the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
08 Mar 2021
08 Mar 2021
Seasonal ecosystem vulnerability to climatic anomalies in the Mediterranean
- 1Institute of Environmental Science and Geography, University of Potsdam, Potsdam, Germany
- 2Institute of Ecology, Technical University of Berlin, Berlin, Germany
- 3Global Water Partnership, Geneva, Switzerland
- 1Institute of Environmental Science and Geography, University of Potsdam, Potsdam, Germany
- 2Institute of Ecology, Technical University of Berlin, Berlin, Germany
- 3Global Water Partnership, Geneva, Switzerland
Abstract. Mediterranean ecosystems are particularly vulnerable to climate change and the associated increase in climate extremes. This study investigates extreme ecosystem responses evoked by climatic drivers in the Mediterranean Basin for the time span 1999–2019 with a specific focus on seasonal variations, as the seasonal timing of climatic anomalies is considered essential for impact and vulnerability assessment. A bivariate vulnerability analysis is performed for each month of the year to quantify which combinations of the drivers temperature (obtained from ER5 Land) and soil moisture (obtained from ESA CCI and ERA5 Land) lead to extreme reductions of ecosystem productivity using the fraction of absorbed photosynthetically active radiation (FAPAR; obtained from Copernicus Global Land Service) as a proxy.
The bivariate analysis clearly showed that, in many cases, it is not just one but a combination of both drivers that causes ecosystem vulnerability. The overall pattern shows that Mediterranean ecosystems are prone to three soil moisture regimes during the yearly cycle: They are vulnerable to hot and dry conditions from May to July, to cold and dry conditions from August to October, and to cold conditions from November to April, illustrating the shift from a soil moisture-limited regime in summer to an energy-limited regime in winter. In late spring, a month with significant vulnerability to hot conditions only often precedes the next stage of vulnerability to both hot and dry conditions, suggesting that high temperatures lead to critically low soil moisture levels with a certain time lag. In the eastern Mediterranean, the period of vulnerability to hot and dry conditions within the year is much longer than in the western Mediterranean. Our results show that it is crucial to account for both spatial and temporal variability to adequately assess ecosystem vulnerability. The seasonal vulnerability approach presented in this study helps to provide detailed insights regarding the specific phenological stage of the year in which ecosystem vulnerability to a certain climatic condition occurs.
Johannes Vogel et al.
Status: closed
-
RC1: 'Comment on bg-2021-47', Anonymous Referee #1, 29 Mar 2021
Review of: Seasonal ecosystem vulnerability to climatic anomalies in the Mediterranean. This study analyses the influence of soil moisture and temperature on vegetation activity in the Mediterranean. The research topic is highly relevant since the vegetation of the Mediterranean is highly sensitive to the interannual climate variability. Nevertheless, I find the study affected by several methodological problems related to the choice of data and the spatial aggregation of the different sources of information and also by some statistical issues. Given these limitations, I would not recommend the publication of this manuscript in Biogeosciences. I am providing detailed assessment for these issues below.
20. This statement should be better described. The authors should refer if this is related to future projections or what are the impacts on ecosystems. In the Mediterranean region there is an observed increase of the plant coverage, forest surface and vegetation activity, so the detrimental impacts on the ecosystems should be explained and supported based on literature.
24-25. there is strong diversity of climate regimes in the Mediterranean region, whith areas showing maximum of precipitation in summer months (e.g. NE Spain), so the spatial diversity/complexity is also valid in terms of climate seasonality.
26-30. There is large amount of literature focusing on impacts of climate extremes, particularly droughts, on Mediterranean ecosystems. See e.g. the studies by J. Peñuelas, J.J. Camarero, Dimitros Sarris, etc. This should be acknowledged in this section.
