Dear Dr. Trevors,
We have addressed the comments of the two reviewers of WATE4526, “Spatial and temporal variation in calcium and aluminum in northern hardwood forest floors,” as detailed below.
Reviewer # 1
The reviewer asked that we report our elemental concentrations, in addition to change over time. Fig. 3 has some actual data, but it only includes acid-extractable data for six sites, and not exchangeable data for those sites nor extractable data for 13 other sites. It would seem to be fairly easy for the authors to include a table of means for all or at least a substantial subset of their data. True! Means are now presented in Table 1, which formerly presented only the detectable differences.
p.2, line 9. Although the authors discuss the variation that they found in soil Ca and Al (and tabulate it in Table 1), I believe that they fail to stress that this variation was not based on individual samples, but on composites of 10 samples(e.g. title to Table 1). They should note somewhere that individual samples, or composites composed of less than 10 samples, would result in even greater measured variation and/or difficulty in detecting trends. Thank you for this comment. This point was made in the abstract, but we have added it to the title to Table 1, and added emphasis to this point in the Results, which now reads “These estimates are specific to our sampling methods, which are rather intensive, involving five or six transects per stand, each of which combines ten 10 cm x 10 cm forest floor blocks.” In the Discussion, we said “Changes …were mostly too small to be statistically significant (Table 1), even though our sampling intensity was relatively high (Yanai et al., 2003).”
p.2, line 17.
Initially I was not too impressed about the statement regarding “sampling at different depths”, because I and many others often collect the entire forest floor (therefore mixing the depths). However, this issue becomes important because the authors have included the A horizon in their study, and the inclusion of varying amounts of A into a “forest floor” sample will clearly (as their data show) influence the resulting concentrations. This problem of differentiating O from A is probably more important than the mixing of the various O subhorizons. We revised this sentence in the abstract to read, “Therefore, a small change in sampling depth, or the inclusion of more or less A horizon material in the forest floor, could cause large differences in measured Ca and Al concentrations.”
The authors log-transformed their concentration data (p.5
line 23). However, in their discussions regarding variation and the number of
required samples necessary to detect differences (e.g. p.6, line 11-12, Table
1) they do not make it clear if they are discussing log-transformed data or
untransformed data. Specifically, did they perform their power test with logs?
The “… change in Oa horizon exchangeable Al required to be significant...
ranged from 50% to 175%” (p.6, line 11-12) – was that the change in
untransformed numbers after conversion from logs, or the change in logs? Or did
they perform the power tests on untransformed data? This makes quite a
difference (at least to me). The new caption to Figure 1, in which these
data are presented, makes clear that the detectable differences are calculated
on means of log-transformed data. We
also added this point to the title of Table 1.
The reviewer suggested that Fig. 3 could be presented as a log-log plot; this was a great suggestion and we implemented it.
The reviewer suggested that concentration data might better be presented on a volumetric basis. The question of what units to use to report forest floor concentrations is one that interests me; in the tradition of Tony Federer, I have advocated reporting concentrations on an organic mass basis. This probably provides a similar correction to the volumetric basis, because of the close relationship of organic matter content to bulk density in the forest floor (Federer et al. 1993). I don’t agree that this question warrants further exploration in this paper. I’m confident that it would not change our results: specifically, no change in units will make our positive changes negative or vice versa, and our test was for increases in Al corresponding to decreases in Ca.
In summary, I would like to see more “real data” and a clarification of the use of log-transforms in power tests and similar interpretations. Thank you for the suggestions, which resulted in a significant improvement to the clarity and usefulness of the paper.
2. “Are the interpretations/conclusions sound and justified by the data?” This is really a presentation of data – there aren’t many conclusions. I interpret this comment to mean that the statements in our conclusions section did not raise any alarms.
6. “Can you suggest brief additions or amendments (word, phrases) or an introductory statement that will increase the value of this paper for an international audience?” Acidification is a global problem; nutrient losses from acidification affects forest health – could tie into C cycle. “Acid rain” and “cation depletion” are key words; it would be misleading to suggest that our research answers questions of “forest health” or “C cyling.” Our first paragraph links acid rain to plant growth and forest productivity.
9. Where are the error bars in Figs. 1 and 2? Earlier versions of Figures 1 and 2 had error bars, which made the figures difficult to read. They also drew attention to the question of whether individual sites had changed, when our experimental unit is the stand and the appropriate test of our hypothesis involves the mean of each stand, not the variation within the stand. We misled the reviewer by failing to remove the reference to these error bars in our figure captions. They have now been removed. Table 1 has been expanded to present means and standard error at each sampling data. The variance relevant to assessing confidence in change over time is presented in the form of detectable differences, also in Table 1.
Table 1: from what I
understand, these values are the minimum change which is detectable @ a 0.05. Caption should be more precisely described. The table has been expanded. The caption now reads “Exchangeable and
extractable Ca and Al in forest floors of
The data described in the last paragraph of the results section should be in a new table. The means have been added to Table 1 and the data removed from the paragraph.
Also, the regression data might be more useful in a table. I don’t find Figure 3 valuable. Figure 3 has been improved according to the suggestion by Reviewer #1; the log-log plot is more legible because the points are better distributed.
The mechanism of
mineral soil Al replacing
Would we expect to see significant acidification (Ca loss) from the relatively small time window of ~ 1980 to ~ 1995? Maybe if it were 1965 to 1980…. the time when most acidification occurred. In the Discussion, we say “It is also possible that greater losses in exchangeable cations took place before our earliest collection date. Atmospheric deposition of acids was highest in the 1970s (Driscoll et al., 2001) and reductions in exchangeable base cations result in reduced rates of loss (Kirchner and Lyderson, 1995). Studies conducted before 1950 reported higher exchangeable Ca and Mg in forest floors of red spruce stands than did studies conducted after 1970 (Shortle and Bondietti, 1992).”
We are returning three hard copies and an electronic copy of the revised manuscript.
Thank you for this opportunity to present an improved paper.
Sincerely,
Ruth Yanai
Associate Professor
Return to Ruth's homepage