If you're wondering whether the disagreement between different methods is severe enough to warrant an entire paper, please take a look at this figure. (I spent a whole weekend making this figure. I re-listened to the first season of Blowback plus the first 4 episodes of ALAB while making this figure. I am begging you to look at this figure.) I hope this illustration also provides more food for thought about the value of a comparative approach.

7/8

The bars in this figure show the number of fossils we have from different terrestrial arthropod groups throughout the Silurian–Permian (i.e., before dinosaurs). The diamonds at the top show ages for winged insects estimated from molecular phylogenetic studies. You can see that they yield ages from times when we have plenty of fossilized terrestrial arthropods, but no winged insects. (This is foreshadowing, btw....)

5/8

To quantify/simplify this further, you can calculate standardized mean differences for the covariate values of only the matched sites among your two categories (N vs S latitude, in this case). When you have sufficient data to properly match your sites, the standardized mean difference for covariate values of matched sites from each category should be below 0.1. This directly tells you if you have enough data. (Figure 7 of my paper.)

8/9

Whether the best matches between your two groups are really all that close will depend on the distribution of your covariates. Here you can see that, for the paleobotanical study I re-analyzed, the covariates are very different between the two hemispheres. (Figure 6 of my paper.)

7/9

Returning to fossil plants specifically, here's Figure 2 from my new paper, showing fossil sites where leaves have been examined for insect herbivory. Do we have enough data to examine change over time while accounting for latitudinal differences, and vice-versa? I'd say no, but a recent paper did exactly that, so clearly some of my colleagues (authors & reviewers) think the answer is yes.

4/9