Interrelationships of Chemical, Physical and Biological Soil Health Indicators in Beef-Pastures of Southern Piedmont, Georgia

Round 1

Reviewer 1 Report

Topics covered in the article „Interrelationships of Chemical, Physical and Biological Soil Health Indicators in Beef-Pastures of Southern Piedmont, Georgia” concern research on soil ecosystem in the context of optimizing nitrogen fertilization and nutrient cycling in pasture areas.

The ability to store and release nitrogen in soils is a current and important area of research for optimizing environmentally safe fertilizer recommendations. It is important to consider biological parameters to assess N availability.

Methodological description of the analyses performed is detailed and comprehensive. Soil samples were taken from six research sites and the results are presented in Table 2. The range of the presented results is wide, which indicates high variability of the tested values.

However, despite the information on average climatic conditions, there is no data on soil moisture conditions to better understand soil nitrogen transformations especially since samples were collected in different years and dates.

Issues needing explanations:

 Why does soil microbial biomass (SMBC) determined on research pastures  treatements have much higher values of ( 168 mg CO₂-C kg^-1 ) than on Farmers’ Field ( 25,9 mg CO₂-C kg^-1 ).

Table 2 Summary statistics of variables contains a mistske,  as the minimal value of Active Carbon (POXC mg ka^-1 2488), is higher than the maximum value (mg ka^-1 1355). This is most likely a technical error.

Why were the soil samples taken at different depths?:

For Research Pastures (Inorganic and/or No Fertilizer) 0-5, 5-10, 10-20 cm.

For Farmers’ Field (Broiler Litter Fertilized) 0-10 cm.

According to the results presented in Table 2, the magnitude of the studied indicators strictly depends on the depth of the soil samples. The values of the studied indicators are higher in the 0-5 level, e.g. mediana Active Carbon (POXC) (mg kg -1 ) is 722, while at the depths of 5-10  and 10—20, respectively less, 299 and 180. 

Why was sampling depth not included as an explanatory variable in the analysis ?

Furthermore, in order to compare Research Pastures (Inorganic and/or No Fertilizer)z Farmers’ Field (Broiler Litter Fertilized)  the authors should recalculate the data for the same depth range of 0-10 cm.

They require a rework.

The work should be thoroughly revised and subjected to a new evaluation.

Author Response

Topics covered in the article „Interrelationships of Chemical, Physical and Biological Soil Health Indicators in Beef-Pastures of Southern Piedmont, Georgia” concern research on soil ecosystem in the context of optimizing nitrogen fertilization and nutrient cycling in pasture areas.

The ability to store and release nitrogen in soils is a current and important area of research for optimizing environmentally safe fertilizer recommendations. It is important to consider biological parameters to assess N availability.

 Methodological description of the analyses performed is detailed and comprehensive. Soil samples were taken from six research sites and the results are presented in Table 2. The range of the presented results is wide, which indicates high variability of the tested values.

Thank you for your comment.

 However, despite the information on average climatic conditions, there is no data on soil moisture conditions to better understand soil nitrogen transformations especially since samples were collected in different years and dates.

 Thank you for your comment. All soil analysis was conducted in air-dried soils with minimal moisture (0.001-0.09 %) which is the standard practice for laboratory soil analysis. All values were corrected for moisture during laboratory measurements which is also a standard practice.

Issues needing explanations:

 Why does soil microbial biomass (SMBC) determined on research pastures  treatements have much higher values of ( 168 mg CO₂-C kg^-1 ) than on Farmers’ Field ( 25,9 mg CO₂-C kg^-1 ).

Thank you for your comment. It could be due to interaction of nutrients with LOI and Clay content. The Farmers field higher LOI and P, and lower Clay content. Farmer’s fields were fertilized with broiler litter which could have had greater C to N ratio or had arsenic or copper.  We did not measure those paramethers.

Table 2 Summary statistics of variables contains a mistake,  as the minimal value of Active Carbon (POXC mg ka^-1 2488), is higher than the maximum value (mg ka^-1 1355). This is most likely a technical error.

 Great catch. Thank you for pointing this out. It is now fixed to show the actual value.

Why were the soil samples taken at different depths?:

For Research Pastures (Inorganic and/or No Fertilizer) 0-5, 5-10, 10-20 cm.