40-41. The authors should note that atmospheric dynamic is the main driver of soil moisture variability and heat waves in the Mediterranean. For example, advections from the Sahara are main driver of heat waves in the region: Saharan air intrusions as a relevant mechanism for Iberian heatwaves: The record breaking events of August 2018 and June 2019 Sousa, P.M., Barriopedro, D., Ramos, A.M., (...), Espírito-Santo, F., Trigo, R.M. 2019 Weather and Climate Extremes. 26,100224 in comparison to the role of moisture anomalies, which explain less than 30% of the variability of extreme temperatures in the region: The synergy between drought and extremely hot summers in the Mediterranean. Russo, A., Gouveia, C.M., Dutra, E., Soares, P.M.M., Trigo, R.M. 2019 Environmental Research Letters 14(1),014011
45-46. Soil moisture is particularly relevant for crops during winter and early spring in the Mediterranean region, a period in which heat is irrelevant. See https://digital.csic.es/handle/10261/101367 In summer, the soil moisture is always very low, and vegetation activity very small (except in forest areas located in the mountain slopes) (Lasanta, T., & Vicente-Serrano, S. M. (2012). Complex Land Cover Change Processes ... Images in Northeast Spain. Remote Sensing of Environment, 124, 1-14.)
45-54: Soil moisture variability is mostly responding to precipitation variability (see cited experimental study by Austin et al. 1998 above). In addition, soil moisture remote sensing estimations are strongly uncertain given temporal inhomogeneities between satellites.
Introduction in general, I cannot find specific focus and the research gap that authors want to address with this research, the reference to the several previous research on drought impacts on ecosystems in the Mediterranean is mostly missing.
69: The classification and aggregation followed by the authors suggest a general lack of knowledge of the diversity of vegetation conditions, seasonality and characteristics in the Mediterranean. The classification is so simple that is absolutely unuseful. The authors are mixing annual croplands (which mostly correspond to wheat and barley in the region, which are active between January and June, with olive trees, vineyards, fruit trees (e.g. almonds) which show longer active periods, from April to October, including summer time). It is not possible to establish reliable results including in a single category vegetation types affected by soil moisture limitations and heat so differently in different seasons. The same happens with grasslands, some of them active in summertime (alpine grasslands in the mountains above 1600 m) with others (e.g. those located in semiarid steppe areas, mostly active in winter and early spring. Also broadleaved forests show extremely different characteristics (from a range of Q. Ilex forests active during all of the year to Abies alba forest only active in summer). I would say that these artificial mixtures in land cover categories mostly invalidate the results obtained by the authors.
75-76: The authors should stress the large limitations of these data, mostly in terms of temporal homogeneity.
75-80: What is the interest of the surface soil moisture to assess drought effects in summer? In this season, main activity and drought impacts are recorded in forests, but given deeper root systems, it would be expected that sensitivity to the surface soil moisture is limited. Thus, even annual crops have deeper root layers (e.g. wheat 50cm).
80-81. What is the relationship between these two soil moisture products.
83-84. Is this introducing inhomogenities between satellite products?
86. This is an oversimplification. there are vegetation types affected by moisture limitations in winter (crops and grasslands of several regions).
85-96: non well-organised description of FPAR, soil moisture. This part of text should be improved.
100-105: Are the series of the different variables normally distributed? On the contrary this approach is not correct. e.g. for biased variables as precipitation, it causes an artificial low frequency of humid conditions and very high frequency of dry conditions. Authors should provide further imformation on the reliability of the chosen standardization approach and this may strongly affect the obtained results as they are focusing on the relationship between variables in the lower tail of the distribution for soil moisture and in the upper tail for temperature.
Figure 5. Given classification used, I think the results of the monthly vulnerability of the different vegetation categories to soil moisture deficits and heat are probably non-reliable. There are some rare results as sensitivity of crops to hot in May (in which high temperatures favor grain filling in cereals) and not to dry conditions (determinant for final crop yields).
Figure 6. The classification for climate regions is also not very fortunate as it is merging areas with vegetation types characterized by different levels of activity across the different months of the year.
- AC1: 'Reply on RC1', Johannes Vogel, 20 May 2021
-
RC2: 'Comment on bg-2021-47', Anonymous Referee #2, 10 Apr 2021
Review of ‘Seasonal ecosystem vulnerability to climatic anomalies in the Mediterranean’ by Vogel et al.
General comments
In this paper, the authors quantify vulnerability of FAPAR, as a proxy for ecosystem productivity, to variability in temperature and soil moisture in the Mediterranean. They disentangle the vulnerability on seasonal timescales, with land cover type and regionally. They clearly show how, in general, FAPAR is reduced due to cold temperatures in the winter, then to hot temperatures in early spring, followed reductions due to hot and dry conditions in late spring and summer, although productivity is small in summer due to soil moisture limitation. These are clear albeit, not surprising, results.
