Fig. 6

Myosin evolution in view of the history of Earth. To correlate major events in myosin evolution with geological times, we determined the density of events within time intervals of 200 million years. We distinguish three types of events: 1) Myosin gain events, which represent the appearance of new classes of myosins with novel domain architectures correlated with new cellular functions; 2) myosin loss events; 3) myosin duplications representing the generation of variants of the same myosin class, having identical or very similar domain architectures and most likely leading to subfunctionalization. For each time interval we summed up the gain and loss events denoted in Figs. 4 and 5; Additional file 1: Figure S15. Events, which could not be attributed to dated splits, were ignored. Depending on the taxonomic sampling of the study, the time intervals for subsequent speciations vary considerably across the tree of the eukaryotes (Fig. 4), and thus the timing can be very different for assigning events to stem or crown age. Therefore, we plotted the earliest possible appearance of a myosin class (stem age, data obtained from Additional file 1: Figure S16B; represented by stars) and, as a more conservative estimation, the latest possible date of myosin invention (crown age, data obtained from Fig. 4; diamonds) for comparison. The density of myosin loss events is displayed by circles. Myosin variants were identified by inspecting the phylogenetic trees and are represented by squares (vertebrate duplications can easily be referred from Additional file 1: Figure S11). Before the Paleozoic, there are only a few myosin duplications within myosin classes resulting in myosin variants, and the respective myosin classes are indicated for orientation. Myosin duplications in the Paleozoic are mainly the result of whole genome duplications (WGDs) affecting all myosin classes present in the respective ancient species. Therefore, not the affected classes but only the WGDs are indicated. All numbers represent rough estimates with respect to the accuracy and the lack of many divergence time estimates. Shaded areas were drawn around the estimated densities of events for simplified orientation