In this study we compared phylogeographic patterns and demographic histories of species with similar feeding preferences inhabiting the same geographical area of Lake Baikal. Overall, we found three different phylogeographic patterns in the five species investigated. Both M. herderiana and G. fasciatus exhibit haplotype networks in which a central haplotype is very abundant and widespread, and to which several less common haplotypes are closely related. Baicalia carinatocostata and B. turriformis display very different pattern, with the same haplotype never found in more than one locality and lacking a central and most abundant haplotype. Baicalia carinata shows an intermediate picture, with some relatively abundant haplotypes being found in different localities and rare haplotypes restricted to single sampling localities.
While phylogeographic patterns often reflect habitat availability and connectivity [8, 48] our results suggest that intrinsic biological factors might play an important role in shaping the genetic structure of the species analyzed. The southwestern shore of Lake Baikal, which was sampled for this study, has relatively few sandy areas, with a mostly steep, rocky shoreline (Additional file 4). One would thus expect that species inhabiting mostly sandy bottoms would show high geographic substructuring, while species that prefer rocky habitats would exhibit a pattern indicative of relatively uninterrupted gene flow. Our results, however, are only partially supportive of this hypothesis. Concerning M. herderiana, it should be noted that although inhabiting mostly rocky areas, this species has been found in both sandy and silted areas . In this regard, M. herderiana resembles the generalist amphipod G. fasciatus, which lives in both sandy and rocky substrates [49, 50] and is further known to easily invade new habitats and occupy places in ecosystems [42, 43, 51]. These two species show remarkably similar phylogeographic patterns, with our data suggesting high degree of gene flow throughout the studied geographic range. The inferred patterns for the remaining three species analysed, however, highlight the importance of specific biological characteristics other than preferred habitat type. In fact, B. carinata and B. carinatocostata both live in sandy bottoms, but show rather different phylogeographic patterns. It was reported that B. carinata increases its dispersal by laying eggs on the shells of its conspecifics [40, 41], and this could explain the difference between phylogeographic structures. Similarly, the rock-dweller B. turriformis displays high degree of geographical substructuring, even though significant geographical barriers between rocky habitats in the sampled shore seem absent. This more sedentary species mostly inhabits the surface of steep rocks and cliffs with individuals hanging on each other [41, 44] and is known as a strict specialist in regard to its feeding behaviour and overhanging slopes . This high degree of specialization might reduce the dispersal ability of B. turriformis due to the lack of suitable habitats available.
For the reconstruction of the demographic history of populations it is important to identify if a sampling set represents a single population. For M. herderiana we sampled most of the range of the shore where this species occurs  and found very little genetic differentiation. Likewise, for this species most FST values between localities were non-significant. Baicalia carinata and G. fasciatus occur along whole shoreline of the lake and previous studies [53, 54] involving samples from outside the area of the current study showed that individuals of each of the species form single populations along the southwestern shore. In our analysis, not a single pairwise comparison between localities where G. fasciatus was found exhibited significant FST values. For B. carinata, significant FST values were found between some localities (notably, between comparison involving localities 11 and 12). Similarly, FST values estimated between localities of B. carinatocostata exhibited only few significant results. Conversely, genetic differentiation was higher in B. turriformis, despite the smaller sample sizes used in this study, and FST analysis revealed significant genetic differentiation between most localities. These results confirm that samples of M. herderiana, B. carinata, B. carinatocostata and G. fasciatus represent populations without strong geographical substructuring, and thus are appropriate for reconstruction of demographic histories. Given the higher genetic differentiation in B. turriformis, the reconstruction of demographic histories for this species should be interpreted with caution. Recent results (Peretolchina et al. in preparation) suggest that the co-occurring populations of B. carinata, B. turriformis and B. carinatocostata were not influenced by interspecific geneflow during the time period covered by the current study.
