Technical comment on Condamine et al. (2019): a cautionary note for users of linear diversification dependencies

Abstract

Condamine et al. (2019; 22: 1900–1912) fitted linear and exponential functions of time‐dependent, diversity‐dependent and temperature‐dependent diversification to investigate diversification dynamics of tetrapod families. Here I highlight potential misinterpretations when using linear diversification dependencies and provide some clarifications.

Response to technical comment ‘A cautionary note for users of linear diversification dependencies’

“…These clarifications being given, we agree with Gamisch (2020) that our sentence ( A positive reflects a slowdown of speciation (extinction) towards the present, while a negative reflects a speed‐up of speciation (extinction) towards the present p. 1902) is imprecise. Similarly, the sentence ( A positive indicates that speciation (extinction) rates are higher under warm climatic periods, while a negative indicates that speciation (extinction) rates are higher under cold climatic periods p. 1903) is imprecise.…”

“…In our paper ‘Assessing the causes of diversification slowdowns: temperature‐dependent and diversity‐dependent models receive equivalent support’ (Condamine et al 2019), we fitted several diversification models to 218 phylogenies of tetrapod families to confirm that diversification slowdowns are common, and to show that this widespread pattern may in part be associated with cooling temperatures during the Cenozoic. Gamisch (2020) does not question these main results, but comments on the interpretation of linear diversification dependencies fitted in the R‐package RPANDA (Morlon et al 2016). Here we take this opportunity to provide further clarifications and recommendations for users.…”

“…Similarly, in the case of an exponential time dependency, if the algorithm returns a negative estimate for , the time dependency should be interpreted as . In the case of the linear time dependency that Gamisch (2020) focuses on, the time dependency should similarly be interpreted as . While in the case of an exponential dependency the interpretation of the sign of is not affected by the transformation (a positive reflects a slowdown in speciation), in the case of a linear dependency it can be affected.…”

Ambiguous support for extra-tropical accelerated diversification of rosids

“…A final issue relates to the recent criticism of linear diversity dependencies in RPANDA 21 , which is therefore relevant not only to the study of Sun et al 1 but also to similar ones using this R package. In short, linear diversification dependencies can lead to negative λ and/or µ rates that would be positivised in the current implementation of RPANDA 21,22 . In such cases, a rate curve is fitted that is first decreasing backwards in time and then increasing again (e.g.…”

“…Supplementary Fig. 1), leading to biologically questionable scenarios 21,22 . When performing initial test runs, I identified that this linear diversitydependency issue in RPANDA 21 also applies to some of the models tested by the authors (e.g.…”