For Farmers’ Field (Broiler Litter Fertilized) 0-10 cm.

According to the results presented in Table 2, the magnitude of the studied indicators strictly depends on the depth of the soil samples. The values of the studied indicators are higher in the 0-5 level, e.g. mediana Active Carbon (POXC) (mg kg -1 ) is 722, while at the depths of 5-10  and 10—20, respectively less, 299 and 180. 

You are correct, however the equations we showed works regardless of depth. Use of different depths was to increase the range of values. As soon as we average the 0-5 and 5-10 to make a 0-10 cm depth, we lose the variability presented by our original data.

We actually did an analysis using the averaged 0-10 cm data, and the lack of range causes the datapoints to cluster in a small group. We did not include add that analysis on the manuscript because we believe that analysis does not add value to what we have presented now. We have now included figures in the supplementary section (Figure S3- Figure S5) showing the analysis with 0-10 cm depth for both Inorganic and Broiler Litter Pastures.

Why was sampling depth not included as an explanatory variable in the analysis ?

Thank you for your comment. This is a legitimate concern, however we collected to soil samples at different depths to get a wide range of values for all measures of soil health. It is quite obvious to see greater activity and better nutritional content in the upper soil layers.

We ask you to  consider what the focus of this  manuscript is. Please refer to the objectives (line 53-57). The main goal of the research was to understand the relationship between soil health parameters, and not direct comparison of two systems directly.

For example, the goal of Figure 4D is to establish a relationship between POXC and Inorganic N regardless of soil depth and pasture management system. One simple question we are trying answer is; If we collect a soil sample from an inorganic pasture and we could only measure POXC, the equations we developed could provide an estimate (regardless of depth).

If the equation works regardless of depth, isn’t it better? We fitted different equations for each depth and the relationships were quite similar. Hence, we decided to fit a generalized model. If you feel such analysis adds value, we can add it the manuscript, however, we believe it is not required.

We added a PCA analysis (Figure 2, Figure S1 and Figure S2) to improve clarity.

Furthermore, in order to compare Research Pastures (Inorganic and/or No Fertilizer)z Farmers’ Field (Broiler Litter Fertilized)  the authors should recalculate the data for the same depth range of 0-10 cm.

Thank you for your comment. We have now included figures in the supplementary section (Figure S3- Figure S5) showing the analysis with 0-10 cm depth for both Inorganic and Broiler Litter Pastures. Also, please refer to our response to your first comment.

They require a rework. The work should be thoroughly revised and subjected to a new evaluation.

Thank you for your comment. We tried to answer your queries and modified the manuscript to add more details. We look forward to response and a positive evaluation of our manuscript.

Reviewer 2 Report

This is a well written paper that attempts to link relationships between chemical, physical and biological soil properties as soil health indicators. Although this is a commendable attempt to gain a better understanding of these relationships from a soil health standpoint, there are issues I this manuscript that seem to confuse the results.  The specific comments are:

1. Soils were sampled at 0-5-, 5-10-, 10-20-cm at the UGA fields and 0-10-cm depths for the farmer fields. This created four distinct populations of samples. This is evident from the data in Table 1 and Figures 2-4. This is also obvious from clustering of the distribution of the data points in the graphs. To try to fit a regression-type of line to this data is impossible and probably has questionable meaning.
2. Trying to compare indicators for 0-5-,5-10-, 10-20- and 0-10-cm depths is nearly impossible because of differences in available biological substrate (C & N) quantity and quality due to other factors not studied (e.g., plant root density and distribution, natural leaching or retention of substrates depending on type of substrate and natural precipitation). It is also not clear if the samples were collected at close to the same time or distributed across the growing seasons. Time of sampling also has an effect. If there is a way of normalizing the data (i.e., weighting by depth), then, perhaps, the approach used by the authors might have some utility. We pretty well know that biological activity decreases with soil depth due to a number of factors (e.g., oxygen availability, substrate availability, moisture) and the data would support that. It would be best to separate the different depths out and data analyzed by depth.  If the authors are comparing the UGA fields with farmer fields then the UGA fields should be averaged between the 0-5- and 5-10-cm depths, say for example, for POXC, UGA pastures have a mean of 526 mg kg^-1 while farmers fields had POXC of 596 which may result of less significance in the difference between the systems.  Also, why are SMBC and soil respiration not reported on the same depth increments? Yet they appear to be compared!
3. It might be useful to use some “newer” statistical approaches such as redundancy analysis (RDA) that could be used to analyze the data and its relationship with environmental variables.
4. The comparisons between sites represented by Figure 5 appear to be reasonable.

