Wolbachia, along with all those from the control populations. The previously reported inefficacy of tetracycline in curing Folsomia of Wolbachia [13, 17] is thus supported.
Population parameters
Clear differences in the population growth rates of high-dose rifampicin treatments compared with high-dose tetracycline treatments were also evident, although these differences did not emerge until the second generation (Figure 1). The growth rate of rifampicin populations differed significantly between the two generations (F
(1,110) = 53.5, P < 0.001), because of a second-generation dose effect whereby growth in the 2.7% and 0.9% treatments was significantly lower than that in the 0.3% treatments which, in turn, was lower than those in the 0.03% or 0% treatments (Figure 1). No other differences in growth rate among the treatments and/or generations were detectable.
Slight differences in growth rate, clutch size and hatching rate were found among some clones. However, as these differences did not significantly interact with the effects of interest (that is, dose, antibiotic type and generational effects), data have been pooled and clone-specific differences are not presented.
Clutch size
At week 4 of the first-generation, individuals from the tetracycline treatments had a significantly larger mean clutch size of 44.25 ± 2.39 than individuals from the rifampicin treatments, which had a mean clutch size of 33.95 ± 1.56 (F
(1,118) = 10.29, P = 0.002). No dose effect on clutch size was detectable in the tetracycline treatments (F
4,55 = 1.79, P = 0.15). However, clutch size was significantly affected by rifampicin dose (F
4,55 = 7.64, P < 0.001), reaching sizes as low as 27.00 ± 2.15 at the highest 2.7% dose.
Over the 6-month observation period (3 months in which the diet included 2.7% rifampicin followed by 3 months in which the diet was antibiotic-free), the clutch size of F. candida differed significantly by month (F
5,24 = 5.37, P = 0.002). While clutch size did not differ among the first 4 months (being 27.45 ± 1.56 on average), the sizes observed in the final 2 months (40.4 ± 3.37 and 42.8 ± 2.13) were significantly greater than those recorded in the preceding months. In F. fimetaria, the average clutch size of 13.03 ± 0.68 did not fluctuate significantly across the 6 months.
Hatching rate
The average hatching rate for clutches laid at week 4 of the first-generation tetracycline treatments was 56.67 ± 3.06%. Hatching rate was not found to be significantly affected by dose in the tetracycline treatments (F
4,55 = 1.82, P = 0.14). However, dose was significant for hatching rate in the rifampicin populations (F
4,55 = 2.76, P = 0.04). Hatching rates at week 4 of the first generation fell from around 50% for doses of 0.03% and 0.3% to just 28.2 ± 3.9% for the 2.7% rifampicin dose. Hatching rates were not systematically recorded in the second generation but this rate was observed to decline rapidly to zero in the 2.7% rifampicin treatments.
Nymph:adult ratio
Marked demographic changes occurred because of prolonged exposure to high doses of rifampicin. In the first generation, the average nymph:adult ratio (in which higher values indicate a younger, faster-growing population) of the tetracycline treatments was very similar to that of rifampicin treatments (9.12 ± 0.31 versus 8.95 ± 0.29). Four weeks into the second generation, the average nymph:adult ratio in the tetracycline treatments was 12.31 ± 0.48, reflecting the increase expected from young, newly founded populations. In contrast, the nymph:adult ratio in the second generation of rifampicin populations at the same stage had dropped to just 6.96 ± 0.67. This treatment difference in week 4 of the second generation was already highly significant (F
1,118 = 42.13, P < 0.001) and it had become absolute by 8 weeks, when nymphs were entirely absent from all the populations exposed to 2.7% rifampcin and 11 of the 16 populations exposed to 0.9% rifampicin. The individuals of these populations were both devoid of Wolbachia and totally sterile.
During the 6-month observation period, the nymph:adult ratio remained at 0 for the entire period in F. candida (that is, zero viable nymphs emerged from the hundreds of eggs that were laid). In F. fimetaria, nymphs of all instars were present for all 6 months: the nymph:adult ratio did not differ significantly across months, being 3.53 ± 0.10 on average.
Males
Although males were easily identified in populations of F. fimetaria (the sexual species), they were never found in any F. candida population.