Changes of Tree Species Composition and Distribution Patterns in Mts. Jiri and Baegun, Republic of Korea over 15 Years

Round 1

Reviewer 1 Report

Congratulations on this extensive survey and the sampling efforts. The modeling approach to understand the driver of community change is also really interesting. This article would sure help to understand better the dynamic of vegetal community in East asia and could be of great supprot for ecosystem management. From my point of view we miss the history of the forest on these plots and how disturbed they were at the beginning of the survey. 

L40: Which change of forest ecosystems are we speaking about? I think you should condense the first two sentences to make clear from the beginning which type of change you speak about.

L48-49: Was there a clear trend from this study? An increase in diversity? Was it a study linked to the abandonment of the pasture in Puerto Rico and describing the change from a grazed to a forest ecosystem? It could be interesting to hear about the main results if they were clear rather than the methodology.

L51: Which type of regional disturbance are we speaking about? But it is interesting to hear about the main driver identified in the study.

L61-62: Why? Lack of precision in the measurement done (not at species and age class level)? Too small geographical area?

L63: Precise what you mean by snapshots of community change. Is it an issue about the geographical scale? The temporal scale used in these studies?

L71: Suggestion: use drivers rather than defining characteristics

L81: Precise what is meant by life history attribute

L97: Any information on the history of land use of this area? It would be really important to have an idea of the type of ecosystems met, if it’s closer from a primary forest or pasture land recovering. I understand it might change a lot along the plots, maybe just give a general description.

L108: any comments on possible sample bias regarding the slope (more plot in the less sloppy area for example)? If it is the case, it is understandable, regarding difficulties to establish plot on a 40-degree slope but should be precise.

L110:112: any reference to support this classification? Precision on the analysis made to classify the soil sample?

L116: which software has been used to develop the MLR models?

L133: what is the margin to consider a species as fluctuating rather than increasing or fluctuating?  A threshold of difference between initial and final effective for which you consider the species as increasing or decreasing?

L178-181: what the reasons for quoting these species specifically? The most abundant in an initial assessment? Hold a particular role in the ecosystem?

L184: Contradicting statement of line 181 182 saying that species diversity showed relatively negligible change.

L193-194: total tree abundance? I am slightly confused, the total number of trees?

L209: Are we speaking about the total number of species?

L210: what is considered a trivial change?

L220: Please include the unit.

L221: Please include the unit.   

L223: how can you link the shift in tree composition and the increase in the number of trees, positive growths pole and larger adult stature of trees? Moreover, you link this increase of growths to another hypothesis in the discussion

L284: maybe could you expand a bit longer on past of these forests, have been heavily logged, saw slash’n’burn agriculture, have been damaged during war?

L319: Would be interesting to know when the succession began

L327: is it an increased concentration of CO2 or a more mature ecosystem which provides better nutrition to the tree and expose less nutrient deficiency? Not contesting your statement, but if there is another likely theory, and maybe some link to the successional state of the forest, it would be good to present them.

L347: Suggestion gives an example of biotic factors that could have been proven as significant in other studies if it’s possible to find it.

Author Response

Congratulations on this extensive survey and the sampling efforts. The modeling approach to understand the driver of community change is also really interesting. This article would sure help to understand better the dynamic of vegetal community in East Asia and could be of great support for ecosystem management. From my point of view we miss the history of the forest on these plots and how disturbed they were at the beginning of the survey. 

L40: Which change of forest ecosystems are we speaking about? I think you should condense the first two sentences to make clear from the beginning which type of change you speak about.

=> The first two sentences were combined and mentioned our study focus with regard to change of species composition, abundance, and forest structure as highlighted in L40-59.

L48-49: Was there a clear trend from this study? An increase in diversity? Was it a study linked to the abandonment of the pasture in Puerto Rico and describing the change from a grazed to a forest ecosystem? It could be interesting to hear about the main results if they were clear rather than the methodology.

=> The main results were added in L49-52.

L51: Which type of regional disturbance are we speaking about? But it is interesting to hear about the main driver identified in the study.

=> The description of main drivers were added in L52-55.

L61-62: Why? Lack of precision in the measurement done (not at species and age class level)? Too small geographical area?

=> Because of small-scale monitoring (geographical) area.

L63: Precise what you mean by snapshots of community change. Is it an issue about the geographical scale? The temporal scale used in these studies?

=> Additional description was presented in L65-67.

L71: Suggestion: use drivers rather than defining characteristics

=> It was replaced with ‘drivers’ as highlighted in L75.

