The authors present an R package that simulates population trajectories and then samples phylogenetic trees from the generated population history. The authors present a compelling case for why the package is better than the currently available alternatives, both through timing benchmarks and through evaluating the accuracy of the simulation packages in comparison to a "true" fully sampled phylogeny. While I am not familiar with the Gillespie algorithm that the authors use for simulation, the evaluation seems exhaustive and enough to convince me that the approach they use is appropriate.

The manuscript also mentions the Bayesian tool MASTER, but does not specify why the TiPS package still adds a lot of value to the community. I would argue here that while MASTER is a very powerful tool, not everyone in the community is comfortable with using the BEAST2 platform, which has a steep learning curve. Moreover, a lot of scientists in the field use R as their primary computational tool, meaning that the package will have a user base pool even if MASTER is a perfectly viable alternative.

The paper describes the method in sufficient detail, and the supplemental material completes the description. The paper is well-written and easy to read, with only two typos that I have noticed (main text table 1 “SEAIR” instead of “SEIR”, supplementary figure S7 caption “Lineage-Throuht-Time” instead of “Lineage-Through-Time”). The supplementary material also provides descriptive examples of the use of the package on two realistic examples on two drastically different scales – one epidemiological for a virus with two disease stages, acute and chronic, and one ecological for a population of elephants that needs to be controlled in order to not outgrow its habitat.

The TiPS package is a well-integrated part of the R standard package repository and was easy to install even for someone who never uses R. The documentation is easily available and clear. The authors also provide an R markdown file that contains runnable code for the package which makes using and understanding the package straightforward. The package additionally uses R capabilities for interfacing with c++ which allows it to benefit from high computational speed without the user needing to compile any c++ sources which can be messy on different systems. The package ran seamlessly on Mac OS, which is not often the case with other, non-standard, software packages.

The only concern that this package raises is the lack of visible testing. The links to the code repository are unfortunately broken and I was unable to find the repository with package code, so even if there are tests I was not able to see them. I would appreciate being able to see the code and verify the testing of the package.