Responses to comments in the open discussion appear to be satisfactory. When preparing your revised version, please make all associated changes to the paper. In addition, I would again request a few changes as requested in my earlier comments, one minor and one major.

1) As requested earlier, please explain how forest area is estimated. This is such a critical element of the present analysis that it deserves a fuller description here. Your earlier response in the initial review phase was fitting but it was not included in the main manuscript. Please include it this time where appropriate, similar to what you wrote...: “Forest area was estimated by the ratio of “forest” plots to the total plots in the systematic sampling across China.”

2) As noted in my earlier review, one weakness in the use of the forest identity approach for this attribution exercise is the way in which it relies on net area changes rather than gross changes in forest area (e.g. gross gain and gross loss), and thus the way in which it assumes that all forest areas can be assigned the average forest biomass. In reality, a gain in forest area invariably involves the addition of a lower than average biomass density in the stand, as gains involve recruitment of young forests with low stocks. Conversely, a loss in forest area typically involves the clearing of an area with higher than average biomass density. The forest identity approach, and this present analysis, appears to ignore this important distinction. It would be possible to fix this by utilizing the full power of the forest inventory data, which records volume at the plot scale. These plot scale data could be used directly to compute gains in biomass from the gain in forest area over an interval of time, as well as gains in biomass from growth in areas that remained forest over that same interval of time. In the absence of this level of detail, the attribution of carbon stock changes to area change seems almost surely biased, with a net gain in area being assigned a larger fraction of total stock changes than it would appear to deserve. What in the present analysis protects against this bias? Can you obtain a more accurate estimate of the area versus density attribution by using the full detail of the plot data? Such an improvement in the analysis would significantly strengthen the work in my opinion so I would ask the authors to address this as wholly as possible before the manuscript is accepted for final publication.

There appears to be an error in the abstract’s statement: “... areal expansion of forests was a larger contributor to C sinks than forest growth for all forests... (74.6% vs. 25.4% for all forests, ...)”. This cannot be correct given the numbers in Figure 5. Density and area each contribute equally to the total C sink in forest biomass in China (49.6% vs. 50.4%). The same error appears at line 161 to 164. I wonder if percentages were added over regions with different forested areas instead of adding absolute C numbers over regions and then calculating percentages, which is the correct way of computing total contributions. Whatever the case, this must be corrected before the work can be accepted for publication.

Tg C per 30 yrs density area total
planted 309 510 819
natural 539 353 892
all 848 863 1711


% of total density area total
planted 37.7% 62.3% 100%
natural 60.4% 39.6% 100%
all 49.6% 50.4% 100%

Thank you for providing a full report of the results in Table 1 and for working through the associated calculations in the response. The results clearly show a conundrum that derives from a problem with using average C density for the attribution analysis. As I noted before, the conundrum is that the carbon sinks attributed to areal expansion and density increases for planted forests plus natural forests does not sum to the carbon sink attribution for all forests. This is not logical nor quantitatively sound and it derives from the use of mean C density applied to the areal expansion for all forests when most of the areal expansion was in fact in the planted forests which have lower carbon density. This is the same problem I raised at the beginning of the review process and throughout. Here I illustrate the problem in two ways.

Again, the attribution to area versus density for All Forests uses an average carbon density which is incorrect and leads to erroneous interpretation of the relative importance of area versus density in the contributing to the large forest C sink in China. It inflates the apparent importance of areal expansion by using an unrepresentative carbon density for the planted forests which are the forests that experienced most of the areal expansion. Those forests have a density of about (26.7 + 15.6) / 2 = 21.15 MgC ha-1 while all forests have more than double the average density at about 44 MgC ha-1. Thus, areal expansion of planted forests does not add the carbon density of an average forest but rather only the carbon density of the planted forests, which hold less carbon.

Another, even more persuasive, illustration of this flaw is seen in the apparent increase in average density over all forests, which is lower than the actual increase in average density of either planted forests or natural forests. Quantitatively, the average density increase over all forests was only 3.1 MgC ha-1 over the interval (41.3 – 28.2 MgC ha-1 from Table 1), while Natural Forests, which account for most of the forestland, saw an increase of 4.9 MgC ha-1, and Planted Forests saw an even greater density increase of 11.1 MgC ha-1. How could both the planted and natural forests have a larger C density increase than All Forests which includes both categories? Again, this derives from a misleading use of averaging. Because the planted forests saw such a large increase in area, their proportion of all forests rose substantially. Because they have a lower carbon density, this makes it look like carbon density increased less than it did in either of the two components (planted and natural) because the increase in carbon density gets diluted by the increasing proportion of all forests that is of low density (planted). This clearly misattributes the rate of carbon density increase, making it look less important than it actually was.

These flaws are of major significance to the conclusions of the work and therefore the must be corrected before the work can be accepted for publication.

The way to correct these issues is to add up the carbon sinks attributed separately for planted forests and natural forests to get the combined carbon sink strengths inferred for All Forests. This will show that area and density increases had about equal importance. The importance of afforestation will still be true. The abstract, discussion, and conclusions can all be adjusted to match this corrected analysis and interpretation. Table 1 should be included in the main manuscript. However, the results for Ma and Md for “All Forests” should be replaced with the sum of the results for “Planted Forests” and “Natural Forests”. Figure 5 could be left as either a bar chart (not histogram) or pie chart, which are equally effective in my opinion and which effectively convey the same information.

I apologize for this painstaking and enduring scrutiny and I thank you for your patience and diligence in addressing these concerns. I truly believe that this is important work that needs to be published not only for science but also for society given its relevance, for example, to national and international forest and carbon management policies. This is why the work is receiving such a carefully review, ensuring that the analysis holds up and that the interpretation of the drivers of the aggregate carbon sink in China is a correct reflection of reality and what the data show.