L81: Precise what is meant by life history attribute

=> Additional description was added as highlighted in L85.

L97: Any information on the history of land use of this area? It would be really important to have an idea of the type of ecosystems met, if it’s closer from a primary forest or pasture land recovering. I understand it might change a lot along the plots, maybe just give a general description.

=> Our study area was established by the Seoul National University Forestry since 1964 and maintained as natural forest mountains for forest ecosystem researches without human disturbances.

L108: any comments on possible sample bias regarding the slope (more plot in the less sloppy area for example)? If it is the case, it is understandable, regarding difficulties to establish plot on a 40-degree slope but should be precise.

=> As shown in marks along the x-axis in Fig 6d, most of plots were evenly located between 10 and 35-degree slopes and a few plots were located in 5-6 degree and 38-40 degree slopes.

=> Additional description was presented in L114-115.

L110:112: any reference to support this classification? Precision on the analysis made to classify the soil sample?

=> The reference was added in L121.

=> The precision of soil classification was followed by a soil texture triangle (soil survey manual, 2017)

L116: which software has been used to develop the MLR models?

=> With regard to software, additional description and reference were added in L126-127.

L133: what is the margin to consider a species as fluctuating rather than increasing or fluctuating?  A threshold of difference between initial and final effective for which you consider the species as increasing or decreasing?

=> There is no margin or threshold of difference to differentiate the three groups of species (increasing, decreasing, and fluctuating). If tree numbers of each species show the increasing and decreasing pattern over the study periods, it was grouped as increasing and decreasing species, respectively. Otherwise, it was grouped as fluctuating species, having no clear increasing or decreasing pattern over the periods.

L178-181: what the reasons for quoting these species specifically? The most abundant in an initial assessment? Hold a particular role in the ecosystem?

=> The reason presented the specific names of 19 eliminated and 27 recruited species is to provide a clear explanation what kinds of species had been contributed to overall change of species richness in our study area.

L184: Contradicting statement of line 181 182 saying that species diversity showed relatively negligible change.

=> The contradicting statement was deleted as highlighted in L195.

L193-194: total tree abundance? I am slightly confused, the total number of trees?

=> The total tree abundance means the total number of trees and was replaced as highlighted in L205.

L209: Are we speaking about the total number of species?

=> No. Here we compared changes among number of individual species over the periods.

L210: what is considered a trivial change?

=> A typo (balance) in L220 was corrected as ‘rest’ highlighted.

=> A trivial change was deleted for more clear explanation as highlighted in L221.

L220: Please include the unit.

=> Since the rate of change (i.e. slope) was obtained by a regression with pole growth rate vs. census periods, there is no unit.

=> The overall mean rate of change was corrected as 0.109 as shown in L230.

L221: Please include the unit.  

 => Since the rate of change (i.e. slope) was obtained by a regression with adult growth rate vs. census periods, there is no unit.

=> The overall mean rate of change was corrected as 0.299 as shown in L232.

 

L223: how can you link the shift in tree composition and the increase in the number of trees, positive growths pole and larger adult stature of trees? Moreover, you link this increase of growths to another hypothesis in the discussion

=> In this study, the shift in tree composition across the 130 0.1-ha subplots were verified in terms of the annual rates of change in the number of trees, pole growth rate, adult growth rate, and adult stature. As shown in Fig.4, tree composition in more than 50% of subplots has been shifted toward increased abundances of species with positive pole and adult growth rates and adult stature.

L284: maybe could you expand a bit longer on past of these forests, have been heavily logged, saw slash’n’burn agriculture, have been damaged during war?

=> Since the survey was started in 1998, it is impossible to expand on the analysis of past of these forests.

=> These forests were heavily damaged during the Korean war (1950-1953).

L319: Would be interesting to know when the succession began

=> Unfortunately, we do not have information on successional state of our study sites.

L327: is it an increased concentration of CO2 or a more mature ecosystem which provides better nutrition to the tree and expose less nutrient deficiency? Not contesting your statement, but if there is another likely theory, and maybe some link to the successional state of the forest, it would be good to present them.

=> Since we do not have any information on the successional state of the forest, there is difficulty to present another likely theory.

L347: Suggestion gives an example of biotic factors that could have been proven as significant in other studies if it’s possible to find it. 

=> The additional explanation and reference of biotic factors on species diversity were presented in L365.

