Quantifying Regulating Ecosystem Services with Increased Tree Densities on European Farmland

Round 1

Reviewer 1 Report

The authors examined the effect of the density of trees on three Regulating Ecosystem Services (soil erosion, nitrate leaching and carbon sequestration) in traditional and innovative agroforestry systems through  four case-study sites in Europe. For this purpose, they used a modelling approach . Many variables  involved (weather, soil, radiation, ecc.), were integrated. They found that the presence of trees reduced soil erosion, nitrate leaching and increased carbon sequestration in the soil, with differences among the four sites.

I think the model was very complex because of the many variables to consider. However, I found interesting the aim of the work. It leads to a more sustainable vision of the environment, where different systems can be integrated together to obtain environmental benefits. In my opinion, the work should be accepted after minor revisions:

1)      What does the acronym Yield-SAFE mean? Could you add it in the text?

2)      In the case of the regulation of carbon sequestration (lines 234-245) you mentioned that the inputs to store carbon in the soil are leaf fall, root mortality, timber, but you did not talk about microbial soil biomass. Bacteria and fungi represent a relevant component of the soil and their respiration significantly affects the amount of carbon sequestered in the soil. Could you explain why you omitted this component in your model?

Author Response

Reviewer 1

The authors examined the effect of the density of trees on three Regulating Ecosystem Services (soil erosion, nitrate leaching and carbon sequestration) in traditional and innovative agroforestry systems through four case-study sites in Europe. For this purpose, they used a modelling approach. Many variables involved (weather, soil, radiation, ecc.), were integrated. They found that the presence of trees reduced soil erosion, nitrate leaching and increased carbon sequestration in the soil, with differences among the four sites.

I think the model was very complex because of the many variables to consider. However, I found interesting the aim of the work. It leads to a more sustainable vision of the environment, where different systems can be integrated together to obtain environmental benefits. In my opinion, the work should be accepted after minor revisions:

• What does the acronym Yield-SAFE mean? Could you add it in the text?

The model was developed by van der Werf et al (2007) and was called Yield-SAFE from “Yield Estimator for Long term Design of Silvo-arable Agro-Forestry in Europe”.  This description of the name has been added to the text (line 56-57). 

2)      In the case of the regulation of carbon sequestration (lines 234-245) you mentioned that the inputs to store carbon in the soil are leaf fall, root mortality, timber, but you did not talk about microbial soil biomass. Bacteria and fungi represent a relevant component of the soil and their respiration significantly affects the amount of carbon sequestered in the soil. Could you explain why you omitted this component in your model?

In this study the carbon from microbial biomass is not omitted, on the contrary it is considered. The soil carbon model used already considers the presence of a Microbial biomass carbon compartment. The integration of the carbon model RothC into Yield-SAFE is explained in a specific publication (Palma et al 2017). For this study we do not consider necessary to explain it again. However, as described in Coleman and Jenkinson (2014) in RothC the soil organic carbon is split into four active compartments and a small amount of inert organic matter (IOM). The four active compartments are Decomposable Plant Material (DPM), Resistant Plant Material (RPM), Microbial Biomass (BIO) and Humified Organic Matter (HUM). The initial carbon content is split into these four compartments. Each compartment decomposes by a first-order process with its own characteristic rate. Only the IOM compartment is resistant to decomposition.

References

van der Werf, W.; Keesman, K.; Burgess, P.; Graves, A.; Pilbeam, D.; Incoll, L.D.D.; Metselaar, K.; Mayus, M.; Stappers, R.; van Keulen, H.; et al. Yield-SAFE: A parameter-sparse, process-based dynamic model for predicting resource capture, growth, and production in agroforestry systems. Ecol. Eng. 2007, 29, 419–433, doi:10.1016/j.ecoleng.2006.09.017.

Palma, J.H.N.; Crous-Duran, J.; Graves, A.R.; Garcia de Jalón, S.; Upson, M.; Oliveira, T.S.; Paulo, J.A.; Ferreiro-Dominguez, N.; Moreno, G.; Burgess, P.J. Integrating belowground carbon dynamics into Yield-SAFE, a parameter sparse agroforestry model. Agrofor. Syst. 2017, 92, 1047–1057, doi:10.1007/s10457-017-0123-4.

