The authors have gathered the baseline for a growing database containing laser scan data at a tree species level. With this database and the additional benchmarking of possible models for species identification based on laser scanning data its a big step for the effectiveness of remote sensing strategies to identify tree species and therefore forest ecology. I have no concerns that this manuscript will increase the possibilities of laser scan data for ecologist and foresters alike. Please, find some minor comments below to further improve and clarify the output of the manuscript.
Line 59:
I don`t understand why we need to get the species identification from the laser scan data to calculate carbon storage and growing stock volume – this is something I could gather directly from the laser scan, or? I might oversee something here, but this statement feels counterintuitive.
Line 65:
Tree crown asymmetry because of stand diversity and habitat should be mentioned here to match with the methods (Line 203) and later in the discussion. This could be a major limitation for species identification based on laser scanning data, if not different conditions are captured, e.g. outside Europe.
Line 159: Figure 1:
Please, change the legend to integers, also the scale should start from 1. What does it mean if a point has no cycle around? Less than 690 trees?
Line 160:
Please avoid the abbreviations in this section.
Line 181:
Please, rephrase as it currently sounds like your data set contains 61% of tree species present in temperate forests and so on.
Line 190:
I would suggest using individuals instead of trees behind the numbers.
Line 198: Figure 3:
Does it make sense to distinguish between single and multiscan TLS data, as the resolution is very different.
Line 229:
Why n = 100, when the minimum threshold is 50 to be part of the dataset?
Results/Conclusion:
While the results from the science competition is clear, a recommendation for the most user / computational power friendly algorithm would be helpful as it was highlighted in the introduction that the post processing of the laser scan is still a bottle neck for many users.

I thank the authors for addressing my minor concerns improving the clarity of their work. All my concerns have been carefully accessed. I just add two minor comments below (Line numbers refer to the document with track changes).

Line 41: I suggest changing significant to notable or substantial, as it does not refer to any statistical test.
Line 174: I suggest writing deep learning (DL) instead of just the abbreviation here.