Author Response File: Author Response.docx

Reviewer 2 Report

The data set developed by the authors is moderately interesting (15 years of change in forest composition).  It is difficult to evaluate how much change might be expected since the authors provide virtually no stand history information, so I have no idea what successional stage these forests are in.  If no such historical data are available, perhaps the authors could use dendrochronological analyses to obtain tree age data.  This data set will become progressively more interesting if periodic monitoring of these permanent plots continues in the future. 

I had great difficulty trying to critically evaluate the claims regarding temporal change and patterns of variation.  For most, there was insufficient documentation of statistical analyses and the graphics presented indicate whatever patterns that might exist are weak and of limited ecological importance. 

I have provided specific feedback with yellow highlight and comments in the manuscript PDF.

Comments for author File: Comments.pdf

Author Response

The data set developed by the authors is moderately interesting (15 years of change in forest composition).  It is difficult to evaluate how much change might be expected since the authors provide virtually no stand history information, so I have no idea what successional stage these forests are in.  If no such historical data are available, perhaps the authors could use dendrochronological analyses to obtain tree age data.  This data set will become progressively more interesting if periodic monitoring of these permanent plots continues in the future. 

I had great difficulty trying to critically evaluate the claims regarding temporal change and patterns of variation.  For most, there was insufficient documentation of statistical analyses and the graphics presented indicate whatever patterns that might exist are weak and of limited ecological importance. 

I have provided specific feedback with yellow highlight and comments in the manuscript PDF.

L79-82: Meaning of this sentence is unclear.

=> Additional explanation was presented in L85.

L100-112: This description of environmental context is all well and good, but I cannot interpret the results that describe change in species abundance and community composition over recent years without some background information re: stand history and successional stage.

=> Our study sites were maintained as natural forest for forest researches since the 1950s’ Korean War.

=> Since there has been no monitoring on the succession, we do not have information on successional state of our study sites.

L118-119: I assume that actual weather data from the 15 weather stations were also used to predict weather at each plot. This is unclear from this sentence. This short list of site variables seems inadequate to predict monthly values for the weather variables.

=> In fact, the data from the 15 weather station were not used to predict weather at each plot, but were used to check the accuracy of MLR models for corresponding 15 weather stations by comparing the predicted values of each climate values.

=> We found incorrect statement about deriving climate data. The statement was corrected as highlighted in L122-134.

L120-121: If weather station data were used to predict plot weather, this seems like circular reasoning. This explanation for how plot weather was determined was not clear to me.

=> As mentioned just above, the data from the 15 weather station were not used to predict weather at each plot, but were used to check the accuracy of MLR models for corresponding 15 weather stations by comparing the predicted values of each climate values.

=> In our study, MLR models were first developed to derive four climate variables (annual mean monthly minimum, maximum, and mean temperatures, total precipitation) for each census year (1998-2012) in monthly basis. In the MLR models, the plot-level topographical variables such as latitude, longitude, elevation, slope, sin(aspect), and cos(aspect) were used as predictor variables through a stepwise method. For example, the following MLR models are to derive the annual mean monthly total precipitation and mean temperature variables for each year and month in R software. Here, the predictor variables were selected by significance for each year and month correspondingly.

< Annual mean monthly total precipitation >

Year 1998: Sqrt(total precipitation) ~ α₀ + α₁Elevation + α₂Slope + α₃Latitude + α₄cos(Aspect) + factor(month) Year 1999: Sqrt(total precipitation) ~ α₀ + α₁Elevation + α₂Slope + α₃Latitude + factor(month)

…..

Year 2012: Sqrt(total precipitation) ~ ß₀+ ß₁Elevation + ß₂Slope + ß₃Longitude + ß₄Latitude + ß₅cos(Aspect) + ß₆sin(Aspect) + factor(month)

< Annual mean monthly average temperature >

Year 1998: Avg.Temp ~ α₀ + α₁Elevation + α₂Longitude + α₃cos(Aspect) + α₄sin(Aspect) + factor(month) Year 1999: Avg.Temp ~ α₀ + α₁Elevation + α₂Longitude + α₃Latitude + factor(month)

…..

Year 2009: Avg.Temp ~ ß₀+ ß₁Elevation + ß₂Longitude + factor(month)

….

The accuracy of the predicted climate values were then checked by comparing for 15 weather stations with the monitoring data from 15 meteorological stations.

