Is Schmidtea mediterranea a single species?
The levels of genetic divergence found among all of the populations of S. mediterranea studied, which cover most of its current range of distribution, suggest that it is a single species. This suggestion is based on the large gap between the higher genetic distance values obtained among S. mediterranea populations and those between the S. mediterranea populations and S. polychroa (values of 0.035 for COI, 0.028 for CYB and 0.014 for N13 vs. the means of 0.15, 0.17 and 0.14, respectively). Furthermore, these values are within the range for other Tricladida species. In Dugesia and Microplana (a freshwater and a terrestrial genus, respectively), the COI divergence within species is less than 0.04, whereas among species it is greater than 0.10-0.12 [10, 45]. Such differences are clearly visualised in the first phylogenetic dating analysis, which includes the other three Schmidtea species (Additional file 4).
S. mediterranea is genetically structured. This is evident for the mitochondrial genes and, although not as clearly, for the nuclear intron as well. The haplotype networks for both mitochondrial genes show three well-defined groups that are separated by a similar number of differences, and these groups correspond to the same three clades separated by deep nodes in the phylogenetic analyses (Figures 2 and 3 and Additional file 3). The structure shown by the haplotype networks and phylogenetic trees is confirmed by the SAMOVA results, which show that the same three groups are the ones that best explain the genetic structure in relation to geographical distribution, in this case, for the three markers (Additional file 3). Additionally, genetic differentiation analyses (Table 4) show that the three groups are significantly differentiated both for the mitochondrial and nuclear genes (high FST values and a significant Snn value). The differences in genetic diversity patterns found between nuclear and mitochondrial markers generally result from differences in coalescence times between the two genomes. The fact that in certain analyses the nuclear marker also detects the division into three groups supports the idea that this might be the result of an old isolation; together with the observed lack of gene flow among populations, this finding could indicate that these groups are undergoing speciation or even constitute cryptic species (although the reported inter-fertility among C and S individuals  makes this latter hypothesis less plausible).
The geographical distribution of Schmidtea mediterranea: remnant of an old distribution driven by microplate dispersal or indicator of recent colonisations?
The genetic structure found for S. mediterranea does not reflect a process of recent colonisation from one region to the others. Indeed, the genetic pattern found -- with significant differentiation among geographical groups and even among populations within each group (polymorphism parameters show low levels of polymorphism within populations but high levels when all populations are considered together for mitochondrial genes, which reflects high differentiation among the populations) --indicates better an ancient differentiation among these three geographical groups and even within them. Moreover, the shape of the haplotype networks, which shows the groups to be equidistant, suggests that no region is central to the others. Altogether, these findings strongly suggest that the current distribution of S. mediterranea is a relic of the split from a former larger geographical distribution.
Taking into account the phylogeny of S. mediterranea, could the fragmentation and migration of microplates be the sole reason for its current distribution? Despite the evident complexities of the geology of the western Mediterranean [16, 17, 47], there is a general agreement on the original location of the microplates of Corsica, Sardinia, part of Calabria, the Balearic Islands and both Kabylies (Grande and Petite), which form the CSb, the region that separated from Iberia and Eurasia 30 mya. In the first event, the southern assemblage, which included the Balearic Islands and the Grande Kabylie, broke off and moved clockwise from Iberia, and the northern assemblage, which included Corsica, Sardinia, part of Calabria and the Petite Kabylie, broke off and moved counter clockwise from Eurasia. Next, the crucial event was the breakage between both assemblages that gave rise to the Sardinian Channel at 24 mya, dividing the western from the central and southern areas. Between 16 mya and 10 mya, Corsica and Sardinia reached their present locations and collided with future peninsular Italy (then connected with Sicily). Finally, the Tyrrhenian Sea opened between 8.6 mya and 7.6 mya . In parallel, the Grande and Petite Kabylies broke off from the Balearic Island and Sardinia microplates, respectively, between 25 mya and 20 mya, moved southwards and collided with Africa to reach their final positions approximately 15 mya. Furthermore, the history of the islands, such as the Balearic Islands and Sicily, is even more complex because in the last 20 my, they have endured periods of submersion and emersion, as well as continental contacts. Thus, ~14.8 mya, as attested by the entrance of diverse animal species, Menorca and Mallorca were in contact with the Iberian Peninsula. Later, a marine transgression covered both islands, wiping out most of the Protoligurian fauna and flora, only sparing some species in the highest parts of Mallorca (Serra de Tramuntana). A second contact with the peninsula, approximately 5 mya (during the Messinian crisis, when the sea level dropped dramatically), allowed the entrance of some mammal species. Finally, during the Pleistocene glaciations, the two islands formed a single landmass, easing the migration of the remaining Protoligurian and other fauna from Mallorca to Menorca [48, 49].
