I appreciate the opportunity to review the manuscript “Zooming in on the temporal dimensions of plant-soil feedback: plant sensitivity and microbial dynamics”. The authors synthesized results from three experiments with a focal species known to experience negative conspecific feedback to evaluate how the strength and direction of plant-soil feedback (PSF; home/away comparison) vary with the conditioning and response time. Specifically, the authors hypothesized that the magnitude of PSF would increase with conditioning time due to increase in plant size and depletion of soil nutrients. They also hypothesized that the response time would influence PSF through two mechanisms: (a) size-dependent plant sensitivity to conditioned soil, and (b) compositional change in the microbiome of home vs. away soil.

The authors found that the magnitude of negative feedback varied with conditioning time. Depletion of soil nitrate might be a cause of the variation. They found a generally declining negative PSF with response time, which could be a combined effect of (a) larger plants being less sensitive to conditioned soil, and (b) the fungal community in the away soil becoming more similar to that in the home soil and the compositional change in bacteria community overall relating to less plant growth in the away soil with increasing response time. They also conceptually demonstrated a potential discrepancy between PSF and relative growth rate, and found so in their experimental results.

I found the simultaneous testing of multiple hypotheses very neat. This is the first paper I have read in the field that empirically tested out multiple possible contributors to the temporal patterns of PSF. The authors used three experiments (two were published data) to circumvent the formidable effort if all hypotheses were tested in one experiment. While this is a useful synthesis, I would like to see more justification of combining results from three separate experiments. For example, in the appendix, it would be helpful to compare the experimental conditions (year conducted, duration of each phase, number/concentration of plants in the conditioning phase, proportion of inoculum in the response phase, other important differences in seed source/soil source/experimental protocols, etc.) in a table. It would also be helpful to compare the same measurement across different experiments if available. For example, both experiment 2 and 3 had a sequential harvest in the response phase. If plant biomasses were measured in both cases, did they show similar trends? How did they compare with the results from experiment 1 (varying size rather than time)? Here I am not asking for those specific measurements, but just to give an idea of the type of comparison that could support the consistency of those experiments.

Three between major and minor comments:
The definition of PSF should be clarified early and remain consistent throughout the manuscript. While this manuscript focuses on conspecific feedback (home-away comparison), PSF can refer to other metrics (e.g. pairwise feedback) in the field. Therefore, the definition of PSF should be stated out in the abstract and early in introduction (L43) to avoid this ambiguity. Moreover, Ke et al. 2021 cited in the second paragraph of introduction and in the discussion (L404) did show that the soil biotic effect became more negative with conditioning time in nonlegumes, but they used a different definition of feedback (live- sterile comparison). I found it confusing to mix different definitions of PSF and challenging to compare results stemming from different definitions without additional explanations (L403-404).

I have two questions regarding L85-92. First, my understanding is that the shift in relative microbial abundance over time can only indicate the direction of shift in PSF, but not the starting value of PSF. How did the authors arrive at the conclusion that “positive” PSF (L88) becoming smaller and “negative” PSF (L91) declining? Second, case I and II indicate the overall plant growth potentially declines in “away” soil over time compared to “home” soil, then why would we expect smaller, rather than greater, positive PSF? Did I miss anything?

L313-314: Purely from the description I cannot distinguish between the two cases: (1) for each (home) sample the average dissimilarity from the three samples in the (away) soil was determined, vs. (2) for each (away) sample the average dissimilarity from the three samples in the (home) soil was determined. I don’t think the results would be exactly the same. More importantly, why not using a linear mixed effect model, with the identity (which replicate) of the home soil and of the away soil as the two random effects? In this case, the authors can keep all 9 comparisons for each time point without condensing them with bias (see the above 2 cases). Alternatively, there are also various methods for multivariate analysis (i.e. use the Bray-Curtis dissimilarity matrix as the response variable) the authors could look into.

Below are some minor comments. Overall I think the manuscript can benefit from improved clarity and context.
L7: I think the abstract could also benefit from briefly linking with the hypotheses at the beginning. In this case, readers can understand why each piece of result helps us better understand the temporal patterns of PSF.
L99 -100: The focus on “the negative effect of self-conditioned soil” is without prior context. Only much later the authors mentioned that the focal species usually experienced negative conspecific feedback.
L109: Shouldn’t it be the other way, i.e. “home” divided by “away”?
L156: It would be helpful to provide information about the dominant plant species in the grassland of soil collection.
L167-168: Why did the authors use an additional “microbial soil inoculum” when the bulk soil is already live soil with microbes?
L181-183: “different batches consisting of soil from different pots” is a bit ambiguous, especially given that the soil from pots of the same treatments was homogenized and mixed already. From the context, but not from this sentence, I guessed that the 15 replicates of 8-week conditioning becomes 5 batches each with 3 pots?
L223-224: No need to express “detail” twice
L244: This sentence feels abrupt without context.
L277: I suggest the authors add “in residuals” after “normality”. Normality of the residuals, rather than raw data, is required, and this nuance can often be neglected.
L303: I am not completely sure what does the t test to compare slopes “for all time points combined” look like. Did the authors run one linear regression for all the measurements of each replicate and used that slope (i.e. N=3 for the t-test)? Additional description or sample size should be given.
L394: “beneficial” and “detrimental” are hypothetical. We could not determine the nature of plant-microbe interactions purely based on statistical correlations of their abundance ranks.
L413: “relatively speaking” is a vague phrase. Relative to their current size?
L451-454: Like with PSF, I found it unconvincing to report “similar” results based on different definitions of RGR (live-sterile vs. home-away) without prior explanation of why they could be compared and hypothesis of why we would expect them to be similar (or dissimilar).
L466: Dissimilarity cannot decrease indefinitely and thus technically we should expect a saturating, rather than linear, decline. However, the authors can expect a linear decline over a short time interval.
Fig. 4b caption: I assume the bar plots are also mean plus/minus SE?
L661-662: the second half of the sentence “and the number that […]” seems incomplete

I (previous Reviewer 2) appreciate the chance to read the revised manuscript again and find its clarity and rigor improved. The authors have well addressed my previous questions and concerns on the definition of PSF, the consistency of the experiments, statistical analyses, and writing.

I thank the authors for adding Table S1 and Figure S3. To provide context for the readers, I further recommend them adding the following information they wrote in their response letter to the caption of Figure S3. “Based on this subset of the data (total dry mass in 7 overlapping weeks during the feedback phase), there was only a negative PSF in the experiment of Bezemer et al. 2018. In the original paper, the negative PSF became less and less, and it gradually turned to neutral over time (Bezemer et al. 2018). In Steinauer et al. (2023), there was only a negative PSF at the beginning of the feedback phase (week 3), and then it changed to neutral. The differences in PSF of these two papers may be due to differences in the ammount of inocula that was used and the initial size of the test plants (see summary in Table S1). “

Reviewer 1 and the editor have raised multiple important issues that the authors should address to their satisfaction. Following up the authors’ use of Benjamini-Hochberg correction, I would like to see the false discovery rate they chose reported in the methods.