Finally, monthly climate variables for each plot were estimated with developed MLR models. For example, topographical values (latitude, longitude, elevation, slope, sin(aspect), and cos(aspect)) of each plot were plugged into the developed models to estimate the corresponding monthly total precipitation for each plot;

Year 1998 & Jan:

Sqrt(annual mean monthly total precipitation₁) = Intercept₁ + (Elevation.coef₁*Elevation of each plot)+ +(Slope.coef₁*Slope of each plot) + (Lat.coef₁*Latitude of each plot) + (cos.Aspect.coef₁*cos(Aspect) of each plot)

Year 1998 & Feb:

Sqrt(annual mean monthly total precipitation₂) = Intercept₂ + (Elevation.coef₂*Elevation of each plot)+ +(Slope.coef₂*Slope of each plot) + (Lat.coef₂*Latitude of each plot) + (cos.Aspect.coef₂*cos(Aspect) of each plot)

…

L130-134: Based on this description it seems to me that tree species abundance was based solely on stem counts, regardless of stem size. Hence, a 6cm dbh stem is counted the same as a 60 cm dbh stem. In my experience this is not appropriate. There does not seem to be a distinction between tree species density vs. dominance (basal area). How was relative abundance calculated? Solely on stem counts? Also, there is no mention of the statistical test used to determine if differences are greater than expected due solely to random variation.

=> Yes, tree species abundance was based on stem counts, regardless of stem size, in each plot. Since we were interested in variation of the number of each tree species over 15 yrs, # of stems for each species were counted regardless of stem size as shown in Fig 3.

=> In this study, species density was estimated as # of individual trees per plot, while the basal area (m²) was calculated based on equation of [π*(dbh/2)²]/10000 for each stem and then total basal area was estimated for each plot.

=> The relative abundance was calculated as the percentage of each species by dividing the number of species from one group by the total number of species from all groups present for each census period as shown in Fig 3.

=> We were interested in the pattern of changes (increasing, decreasing, no pattern) in tree abundance between census periods, the statistical test was not needed to verify those patterns.

L148-150: B-C dissimilarity is computed by comparing individual species relative abundance and summing the common abundance across all species. This is NOT a metric that can be used to "identify changing patterns in the mean species diversity". I'm wondering if the authors sometimes refer to "species diversity" when what they actually mean is "species composition", which is the term that refers to the relative abundances of all species in a plot or forest.

=> Yes, it refers to change in species composition. That statement was corrected highlighted in L158-159.

L165: Unclear what has been averaged. Does this refer to the average across the three times when species richness was determined?

=> We found a mistake on description on GAMs. Corrected statement was presented in L159-178. In this study, GAMs were conducted for species diversity.

=> For clarity, additional statement was added in L175-176.

L181-184: Sequential sentences seem contradictory. First, changes in diversity and evenness are "negligible". In the very next sentence the authors report "a steady increase" in these variables. No test of significance is reported and the absolute differences over time are quite small and of little apparent ecological importance.

=> The contradicting statement was deleted as highlighted in L196.

L186: Unclear how these negligible changes have any relationship to "stable forest structure".

=> The statement was replaced as highlighted in L197-198.

L188: There is no apparent statistical test for these slopes. Visually, The bottom panel seems to clearly indicate No Change. Is this panel necessary?

=> Fig 2 is needed to show the overall change patterns of stand density, richness, diversity, and evenness over census periods. Since we were interested in percent change of stand density, richness, diversity, and evenness, the statistical tests were not performed.

L190-197: Given the wide range of tree sizes documented (6 cm to 100cm, I do not think it is appropriate to quantify species abundance simply based on stem counts. Tree saplings cannot be counted as equivalent to large overstory trees. This would not be considered valid by American forest ecologists.

=> In this study, the shift of species abundance was verified by stem counts, which is generally used in forest research. Also, it is not necessary to differentiate the tree size to verify the variation in number of individual species.  

L204: Use of the word "significantly" usually implies some kind of statistical test of significance was performed, but I see no evidence of that.

=> The word was deleted as highlighted in L216.

L209-210: In the absence of statistical tests, how did the authors distinguish "significant" vs. "trivial" changes? Subjective judgment?

=> The word was deleted for clarity as highlighted in L222.

L220-221: Does this statistic have units? Given the small value, this seems like the more appropriate conclusion is "no mean change" to me.

=> No, there are no units. Since the rate of change (i.e. slope) was obtained by a regression, there are no units.

=> We found a mistake and recalculated those values as highlighted in L230-233.

L224: Tree growth can only be positive. Should this be growth RATE?