The first dating analysis (Additional file 4) provides an age of ~43 mya for the divergence of S. mediterranea from its sister species Schmidtea polychroa. This date and the age inferred for the older node of S. mediterranea in our two dating analyses (between ~20 and ~5 mya; Figure 3, Additional file 4) sets a time range for the origin of the species that is congruent with the hypothesis that S. mediterranea, or its most recent ancestor, was already present on the CSb microplate (or microplates ) when the microplate broke off from the continent ~30 mya. Furthermore, the basal position of the W group (Spanish populations) is consistent with the next breakage of the CSb into the northern and southern microplate assemblages. Finally, the splitting of Corsica and Sardinia from Calabria during the opening of the Tyrrhenian Sea (7.6-8.6 mya) and the connection of the latter, via Sicily, with Tunisia would have resulted in the separation of the C and S groups (~7 my of age for the last common ancestor). The spreading of the groups from Calabria to Sicily and Tunisia was likely very hazardous given its low dispersal capability, but it happened during a rainier period than today. Indeed, the Mediterranean climate as we know it, with dry summers, did not appear until 3.2 mya  with the main difference being the frequency of rain, rather than the temperature, which should have resulted in more freshwater courses and, hence, more opportunities for freshwater planarians to disperse. Last but not least, with the present data, we can also postulate that the present asexual populations (W group) are descendants of the original sexual S. mediterranea that remained on the continent or in the Balearic Islands after Sardinia and Corsica broke away, although the dating obtained is too recent to be explained by the geological history (between ~20 and ~4 mya vs. 24 mya). Alternatively, this age could be consistent with a variation of the CSb migration hypothesis  in which the microplate including Corsica-Sardinia and Calabria remained connected to the Paleo-Europe continent during its migration, becoming detached ~5 mya, when the Mediterranean was refilled after the Messinian salinity crisis and simultaneously with the raising of Tuscany. This alternative could explain the closeness of the split between the three geographical groups in our phylogenetic trees and their equidistance in haplotype networks. However, in this case, we would expect S. mediterranea to be distributed across a wider area, including Liguria and the Mediterranean coast of France, but S. mediterranea has never been described in these places or been found there in recent samplings, whereas S. polychroa and other planarian species have. One can consider that the asexual character of the W group populations could have affected the dating analyses, resulting in artefactually younger datings (see the next section).
Compared to its sister genus Dugesia (with more than 20 species described in the Mediterranean), the genus Schmidtea (with a mere 4 species) has a much lower diversity (see Additional file 4). Even in the area where we find S. mediterranea, Dugesia has diversified into at least 4 species (D. hepta, D. benazzii, D. subtentaculata, and D. liguriensis) probably as a consequence of the same geological events described here. Therefore, the results point to Schmidtea as a low diversifying genus and to S. mediterranea as an old species compared to others in the same area, even within the Tricladida. With the exception of S. mediterranea and some populations of S. polychroa in North Africa and Sardinia, most Schmidtea inhabit continental Europe up to Scotland and Sweden. Sexual S. polychroa and S. lugubris lay cocoons between 10°C and 23-25 °C  because higher temperatures are deleterious. Even for sexual populations of S. mediterranea, Harrath et al.  showed a moderate to drastic reduction in the size of the testes, ovaries and the copulatory apparatus begins at over 20 °C and worsens over 25 °C. In other words, Schmidtea seem best adapted to climates with moderate summer temperatures, a situation that is by no means found on the Mediterranean islands and in continental coastal areas. This raises a final intriguing question: could S. mediterranea be considered a survivor trapped in a harsh habitat as a consequence of CSb breakage and migration?
