In this manuscript, the authors ask whether vertical movements of larval zebrafish in response to light are wavelength dependent, and they seek to identify brain regions that are activated by such light and mediate the behavior. They provide evidence that the vertical movement behavior is wavelength-dependent, with stronger responses to blue and green light than to red light. They use calcium imaging and lesion data to support their hypothesis that thalamus and habenula are activated by blue light, and that lesions of the habenula can impair the vertical movements.
Despite some significant concerns about the statistics and interpretation, this is a valuable study and is written clearly with well-organized figures.
Although the paper (in the abstract and throughout) refers to the light as “masking” diel vertical migration, all of the experiments were conducted between 11AM and 6PM, during the light phase. “Masking” of the normal dark-phase behavior can presumably only happen at night. Therefore, it would make sense to eliminate or reduce the reliance on the idea of “masking”, which isn’t directly shown for all aspects of the behaviors studied here.
What is the statistical power for the nonparametric tests used? Underpowered statistical tests can lead to false positive and false negative findings, and tend to inflate the size of any detected effects.
For the comparisons where no difference is reported (i.e. vertical speed under high vs. mid intensity light), a two-tailed Mann-Whitney test was used. For some other tests, where a statistically significant difference is reported (e.g. vertical speed under high vs. low intensity light), a one-tailed test was used. What explains these differences?
For the Discussion section: what could explain why the high intensity light generated a less robust behavioral response than the mid intensity light?
I suggest including the spectra data for the LEDs, since data at website links can easily disappear or change.
Blue light led to much more vertical movement than UV light did at the same intensity. It would be informative to also have a control group that received no light, in order to determine if UV light had any significant effect on vertical movement behaviors.
On pgs. 5-6, the description of the analyses is not entirely clear. What does it mean that “temporal signals” were “averaged by weights of each pixel”? This sentence is also vague because it is not made clear what “real fluorescence signals might be different” from.
What regions or pixels or timepoints do the ICA signals represent? Do they include the initial “wavelength-independent phasic excitation” period?
Interestingly, it appears that there is a lateralization of the blue-light response in the habenula in Fig 5b. This would be good to explore further, especially since the authors go on to lesion the habenula unilaterally or bilaterally in Fig 7. These experiments seem to be somehow logically connected and could be discussed together to give some insight into their rationale.
The abstract claims that the habenula is “tuned to blue light”, but green light is not tested here. Therefore, it would be more accurate to claim that blue light elicits a greater response in the habenula specifically when compared to red light.
In both Fig. 5 and Fig. 6 it would be helpful to state clearly somewhere in the results as well as the methods that the imaging is done using blue light, so it is impossible to remove blue light from the imaging data conditions. Discussion of this caveat would also be useful.
‘d’, ‘e’, and ‘f’ are missing values for the y-axes.
The text (pgs. 6-7) refers to “neuropil” when it appears that it should be “habenula”.
Are the statistical tests mentioned here one-tailed or two-tailed?
The description for ‘e, j, o’ (“correlation between the movement of individual fish”) is not clear.
Given that no thalamo-habenular connection was studied or tested, the results at best “suggest”, rather than “indicate”, that a thalamo-habenula projection is involved in the effects observed. The idea that blue light influences vertical movements via release of neuromodulators by the habenula is, to say the least, speculative.
In the discussion, it is claimed that the thalamic nucleus “involved here is likely to be the nucleus rostrolateralis”. What is the basis for this claim? This nucleus does not seem to appear anywhere in the results.