General comment:
The study presents very interesting and novel data with respect to the quantification of insect migration. It covers a significant time period of several years at several sites. This is the first study estimating the exchange of larger insects between Africa and Europe/Asia. The results allow a first approach comparing the extent of insect migration across a continental range based on the same sampling method. Nevertheless, the study has some shortcomings which need to be considered, revised or explained (see specific comments below).
The result chapter is a complex mixture of results and interpretations which should be clearly separated. I recommend to keep the result section free of any interpretation (e.g. which kind of species might be involved). I have a major concern with the relatively high proportion of wingbeat frequencies below 10Hz. To my experience with this kind of data, this is because the authors did not do any filtering based on the accuracy index provided by the wingbeat algorithm. At least they did not mention any. A reasonable approach is to set the filter at various levels to see whether or not the overall distribution is affected by the filter level (e.g. 0.2, 0.5, 0.8, 0.9).
Specific comments:
Line 119 – 127: I think these statements do not fit to this paragraph headed “Hard- and software”.
Line 113: I missed any statement about the selection criterias available as class probabilities and wingbeat accuracy. As far as I know each the software provide a probability for each class and also a probability of accuracy for the wingbeat frequencies. To my experience wingbeat frequencies for insects of less the 0.5 are not very reliable.
Line 134: It should also be mentioned that detection only starts at about 30m agl and that max range is limited to about 500m for large insects.
Line 140: shape is most-likely the most important feature for separating insects from birds and bats and therefore should be referenced => perhaps this should be referenced https://doi.org/10.1080/01431160500165955 because shape is not easily understood.
Line 150-152: Unclear to me. To my knowledge a main feature for differentiating between insects and vertebrate used by the classifier are the shape factors – see literature above.
Line 152-155: Generally, insect abundance is high between ground level and 100m. Did the authors check whether or not directions of low flying insects (<100m) where more scattered or undirected, respectively? If directions would correspond to those higher up, I would recommend to include these targets.As mentioned above the lowest 30m are anyhow not covered by the radar.
Line 167-169: I would appreciate a more distinct definition of these periods. What is “weak”? Directionality can be scored through circular statistic measures.
Line 171-176: quite unclear – MTR is not radar specific! It was introduced first for moonwatching counts. MTR is a measure used to compare any kind of directed movements quantitatively, by standardizing the passage of birds or insects to a transectline of 1km and a time period of 1h. it’s like any other quantitative measure e.g. nests/ha. Please clarify.
Line 178-184: As I understand this is a characteristic of the hardware and thus belonging to the Hard- and software paragraph.
Line 185ff: How should the reader know what is a “individual MTR factor”? I strongly recommend to rewrite the MTR explanation. Although I am familiar with the system, I do not understand where the 1.47 factor comes from. Start the the individual MTR-factor which considers that the observed insect specific range changes with distance (height) and standardizes a single count to 1km.
Line 196-197: I assume height was also restricted to 100-400m agl. Please mention.
Line 199-200: I do not understand this sentence.
Line 2016: 4 models for each radar? From above it is unclear whether or not MTR’s from the radars were combined?
Line 236: the periods seemed to be chosen quite arbitrarily – e.g. in figure 2 directions in the first 10days of June are still well oriented towards North. I am also wondering why diurnal movements where not considered for defining the movement periods. What I have seen from unpublished data in southern France insects are already strongly northward directed in mid February at fairly good numbers. Looking at figure 4 there are quite high intensities recorded at daytime in early march (e.g. “E.U.G 2017” and “E.L.G 2017”). I think it is of great interest to learn from these data when insects movements start to come in from Africa to Eurasia.
Line 240: As I understood Figure 2 presents only nocturnal data? In addition, directions maybe somewhat less concentrated during summer and winter, but more obviously, means are scattered all across the circle.
Line 246 Figure 2: Interpretation of figures should not go into the captions (at least as I understand it).
Line 253: why “daily/nightly” – I understood figure 2 presents only nightly data.
Line 258: this west-east tendency is not very obvious to me from the figure.
Line 261: the whole chapter should rather be in the discussion than in the results.
Line 264-266: I am not aware of any study showing so low wingbeat frequencies for these kind of butterflies (Nymphalidae). There is one number around for painted-ladies which is much higher (25Hz). There are very few insect species with frequencies below 10Hz. See e.g. Yu, W., Zhang, H., Xu, R., Sun, Y., Wu, K., 2022. Characterization of Wingbeat Frequency of Different Taxa of Migratory Insects in Northeast Asia. Insects 13. https://doi.org/10.3390/insects13060520. The authors should make sure that the only included wingbeat frequencies with a relatively high probability of accuracy. In addition, these kind of interpretations belong to the discussion.
Line 273: the citations given here, rise the impression that the cited papers provide information on the wingbeat frequencies, which I think is not the case.
Line 315-317: The numbers given do not support the statement that autumn migration rate is 5% higher. Please clarify the day – night comparison.
Line 322-332: I would transfer these calculations to the discussion.
Line 355ff: I am sorry to say, but I did not understand what is shown in figure 5. How is insect directionality and downwind direction defined? The captions mention wind direction plots, which I could not identify. Which samples are tested against each other?
Line 403: I could not find “Specified directions on right-hand plots . . . “ and I was fully confused about “North-blowing into the corridor from Northeast Africa, towards Eastern Europe/Western Asia”.
Line 412: it is not clear to me why the authors only compared the diel rates and not also the nocturnal ones.
Line 413-414: Please confirm that the same height range has been extracted.
Line 460: Could this statistically be checked?
Line 486: reference 33 – at least for me, the data presented their doesn't really allow for such a clear statement (sample size = 3 and all tracks stopped before crossing the Alps). And by the way – it has been published in 2022 (no more in press).