Peers over the past: prior predation-risk experience does not dictate antipredator responses of individuals in groups

The central thesis of the paper is a comparison between solitary and group behaviors, as stated in the introduction: 'In contrast to solitary settings, we predicted that the behavioural carryover would not manifest in a group setting due to differences in trade-offs.' However, the authors did not run a solitary treatment in parallel with the group treatments in this study. Instead, they compare their current results to a separate, previously published study ('Rawat et al., 2024'). This is a significant methodological issue because environmental conditions, colony generations, and baseline behaviors can vary between experimental rounds. To robustly claim that 'the group setting eliminated these behavioural differences,' the authors must test solitary and grouped individuals from the same cohort under identical conditions. At a minimum, they must explicitly discuss this limitation and temper their comparative assertions.

The authors repeatedly frame their non-significant findings as proof of absence. For example, they write: 'Our findings revealed that the past predation-risk experience did not influence the behaviour of individuals in a group setting.' However, a non-significant p-value does not prove the null hypothesis. Given the high individual variation typical of behavioral assays (as seen in the wide confidence intervals in Figure 1), the lack of a detectable effect could simply be due to insufficient statistical power. The authors should provide effect size estimates with confidence intervals and discuss the power of their design to detect the expected effect sizes based on their solitary study.

The description of the immediate threat cue is vague. In the Materials and Methods, the authors state: 'The pupae were then recorded in two behavioural assay environments: under immediate threat (using water with predation cues) or no immediate threat environments (control tap water).' It is unclear how this 'water with predation cues' was generated. Was it taken directly from the larval rearing trays where predators were housed in containers, or was it prepared separately? Did it contain alarm cues from crushed conspecifics, or only predator odor? The exact protocol for preparing these chemical cues must be detailed to ensure reproducibility and to rule out confounding chemical differences.

The experiments were conducted in 'two blocks of experiments in February and July 2022.' In India, February and July represent vastly different seasons (cool/dry vs. monsoon/wet), which can affect ambient laboratory humidity, pressure, and the physiological state of the mosquitoes. Table 1 shows a significant block effect for the variation in dive depth (p = 0.025). The authors should discuss how these seasonal differences might have influenced their results and whether the block effect interacted with the treatment effects.

The authors state: 'In contrast to the solitary setting, where experienced pupae exhibited more frequent and shallower dives with greater variability in depth compared to naive pupae, the group setting eliminated these behavioural differences.' While this contrast is the central finding, the manuscript would benefit from a more nuanced discussion of why the 'antipredator shield' of the group specifically suppresses the expression of prior experience. Does the group environment impose a physical constraint on the individual's ability to express their learned behavior, or is the decision to ignore past experience an active, adaptive strategy? The authors touch on this, but a more explicit discussion of the trade-offs—perhaps referencing the energetic costs mentioned in the introduction—would improve the logical flow.

Regarding the interpretation of the results, the authors argue: 'The substantial effect of the immediate threat environment on naive pupae underscores the significance of an innate pupal response against a potential threat, which is amplified in a group.' This is a strong claim. To justify this, the authors should clarify if the magnitude of the response to the threat in the group setting is statistically greater than the response observed in the solitary setting (from their previous work). If the data only show that an effect exists in both, calling it 'amplified' in a group requires a direct statistical comparison of the effect sizes between the two contexts, which is currently missing.

To ensure the larval exposure phase can be replicated, the authors must clarify the physical properties of the predator containers. The manuscript states that larvae were reared 'with two dragonfly nymphs enclosed in 50 ml plastic containers' and that they 'experienced the predation risk through visual and chemical cues'. If the plastic containers were sealed, chemical cues (which are primary mediators of aquatic risk assessment) could not have diffused into the tray water. The authors must explicitly state whether these containers were perforated, the size/number of the perforations, and whether any active water exchange was facilitated. Without these details, the exposure protocol is not reproducible.

The preparation of the immediate threat environment is currently a 'black box' that hinders experimental replication. The authors state that pupae were recorded 'under immediate threat (using water with predation cues)'. Aquatic predator cues are highly variable and depend on predator density, incubation volume, duration of cue accumulation, and the predator's diet (e.g., whether they were recently fed conspecific prey to release alarm cues). The authors must provide a detailed protocol of how this cue water was generated, including the number of dragonfly nymphs used, the volume of water they were held in, the duration of holding, and their feeding regime prior to water collection.

Tracking individual behaviour in a group of identical, unmarked animals is highly challenging and prone to error. The authors note: 'We recorded pupae in groups of five. Three individuals per group were randomly selected and followed for behavioural data'. Because mosquito pupae are morphologically indistinguishable and move rapidly during active diving events, they can easily swap positions. The authors must explain how the observers maintained the identity of the three selected individuals over the 10-minute recording period. If specialized multi-object tracking software was used, its parameters and error rates should be reported; if manual tracking was performed, the authors should clarify if individuals were tracked sequentially from video recordings (e.g., tracking individual 1 for 10 minutes, then re-playing to track individual 2) to prevent identity loss.

In accordance with modern open-science standards and the editorial policies of Biology Letters, all raw data and analysis code must be publicly archived. The manuscript currently refers to a preprint for details ('Detailed experimental protocols are provided in Rawat et al., 2024') and mentions supplementary tables (Tables S1–S4) that are not embedded. To ensure full transparency, the authors should deposit their raw tracking datasets, R scripts for the mixed-effects models, and supplementary tables in a public repository (such as Dryad, Zenodo, or GitHub) and provide the corresponding DOI or temporary access link in the manuscript.