This manuscript requires revision  in the way the data is presented before it can be accepted for publication. Besides, if some of the comparisons are adjusted as above, the study conclusions may change.

Author Response

This is a well written paper that attempts to link relationships between chemical, physical and biological soil properties as soil health indicators. Although this is a commendable attempt to gain a better understanding of these relationships from a soil health standpoint, there are issues I this manuscript that seem to confuse the results.  The specific comments are:

1. Soils were sampled at 0-5-, 5-10-, 10-20-cm at the UGA fields and 0-10-cm depths for the farmer fields. This created four distinct populations of samples. This is evident from the data in Table 1 and Figures 2-4. This is also obvious from clustering of the distribution of the data points in the graphs. To try to fit a regression-type of line to this data is impossible and probably has questionable meaning.

That is a good observation, and we thank the reviewer for the comment. The motivation of collecting soil samples from different soil depths was to increase the range of measurement values. And the motivation of collecting soil samples from different fertilizer management systems was to generalize the results across a bigger range of systems.  We have reorganized the data by combining the 0-5 and 5-10 inorganic soils and have described this in the text (Line200 to 205) as suggested by Reviewer 1 and have placed that information in supplementary material.

1. Trying to compare indicators for 0-5-,5-10-, 10-20- and 0-10-cm depths is nearly impossible because of differences in available biological substrate (C & N) quantity and quality due to other factors not studied (e.g., plant root density and distribution, natural leaching or retention of substrates depending on type of substrate and natural precipitation).

Thank you for your comment. We agree there is a much innate spatial and temporal variability in soil health caused by various biotic an abiotic factors. You are correct that every soil sample is different. Even a soil sample collected from same field, few meters away, can have drastically different soil health values. If we started controlling of all possible factors, we will remain with single observation per group. As we mentioned in the previous answer (Point 1), the motivation for different soil depths was to create a wide range of values.

Furthermore, the goal of the manuscript (refer to line 53-60) is to establish relationship between soil health indicators not direct comparison of depths or fertilizer management systems.

For example, the goal of Figure 5D is to establish a relationship between POXC and Inorganic N regardless of soil depth and pasture management system. One simple question we are trying answer is; If we collect a soil sample from an inorganic pasture and we might we  only measure POXC, the equations we developed could provide an estimate (regardless of depth). If the equation works regardless of depth, isn’t it better?

We added a PCA analysis to show variable importance (lines) and added more analysis in the supplementary section (Figures S3-S5) to show comparison between pasture systems at 0-10 cm depth. We averaged 0-5 cm and 5-10 cm depths to get 0-10 cm values on the inorganic pastures.

Also, it is common practice to use soils of different origins and fit a regression to establish s general relationship. Please refer to Figure 6 of this manuscript (https://soil.copernicus.org/articles/6/53/2020/ ), and Figure 1 of (https://cdnsciencepub.com/doi/full/10.4141/cjss2013-005 ).

1. It is also not clear if the samples were collected at close to the same time or distributed across the growing seasons. Time of sampling also has an effect.

Thank you for your comment. We provided time of soil sampling (Table 1 Column: Sampling Dates). Samples were collected between May-July across two years 2017 and 2018.Again, we want to stress on the fact that we are not interested in direct comparison of soil health indicators but in establishing relationship between them. Our soil sampling timeframe is pretty narrow (3 months around same time every year), but isn’t establishing relationships that works across sampling times better than a relationship that works for only certain timeframes of soil sampling?

In the PCA analysis (Figure 2), year/time of sampling does not show any discernable effect. (Line 145-155)

1. If there is a way of normalizing the data (i.e., weighting by depth), then, perhaps, the approach used by the authors might have some utility. We pretty well know that biological activity decreases with soil depth due to a number of factors (e.g., oxygen availability, substrate availability, moisture) and the data would support that. It would be best to separate the different depths out and data analyzed by depth. 

