Experimental tests of host–virus coevolution in natural killer yeast strains

Abstract Fungi may carry cytoplasmic viruses that encode anticompetitor toxins. These so‐called killer viruses may provide competitive benefits to their host, but also incur metabolic costs associated with viral replication, toxin production and immunity. Mechanisms responsible for the stable maintenance of these endosymbionts are insufficiently understood. Here, we test whether co‐adaptation of host and killer virus underlies their stable maintenance in seven natural and one laboratory strain of the genus Saccharomyces . We employ cross‐transfection of killer viruses, all encoding the K1‐type toxin, to test predictions from host–virus co‐adaptation. These tests support local adaptation of hosts and/or their killer viruses. First, new host–virus combinations have strongly reduced killing ability against a standard sensitive strain when compared with re‐constructed native combinations. Second, viruses are more likely to be lost from new than from original hosts upon repeated bottlenecking or the application of stressful conditions. Third, host fitness is increased after the re‐introduction of native viruses, but decreased after the introduction of new viruses. Finally, rather than a trade‐off, original combinations show a positive correlation between killing ability and fitness. Together, these results suggest that natural yeast killer strains and their viruses have co‐adapted, allowing the transition from a parasitic to a mutualistic symbiosis.