We first focus on scenarios simulated under the SST, with parameters close to those estimated in real data. The first and most striking result is that we systematically observed a recent bottleneck under all simulated scenarios (Table 2, Figures 4, 5, S1, S2 and S3). This result could seem at a first glance surprising and due to an artefact. However, this is not the case, as: i) the signal does not depend on the inferential algorithm chosen to analyse the data (i.e., the stairwayplot), since the normalized spectra showed a deficit in singletons compared to the other low frequency classes (Figures 4, 5, S1 and S2), which is typical of a recent population decline; ii) it is consistent with the distribution of the Inverse Instantaneous Coalescence Rate (IICR) computed in one diploid individual, which shows a signature of decline under similar meta-population models (Chikhi et al., 2018; Mazet, Rodríguez, Grusea, Boitard, & Chikhi, 2016; Rodríguez et al., 2018). The results of our simulations are consistent with the recent bottleneck observed in the three shark species (Figure 3), with its intensity inversely correlated to the estimated Nm (i.e., stronger for C. melanopterus and C. limbatus than for C. amblyrhynchos). In our SST model there is an instantaneous colonization of the array of demes at TCOL, which corresponds also to a demographic expansion (i.e., the total number of individuals in the array of deme is larger than those in the ancestral deme). However, this signature is detected only for Nm≥5 when TCOL is neither too recent nor too old (at equilibrium) (Figures 5, S1, S2 and S3). In these scenarios, the beginning of the expansion retrieved by the stairwayplotbroadly corresponds to the simulated TCOL. This again corroborates the results obtained for the three shark species, since the two species with higher Nm displayed indeed an ancestral expansion in the stairwayplot with a timing consistent with the estimated TCOL (Table 1, Figures 2 and 3). Similarly, it explains why we could not retrieve the ancestral expansion for C. melanopterus nor estimate TCOL under the SST model: this appears to be a property of the coalescence pattern and it is not related to the amount of data available (see below).