The asexual populations of S. mediterranea: Single or recurrent origins? Recent or ancient?
Genetic variability at the mitochondrial level is null within asexual populations and very low when the three populations are taken together (only for CYB did one of the three populations show a different haplotype diverging only at one nucleotide). In contrast, the nuclear marker N13 shows, namely, in the Barcelona population, high haplotype diversity similar to that found for sexual populations in other regions. This value might, however, be rather misleading because the Barcelona population is almost exclusively made up of heterozygous individuals (32 out of 33). When we measure genotype diversity instead of haplotype diversity (Table 3) all asexual populations show lower genetic variabilities than the sexual populations. In summary, whereas in sexual populations the variability is evenly distributed among all genotypes (whether heterozygous or homozygous), in asexual populations variability is exclusively found in specific genotypes. Although the latter situation would be at odds with a panmictic population, it might be anticipated for an asexual population that reproduces by fission.
How is the very low or null variability in asexual fissiparous S. mediterranea explained? Most likely, when a few individuals, or even one, became asexual within a sexual population (assuming that they persisted and outcompeted their sexual counterparts), this genotype would become fixed, which would result in the disappearance of the sexual individuals and of most genetic variability. Present climatic conditions and the types of habitat where these asexual forms are found make this scenario very likely. In Menorca, the asexual forms live in temporary water courses that become dry or are reduced to small ponds during summer, with running water only after rainfall in winter and early spring. Under these conditions and with a fair amount of food available, fissiparous organisms (which duplicate in number after fission and regeneration in 15- 20 days) could reproduce faster than sexual individuals, which have higher energy demands than asexual individuals because they need to develop and maintain reproductive organs, mate, and lay cocoons . In fact, the negative values obtained in the Tajima's D neutrality test, although not significant, provide support for an expansion signal for the two Menorca populations. For the Barcelona population, the situation is different because it is localised in a plant nursery without seasonal water shortages. However, other factors, such as periodic pond cleaning in the plant nursery, may cause some decreases in population size. The significant positive values obtained in the neutrality tests may result from these types of recent and less pronounced bottleneck.
It is important to bear in mind that asexual reproduction by fissiparity poses problems that are very different from the much more common asexual reproduction by parthenogenesis. In parthenogenetic populations, new genetic variants could arise and spread by mutation in the germline, and with the occurrence of meiosis, new combinations could arise by recombination. In fissiparous populations, all mutations are somatic instead, and the probability of a new mutation spreading over a whole individual, and thereafter through the whole population, should be very low and could explain the extremely low rates of substitutions in these populations. Therefore, the genetic composition found in these organisms might be a frozen picture of the genetic pattern that was present in the first animals that became asexual. Although very low rates of substitution (absence of nucleotide diversity) are evident in the data presented here, more in depth studies with new markers are necessary to understand how these populations keep the burdens of a lack of genetic variability at bay.
From the data obtained, is it possible to make an educated guess about when and where asexual populations originated and spread? The close genetic relationship among the Menorca and Barcelona populations, the sharing of mitochondrial haplotypes, and their geographical closeness argues for a single asexualisation event. After the asexual individuals outcompete the original sexual population, they might have spread from the island to the continent or vice versa (the former is more likely given the basal situation of MEN_MER in the W clade) either by passive dispersal or through human activities, that has been proven for several species, freshwater planarians among them (Ribas et al.  for Girardia tigrina; and Lázaro et al.  for Dugesia sicula). However, although the estimated age for the clade, including the BAR_MON and MEN_GOR individuals, is too recent to be explained by the geological events described, it is also too old to be explained by human transport (0.6-0.2 mya). As for the age of onset of the asexual lineage, it is recent according to the point of diversification among fissiparous populations (1.36-0.14 mya).
