Two contrasting hybridogenetic systems in Ticino
Several hybridogenetic systems are known from the Pelophylax model, where the hybrid P. kl. esculentus co-exists with either P. lessonae (L-E system), P. ridibundus (R-E system), sometimes with both (L-R-E system) and sometimes with neither (E-E system), in which case the sexual cycle relies on triploid individuals (see Background). In Ticino, we characterized two putatively different hybridogenetic systems in otherwise closely-related populations.
In northern Ticino, the hybrid P. kl. esculentus was found together with P. lessonae at sites BIA, CAM, GUD, PIA and LOS. At GOL (a river bank) only hybrids were captured but these frogs most likely breed elsewhere, perhaps at the nearby site LOS, where P. lessonae occurs in large numbers. Hence, all these populations fit the expectations of an L-E system, where P. kl. esculentus exclusively produces R gametes and backcrosses P. lessonae to perpetuate (Fig. 7a).
In southern Ticino however, we did not find P. lessonae at any site, despite equivalent search efforts. All populations were exclusively composed of P. kl. esculentus of both sexes, except for five females at site STA that we unexpectedly identified as P. ridibundus (Table 1). Several clues indicate that these P. ridibundus were not parental frogs, but rather the offspring of P. kl. esculentus × P. kl. esculentus hybrid crosses. First, P. ridibundus is not naturally present in Ticino and northern Italy; the closest naturally-connected populations are in Croatia [28]. Second, all five possessed local P. lessonae mitotypes, while parental P. ridibundus normally conserve their maternal lineages throughout hybridogenesis, because of mating preferences: in mixed populations, the large P. ridibundus females are preferentially chosen by the males of the other smaller species, rarely the other way around (e.g. [26]). Third, the nuclear diversity of these females was extremely low (Ho = 0.02), as expected for hemiclonal RR individuals. Fourth, these frogs were much smaller (SVL = 45–65 mm) than regular P. ridibundus (up to 170 mm, [28]). While they could be subadults, all our other observations at the same time of the year in Ticino involved sexually mature frogs, and their small size could rather reflect the mutation load and low fitness expressed by diploid clonal RR genotypes. Although rarely fit, non-hybrid frogs are supposed to arise every year in all-hybrid populations [15, 31]. Hence, southern Ticino could be inhabited by all-hybrid E-E- systems that sometimes produce non-hybrid individuals.
Surprisingly however, we did not find evidence for mixed ploidy. First, L/R-diagnostic SNP alleles received even sequence coverage in all hybrids, as expected for LR diploids, where the L and R alleles should be present in equal quantities. While no confirmed triploid frogs could be included here as controls, the same approach efficiently disentangles diploids from asymmetric polyploids in other hybridogenetic systems (G Lavanchy, pers. com.). Second, no individual was tri-allelic at our diagnostic microsatellites. For the latter, the strong diversity of L alleles should have allowed to detect LLR individuals if these were present. Third, the allelic profiles of our hybrid frogs were far from the range of allele quantity difference reported for LRR frogs at the microsatellite locus RICA1b5 [30]. Hence, the data at hand suggests that the putatively all-hybrid P. kl. esculentus populations of southern Ticino are maintained without triploids, unlike in other parts of Europe. This assumption necessitates confirmation by direct evidence from experimental crosses to trace allele inheritance, and from cytogenetics.
Alternatively, P. lessonae could be cryptically present in southern Ticino, in which case populations would be composed of unnoticed L-E systems. In a bioacoustic survey over 2001–2002, Mattei-Roesli & Maddalena [27] mostly reported P. kl. esculentus in this area, but suspected the presence of P. lessonae at a few sites. A few years before, Vorburger [12] identified two P. lessonae among tens of P. kl. esculentus captured nearby STA (locality Seseglio, now extinct). These observations could represent a recent shift in the composition of these populations (as seen elsewhere [21]), but they could also be the scarce LL offspring from hybrid crosses (rather than parental P. lessonae). Occasional dispersal from northern to southern Ticino is also possible, as illustrated by one “northern” frog caught at site AGR (Fig. 2). Therefore, formally rejecting the hypothetical presence of breeding P. lessonae in southern Ticino will require additional monitoring efforts throughout an entire breeding season.