=> Yes, it should be growth rate. The word ‘rates’ were added as highlighted in L235.

L224-226: Since growth is measured as simple change in DBH, this is obvious since annual tree-ring width deceases exponentially as a function of age and increasing diameter.

=> The statement in L224-226 was deleted.

L239-240: Meaning of "intermittent" unclear. Wording awkward. Perhaps "...high recruitment (5-20 trees) of intermediate or shade-tolerant species,..."

=> The word was changed to “intermediate” as highlighted in L251.

L243: Wording: "...minimal (1-3 trees) recruitment and loss..."

=> The word was changed as highlighted in L254.

L250: "geographic"

=> In general, elevation, longitude, latitude, slope, etc were called as topographic drivers.

L252: variance

=> It should be “deviance” since the percentage of deviance is generally used as a goodness-of-fit statistic in a generalized additive model.

L254-255: Given the extremely limit range of values for longitude and latitude, there is absolutely no ecological reason to expect an association with species richness associated with environmental gradients.

=> Please check ‘2.2 Data Analysis’. We found a mistake and corrected descriptions with regard to GAM as highlighted in L159-179. In this study, GAMs were developed for species diversity instead of species richness.

=> Since our study sites are located within the limited ranges of longitude and latitude, we cannot help it. However, GAM would be good way to identify the non-linear associations between species diversity and environment gradients under the limited range of values for longitude and latitude. As you mentioned, GAM showed that species diversity showed no obvious relationship with longitude and latitude, but it revealed an interesting non-linear relationship.

The fact that the authors claim (without any documentation) that the associations were "significant" causes me to question the validity of their analyses. This concern is further increased by the wide scatter of data points in Figures 6 and 7 that indicated to me virtually no meaningful associations with any of the environmental variables. While the confidence intervals make it seem that predictions have adequate precision, I strongly suspect that the prediction intervals would that the precision of estimates was so poor as to be useless.

=> The p-values were presented for significant variables in L261-263 and L278-279.

=> As you are concerned, the associations between environmental variables and species diversity seem weak based on the deviances explained (50.4% by topographic variables and 18.3% by climate variables). However, the results of GAMs are still critical to reveal the underlying relationships.

L265-270: Wording implies cause-effect, when only a statistical association is presented. Suggested revision: "...were significantly associated with the distribution of species richness." The words "relationship" and "effect" (implying cause-effect) does not belong in the Results section. Replace with either correlation or association throughout.

=> The words were revised as highlighted in L281-286.

L273-276: I do not believe the authors provide sufficient documentation for me to critically evaluate the evidence in support of their claims about "significant" associations. And even if the associations with individual variables are "significant", I would need to see coefficient of determination (R-sq) statistics to evaluate whether the associations were strong enough to be worth noting. The slope of the regression lines vs. the scatter of data points suggests that these individual associations explain so little variation in species diversity as to be meaningless in the real world.

=> The R-sq of GAM with climate variables is 0.164 (18.3% of deviance).

L284-285: Am I missing something? According to my history books the Korean War was in the 1950's. Also, is there no information about historical land use or disturbance? Without this context I find it hard to interpret the study results that describe forest change over time.

=> The year was corrected as highlighted in L98.

=> The additional information was presented with regard to vegetation and disturbance in our study site in L97-100.

L287-289: The authors confuse the word growth with standing crop. Basal areas reported here measure standing crop not growth. The change in basal area among the three measurement periods reflects growth. Basal area increased over time, not growth

=> The word was replaced with ‘basal area’ as highlighted in L299.

L291-292: Growth of dominant trees is often associated with decreased species richness due to competitive exclusion. So why say "In spite of growth"? This is supported by the fact that early successional species were lost what late successional species were gained.

=> The word was changed as highlighted in L303-304.

L308-309: Showed increases in what? Basal area? Stem count? Relative abundance (based on stem count)?

=> It refers to increase in stem count. The additional explanation was presented as highlighted in L321 and L323.

L311: abundance

=> The word was changed as highlighted in L325.

L324-329: I don't understand why this is worthy of note. Of course species whose abundance increased are species with positive growth. Is there a larger point to make here?

=> In Fig. 4, the annual rates of change in pole growth rate across 130 subplots are distributed between -0.35 and 0.35. Approximately 41% of the plots showed negative mean annual rates of change in pole growth. Therefore, it is worth that the shift of tree composition in the rest of subplots had been toward positive.