The paper is very well written considering style and structure. The methods are clearly explained and being simple, they make sense and may easily be applied in other regions (by others). In that sense I do not see major obstacles preventing publication of the paper in BGS.
However, having read the paper, I continue asking myself what point the authors are trying to make. I will explain my remark in two points.
First, the authors use satellite and model data to quantify ecosystem vulnerability in an original and interesting way. However, the paper is rather clean or academic in the sense that it only uses these data streams to apply some statistical/mathematical procedure resulting in maps. Data treatment always produces results, but they are not meaningful without validation. The authors do not confront their results with ground truth data or a soil-vegetation-atmosphere model for validation. How do the readers know that the results actually make sense? (see also my comment to line 231). Additionally, the authors take a rather straightforward route towards the final results, where some reflection is necessary, e.g. when assuming time lags (see my comments to lines 311 and 325).
Second, the authors write about potential applications, e.g. at the end of section 4.1. There the authors suggest that the method may be used to detect (the effect) of trends in temperature and moisture. For the readers, this is where the paper becomes really interesting, but the authors do not take up their own challenge.
The paper as it is now, contains a comprehensive, rounded whole, which is good. I am aware that I might be asking for protruding tentacles to the story, which disturb the entirety of the paper and make it lengthier. However, I think the paper needs those tentacles to appeal to the audience, get cited and make an impact. Therefore I ask the authors to address the two points mentioned before publishing the paper in BGS.
Specific comments
Lines 90-96: Although the introduction and methods are clearly structures in general, in those lines some things are repeated (e.g. the improvements of the somo product, they being ECV’s, etc.)
Line 105: is sigma calculated over the entire year or per month? I think per year, but this does not become entirely clear from the paper.
Section 2.4: ‘The terminology ... is confusing ...’. Yes it is. I have always used the term vulnerability more or less in the sense of sensitivity combined with risk, so FAPAR would be vulnerable to temperature, because it is sensitive and temperature extremes do occur. As such, I had a hard time understanding your definition, and I presume I am not the only one. Could you make an effort to explain better your definition and how it is different from other definitions? This will help many readers to understand the paper.
Line 171: ‘Sparse vegetation never shows vulnerability to hot conditions...’. I was confused here. Since the extremes are defined by the 10th percentile, each pixel must have extremes. Why is there no vulnerability? Do I understand correctly that this is because the extremes in T and somo do not cause a significant change in FAPAR, because the vegetation is dormant? I think these lines would benefit from a better explanation.
Figure 6: Please write out the abbreviations for BALkan, Italy and France, etc., there is enough space in the figure and it would help the readers by not having to turn back to previous pages.
Line 222: ‘... in a transitional or wet system...’: don’t you mean ‘... or dry system...’?
Line 231: ‘... presumably because August is outside of the growing season.’ Why don’t you show this, you have the data.
Line 267: ‘Depending on the plant... beneficial or detrimental.’ Please explain this process wise.
Line 311: Why didn’t you play with the time lag? It would be very interesting if you could show that your method is capable of quantifing the time lag for specific ecosystem types.
Line 325: It seems simple enough to apply different averaging windows to T and somo. Why don’t you take the time to do this properly?
