It was my pleasure to read and review the manuscript "Functional response metrics explain and predict high but differing ecological impacts of juvenile and adult lionfish" for publication in Royal Society Open Science.

This work studies the functional responses of an invasive lionfish on four different prey species. The authors have performed well-designed feeding experiments, and have found that the functional responses, represented as the attack rates and handling times, differ across studied predator-prey pairs. This finding is very interesting for a general ecological audience. Potential stage-dependent feeding differences, and therefore an ontogenetic shift in prey preference and predator impact, are still somewhat understudied topics. From a more applied perspective, this article offers practical inputs for nature conservationists that deal with invasive species.

I commend the authors for a well-written, concise yet solid study. It presents straightforward results that are intruiging and potentially inspiring for other researchers.

I have one more major comment concerning the interpretation and usability of the functional response ratio (FRR). Other comments are rather minor. Please find all numbered below.

1. I have not heard of the FRR concept before, and I feel that it deserves a better description in the text. Specifically, it is rather unclear to me if and how FRR can be understood and compared across different predator-prey pairs, and also across different functional response studies.

The attack rate "a" and the handling time "h" have different units, and therefore the FRR defined as "a/h" is not a unitless measure. The attack rate is typically represented as an area (or volume) searched by predator in unit time. The handling time is typically in unit time that it takes to handle and digest a single prey individual. In this manuscript, the authors mention the volume of experimental tanks (litres) and the experimental duration (hours), but the final units of fitted attack rates and handling times need to be explicitly mentioned, probably in Table 1 and/or in the results or methods sections.

I am not familiar with the "frair" package in R. Are the fitted attack rates and handling times somehow standardized or rescaled, so that units do not matter?

If FRR is not unitless, then it can only be compared across studies that use the same units. However, also within the current study it is not clear if the comparison of FRRs across different predator-prey pairs is valid. The authors mention in lines 296-298 that A. salina and P. varians are generally pelagic and mobile, whereas G. oceanicus and N. norvegicus are benthic with a relatively lower mobility. This means that a lionfish searches for A. salina and P. varians in a 3D volume of water, but it searches for G. oceanicus and N. norvegicus on a 2D bottom surface. This suggests that fitted attack rates are, respectively, in units of volume per unit time (e.g., litre per hour) and in units of area per unit time (e.g., square metre per hour). If this is the case, the comparison between such attack rates - and therefore between FRRs - is not possible.

Moreover, attack rates and handling times depend on other factors, such as prey and predator body sizes and temperature. It could be a good idea then to represent parameters not per individual, but per unit biomass. Also temperature would have to be the same. I have not read in detail the article by Cuthbert et al. (2019), but to compare the FRR values in the reviewed manuscript to Cuthbert et al., it needs to be ensured that such comparison is valid, and that other factors affecting the functional response parameters are accounted for.

A minor issue - FRR works only for a type II functional response. Type I has no handling time (h=0), and in type III the attack rate has a different meaning (if any at all) than in type II.

All in all, the concept and usage of FRR needs at least a much better description throughout the manuscript, especially early on in the introduction (lines 122-128) and in the methods (lines 219-227). Of course, first it needs to checked (and stated in the text as well) that using FRR is valid and meaningful.

1. (Figure 1) It is probably a smaller, semantic issue. The axis labels should include units as well (or be re-labelled), which are probably number of prey offered (x-axis) and number of prey consumed (y-axis). Or is the prey consumption in the y-axis already shown as a rate (e.g., number of eaten prey per hour)? The functional response is formally an instantaneous rate, and it is a little unclear if Fig. 1 shows this, or if it shows fitted model curves to experimental outcome (i.e., number of eaten prey after 3 hours).

2. (lines 97-102) Maybe break this one long sentence into two shorter ones?

3. (line 194) Is the replication number equal to 8 because there were 8 fish per stage used? It could be mentioned here for clarity.

4. (lines 328-329) It is a ittle unclear (or rather not explicitly stated) why this would be a recommendation that stems specifically from this study (see also the last sentence in the abstract). True, the reader gets a feeling that due to high impact of lionfish it would be better to eliminate them all early on, but an extra sentence or two that connect the results to this recommendation would be appreciated.