I don't understand this statement at all. Trees do not exhibit negative growth rates, only STANDS exhibit negative growth (due to tree mortality). Furthermore, there is nothing to support this claim that increased tree growth is in anyway caused by increasing concentrations of CO2. Many factors could cause trees to grow faster. Without further evidence, this speculation should be deleted.

=> As shown in L339, we mentioned the annual change in stand-level growth rates across 130 subplots, not individual tree growth rate.

=> Since we do not have a direct evidence of CO₂ effect on increased tree growth, we just mentioned one of speculations as shown in recent studies [12, 39, 40].

L336: Based on the results presented, I saw insufficient evidence to support this claim. While I can understand how elevation might influence local variation in species diversity, the time range of variation in latitude and longitude cannot be associated with any meaningful variation in environmental conditions that might influence species diversity. The fact that the authors make this claim causes me to question the validity of their statistical analyses.

=> As shown in Figure S1, latitude and longitude were the second most significant variables to explain the variation of species diversity based on a CART model. GAM also showed the significant p-values for variables as presented in L261-263.

=> Although the associations of latitude and longitude with species diversity showed non-linear (no obvious relationships), it does not mean insignificance. Our sample size likely limits to draw any such inference.

Author Response File: Author Response.docx

Reviewer 3 Report

This manuscript examined changes in tree species richness, composition and abundance across three time periods in temperate forests of South Korea. In general, I appreciate the aims of the authors however there are a number of areas in the manuscript that need further attention. Specifically, (1) the introduction is a bit disjointed and misses the opportunity to link the importance of long-term studies, potential drivers of tree diversity, and how this can aid in forest management intervention. (2) There are a number of important methodological details that need further elaboration (see below). Of particular concern is that it does not seem that the differences in mountain regions (Jiri and Baegun) are accounted for in the analyses. A possible alternative would be to reanalyze the data separately for each mountain region and discuss findings, but as it currently stands, these seem to be two distinct areas and plots separated by some geographic distance, thus the generalizable patterns shown in the results are difficult to interpret. (3) The discussion needs to be revised, as the first half is more-or-less a summary of the results. The second half of the discussion would also benefit from a connection back to forest management implications based on long-term forest inventories.

L40-41 – further elaborate on what is meant by combined impacts of abiotic and biotic pressures? Such as?

L47 – here and thereafter, what is meant by forest structure? # of trees/ha?

Consider incorporating 1^st and 2^nd paragraphs to provide better cohesion.

Furthermore, authors need to include defining what abiotic factors and/or gradients are important, particularly in temperate forest systems.

Why are authors looking at both the individual and community level? Bring some rationale into the introduction for this.

L75 – this is the first mention of function and seems out of context.

L100 – can incorporate this sentence with L90.

Methods - How close are these mountains to each other? Further background information is needed on the study areas….for example, what is the dominant vegetation? Are soils similar across this region?

L101 – remove the word ‘quadrats’ as you are measuring at the plot level.

L111 – how were soils collected, processed and analyzed? Further information is needed.

Although there are 130 plots, weather data was used based on 15 stations, thus how were authors able to tease apart changes at the individual or plot level from more regional level weather data (as multiple plots would have the same temp/precip)? Furthermore, authors are using 15 years of weather data, however, from the plot inventories, there are only three distinct periods….a bit more clarity on the methods are needed here.

L134 – what is meant by ‘fluctuating species’?

L137 – isn’t your metric of stand prevalence similar to your metric for stand density? How are the two different?

L138 – what is meant by pole growth rates? How are authors determining/defining individual adult trees? Are all trees found within these systems with >20 cm dbh adults?

L142 – Were all the same trees tagged and sampled during the three periods? If so, why isn’t there a comparison between the 1^st sampling and the 3^rd? On another note, since inventories were sampled across years for a given time period, how was this variability accounted for in the analysis?

L150 – why were GAMs only developed for tree richness? How about relative abundance?

Figures 4 and 5 – plots here are represented as 0.1 ha sub-plots…this seems confusing…why the change to sub-plots?

Figure 5 – which plots are from Mt. Jiri versus Baegun? What is being represented in the left panel?

L264-271 – by ‘species diversity’, authors mean richness? I would use richness explicitly when describing # of species as the terminology is confusing in this paragraph. Same goes for Figures 6 and 7.

Discussion – L278-286, this information should be included in the introduction.

L287-307 – seems to be a summary of results

L327-332 – very speculative with the data presented, language needs to be toned down.

L346 – awk sentence. Authors measured temp/precip….what other factors then might be important in this system that were missed and why?