- AC2: 'Reply on RC2', Johannes Vogel, 20 May 2021
-
EC1: 'Comment on bg-2021-47', Bart van den Hurk, 10 May 2021
Both reviewers raise relevant points of criticism to the methodology and implications of this study, where one reviewer recommends to reject, the other has concrete suggestions for extensions to improve the level of interest of the paper. I would recommend the authors to formulate a new version that at least addresses:
- the use of standard deviation as a normalization factor for non-gaussian distributions does introduce the risk of introducing biases. For non-gaussian distrubutions non-parametric indicators such as inter-quantile ranges are more suitable. However, it is well possible that the time series of monthly residuals resembles a gaussian distribution and justifies the use of a z-score, which could be tested by an indicator such as Kolmogorov-Smirnov or Shapiro-Wilks, or display of transformed time series
- surface soil moisture as such is not a powerful indicator for ecosystem stress, but its success and ability to link to precipitation records, physiological state and deep soil moisture is promoted by the avialability of long duration products in which the propagation and accumulation from shallow (short time scale) surface soil moisture to deeper (long time scale) root zone soil water anomalies can be diagnosed. This time scale argument is well adressed in literature on e.g. soil moisture data assimilation and offline land modelling, and could be part of the rationale to use surface soil moisture data sets in your study
- Earlier studies apparently made effective use of this aggregated ecosystem data set to infer statements on ecosystem responses, despite the large diversity of physiological typologies within a single ecosystem class. Reference to this earlier literature including an explanation of why these authors apparently could extract relevant information on vegetation dynamics would be desirable
- some extension to reply to the suggestion of reviewer 2 (using e.g. ground truthing, or derive interpretations of the findings in the context of past or future trends in driving variables) would be highly welcomed and would be a demonstration of the suggested added value of this study for decision support, but I can understand that the paper should not extend its focused scope too much.
- AC3: 'Reply on EC1', Johannes Vogel, 20 May 2021
Status: closed
-
RC1: 'Comment on bg-2021-47', Anonymous Referee #1, 29 Mar 2021
Review of: Seasonal ecosystem vulnerability to climatic anomalies in the Mediterranean. This study analyses the influence of soil moisture and temperature on vegetation activity in the Mediterranean. The research topic is highly relevant since the vegetation of the Mediterranean is highly sensitive to the interannual climate variability. Nevertheless, I find the study affected by several methodological problems related to the choice of data and the spatial aggregation of the different sources of information and also by some statistical issues. Given these limitations, I would not recommend the publication of this manuscript in Biogeosciences. I am providing detailed assessment for these issues below.
20. This statement should be better described. The authors should refer if this is related to future projections or what are the impacts on ecosystems. In the Mediterranean region there is an observed increase of the plant coverage, forest surface and vegetation activity, so the detrimental impacts on the ecosystems should be explained and supported based on literature.
24-25. there is strong diversity of climate regimes in the Mediterranean region, whith areas showing maximum of precipitation in summer months (e.g. NE Spain), so the spatial diversity/complexity is also valid in terms of climate seasonality.
26-30. There is large amount of literature focusing on impacts of climate extremes, particularly droughts, on Mediterranean ecosystems. See e.g. the studies by J. Peñuelas, J.J. Camarero, Dimitros Sarris, etc. This should be acknowledged in this section.
40-41. The authors should note that atmospheric dynamic is the main driver of soil moisture variability and heat waves in the Mediterranean. For example, advections from the Sahara are main driver of heat waves in the region: Saharan air intrusions as a relevant mechanism for Iberian heatwaves: The record breaking events of August 2018 and June 2019 Sousa, P.M., Barriopedro, D., Ramos, A.M., (...), Espírito-Santo, F., Trigo, R.M. 2019 Weather and Climate Extremes. 26,100224 in comparison to the role of moisture anomalies, which explain less than 30% of the variability of extreme temperatures in the region: The synergy between drought and extremely hot summers in the Mediterranean. Russo, A., Gouveia, C.M., Dutra, E., Soares, P.M.M., Trigo, R.M. 2019 Environmental Research Letters 14(1),014011
45-46. Soil moisture is particularly relevant for crops during winter and early spring in the Mediterranean region, a period in which heat is irrelevant. See https://digital.csic.es/handle/10261/101367 In summer, the soil moisture is always very low, and vegetation activity very small (except in forest areas located in the mountain slopes) (Lasanta, T., & Vicente-Serrano, S. M. (2012). Complex Land Cover Change Processes ... Images in Northeast Spain. Remote Sensing of Environment, 124, 1-14.)
45-54: Soil moisture variability is mostly responding to precipitation variability (see cited experimental study by Austin et al. 1998 above). In addition, soil moisture remote sensing estimations are strongly uncertain given temporal inhomogeneities between satellites.
Introduction in general, I cannot find specific focus and the research gap that authors want to address with this research, the reference to the several previous research on drought impacts on ecosystems in the Mediterranean is mostly missing.
