Evolution of olfactory sensitivity, preferences and behavioral responses in Mexican cavefish: fish personality matters

Reviewer #1 (Public Review):

Summary:
The authors posed a research question about how an animal integrates sensory information to optimize its behavioral outputs and how this process evolved. Their data (behavioral output analysis with detailed categories in response to the different odors in different concentrations by comparing surface and cave populations and their hybrid) partially answer this tough question. They built a new low-disturbance system to answer the question. They also found that the personality of individual fish is a good predictor of behavioral outputs against odor response. They concluded that cavefish evolved to specialize their response to alanine and histidine while surface fish are more general responders, which was supported by their data.

Strengths:
With their new system, the authors could generate clearer results without mechanical disturbances. The authors characterize multiple measurements to score the odor response behaviors, and also brought a new personality analysis. Their conclusion that cavefish evolved as a specialist to sense alanine and histidine among 6 tested amino acids was well supported by their data.

Weaknesses:
The authors posed a big research question: How do animals evolve the processes of sensory integration to optimize their behavioral outputs? I personally feel that, to answer the questions about how sensory integration generates proper (evolved) behavior, the authors at least need to show the ecological relevance of their response. For the alanine/histidine preference in cavefish, they need data for the alanine and other amino acid concentrations in the local cave water and compare them with those of surface water.

Also, as for "personality matters", I read that personality explains a large variation in surface fish. Also, thigmotaxis or wall-following cavefish individuals are exceeded to respond well to odorants compared with circling and random swimming cavefish individuals. However, I failed to understand the authors' point about how much percentages of the odorant-response variations are explained (PVE) by personality. Association (= correlation) was good to show as the authors presented, but showing proper PVE or the effect size of personality to predict the behavioral outputs is important to conclude "personality is matter"; otherwise, the conclusion is not so supported.

From the above, I recommend the authors reconsider the title also their research questions well. At this moment, I feel that the authors' conclusions and their research questions are a little too exaggerated, with less supportive evidence.

Also, for the statistical method, Fisher's exact test is not appropriate for the compositional data (such as Figure 2B). The authors may quickly check it at https://en.wikipedia.org/wiki/Compositional_data or https://www.annualreviews.org/doi/pdf/10.1146/annurev-statistics-042720-124436.

The authors may want to use centered log transformation or other appropriate transformations (R-package could be: https://doi.org/10.1016/j.cageo.2006.11.017). According to changing the statistical tests, the authors' conclusion may not be supported.

Reviewer #2 (Public Review):

In their submitted manuscript, Blin et al. describe differences in the olfactory-driven behaviors of river-dwelling surface forms and cave-dwelling blind forms of the Mexican tetra, Astyanax mexicanus. They provide a dataset of unprecedented detail, that compares not only the behaviors of the two morphs but also that of a significant number of F2 hybrids, therefore also demonstrating that many of the differences observed between the two populations have a clear (and probably relatively simple) genetic underpinning.

To complete the monumental task of behaviorally testing 425 six-week-old Astyanax larvae, the authors created a setup that allows for the simultaneous behavioral monitoring of multiple larvae and the infusion of different odorants without introducing physical perturbations into the system, thus biasing the responses of cavefish that are particularly fine-tuned for this sensory modality. During the optimization of their protocol, the authors also found that for cave-dwelling forms one hour of habituation was insufficient and a full 24 hours were necessary to allow them to revert to their natural behavior. It is also noteworthy that this extremely large dataset can help us see that population averages of different morphs can mask quite significant variations in individual behaviors.

Testing with different amino-acids (applied as relevant food-related odorant cues) shows that cavefish are alanine- and histidine-specialists, while surface fish elicit the strongest behavioral responses to cysteine. It is interesting that the two forms also react differently after odor detection: while cave-dwelling fish decrease their locomotory activity, surface fish increase it. These differences are probably related to different foraging strategies used by the two populations, although, as the observations were made in the dark, it would be also interesting to see if surface fish elicit the same changes in light as well.

Further work will be needed to pinpoint the exact nature of the genetic changes that underlie the differences between the two forms. Such experimental work will also reveal how natural selection acted on existing behavioral variations already present in the SF population.

It will be equally interesting, however, to understand what lies behind the large individual variation of behaviors observed both in the case surface and cave populations. Are these differences purely genetic, or perhaps environmental cues also contribute to their development? Does stochasticity provided by the developmental process has also a role in this? Answering these questions will reveal if the evolvability of Astyanax behavior was an important factor in the repeated successful colonization of underground caves.

Reviewer #3 (Public Review):

Summary:
The paper explores chemosensory behaviour in surface and cave morphs and F2 hybrids in the Mexican cavefish Astyanax mexicanus. The authors develop a new behavioural assay for the long-term imaging of individual fish in a parallel high-throughput setup. The authors first demonstrate that the different morphs show different basal exploratory swimming patterns and that these patterns are stable for individual fish. Next, the authors test the attraction of fish to various concentrations of alanine and other amino acids. They find that the cave morph is a lot more sensitive to chemicals and shows directional chemotaxis along a diffusion gradient of amino acids. For surface fish, although they can detect the chemicals, they do not show marked chemotaxis behaviour and have an overall lower sensitivity. These differences have been reported previously but the authors report longer-term observations on many individual fish of both morphs and their F2 hybrids. The data also indicate that the observed behavior is a quantitative genetic trait. The approach presented will allow the mapping of genes' contribution to these traits. The work will be of general interest to behavioural neuroscientists and those interested in olfactory behaviours and the individual variability in behavioural patterns.

Strengths:
A particular strength of this paper is the development of a new and improved setup for the behavioural imaging of individual fish for extended periods and under chemosensory stimulation. The authors show that cavefish need up to 24 h of habituation to display a behavioural pattern that is consistent and unlikely to be due to the stressed state of the animals. The setup also uses relatively large tanks that allow the build-up of chemical gradients that are apparently present for at least 30 min.

The paper is well written, and the presentation of the data and the analyses are clear and to a high standard.

Weaknesses:
One point that would benefit from some clarification or additional experiments is the diffusion of chemicals within the behavioural chamber. The behavioural data suggest that the chemical gradient is stable for up to 30 min, which is quite surprising. It would be great if the authors could quantify e.g. by the use of a dye the diffusion and stability of chemical gradients.

The paper starts with a statement that reflects a simplified input-output (sensory-motor) view of the organisation of nervous systems. "Their brains perceive the external world via their sensory systems, compute information and generate appropriate behavioral outputs." The authors' data also clearly show that this is a biased perspective. There is a lot of spontaneous organised activity even in fish that are not exposed to sensory stimulation. This sentence should be reworded, e.g. "The nervous system generates autonomous activity that is modified by sensory systems to adapt the behavioural pattern to the external world." or something along these lines.