Author Response

This manuscript examined changes in tree species richness, composition and abundance across three time periods in temperate forests of South Korea. In general, I appreciate the aims of the authors however there are a number of areas in the manuscript that need further attention. Specifically, (1) the introduction is a bit disjointed and misses the opportunity to link the importance of long-term studies, potential drivers of tree diversity, and how this can aid in forest management intervention. (2) There are a number of important methodological details that need further elaboration (see below). Of particular concern is that it does not seem that the differences in mountain regions (Jiri and Baegun) are accounted for in the analyses. A possible alternative would be to reanalyze the data separately for each mountain region and discuss findings, but as it currently stands, these seem to be two distinct areas and plots separated by some geographic distance, thus the generalizable patterns shown in the results are difficult to interpret. (3) The discussion needs to be revised, as the first half is more-or-less a summary of the results. The second half of the discussion would also benefit from a connection back to forest management implications based on long-term forest inventories.

L40-41 – further elaborate on what is meant by combined impacts of abiotic and biotic pressures? Such as?

=> Additional description was presented in L40-45.

L47 – here and thereafter, what is meant by forest structure? # of trees/ha?

=> It means the forest stages by tree age (diameter) and # of trees/ha.

Consider incorporating 1^st and 2^nd paragraphs to provide better cohesion.

=> Two paragraphs were incorporated as shown in L40-59.

Furthermore, authors need to include defining what abiotic factors and/or gradients are important, particularly in temperate forest systems.

=> Additional description was presented in L41-42.

Why are authors looking at both the individual and community level? Bring some rationale into the introduction for this.

=> The rationale was presented in L64-67.

L75 – this is the first mention of function and seems out of context.

=> It was deleted as highlighted in L79.

L100 – can incorporate this sentence with L90.

Methods - How close are these mountains to each other? Further background information is needed on the study areas….for example, what is the dominant vegetation? Are soils similar across this region?

=> Since the paragraph of L90 is a description of study site, the paragraph of L100 would be better to be separated.

=> Further background of study areas was presented in L94-102.

L101 – remove the word ‘quadrats’ as you are measuring at the plot level.

=> It was deleted as highlighted in L108.

L111 – how were soils collected, processed and analyzed? Further information is needed.

=> Further information were presented in L117-121.

Although there are 130 plots, weather data was used based on 15 stations, thus how were authors able to tease apart changes at the individual or plot level from more regional level weather data (as multiple plots would have the same temp/precip)? Furthermore, authors are using 15 years of weather data, however, from the plot inventories, there are only three distinct periods….a bit more clarity on the methods are needed here.

=> We found incorrect statement about deriving climate data. The statement was corrected as highlighted in line 122-134.

=> In fact, the data from the 15 weather station were not used to predict weather at each plot, but were used to check the accuracy of MLR models for corresponding 15 weather stations by comparing the predicted values of climate values.

=> In our study, MLR models were first developed to derive four climate variables (annual mean monthly minimum, maximum, and mean temperatures, total precipitation) for each census year (1998-2012) in monthly basis. In the MLR models, the plot-level topographical variables such as latitude, longitude, elevation, slope, sin(aspect), and cos(aspect) were used as predictor variables through a stepwise method. For example, the following MLR models are to derive the Annual mean monthly total precipitation and mean temperature variables for each year and month in R software. Here, the predictor variables were selected by significance for each year and month correspondingly.

< Annual mean monthly total precipitation >

Year 1998: Sqrt(total precipitation) ~ α₀ + α₁Elevation + α₂Slope + α₃Latitude + α₄cos(Aspect) + factor(month) Year 1999: Sqrt(total precipitation) ~ α₀ + α₁Elevation + α₂Slope + α₃Latitude + factor(month)

…..

Year 2012: Sqrt(total precipitation) ~ ß₀+ ß₁Elevation + ß₂Slope + ß₃Longitude + ß₄Latitude + ß₅cos(Aspect) + ß₆sin(Aspect) + factor(month)

< Annual mean monthly average temperature >

Year 1998: Avg.Temp ~ α₀ + α₁Elevation + α₂Longitude + α₃cos(Aspect) + α₄sin(Aspect) + factor(month) Year 1999: Avg.Temp ~ α₀ + α₁Elevation + α₂Longitude + α₃Latitude + factor(month)

…..

Year 2009: Avg.Temp ~ ß₀+ ß₁Elevation + ß₂Longitude + factor(month)

….