69: The classification and aggregation followed by the authors suggest a general lack of knowledge of the diversity of vegetation conditions, seasonality and characteristics in the Mediterranean. The classification is so simple that is absolutely unuseful. The authors are mixing annual croplands (which mostly correspond to wheat and barley in the region, which are active between January and June, with olive trees, vineyards, fruit trees (e.g. almonds) which show longer active periods, from April to October, including summer time). It is not possible to establish reliable results including in a single category vegetation types affected by soil moisture limitations and heat so differently in different seasons. The same happens with grasslands, some of them active in summertime (alpine grasslands in the mountains above 1600 m) with others (e.g. those located in semiarid steppe areas, mostly active in winter and early spring. Also broadleaved forests show extremely different characteristics (from a range of Q. Ilex forests active during all of the year to Abies alba forest only active in summer). I would say that these artificial mixtures in land cover categories mostly invalidate the results obtained by the authors.
75-76: The authors should stress the large limitations of these data, mostly in terms of temporal homogeneity.
75-80: What is the interest of the surface soil moisture to assess drought effects in summer? In this season, main activity and drought impacts are recorded in forests, but given deeper root systems, it would be expected that sensitivity to the surface soil moisture is limited. Thus, even annual crops have deeper root layers (e.g. wheat 50cm).
80-81. What is the relationship between these two soil moisture products.
83-84. Is this introducing inhomogenities between satellite products?
86. This is an oversimplification. there are vegetation types affected by moisture limitations in winter (crops and grasslands of several regions).
85-96: non well-organised description of FPAR, soil moisture. This part of text should be improved.
100-105: Are the series of the different variables normally distributed? On the contrary this approach is not correct. e.g. for biased variables as precipitation, it causes an artificial low frequency of humid conditions and very high frequency of dry conditions. Authors should provide further imformation on the reliability of the chosen standardization approach and this may strongly affect the obtained results as they are focusing on the relationship between variables in the lower tail of the distribution for soil moisture and in the upper tail for temperature.
Figure 5. Given classification used, I think the results of the monthly vulnerability of the different vegetation categories to soil moisture deficits and heat are probably non-reliable. There are some rare results as sensitivity of crops to hot in May (in which high temperatures favor grain filling in cereals) and not to dry conditions (determinant for final crop yields).
Figure 6. The classification for climate regions is also not very fortunate as it is merging areas with vegetation types characterized by different levels of activity across the different months of the year.
- AC1: 'Reply on RC1', Johannes Vogel, 20 May 2021
-
RC2: 'Comment on bg-2021-47', Anonymous Referee #2, 10 Apr 2021
Review of ‘Seasonal ecosystem vulnerability to climatic anomalies in the Mediterranean’ by Vogel et al.
General comments
In this paper, the authors quantify vulnerability of FAPAR, as a proxy for ecosystem productivity, to variability in temperature and soil moisture in the Mediterranean. They disentangle the vulnerability on seasonal timescales, with land cover type and regionally. They clearly show how, in general, FAPAR is reduced due to cold temperatures in the winter, then to hot temperatures in early spring, followed reductions due to hot and dry conditions in late spring and summer, although productivity is small in summer due to soil moisture limitation. These are clear albeit, not surprising, results.
The paper is very well written considering style and structure. The methods are clearly explained and being simple, they make sense and may easily be applied in other regions (by others). In that sense I do not see major obstacles preventing publication of the paper in BGS.
However, having read the paper, I continue asking myself what point the authors are trying to make. I will explain my remark in two points.
First, the authors use satellite and model data to quantify ecosystem vulnerability in an original and interesting way. However, the paper is rather clean or academic in the sense that it only uses these data streams to apply some statistical/mathematical procedure resulting in maps. Data treatment always produces results, but they are not meaningful without validation. The authors do not confront their results with ground truth data or a soil-vegetation-atmosphere model for validation. How do the readers know that the results actually make sense? (see also my comment to line 231). Additionally, the authors take a rather straightforward route towards the final results, where some reflection is necessary, e.g. when assuming time lags (see my comments to lines 311 and 325).
Second, the authors write about potential applications, e.g. at the end of section 4.1. There the authors suggest that the method may be used to detect (the effect) of trends in temperature and moisture. For the readers, this is where the paper becomes really interesting, but the authors do not take up their own challenge.