Coleman, K.; Jenkinson, D.. S. RothC - A model for the turnover of carbon in soil. Eval. Soil Org. Matter Model. 2014, 1–44, doi:10.1007/978-3-642-61094-3_17.

Reviewer 2 Report

REVIEW:  Quantifying Regulating Ecosystem Services with Increased Tree Densities on European Farm land Fruit agroforestry system. 

This modeling studying was well done, and the manuscript was clear and understandable.  The study should provide a useful contribution to the field of agroforestry.  Below are comments that I think need to be addressed for publication.

It is appreciated that the authors compared the results to other studies as a cross check.  That said, several comparisons were made to other modeling studies with Yield-SAFE.  While that may be the only results available for comparison, it should be recognized that comparing modeling results from the same model has significant limitations and should be noted.  

In similar vein, it would be worthwhile to strengthen caveats associated with quantitative results from a model with limited field verification.  The main value of the results is in the relative comparison within and across agroforestry systems. The study provides a good relative comparison across systems due to significant differences in the underlying variables (i.e., fertilized vs unfertilized systems). 

It is recognized that a previous paper (Crous-Duran et al. 2019) addressed the provisioning services of these agroforestry systems based on changes in tree density.  Is it possible to pull in some of the results to get a better sense of the trade-offs between the different ecosystem services?  While it was addressed conceptually in the Linking Provisions section, it could be worthwhile to expand.

Soil erosion

In Table 4, it appears mixed use of commas and periods to indicate decimal point (K and L-S vs C factors).  In regard to L-S, it appears all of the sites have the same L-S values?  This seems highly unlikely and contradicts a statement in LN 369 about the high L-S values at the Swiss site.  Can you please explain? 

Also, the L-S map that was used for the Portuguese, English and German sites was based on 25 m grid cell and as Wu et al. 2005 points out, this resolution can effect calculation of L-S values, particularly max values compared to higher resolution data sets (i.e., 10 m).  While using a consistent data source is worthwhile, it may be worth mentioning that the calculated erosion rates are best as used as comparison tool between alternatives rather than as predicted specific rate.

C factors were based on Panagos et al 2015.  Based on this paper, the range is 0.05–0.15 and 0.15 was selected for this study.  Why was this value selected?  For forest cover, a C factor of 0.03 was used.  How was this derived from the Panagos?  Agroforestry areas were cited with a range from 0.03–0.13.   Forests had a range from 0.0001–0.003.  For modeling purposes, I understand the desire to have a single value for forest cover however in actuality, C factor is more dynamic.  Just applying it as a static multiplier to area under tree canopy does not consider some of the changes and impacts on erosion processes as density increases.  While this may not be addressed easily in the study, it should be mentioned.

Wu, S.; Li, J.; Huang, G. An evaluation of grid size uncertainty in empirical soil loss modeling with digital elevation models. Environ. Model. Assess. 2005, 10, 33–42.

Proof reading comments

LN 161 Reference not found

LN 180-181 Reference not found

Ln 470 change to dependent

Ln 590 I don’t think climate change should be capitalized

Author Response

Reviewer 2

This modeling studying was well done, and the manuscript was clear and understandable.  The study should provide a useful contribution to the field of agroforestry.  Below are comments that I think need to be addressed for publication.

It is appreciated that the authors compared the results to other studies as a cross check.  That said, several comparisons were made to other modeling studies with Yield-SAFE.  While that may be the only results available for comparison, it should be recognized that comparing modeling results from the same model has significant limitations and should be noted. 

One of the main constraints of this study was the lack of similar studies and the lack of information related to the systems and alternatives analyzed. Therefore, we acknowledge the lack of information and we do agree that the validation of the results was limited by the amount of information available. A sentence was added in Conclusion section (line 578... Considering the limited information and quality of data available for comparison, and the restrictions derived from the implementation process of the methodologies to Yield-SAFE, the results …) referencing this point.

