The authors have added to the existing literature on the effects of predation risk on prey cognition, by assessing how chronic predation risk impacts both short- and long-term memory of a wild population of free-living white-footed mice using a modified Morris water maze. Overall the manuscript is well written, and the goals of the experiment are generally clear. Although the experiment is mostly well designed and the question is compelling, I have some concerns with this manuscript regarding the experimental design, statistical analyses, and points made in the discussion. Without addressing these concerns, I cannot recommend that this manuscript be accepted for publication. Below, I address major concerns first, and then move on to minor concerns.

Major:

Experimental design:
-The authors effectively only test two 150m x 150m grids per predation treatment (n=2). Can the authors definitively extrapolate this to the whole Bioreserve? There are several studies highlighting the importance of habitat complexity on perceived predation risk- has this been considered at all? Could the selected grids differ in the complexity in terms of leaf litter/shrubbery/fallen trees/tree density? Might this not have an impact on the effects of the control vs. predation treatment?

-The authors used acoustic playback as a proxy for predation risk. How confident are the authors that habituation to a riskless sound did not occur over the 24 day period? The authors could at least mention this in the methods. This point is also pertinent in the discussion, e.g., L319-320: Perception of risk can also be shaped by multimodal cues, potentially auditory cues alone may not have been perceived as particularly high risk?

Statistical analysis & Results:
-I am generally unclear why the results have been presented the way the have, structured by main effects and not by response variable (and highlighting possible interactions within each).
-The authors appear to present results of what I interpret to be posthoc analyses of interactions, without presenting whether or not that interaction was significant. Instead, the authors only describe main effects (e.g. effect of predation), and then go straight into the posthocs results. For example, the interaction of predation and trial was not reported, yet they report the effect of trial 1 vs. 5, and trial 5 vs. 6, within the predation and control treatments (L251-254; L257-261). Overall, I am not confident if these posthocs are appropriate to report without the full information.
-I am also not sure about the wording used by the authors when interpreting their results. E.g., L252: What does “Retained 74% of memory” mean? This is never explained. Does this translate to the difference distance travelled in Trial 1 vs. 6 was 74% of the difference in distance between trial 1 & 5?
-I was unable to verify the statistical results without any tables summarizing the model outputs, and without a functioning R code. The provided R code doesn’t run with provided .csv data, it appears the code requires an excel file with an additional data column compared to the file provided.

Discussion:
-Generally, the distinction between adaptive benefits of short-term and long-term memory is unclear. For example, L334-340 could be interpreted to also apply to long-term memory. How is this distinct? Potentially this could be clarified if the authors add a paragraph in the introduction about distinctions between short- and long-term memory, and why it is important to test for both. The way the manuscript currently reads, this distinction is not clear and consequently the interpretation of the results seems somewhat arbitrary. I am confident the authors can edit their manuscript to resolve this issue.

Minor:

Title: Although the authors make their point about “how much is too much” in their conclusion, their experimental design doesn’t address this question. Therefore this title doesn’t reflect the papers findings, and is perhaps not the most appropriate. The title should instead emphasize the findings learning, and on short- vs. long-term memory.

Introduction:
-L77: “maze” should be capitalized to be consistent with rest of manuscript. Likewise captions of Figure 2 & 3 are missing “Maze” after “Morris Water”.

Methods:
-L136: Could the authors explain why they chose to expose prey to acoustic cues in a 4 days on vs. 4 days off cycle? The authors cite a paper, but offer no explanation.
-L175: Between each trial, both the platform and the mouse entry point were moved. How exactly were these new locations chosen? Were there preset locations (e.g., quadrants) or was it arbitrary? If arbitrary, how were locations chosen, how distant were they from other locations? If preset, were locations randomly chosen?
-L174: The authors mention they guided mice which failed their task within the 60s. Does this mean the observers were present during the whole trial, or were they absent until the 60 s trial was over?
-L 194: Due to the journal formatting, this reads as an incomplete sentence. I would modify this to say “… by First Author and colleagues [64].”
-The statistical software used was never mentioned. I can extrapolate that R was used, but this should be explicit along with the version used.

Results:
-L255: Instead of “Supplementary Material 1”, Specify “Supplementary Material Figure S1”. Same applies to L262, where it should be “Supplementary Material Figure S2”

Discussion:
-L320-323: This sentence is a bit convoluted and could be worded a bit more clearly.
-L342-343: For clarity I would rephrase the second sentence as “These results also differ from most findings of laboratory based studies”.

