This study uses 14-yr dataseries of reef fishes from two regions in the SW Atlantic transition zone to estimate how richness and functional diversity change over time. After reading the comments of the first-round reviewers (I am not one of them), it is clear that the manuscript has improved. However, several issues still need to be addressed, mainly related to Methodology. In relation to this, the methods are detailed in a somewhat disorganized and imprecise way, causing some doubts in the reader. Possibly the most critical point is the need to control for the effect of species richness. The authors use an unusual approach instead of the method with which we are all familiar: the calculation of standardized effect sizes using null models. The methodology employed for the calculation of functional diversity is also unclear, and is chopped up in different parts here and there. I also suggest the use of models more suitable for the calculation of temporal trends (GAM) and analysis specifically aimed at determining the identity of those species with a larger contribution to overall compositional dissimilarity.

Title: The existence of replacement is somewhat foreseeable so I would remove this from the title.

L94 I would use species and trait diversity here (instead of “richness”).

L135 Please, specify the number of traits considered in the present study.

L132-144 It is not clear to me the approach employed to characterize the functional space of each plot. A common approach is obtaining a matrix of functional distances (Euclidean or Gower’s distance), to be used as input in a PCoA. Then, the resulting 2-4 axes can used to build a multidimensional functional space and the species coordinates be employed for computing functional diversity indices. Later, I would assess the quality of my functional space (Maire et al., 2015. Global Ecology and Biogeography).
L146 This section (Standardization of sampling effort) should be after the one devoted to censuses data.
L160 Species and trait signs of homogenization: This section is a bit messy. The authors mix several things here. The presentation is not smooth and the authors mix up variation in functional diversity indices with variation in beta diversity. I would devote a section to the calculation of functional diversity after the description of the results.
L168 Although one of the Reviewers of the first round suggested sticking to a single term and using “functional” or “trait” consistently throughout the manuscript, I think both terms are not totally homologous. For instance, in my opinion, trait richness (evenness) sounds odd. The most common term in the literature is functional richness/evenness.

L167 Functional richness is not independent of species richness, to deal with it a common approach is computing SESFric (standardized effect sizes of functional richness) using null models with a considerable (e.g. 999) number of iterations. I would like to see the temporal variation of SESFric instead of raw FRic values in Figure 2.

L168 Also, authors need to justify why they chose these indices and not others (e.g. functional divergence).

I strongly suggest using generalized additive models (GAM) instead of basic linear models to assess the temporal variation of the obtained indices.

L180 Beta diversity: this should be a separate paragraph.

L181 Did you use the beta.pair function (pairwise dissimilarities) of ‘betapart’ and computed an average value for each site? Please, explain. The Methods are described in a rather vague way, without much detail.

L182 OK, this is what I was referring to a few lines ago. Thus, the authors computed a dissimilarity matrix (functional distance between species) that was used to (a) build a multidimensional functional space and compute functional diversity indices and (b) to compute n-dimensional hypervolumes and estimates functional beta diversity using the kernel.beta function. Right? I recommend the inclusion of a section devoted to explain both facets: functional (alpha) diversity and functional beta diversity for the sake of clarity. A diagram would be helpful too.

L193 Perhaps the authors could calculate the temporal beta diversity for the period 2008-2022 following the approach devised by P. Legendre.

It would also be interesting to explore if the nestedness component of beta diversity (ordered species loss) varies across time and between regions using the Baselga’s partitioning method.

Furthermore, I suggest computing the species contribution to beta diversity (SCBD) for each region to identify those species with a larger contribution to overall beta diversity and determine if these species are consistently the same across the study period.

Also, I suggest addressing the existence of directionality of changes in community composition using the rate of community change. This metric assesses the rate and pattern of variability within a community, which indicates whether species reordering over time is resulting in directional change. The function rate_change of the ‘codyn’ package calculates differences in species composition between samples at increasing time intervals (see Collins et al., 2000 for details).

L198-203 Excessively long and awkward sentence. Please, reword.

L201 The functional singularity or distinctiveness of a species depends on the species pool. Thus, a species with a high uniqueness value for a certain year can have a much lower value in the subsequent one if the identity of the remaining species changes substantially between years. It is not clear if the authors computed an average value for each species across years or if they obtained a single value after pooling all assemblages (years) together.

L205-208 Sorry, but I don’t understand well this analysis.

Also, in my opinion, authors should use standardized values (z-scores) when analyzing variation across years.

L227-231 I think this is not the most common and elegant way to consider the influence of species richness.

L238 Again, fitting GAMM models would be more appropriate here.

I am happy with this new version of the manuscript. The authors addressed successfully all the issues identified in the previous round. I have no further comments.