Using reciprocally retained gene families to detect whole-genome multiplications in plants

Traces of ancient whole-genome multiplications (WGMs) have been observed all over the plant kingdom and have been associated with various evolutionary processes, such as increased evolvability, speciation, adaptation to changing environments, domestication and the origin of evolutionary novelties. However, understanding the impact of WGMs on plant evolution requires accurate detection of WGM events, which is challenging because of rapid signal erosion due to genome rearrangements, sequence divergence and the occurrence of additional large- and small-scale duplications (SSDs). Here, we investigate whether reciprocally retained gene families (RR GFs), i.e. GFs that preferentially expand through WGM and rarely undergo SSDs, can be used as WGM markers. Using stochastic birth-death (BD) modeling of GF gene count data to test for WGM presence or absence, we demonstrate that strongly RR GFs have higher power to detect true WGMs and to reject false WGMs than non-RR GFs. However, none of the RR GFs is a perfect WGM marker on its own, and different GFs perform better in different plant clades, prohibiting the use of a fixed small set of RR GFs as WGM markers across all angiosperms. Instead, we show that using an extended set of RR GFs rather than whole paranomes as input for BD and KS distribution modeling approaches leads to improved WGM detection performance.