This paper examines patterns of liana infestation at the plot, species, and individual tree level using data from 27 forest inventory plots along the southern frontier of the Amazon. The paper is well written and easy to read. The analyses are well designed and results are fairly clear. I like the paper overall and think that it can be solid contribution to our understanding of tropical lianas. That said, I do have some questions and concerns, and think that there are several places where the paper could be improved.

L33-36: Large trees of slow growing species have highest infestation – so is infestation just a matter of time?

L55: can be simplified to “For example, once they reach the canopy, lianas can create dense foliage that reduces the light available for the trees below”

L61: here you say that lianas can increase light availability but you just stated above that lianas decrease light availability. I think I understand what you are saying here but please make sure that the meaning of this apparent contradiction is clear to readers.

L90: You already explained that changing climate may contribute to liana increase – changing climate is due in part to “elevated atmospheric CO2 concentrations” so please make the distinction between direct vs indirect effects of CO2 clearer.

L118: I think that the structure of the sentence could be simplified to improve clarity.

L120: Inconsistent verb tense: “we evaluated” “weuse”

L134: H2 and H8 seem very similar. I assume that the distinction is scale – i.e., species vs individual level – but please make this clearer.

L143: “Mean annual precipitation, maximum 143 climatological water deficit and the mean monthly temperature ranges from 1,511 to 2,353 mm, -328 to -498 mm and 24.1 to 27.3°C, respectively” – you have not defined CWD yet so these numbers will be meaningless to most readers. These values are too “precise” especially using a large scale extrapolated climate data base such as worldclim - 1551-2353 mm? Better to say something like ~1500-2400 mm. What resolution of worldclim data? Worldclim does not provide CWD so you had to calculate this from the monthly data correct? What PE values were used?

L160: Were plots originally established with the motivation of investigating liana infestation?

L161: How were they randomized within habitats? What criteria were used for selecting plot locations?

L172-L178: Give formula for estimating CWD. Was CWD calculated for each year and then averaged across all years? When calculating CWD you need to account for last year’s rainfall so there is a built in autocorrelation correct? What evapotranspiration did you use?

L177: It would be better to say that you are calculating Maximum Climatological Water deficit (MCWD) than to say that you are looking at CWD and then define CWD as the most negative value of CWD.

L179: Distance to edge is a better quantification of edge effect than fragmentation. Fragmentation and edge effect are related but are not interchangeable. Please reword sentence so it doesn’t end in “was”.

L194: I am not clear what OM and BS are – please define and explain better.

L197: Were liana measured and counted?

L213: Were Lianas identified or just trees?

L219: Was wood density estimates available for all of you tree species? How do you deal with species without wood density estimates?

L221: was the percentage of broken trees determined in each census or just once?

L225: You say above that all plots were established between 2008-2016 and that they were all recensused every 2-3 years, so why is there not two censuses of data available for all 27 plots?

L229: so potential growth rate of the species is based on only 2-3 years data? Is there any way to extend this? Can you use forest plot data from other sites? Is there any reason to use just the uninfested trees? Do infested trees ever grow faster than uninfested trees? I think that you should only include species for which you have multiple uninfested trees but then you should include both uninfested and infested trees in your calculation of potential growth rate.

L250: what is liana infestation? Just the percent of trees infested or does it incorporate level of infestation as quantified by your crown occupancy index?

L303: is it right to say that a tree was infested if it only has 1% COI?

L345: It appears that DBH should be on a log scale

L356: Please provide families after the species names

L373: So wood density is associated with potential growth rate but not individual growth rate. This suggests that individual conditions are more important to individual growth than species-level traits (wood density was measured at species level, correct?).

L375: (>75% COI)

L435: font problem

L439: or that humans chose to live in places that have higher intrinsic rate of infestation. You do not have the data to distinguish between these hypotheses so you should avoid making such as bold claim.

L441: if there is no effect of precipitation and fragmentation on liana infestation, then what explains the increasing abundance of lianas through time? Tie these results back in with some of the original ideas discussed in the introduction.

L480: An alternative explanation for this and some of your other results is survivorship bias. If slender trees are more susceptible to breakage or mortality from infestation, then you would tend to only observe the stout trees that survive infestation, giving you the wrong impression that stout trees are more infested than slender trees. Please discuss and account for survivorship bias.

L491: You have not discussed before the idea of pioneer species. Are your fast growing species pioneers? Could you classify species as pioneer or not and test for a difference in infestation? I am surprised to see no mention of bark characteristics. Haven’t previous studies found that infestation rates are lowest in species with smooth or shedding bark? Are differences in growth rates associated with differences in bark characteristics/texture?

L496: Glad to see survivorship bias brought up here but it should be brought up in relation to other observations as well. The lower infestation of fast growing species may have nothing to do with their ability to survive infestation – maybe fast growing trees die quicker so there is just less time for them to be infested. Indeed all of your results point to exposure time being a potentially important driver of infestation rates.

L482-499: this paragraph is not as well written as the rest of the paper (overall this is a very well written paper) and includes several awkward sentences. I think that this paragraph should be reworked.

L510: How would this affect “the capacity of tropical forests to sequester CO2”? Can we make this statement yet without knowing anything about the impact of liana infestation on mortality?

L540: Do you know that liana affected tree growth or could it be that slower growing individuals have a higher risk of infestation. Do your results allow you to infer causality?

L527: I was very surprised and disappointed that you did not discuss how your results may help explain the observed increases in liana abundances or what ongoing global change will mean for liana abundances. You need to connect the discussion and introduction sections better.

Table S1. Please provide the census years for each plot and indicate which plots were included in the 13 recensused plots used for growth rate estimates. I see that all of your FSI lots have the same DE and that their DE is significantly lower than in the other habitat types– so how do you account for this in your models? You state several times that seasonally flooded forests have higher infestation rates – is that true or is it just that forest close the edge have high infestation rates Can you distinguish between these ideas. Why not discuss DE in your discussion lines 408-427.