Table 2. Statistical probabilities of line loss.

CV	p	O. tauri		O. mediterraneus		B. prasinos		M. pusilla
		Lexp	P(L ≥ Lobs)	Lexp	P(L ≥ Lobs)	Lexp	P(L ≥ Lobs)	Lexp	P(L ≥ Lobs)
0	0.0025	2.7	0	2.1	0	2.4	0	1.7	0
0.05	0.0026	2.8	0	2.2	0	2.5	0	1.8	0
0.4	0.0150	16.2	0.09	12.6	0	14.4	0	10.2	0
0.5	0.0260	28.1	0.89	21.8	0.5	25.0	0.0000	17.7	0.0033

Statistical probabilities of line loss, with p the probability of line loss at each bottleneck, L_exp the expected number of line losses for each experiment, and L_obs the number of observed line losses. Probability of observing L_obs or more line losses, as a function of the number of lines, the number of bottlenecks, t (16, 21, and 27 bottlenecks depending on species), and the coefficient of variation of the sampling error (γ distribution with average 1 and Coefficient of Variation CV). As an example, for O. tauri, the probability of obtaining the observed line loss, L_obs, over the number of bottlenecks performed, with a CV of 0.04, is 0.09 [P(L ≥ L_obs)], the expected line loss, L_exp, being 2.8.