The study was conducted to examine the accuracy of fish size estimates by a baited single camera, facing down with a mirrored surface at the base of the stand. Firstly, sizes of plastic objects with known size were estimated underwater by the calibrated MBUV, with the camera set at 22 different controlled heights from the mirrored surface, and error rates were estimated. distance between the mirror and camera the camera set at different high to calculate error rates. Secondly, case studies were conducted on three reefs, and the abundance and size of A. patachonicus were recorded. The study found that the size of carnivorous fish could be estimated accurately and precisely using MBUV, with a slight underestimation (-0.6% error). This is a great study as the limitation of the single- BRUVS (which is the inaccurate length estimates) were overcome by the slight modification of a simpler, cheaper, and smaller single- BRUVS using a mirror. This achievement can provide great opportunities to conduct BRUVS studies with less cost for researchers, or with the same cost but with more replicates.

The design of the camera, including the bait types, was well described in text as well as in a figure, and appropriate for the aim of the study (to test the accuracy of MBUV's size estimation).
“Empirical” calibration methods at different distance from the camera were also well designed and described. The variables for calibrations were defined and measured appropriately.

However, there are 2 major issues in the design of this study that need to be carefully considered and fixed for the acceptance of the manuscript.

Firstly, this study design was valid and reliable to examine the accuracy of size estimate of carnivorous fish, but not to examine differences in the abundance, size structures, between species or reefs. However the authors made the latter examinations. If the aim of the study was solely to test the accuracy of the length estimates of MBUV, the study design was appropriate (although replicates of the reefs were unnecessary). If the aim was to examine the differences in abundance and species compositions between reefs, or the differences in the accuracy of MBUV between reefs, then this extra aim(s) needs to be stated in the final paragraph of the introduction, and the study needs to be redesigned in the following 4 ways; (1) more replicates of the videos in each reefs are needed, (2) each video must be deployed further apart to avoid bait plume dispersal (in this study, it was only 25m separation. More than 500m separation is recommended (Cappo et al. 2004)), (3) longer deployment than 15min per video, and (4) the differences in the measured variables (i.e. size, abundance) between reefs must be statistically tested. For instance, cameras were deployed with the separation of >25m, which is too close from each other that there would be effects of bait plume from each camera. Fish species/individuals with wider habitat range, greater swimming performance, and/or greater bold personality, can easily swim from one video to another with such short distance between the replicates, and therefore the abundance can be overestimate. Therefore, the following statement that the authors made is not valid with this study design; “A. patachonicus was the most abundant fish in the surveyed reefs” (the end of page 97). In order to accurately compare the abundance of different fish species, these 4 conditions above (replicate, deployment distance, and statistics) need be satisfied.

The second major weakness is that the statistical analyses were poorly designed/ described. For instance, only the statistical results presented was the Wilcoxon test of abundance, but the results of general linear model were not presented even though it was mentioned in the materials and methods. Also, authors described the patterns of figures (sometimes even with the word “significantly”) only based on the visual analyses of figures without any statistical tests. i.e. “the naïve method overestimated significantly the length of 106 fish (of 132), >5% and 91%, with respect to the values obtained with the MBUV (Fig 5a).” “length ranges of fish were simlar in different reefs, except for reef B, in which a larger abundance of smaller sea basses was recorded (Fig 4)”. In the latter case, the size in reef B is probably not statistically significantly smaller than those on other reefs (assuming from the large error bars of the box plots overlapping). Please consider the following 4 points in the revised manuscript. 1) The fish abundance of MBUV and other single video were compared using Wilcoxon signed- rank test. Why did the author use the non- parametric Wilcoxon test? Because of the violation of parametric test assumptions? Because of the low number of samples? How were the assumptions tested? These information needs to be explained in the materials and methods. 2) Authors statistically tested the difference in the number of individuals, but not the length. Why? Comparisons of length estimates between types off measures/methods were well displayed in Fig 4 and 5, but no statistical analyses were carried out. 3) It is stated that “the models were fitted using the glm function of the package ‘stats’ included in the R software” to test the relationship of Gif 2. However, only R2 values of the relationships were stated in the results but no statistical results (i.e. F-statistic, P value?). 4) Relative errors was calculated by taking snapshots of plastic silhouettes of 3 different known sizes and taking their measurements. The average relative errors were calculated but the comparisons between the 3 sizes were not made. Authors should make the comparison as it gives the information of how the size of fish can affect the accuracy of size estimates.

Here are also some other specific comments that I would like the authors to revise;

1. This main finding was clearly stated in the abstract; “estimates were highly accurate (mean relative error = -0.6%) and precise (mean coefficient of variation = 3.3 %), even in the range of those obtained with stero-viddeo methods”. However this finding was not stated in the first paragraph of the result section, therefore it doesn’t stand out as the “major” finding. I recommend the authors to make the new first paragraph which states the main finding of the study.
2. In the legend of the figure 1 (b), the authors stated that 4 fish could be measured from the snapshot including the fish #4 whose whole body was not over the mirror. 3 other fish on the right edge of the mirror are also not fully reflected in the mirror, but authors did not count them as measurable fish. Why? Specifically, the fish right next to the fish #3 has clear image of the top half of the body (from the head to the pectoral fin) reflected in the mirror, just like the Fish #4. Why can it not be sized?
3. Why does the number of snapshots of a sphere at each height vary between 2 and 9? It would have been better to systematically take the same number of snapshots at each height. Please specify how the numbers were decided (systematically or randomly?).
4. The consideration to barrel distortions was well-designed. However the method of removing barrel distortion is unclear for readers (including myself) who are not familiar with photography and the software PTLens. Please provide more details.
5. Most of the results aligned with the aim of the study, but the study deviated from the scope of the study when authors made the comparison of length frequency and abundance of fish between methods and reefs. The study was not aimed for this, and more importantly it was not designed properly to test these (details explained above). In the same context, Fig 4 (comparisons of length frequency between reefs and methods) is out of the scope of the study. I recommend the authors to exclude these parts of the results unless they are able to improve the sampling designs and statistical analyses.
6. Units, rounding, and number of decimals were appropriately used for all the figures and tables except for Fig 4 which needs a couple of improvement; (1) X axis label of Fig 4 is missing, which should be labelled i.e. “Estimated fish length (mm)”, and (2) The type of error bar is not specified (i.e. Confidence interval? Standard error?).
7. Fig 5 presents relative the overall error of all the measurements of the plastic silhouettes (3 known sizes). As mentioned in my review on the materials and methods above, comparison of the errors between the different sizes should be made to show the differences in the accuracy depending on the size of the target objects. Therefore separate figures for each size of silhouettes would be better than the current figure.