Spontaneous deleterious mutations are expected to accumulate through Muller's ratchet in clonally reproducing organisms and may lead to their extinction. We study deleterious mutations and their effects in a system of European frogs. Rana esculenta (RL), natural hybrids R. ridibunda (RR) × R. lessonae (LL), reproduce hemiclonally; both sexes exclude the L genome in the germ line and produce unrecombined R gametes; hybridity is restored each generation by matings of RL with coexisting LL. Different allozyme-defined hybrid hemiclones (R genome haplotypes) are thought to have originated independently from primary hybridizations RR × LL. Natural matings between two hybrids usually lead to inviable RR tadpoles. This inviability is thought to result from unmasked deleterious alleles on the clonally transmitted R genomes. Most simply it reflects homozygosity for recessive deleterious alleles at particular loci; alternatively (consistent with absence of RR adults in multiclonal populations) it may reflect hemiclone-specific sets of incompletely recessive deleterious mutations that cumulatively cause inviability when two such genomes are combined. If inviability results from the former, progeny of two hybrids of different hemiclones, whether allopatric or coexisting, should be viable, because it is improbable that their R genomes share recessive deleterious alleles at the same set of loci; if inviability results from the latter, progeny of hybrids of different hemiclones should be inviable, especially when hybrid lineages are old. We tested these hypotheses in artificial crosses, using frogs from three regions: hemiclonal hybrids outside R. ridibunda's range from northern Switzerland (two abundant coexisting allozyme-defined hemiclones; estimated lineage age ≤5000 generations) and from Sicily, Italy (one hemiclone; estimated age ≥25,000 generations) and R. ridibunda from Poland. We generated RR progeny, which we reared under benign conditions in the laboratory, by crossing (1) two hybrids from the same region (H × H local); (2) two hybrids from different regions (H × H foreign); (3) hybrids and R. ridibunda (H × R); and (4) two R. ridibunda (R × R). Survival to metamorphosis was similar and high for R × R, H × H foreign, and H × R, whereas all tadpoles of H × H local died before metamorphosis. This supports the hypothesis that homozygosity for recessive deleterious mutations at particular loci causes inviability. Crosses within and between the two coexisting hemiclones from Switzerland were, however, equally inviable. This result may reflect episodic sexual recombination in RR progeny from exceptional successful interclonal hybrid × hybrid matings, followed by matings of such RR with LL. This process would both slow down or halt Muller's ratchet and disrupt genetic independence of coexisting hemiclones, so that the same remaining deleterious R alleles could exist in different allozyme-defined hemiclones. Whereas all data are consistent with the prediction of Muller's ratchet operating on clonally transmitted R genomes of natural hybrid lineages, they are insufficient to demonstrate such operation, because deleterious recessives that mutated after clone formation and those that preexisted in the R. ridibunda source populations that formed the hemiclonal lineages are not distinguished. The possibility of episodic sexual recombination must be carefully taken into account when studying Muller's ratchet in natural populations of this Rana system.

Corresponding Editor: C. López-Fanjul