The accuracy of the predicted climate values were then checked by comparing for 15 weather stations with the monitoring data from 15 meteorological stations.

Finally, monthly climate variables for each plot were estimated with developed MLR models. For example, topographical values (latitude, longitude, elevation, slope, sin(aspect), and cos(aspect)) of each plot were plugged into the developed models to estimate the corresponding monthly total precipitation for each plot;

Year 1998 & Jan:

Sqrt(annual mean monthly total precipitation₁) = Intercept₁ + (Elevation.coef₁*Elevation of each plot)+ +(Slope.coef₁*Slope of each plot) + (Lat.coef₁*Latitude of each plot) + (cos.Aspect.coef₁*cos(Aspect) of each plot)

Year 1998 & Feb:

Sqrt(annual mean monthly total precipitation₂) = Intercept₂ + (Elevation.coef₂*Elevation of each plot)+ +(Slope.coef₂*Slope of each plot) + (Lat.coef₂*Latitude of each plot) + (cos.Aspect.coef₂*cos(Aspect) of each plot)

…

L134 – what is meant by ‘fluctuating species’?

=> It means that those species showed mixed patterns of increasing and decreasing between study periods. In other words, it does not show continuous increasing or decreasing patterns over periods.

L137 – isn’t your metric of stand prevalence similar to your metric for stand density? How are the two different?

=> Stand prevalence is similar to tree abundance (# of trees in a plot) in a log scale. Stand density was estimated as # of trees per unit plot.

L138 – what is meant by pole growth rates? How are authors determining/defining individual adult trees? Are all trees found within these systems with >20 cm dbh adults?

=> Pole growth rate is the mean change in stem diameter across all individual trees with a dbh of 6-20cm in each plot over census periods.

=> In Korea, adult trees are considered as trees with a dbh >20cm based on the general guidance of stand-level growth rates.

=> Yes. All trees with a dbh>20cm were defined as adult trees.

L142 – Were all the same trees tagged and sampled during the three periods? If so, why isn’t there a comparison between the 1^st sampling and the 3^rd? On another note, since inventories were sampled across years for a given time period, how was this variability accounted for in the analysis?

=> No. Trees were not tagged and sampled over the census periods. Without tagging trees, they sampled all trees in the plot during each census period.

L150 – why were GAMs only developed for tree richness? How about relative abundance?

=> We found a mistake on description on GAMs. Corrected statement was presented in L159-178.

=> In this study, we developed several GAMs with regard to species richness, species diversity, and species abundance. Since we’d like to examine the patterns of species diversity along env drivers, we only presented GAMs for species diversity.

Figures 4 and 5 – plots here are represented as 0.1 ha sub-plots…this seems confusing…why the change to sub-plots?

=> Since all plots were 20m x 50m (0.1 ha), the change of species composition in community-level was conducted in each plot (0.1 ha).

Figure 5 – which plots are from Mt. Jiri versus Baegun? What is being represented in the left panel?

=> Since we used census data from both Mts. Jiri and Baegun, Fig 5 represents the BC index of 130 plots from both Mt.Jiri and Baegun.

=> The left panel represents the relative distribution (frequency) of a BC index of 130 plots. For example, 28 out of 130 plots show to have average BC value of 0.28.

L264-271 – by ‘species diversity’, authors mean richness? I would use richness explicitly when describing # of species as the terminology is confusing in this paragraph. Same goes for Figures 6 and 7.

=> Since GAMs were conducted for species diversity (L159-178), the terminology was all replaced with ‘species diversity’ as shown in L259-291.

Discussion – L278-286, this information should be included in the introduction.

=> For coherence, part of sentences were deleted as shown in L293-297 and some were incorporated in the description of study site as shown in L97-98.

L287-307 – seems to be a summary of results

=> I do not agree with your comment. In this paragraph, we presented more detail discussion on the overall changes in stand density, richness, diversity, and evenness.  

L327-332 – very speculative with the data presented, language needs to be toned down.

=> The words were toned down as highlighted in L340-341.

L346 – awk sentence. Authors measured temp/precip….what other factors then might be important in this system that were missed and why?

=> Additional description was presented in L359-367.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

Thank you for the effort put forth in revising the manuscript. While having reservations for its suitability in the previous round, I am for the most part satisfied with the updates made throughout the manuscript. This has increased its clarity and resolved the questions and queries posed by the reviewers, particularly the methods section stats. Also, I appreciated the detailed answers to the reviewer queries; these were very helpful.