However, this and the recent dating obtained for the basal splitting in the species tree (Figure 3) might be underestimates caused by the anomalously low genetic substitution rates of fissiparous organisms misleading the methodologies used in the dating analyses. To test this hypothesis further, we performed a test and attempted to model the change in the evolutionary rate in the asexual lineage by using BEAST. To do this, we repeated our second dating analysis, and this time we considered the asexual sequences as fossils so that at the point in the past where they first appear, the lineage stops accumulating changes (Additional file 5). The results show that, if we assign an old age to the asexuals (around 20 mya) and use the rate obtained in our first dating analysis, we recover splitting ages among geographical groups that are close to those that would be expected if the CSb hypothesis were true. The low substitution rates of fissiparous populations could spare them the burden of accumulating deleterious mutations and explain this longevity, although they still need to avoid the problem of a lack of variability. Further genetic analyses, including more coding regions, could help to ascertain whether these hypotheses are valid and to understand how mutations spread within individuals and became fixed in populations of these asexuals.
A final conundrum of the asexual populations of S. mediterranea is the presence of the fissiparous triploid population (MEN_MER) that bears the translocation in one of its three chromosomal sets. In freshwater planarians, triploid individuals are known to form in sexually reproducing populations (e.g., in S. polychroa [54, 55]) from the union of unreduced diploid ovules (or sperm) and haploid sperm (or ovules). The MEN_MER population bears two normal chromosomal sets with one set including the translocation between the 1st and 3rd chromosomes. Therefore, this population could have originated from the union between an unreduced diploid ovule from a sexual diploid and a sperm bearing the chromosomal translocation from an asexual individual. In laboratory cultures, large (e.g., >20-25 mm in length) individuals from asexual fissiparous populations often develop testes, large ovaries, and the whole copulatory complex [6, 56] and have been reported to mate with sexual diploids, although the cocoons that are laid are usually sterile. Although fissiparous specimens do not usually reach this size in nature, such an event is the more reasonable explanation for the origin of fissiparous triploids in S. mediterranea.
The sexual populations of S. mediterranea
The sexual populations studied showed extremely low mitochondrial variabilities (Table 3), with two of them (one in Sardinia, one in Sicily) having no variability. Data on intrapopulation variability for other triclads are scarce, but two terrestrial planarian species from the Atlantic forest in Brazil have notably higher values of π (0.01 and 0.017 for COI ). Although other groups of animals show very low values of π [e.g., Palinurus elephas (0.0013-0.0030 for COI, ); and the lepidopter Helicoverpa armigera (0.0017-0.0038 for COI, )], the values are not as low as those found for S. mediterranea. For the nuclear marker, intrapopulation variability is higher than for mitochondrial genes, although again two populations (from Tunisia and Sicily) show no variability. However, as pointed out above, the high diversity found among geographical regions, even for the nuclear marker, suggests an old origin for the present distribution of planarians and, hence, for their populations.
Because old populations are expected to contain high variability (because long stable populations accumulate changes), the low levels determined here could be explained by demographic events affecting neutrality. The neutrality tests applied (Table 6) show that many of these populations demonstrate a significant result against the hypothesis of neutrality, with the estimated values on the positive side of the interval obtained in the permutation test. This finding may be interpreted as a consequence of recent bottlenecks (that would also explain the significant genetic differentiation found among populations; data not shown). This interpretation is in agreement with the habitat and climate conditions in which these populations live, even though their demographic conditions are not as harsh as those for the fissiparous populations from Menorca, and the sexual character of the C and S populations results in a different outcome (here, there are no signs of population expansion).
The relationship between Sicilian and Tunisian populations merits a final comment. The close relationship between the Tunisian population and some individuals from the Sicilian populations may indicate that the only Tunisian population (despite extensive sampling [12, 60]) was recently introduced by human activities. However, this is unlikely because the Sicilian and Tunisian populations do not share a mitochondrial haplotype and because the age of the node giving rise to the Tunisian population (Figure 3) far exceeds the age in which human activities began in the Mediterranean. Therefore, the scarcity of S. mediterranea in Africa (it has so far not been reported from Algeria to the Canary Islands) may be explained by competition between the abundant fissiparous triploids of D. sicula  and S. mediterranea. Fissiparous populations of D. sicula are present all around the Mediterranean coastal areas (from Greece and Israel to the Canary Islands ; E. Solà and M. Riutort, unpublished data) where they seem well adapted to high temperatures and dry conditions [60, 61]. It is not surprising that D. sicula may outcompete S. mediterranea, which at present has been reduced to a single locality found in Tunisia, despite a reasonable sampling coverage.