However, we would like to clarify that the present study is the first one using the implementation into the Yield-SAFE model of largely-used methodologies for the quantification of soil erosion such as the RUSLE question; the method followed by Palma et al 2007 for nitrate leaching estimations and/or the carbon soil model RothC for considering the total carbon sequestered in biomass and soil.

 On the other studies, similar methodologies were used but always in separate terms. In this sense, modeling studies with Yield-SAFE, sensu stricto were only compared for bio-physical results giving estimations of the carbon sequestration form biomass growth that is the original output of the model. For soil erosion, results compared were estimated mostly using the RUSLE equation and the nitrate leaching the Palma et al 2007 method.

References

Palma, J.H.N.; Graves, A.R.; Bunce, R.G.H.; Burgess, P.J.; de Filippi, R.; Keesman, K.J.; van Keulen, H.; Liagre, F.; Mayus, M.; Moreno, G.; et al. Modeling environmental benefits of silvoarable agroforestry in Europe. Agric. Ecosyst. Environ. 2007, 119, 320–334, doi:10.1016/j.agee.2006.07.021.

In similar vein, it would be worthwhile to strengthen caveats associated with quantitative results from a model with limited field verification.  The main value of the results is in the relative comparison within and across agroforestry systems. The study provides a good relative comparison across systems due to significant differences in the underlying variables (i.e., fertilized vs unfertilized systems). 

We can just partially agree with the sentence of the model being used with limited field verification. The Yield-SAFE model has been largely tested and validated with field measurements in previous studies (Crous-Duran 2018). It is s not the main objective of the present study to show the validation process again and even more considering the systems analysed are the same as in the previous publication. However, with the present study the main objective was to show the methodologies implemented for the quantification of the Regulating ES were giving similar results to those obtained previously separately form the process-based model and this despite in this study being implemented to the model being used after for estimating Regulating ES in different tree density, from arable to forestry alternatives with mid-term tree densities considered as agroforestry practices.

Reference:

Crous-Duran, J.; Graves, A.R.; Paulo, J.A.; Mirck, J.; Oliveira, T.S.; Kay, S.; García-de-Jalón, S.; Palma, J.H.N. Modelling tree density effects on provisioning ecosystem services in Europe. Agrofor. Syst. 2018, 4, doi:10.1007/s10457-018-0297-4.

It is recognized that a previous paper (Crous-Duran et al. 2019) addressed the provisioning services of these agroforestry systems based on changes in tree density.  Is it possible to pull in some of the results to get a better sense of the trade-offs between the different ecosystem services?  While it was addressed conceptually in the Linking Provisions section, it could be worthwhile to expand.

We did not find necessary to explain again the results of a previous paper more than we did in the present study. We considered the present paper is complex enough as it considers 3 Regulating ES in four different agroforestry systems with 6 different tree-densities and explaining more in detail the results found in a previous paper would increase even more this complexity and capacity of the reader to understand the present study.  Another aspect considered was that in a future paper, as suggested by the reviewer, the trade-offs between the Provisioning ES and the Regulating ES, will be further analysed and described.

Soil erosion

In Table 4, it appears mixed use of commas and periods to indicate decimal point (K and L-S vs C factors).  In regard to L-S, it appears all of the sites have the same L-S values?  This seems highly unlikely and contradicts a statement in LN 369 about the high L-S values at the Swiss site.  Can you please explain? 

The LS-Values were extracted from Panagos et al (2015) and Kay et al 2017. For the PT, UK and DE case studies we considered all three were located in the interval of LS between 0.1 and 0.5. For the Swiss case study, we asked directly the author of the paper (Dr. Sonja Kay) answering the same value (LS=0.3). 

The high LS factor mentioned for the Swiss case was an error from previous version of the document. And the reference in the text mentioning it was deleted. At first version we considered a higher LS factor for the Swiss case but the author of Kay et al 2017, confirmed LS factor for the Swiss case was also 0,3 meaning the high levels of soil erosion are mainly due to the higher value of erodibility (K-factor).

Part of the sentence referred was deleted and modified to Line 439-441: The particularly high values at the Swiss case study site were then mostly as a result of the high soil erodibility (K) (García-Ruiz et al. 2015).