On the causes and consequences of a putative diploid all-hybrid system
How could a diploid all-hybrid E-E hybridogenetic system perpetuate? To our knowledge, such situation has never been reported. Importantly, because sex is supposedly determined by an XY system [32], and because primary hybridization events preferentially involve P. lessonae males (LxLy) with P. ridibundus females (RxRx), the L hemiclone of P. kl. esculentus hybrids can carry either an X or a Y, while the R is strictly X-linked, i. e. hybrid males are LyRx and hybrids females are LxRx [33]. Therefore, both sexes must provide L and R gametes so sons and daughters can be generated (Fig. 7b). As a consequence, RxRx females (like the ones we found in STA, see also [31]) and LxLy males should also be produced. However, it is worth noting that Vorburger [12] failed to obtain any viable metamorphs from a few inter-hybrid crosses from the extinct Seseglio site. This suggests an important hybridogenetic load and if they exist, that hybrid frogs producing both L and R gametes might be rather infrequent. In counterpart, the occasional LL and RR individuals would provide opportunities for recombination to purge deleterious mutations from hemiclones.
The pre-requisites of this putative diploid E-E system are difficult to reconcile given our current knowledge of Pelophylax gametogenesis. In R-E populations, P. kl. esculentus males can produce sperm of either hemiclones [34], or sometimes both simultaneously by hybrid amphispermy [35, 36]. We are not aware of reciprocal dynamics in female hybridogens, which either transmit their R hemiclone (in L-E systems), or both the R and L within diploid eggs (in regular E-E- systems with mixed ploidy) [34]. Alternated L or R genome exclusion between hybrid females, or between the germ cells of the same female – what could be referred to as “hybrid amphigamy” – is yet to be documented. Nevertheless, gametogenesis and notably oogenesis might be more labile than previously assumed in diploid P. kl. esculentus [34], including the mechanism and timing of genome exclusion [37].
How could such a system arise? Lability in genome elimination could stem from a mixed origin of these frogs, involving secondary contact between L- and R-eliminating lineages. The amphispermic hybrid reported by Ragghianti et al. [36] was a cross between P. lessonae from a L-E system and P. ridibundus from a R-E system. Water frogs most likely colonized the south-alpine region from a Central European refugia, where a great diversity of hybridogenetic systems (including L-E and R-E populations) and clonal lineages co-exist [21].
The effective maintenance of a diploid P. kl. esculentus system, although pending further investigations to confirm the total absence of P. lessonae and of triploids, contributes to the ongoing debate of the evolutionary fate of hybridogenetic hybrids. First, since both genomes would be transmitted, this system challenges the usual view that hybridogenetic hybrids are sexual parasites [38]. Second, it suggests that polyploidization is not required to become reproductively independent, as a preliminary stage of hybrid speciation. Triploidy is often seen as a springboard towards tetraploidy, from which hybrid species are easier to evolve [39]. The diploid hybridogens from Ticino would emphasize an alternative pathway, although whether they are truly self-sustainable (if they reproduce by “hybrid amphigamy”) or rely on interdependent L- and R-eliminating lineages, remains an open question. In a later step, these populations could eventually evolve reproductive isolation from P. lessonae and P. ridibundus by allopatric divergence, leading to homoploid hybrid speciation. Such outcome yet appears unlikely given the frequent reshuffling of amphibian distributions throughout the Quaternary. For the time being, water frogs from Ticino represent some of the last genuine Pelophylax assemblages in Western Europe and offer a promising framework to study these fascinating aspects of hybridogenesis.
No evidence for a south-alpine endemic water frog lineage
In contrast to our previous investigations based on the SAI-1 intronic marker [29], we did not recover a south-alpine lineage endemic to Ticino and northern Italy using genome-wide RAD data. Instead, all pool frogs belonged to a monophyletic, well-supported P. lessonae clade (Fig. 4) and all possessed P. lessonae mitotypes (Fig. 3). SAI-1 thus features strong ancestral polymorphism and does not always seem representative of the evolutionary history of species, perhaps because it contains a retro-transposon [40]. Hence, we recommend that phylogenetic and phylogeographic inferences relying on this marker to be treated with caution (e.g. [29, 41, 42]).
In particular, we previously hypothesized that the hemiclone of Italian hybrid frogs P. kl. hispanicus is related to an undescribed extinct lineage of Anatolian origin, based on SAI-1 variation (“P. n. t. 1”, clearly differing from the north-Italian hemiclones; [29]). Alternatively, this "lineage" could thus simply represent intraspecific SAI-1 alleles of P. ridibundus, or of a related Middle Eastern taxon. Yet, despite several molecular surveys focusing on this region [41, 42], such alleles have still never been reported from extant populations. Similarly, the Cyprus endemic P. cypriensis, which was described from mtDNA and nuclear SAI-1 divergence [42], deserves a re-evaluation. Pelophylax phylogeography and systematics are in clear need for more comprehensive molecular analyses, as offered by genome-wide loci.