The authors have made an effort to address reviewer comments, and the clarity of the manuscript is much improved. I appreciate the added definitions of cognition, short- and long-term memory, as well as the additions of Tables 1 & 2 and Figs. 2&3. I do still have some major concerns regarding the analysis and attribution of their results to experimental manipulation, and so I cannot recommend this manuscript for publication without further revision.

Major concerns:
For the short-term memory results: It is important to note that your “treatment” in your model actually refers to grid site. In other words, on day 0, “treatment” should have no effect in order for these “pre-experiment” trials to function as an effective control. In Figure 2A, there seems to be a difference in learning and short-term memory (distance travelled) between control and predator sites on day 0. The authors never test for the effect of treatment on day 0 results, and therefore it is unclear if the effects found on day 24 (L349-361) are attributable to the experimental manipulation, or some innate difference in cognitive abilities between grid sites. The authors should address this in their analysis, results, and discussion. If there does happen to be an effect of “treatment” on day 0, the authors should reconsider their analyses to take this into account.

Similarly on L489-495 (and 708 in discussion), the explanation is unclear. Are these results referring to day 24 specifically, or overall? This should be explicitly stated. Like mentioned above, the authors should test for an effect of treatment on day 0, in order to be confident in these results being attributable to predator treatment. This would affect the interpretation, e.g. differences in exploratory behaviour between sites regardless of treatment, vs. difference in exploratory behaviour as a result of treatment. Additionally, the authors mention that “experiment days was not significant” (L489), but should add some interpretation for what this means and why it is important.

Minor comments:

Title: The new title is still unclear: “is predation risk enough?”… enough for what exactly? However, the rest of the title is clearer than before in reflecting the contents of the manuscript. I leave further changes up to the discretion of the authors and editor, but I am still not sure the current catch phrase is adding anything to the title.

L20, L821: The authors should consider updating the motto “how much is too much” mentioned in the text to better reflect new title? This should depend on further changes made to the title.

L191: The phrasing here is awkward. Could be simplified here, perhaps: “the main reason we opted to use no-sound controls rather than sound controls….”

L233-234: Another minor wording issue, do you mean “same entry point and order of platform position”?

L339: To help with manuscript clarity, this subheading could specify short-term memory rather than just “memory”.

L830: Should be “healthy”, not “health”.

The authors have generally made improvements to the manuscript. Nevertheless, I do not feel that the statistical analysis on the day 0 data is adequate to establish similar baselines in cognitive abilities between the control and predator treatment sites:
- The authors have not included any methodology related to this statistical testing of baseline behaviours on day 0, and this should be detailed in the “Statistical Analysis” section in their methods.
-The authors have attempted to present results from such a test on L327-329, and the test appears to assess day 0 differences between control and predator sites in Trial 1 and overall performance (distance travelled). However, the authors specify on L214-216 that they test for learning by comparing distance travelled in trial 1 to distance in trial 6, and short-term memory by comparing distance travelled in trial 5 to distance in trial 6. Why have the authors not done this exact test for day 0 to test for the baseline effect of site (coded as predator treatment) on these two metrics of cognition? The results from this analysis would fit nicely in the learning and short-term memory section of the results (L337-373). Based on Fig. 1, it appears that individuals tested in the control treatment sites learned better than those in the predator treatment sites, while short-term memory doesn’t appear to differ. Again, this analysis is crucial for the interpretation of the results from day 24 and 36.

Similarly, this applies to success rate. There is no apparent test of baseline differences in success rate. One should be done for day 0 similar to what is done on L396-400 on day 24. This would further bolster the argument about exploratory behaviour (which may or may not be plastic based on predation risk).

Other minor comments:

L45: Extra space between of and new.

L212: Don’t need comma after “both”

L300: For clarity, maybe: “relative initial individual performance”

L301: For clarity “treatment (not day 0) was set…”. Also it would be useful to know how many individuals, of the total number observed, were first captured and tested during the treatment phase. This is mentioned in Table 1, so it should be referenced here.

L808-813: The authors should clarify whether this test done on all mice, or only on the mice that were tested 3 times.

L992: Did you mean “cognitive individual differences”?