References

Panagos, P.; Borrelli, P.; Meusburger, K. A New European Slope Length and Steepness Factor (LS-Factor) for Modeling Soil Erosion by Water. Geosciences 2015, 5, 117–126, doi:10.3390/geosciences5020117.

Kay, S.; Crous-Duran, J.; Garcia-de-Jalón, S.; Szerencsits, E.; Herzog, F. Landscape-scale modelling of agroforestry ecosystems services: A methodological approach. Manuscr. Prep. 2017, 3, doi:10.1007/s10980-018-0691-3.

Also, the L-S map that was used for the Portuguese, English and German sites was based on 25 m grid cell and as Wu et al. 2005 points out, this resolution can effect calculation of L-S values, particularly max values compared to higher resolution data sets (i.e., 10 m).  While using a consistent data source is worthwhile, it may be worth mentioning that the calculated erosion rates are best as used as comparison tool between alternatives rather than as predicted specific rate.

We do agree in the suggestion made. In modelling exercises, outputs are as good as are the inputs used. In this case, for undertaking a European scale comparison exercise for soil erosion, European maps for each of the soil elements were used. The positive aspects of this was the consistency in the methodology for the different factors of the RUSLE equation and the consistency of the results for the three case studies considered (PT, UK and DE). But of course the negative aspect of the scale (25m) being quite large and of course with room for the improvement. In this sense we consider the sentence added in the Conclusion section (line 578... Considering the limited information and quality of data available for comparison, and the restrictions derived from the implementation process of the methodologies to Yield-SAFE, the results …) references this aspect.

C factors were based on Panagos et al 2015.  Based on this paper, the range is 0.05–0.15 and 0.15 was selected for this study.  Why was this value selected?  For forest cover, a C factor of 0.03 was used.  How was this derived from the Panagos?  Agroforestry areas were cited with a range from 0.03–0.13.   Forests had a range from 0.0001–0.003.  For modeling purposes, I understand the desire to have a single value for forest cover however in actuality, C factor is more dynamic.   Just applying it as a static multiplier to area under tree canopy does not consider some of the changes and impacts on erosion processes as density increases.  While this may not be addressed easily in the study, it should be mentioned.

 Wu, S.; Li, J.; Huang, G. An evaluation of grid size uncertainty in empirical soil loss modeling with digital elevation models. Environ. Model. Assess. 2005, 10, 33–42.

Of course, it was difficult to determine the same C factor value for pastures in Portugal and Switzerland. The range identified was from 0.05 to 0.15 but other references were giving higher values  (C=0,2 for pastures for extensive grazing, Bakker et al. 2008). Therefore, we opted for the higher value of the range.

Related to the tree cover, we did not consider the “forestry” alternatives, proper forested areas as the soil and shrub layers suffer from some human management operations. Instead we considered these “forestry” alternatives as high density agroforestry systems and therefore we considered the lowest value presented by agroforestry practices form the range 0,03 – 0,13. In this sense we are considering a high tree density agroforestry system.

Of course, the method followed for soil erosion estimation could be improved and would deserve a scientific publication in its own. But in this case, a more complex methodology considering additional processes taken place could not be so easily implemented into a biophysical process-based model and would not have been possible to compare the effects on soil erosion of different tree densities growth for 4 agroforestry systems in Europe.

References

Bakker MM, Govers G, van Doorn A, et al (2008) The response of soil erosion and sediment export to land-use change in four areas of Europe: The importance of landscape pattern. Geomorphology 98:213–226. doi: 10.1016/j.geomorph.2006.12.027

Proof reading comments

LN 161 Reference not found

The reference to Equation 1 was added. Line 161 “(1)” was added.

LN 180-181 Reference not found

The reference to Table 4 was added. Line 180 “Table 4” was added.

Ln 470 change to dependent

The text was modified accordingly to what is suggested. Line 470 “depend” changed to “dependent”.

Ln 590 I don’t think climate change should be capitalized

The text has been changed accordingly to the suggestion. Line 590 “Climate Change” change to “climate change”.