Thank you for your comment. Please refer to response to Questions 1 and  2.

Every soil sample is different. Even a soil sample collected from same field, few meters away, can have drastically different soil health values. If we started controlling of all possible factors, we will remain with single observation per group. As we mentioned in the previous answer (Question1), the motivation for different soil depths was to create a wide range of values.

Furthermore, the goal of the manuscript (refer to line 53-60) is to establish relationship between soil health indicators not direct comparison of depths or fertilizer management systems.

We fitted different equations for each depth and the relationships were quite similar. Hence, we decided to fit a generalized model. If you feel such analysis adds value, we can add it the manuscript, however, we believe it is not required.

1. If the authors are comparing the UGA fields with farmer fields then the UGA fields should be averaged between the 0-5- and 5-10-cm depths, say for example, for POXC, UGA pastures have a mean of 526 mg kg^-1 while farmers fields had POXC of 596 which may result of less significance in the difference between the systems. 

Thank you for your comment. Again, please refer to our response of Question 1. You are correct that if we do a direct comparison of soil health values, after we average the 0-5 and 5-10 cm to match it with farmers’ field soil samples, we would see a significant difference. However, that is not the goal of this manuscript. We are interested in the relationship between soil health indicators rather than absolute values.

1. Also, why are SMBC and soil respiration not reported on the same depth increments? Yet they appear to be compared!

SMBC and respiration were not directly compared. Please refer to Table 3.

1. It might be useful to use some “newer” statistical approaches such as redundancy analysis (RDA) that could be used to analyze the data and its relationship with environmental variables.

Thank you for your suggestion. We conducted a Principal Component Analysis and a factor analysis of mixed data PCA/FAMD to understand the relationship better. Please refer to lines 126-130 and lines 145-155)

1. The comparisons between sites represented by Figure 5 appear to be reasonable.

Thank you for your comment.

This manuscript requires revision  in the way the data is presented before it can be accepted for publication. Besides, if some of the comparisons are adjusted as above, the study conclusions may change.

Round 2

Reviewer 1 Report

After reading the paper, I have no major reservations about the methods used and the way of developing the results. The main remark concerns the inclusion of all observations to determine the regression equations (Fig. 3), although, on the basis of the principal components analysis (Fig. S1), it was found that the relationships between the studied characteristics differ for samples from pastures without organic fertilization and samples from pastures fertilized with chicken manure. . Of course, the authors emphasize this fact, as evidenced by graphs S3, S4 and S5, and they should be discussed in more detail in the text of the work, not in a supplement, as well as biplot (S2). One more small remark - in Table 2, instead of the standard deviation, it may be better to introduce the coefficient of variation (cv (%) = 100SD / Mean), then the variability of the analyzed features can be compared.

Author Response

Reviewer 1

 After reading the paper, I have no major reservations about the methods used and the way of developing the results. 

Thank you for your comment.

The main remark concerns the inclusion of all observations to determine the regression equations (Fig. 3), although, on the basis of the principal component analysis (Fig. S1), it was found that the relationships between the studied characteristics differ for samples from pastures without organic fertilization and samples from pastures fertilized with chicken manure. . Of course, the authors emphasize this fact, as evidenced by graphs S3, S4 and S5, and they should be discussed in more detail in the text of the work, not in a supplement, as well as biplot (S2).

Thank you for your suggestion. First, regression equations for all soil health parameters were fit for each pasture system (Broiler litter and Inorganic) separately (Fig. S3, S4, S5). If a combined equation had better R2 and slope of equation did not differ significantly between two systems, combined equation was selected and reported.

We added more details in the text (Statistical Analysis Section) as suggested to improve clarity.

Biplot and PCA: Lines 145-159.

Differential relationships of soil health indicators in two pasture systems. Lines 125-128 and Line 174-176.

One more small remark - in Table 2, instead of the standard deviation, it may be better to introduce the coefficient of variation (cv (%) = 100SD / Mean), then the variability of the analyzed features can be compared.

Thank you for your suggestion. We now have added %CV as a measure of variability among measured soil health indices. (Line 142: Table 2)

Thank you for your comments which we believe have improved the readability and understanding of this work.

Reviewer 2 Report

The authors have addressed the concerns of this reviewer.

Author Response

Reviewer 2

Thank you very much for your comment.