The paper as it is now, contains a comprehensive, rounded whole, which is good. I am aware that I might be asking for protruding tentacles to the story, which disturb the entirety of the paper and make it lengthier. However, I think the paper needs those tentacles to appeal to the audience, get cited and make an impact. Therefore I ask the authors to address the two points mentioned before publishing the paper in BGS.
Specific comments
Lines 90-96: Although the introduction and methods are clearly structures in general, in those lines some things are repeated (e.g. the improvements of the somo product, they being ECV’s, etc.)
Line 105: is sigma calculated over the entire year or per month? I think per year, but this does not become entirely clear from the paper.
Section 2.4: ‘The terminology ... is confusing ...’. Yes it is. I have always used the term vulnerability more or less in the sense of sensitivity combined with risk, so FAPAR would be vulnerable to temperature, because it is sensitive and temperature extremes do occur. As such, I had a hard time understanding your definition, and I presume I am not the only one. Could you make an effort to explain better your definition and how it is different from other definitions? This will help many readers to understand the paper.
Line 171: ‘Sparse vegetation never shows vulnerability to hot conditions...’. I was confused here. Since the extremes are defined by the 10th percentile, each pixel must have extremes. Why is there no vulnerability? Do I understand correctly that this is because the extremes in T and somo do not cause a significant change in FAPAR, because the vegetation is dormant? I think these lines would benefit from a better explanation.
Figure 6: Please write out the abbreviations for BALkan, Italy and France, etc., there is enough space in the figure and it would help the readers by not having to turn back to previous pages.
Line 222: ‘... in a transitional or wet system...’: don’t you mean ‘... or dry system...’?
Line 231: ‘... presumably because August is outside of the growing season.’ Why don’t you show this, you have the data.
Line 267: ‘Depending on the plant... beneficial or detrimental.’ Please explain this process wise.
Line 311: Why didn’t you play with the time lag? It would be very interesting if you could show that your method is capable of quantifing the time lag for specific ecosystem types.
Line 325: It seems simple enough to apply different averaging windows to T and somo. Why don’t you take the time to do this properly?
- AC2: 'Reply on RC2', Johannes Vogel, 20 May 2021
-
EC1: 'Comment on bg-2021-47', Bart van den Hurk, 10 May 2021
Both reviewers raise relevant points of criticism to the methodology and implications of this study, where one reviewer recommends to reject, the other has concrete suggestions for extensions to improve the level of interest of the paper. I would recommend the authors to formulate a new version that at least addresses:
- the use of standard deviation as a normalization factor for non-gaussian distributions does introduce the risk of introducing biases. For non-gaussian distrubutions non-parametric indicators such as inter-quantile ranges are more suitable. However, it is well possible that the time series of monthly residuals resembles a gaussian distribution and justifies the use of a z-score, which could be tested by an indicator such as Kolmogorov-Smirnov or Shapiro-Wilks, or display of transformed time series
- surface soil moisture as such is not a powerful indicator for ecosystem stress, but its success and ability to link to precipitation records, physiological state and deep soil moisture is promoted by the avialability of long duration products in which the propagation and accumulation from shallow (short time scale) surface soil moisture to deeper (long time scale) root zone soil water anomalies can be diagnosed. This time scale argument is well adressed in literature on e.g. soil moisture data assimilation and offline land modelling, and could be part of the rationale to use surface soil moisture data sets in your study
- Earlier studies apparently made effective use of this aggregated ecosystem data set to infer statements on ecosystem responses, despite the large diversity of physiological typologies within a single ecosystem class. Reference to this earlier literature including an explanation of why these authors apparently could extract relevant information on vegetation dynamics would be desirable
- some extension to reply to the suggestion of reviewer 2 (using e.g. ground truthing, or derive interpretations of the findings in the context of past or future trends in driving variables) would be highly welcomed and would be a demonstration of the suggested added value of this study for decision support, but I can understand that the paper should not extend its focused scope too much.
- AC3: 'Reply on EC1', Johannes Vogel, 20 May 2021
Johannes Vogel et al.
Johannes Vogel et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 384 | 189 | 16 | 589 | 4 | 2 |
- HTML: 384
- PDF: 189
- XML: 16
- Total: 589
- BibTeX: 4
- EndNote: 2
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1