Classic neutrality tests did not detect significant departures from neutrality for any of the datasets. However, the most powerful Ramos-Onsins R2 test  detected population growth of M. herderiana. For G. fasciatus, results of neutrality tests were not significant, but had small p-values (p = 0.08 for R2 and p = 0.09 for Tajima's D test). The structure of the haplotype networks of M. herderiana and G. fasciatus, with a central abundant haplotype and a number of singleton haplotypes, also suggests population growth for these species.
Our demographic reconstructions suggest that population sizes in B. turriformis, B. carinata and B. carinatocostata were rather stable during their evolutionary histories. There are slight trends towards a decline for B. turriformis and B. carinatocostata as well as slight trend towards population growth for B. carinata. However, these slight trends cannot not be taken as evidence for changes in population size because as they appear, the posterior distributions widen. Conversely, BSPs suggest moderate growth for G. fasciatus, and dramatic expansion for M. herderiana. Figure 3 (d) allows to compare the duration of demographic histories for all species, and one could see that demographic histories of M. herderiana and G. fasciatus are short, contrary to demographic histories of B. turriformis, B. carinata and B. carinatocostata. Long demographic histories of B. carinata, B. turriformis and B. carinatocostata do not show response to the climatic fluctuations that are known from the paleo-record of the lake, while shorter demographic histories of M. herderiana and G. fasciatus exhibit strong to moderate growth. It is thus plausible that M. herderiana and G. fasciatus are relatively recent colonizers of the southwestern shore of Lake Baikal, while the remaining species analysed represent more ancient inhabitants of this area. Alternatively, G. fasciatus and M. herderiana populations may have recently undergone strong bottlenecks, with the growth detected reflecting the recent recovery from such bottlenecks, while the remaining species could have maintained relatively constant population sizes throughout their histories. To elucidate this, future work could focus on the analysis of nuclear gene diversity, as autosomal and mitochondrial DNA diversity are expected to show different rates of recovery from bottlenecks .
Calibration of demographic histories based on molecular sequences is notoriously difficult, particularly when specific rates of molecular evolution are unavailable [57, 58]. Nevertheless, such dating can often provide rough time estimates for important events of a species' evolutionary history. After we calibrated demographic histories for populations of M. herderiana and G. fasciatus by applying available rates of molecular evolution, we found that the start of expansion of populations of these species coincide, and could be estimated to 25-50 Kyr BP (Figure 4). Urabe et al.  inferred lake-level variations from seismic surveying and core sampling of the floor of the lake, which appeared to be correlated to changes of the global climate represented by MIS. However, there is no evidence that the drop of the water level due to climate cooling could separate basins of the lake or result in any kind of geographical separation of the fauna inhabiting the southwestern shore. Diatom abundance, that could directly indicate amount of food items available for both species, is shown in Figure 4 (c). The sedimentary core BDP-93-2 from Buguldeika Saddle [21, 23] in concordance with cores st2 and st2-PC-2001 from Akademichesky Ridge  demonstrate a strongly pronounced interstadial peak at the time c. 25-60 Kyr BP. This suggests that populations of M. herderiana and G. fasciatus in the southwestern shore of Lake Baikal started expanding during a warm period of relatively high water level, and when the amount of food available was also rather high. While this would indicate that food availability played an important role in the population growth of these species, it should be mentioned that from c. 24 to c. 14 Kyr BP the amount of diatoms in the lake was very much reduced, however the populations of M. herderiana and G. fasciatus do not appear to have stopped expanding. Data on sedimentary photosynthetic pigments suggests that, despite the reduced bioproductivity of the lake, green algae, diatoms and dinoflagellates were still present in the lake between 16 and 27 Kyr BP . Therefore, it is possible that during this period the abovementioned species relied on other food items. At any rate, the simultaneous growth detected in M. herderiana and G. fasciatus suggests that environmental factors promoted the population growth of these species in the southwestern shore of Lake Baikal. High resemblance of demographic histories of M. herderiana and G. fasciatus, a species known to be of high invasive capability, highlights the strong dispersal potential of M. herderiana and its ability to expand its population size when environmental conditions are favorable.