Lauren FrankelMy first paper from my PhD w/ @cecileane is available on bioRxiv! We found that variation in substitution rates impacts the power and false discovery rate of the D-statistic, D3, and HyDe. 🧵 1/4
Variation of rates between lineages had the largest impact on false discovery of reticulation (type-1 error) across all three methods. D3 was the most sensitive with ~80% type-1 error.
We also found the power to detect hybridization events decreased when there were more hybridizations, for all three tests. So it seems that hybridization events can "hide" each another if they occur within a small subset of taxa. (3/4)
So! Evolutionary biologists: be careful using hybridization summary statistics if you have substitution rate variation, and consider using methods that do not assume constant substitution rates. (4/4)
grimmigesVery nice that finally somebody looked into this. Simply love it!
The suspicion(s) has (have) been out there for quite some time.
Remember discussing this as a potential issue on the #PhyloNetworks workshop in Leiden a few years ago
Distinguishability in Phylogenetic Networks
https://phylonetworks.blogspot.com/2018/08/distinguishability-in-phylogenetic_22.html
Nice, we now know, we worried not for vain ;)
