Van der Hedjen et al.’s manuscript aimed at reviewing the current contribution of remote sensing, and its promise, to liana ecology. The paper is well-written and is, in my opinion, interesting beyond the remote sensing aspects (the authors did a great work in reviewing many aspects recently developped in the field of liana ecology, adding values to the previous reviews done on liana ecology). It was a pleasure to review this paper that, in my opinion, deserves a publication in Journal of Ecology. I do have only a limited number of comments, mostly minors, listed below. Congratulations to the authors for this work.

(Not really) major comments

1) The authors discuss the differences in traits between lianas and trees, and their implications to separate lianas from trees spectrally. However, lianas originated from diverse clades and display a large interspecific variability in their traits (e.g. Meunier et al. 2020), which is expected to translate into a large spectral diversity too, overlaping with that of trees. Thus, generally assuming that liana have a distinct spectral signature than that of trees may be tricky. I agree that the signal may be significantly different on average between trees and lianas, as shown in Fig. 3, but the large overlap (not illustrated here) does not allow an accurate pixel-based classification in my opinion. However, the probability to get distinct spectral signatures between lianas and trees dramatically increases at the individual crown level, simply because we are theroretically contrasting two distinct signals and not two mixture of signals. Thus, in my opinion, a promising avenue to map lianas is to classify pixels based on their local context rather than on a hypothetical generic spectral signature of lianas. To do this, deep learning approaches on very high resolution images (from drone, airborne or satellite) may in a next future really help. Another major advantage of such approach is that it has the potential to be much less affected by the influence of environmental gradients and forest types, because it relies on local relative responses rather than absolute generic response. This maybe an aspect to discuss in that paper.

2) In section 2.1, it is stated that remote sensing can help to follow up the work by Phillips et al. 2002, by checking whether the rate of increase in liana stem number has changed from the pre-2000 period (studied by Phillips et al.) to today. I am quite skeptical that the link between such ground and remote sensing observations can be accurate enough to compare these rates of increase and no RS data is accurate enough to do that before 2000.

3) During my reading I found that the different sections entitled « Current contribution of remote sensing » were not concrete enough on what potential RS data can be useful for different purpose. It, however, comes later in section 3 but I would have liked to see more concrete RS elements before. For instance, it is stated lines 544-545 : « The coarse-scale resolution of current satellite datasets can be problematic as one single pixel may be occupied by multiple tree crowns. » This is not true for all satelitte sensors. We, e.g., currently have access to plenty of very high resolution satelitte images through private agencies.

Minor comments

Lines 23-24 : Not yet demonstrated in the paleotropics...

Line 28 : canopy → crowns ?

Lines 64-66 : Are those numbers really so common when similar minimum dbh between trees and lianas are compared ?

Lines 66-67 : 40 % over 1 ha? Locally, it can easily be 100 %…

Line 90 : « can be far longer » → « can be very long » ?!

Line 100 : which suggest or suggesting ?

Line 116 : I would put « strongly compete », aggressively is a bit harsh…

Line 155 : often plot-based → often field plot-based ?!

Lines 157-159 : See also Cox et al, 2019 Austral Ecology.

Line 171 : «which are not feasible with» → « which is very challenging with »

Line 182 : reformulate ?! It seems that some words are missing here.

Line 202 : Why relative to trees ?

Line 202 : a year → per year ?

Lines 203-205 : Maybe worth mentioning that these studies were done in Africa and that we, up to know, lack data from Asia ?!

Lines 246-247 : But also for people studying trees because liana act as a kind of signal pollution… Typically in spectroscopical works that aim at identifying tree species. Ok got it later line 468 !

Line 253 : I would add : « the view from above must allow us to discriminate the lianas from the trees in the scene at an appropriate spatial resolution »

Line 261-262 : In Marvin et al. 2016, only high infestations were accurately classified.

Line 278 : Combining hyperspectral data with what ? Unclear.

Line 286 : « unmanned aerial vehicles » is a more common naming.

Lines 292-295 : And it may be very useful to calibrate airborne or satelitte sensors, acting as a bridge.

Line 329 : and probably disturbance history.

Lines 332-340 : Apart stem lenght, stem wood density is also ignored while highly variable between liana species (but that is something that TLS won’t solve).

Lines 340-350 : This may explain why Cox et al. Austral Ecology did not find any link between diameter-based estimates and leaf coverage in the canopy.

Line 354 : aboveground volume, not biomass. Need for wood density then to convert into Biomass and this has nothing to do with TLS.

Lines 370-372 : I would not say that TLS can assess the complete above-ground liana structure but rather the ground to canopy one because, in the canopy, occlusion is far too important to accurately detect lianas. So at the end, we still miss the full volume and the allocation patterns (or we would need to combine TLS with top of canopy measurements to assess, e.g. the liana leaf area).

Lines 380-382 : But given that no increase is observed in Africa, the global increase in CO2 in the atmosphere unlikely explains the liana increase in the neotropics.

Line 383 : positive response

Lines 401-403 : So acquiring spectral data during this time of the day would also optimize the differences in liana versus tree specific response ?!

Line 407 : is it true for CO2 ?

Line 421 : What about the study of liana leaf phenology, notably in response to local climate variation ?

Lines 448-451 : The sentence sounds weird, to be reformulated.

Lines 462-465 : Yes agree. A major challenge would be to build a reference dataset where liana leave pixels can be assigned a liana species. This would be possible only on trees that are infested by a single identified liana or would need drone acquisition with a 1 cm resolution (at the very minimum) to distinguish different liana species, which would be quite risky for the UAV system…

Line 458 : Field-based estimate of liana infestation is clearly less accurately estimated than with drone data, as said above. The same for biomass etc. Why keeping suggesting that ? I would personally suggest to acquire field data on : liana abundance, diameter (as we can’t do much better with classical ground inventories), liana identification and liana wood and leaf traits (I know that you mentionned the traits in point 1 but this is also field measurements, no?).

Lines 499-500 : Forest Observation System only aims at providing aggregated measurements at the plot level, such as BA, AGB, Ntree, etc. , not raw data and as for now nothing in relation to lianas.

Line 510 : You may look at Réjou-Méchain et al. 2014 Biogeosciences

Lines 511-512 : Agree. This can be tested e.g. with forestgeoplot that have all lianas measured above 1 or 2 cm over large plots.

Lines 514-516 : It seems a bit unrealistic to measure all liana species… Let first focus on the most abundant ones ?! and then e.g. test whether the spectral signatures are phylogenetically conserved.

Lines 543-545 : But at 10-m resolution, Sentinel is not that bad for tropical forests ! We can also potentially rely on very-high resolution images, such as worldview3, which is quite spectrally rich.

Lines 630-631 : Seems a bit risky to realy on citizen science for such purpose. Identifying lianas in canopies needs experts...

Fig. 3 : The illustration on the right is seemingly wrong. The lianas having a 10 cm diameter stem should be heavier when they are longer (which is the reverse in the figure). If there are no error, we would need an explanation...

Several typos in the reference list.