the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 05 Jan 2022
The use of paleoclimatic simulations to refine the environmental and chronological context of archaeological/paleontological sites
Abstract. To reconstruct the paleoenvironmental and chronological context of archaeological/paleontological sites is a key step to understand the evolutionary history of past organisms. Commonly used method to infer paleoenvironments rely on varied proxies such as faunal assemblages and isotopes. However, those proxies often show some inconsistencies. Regarding estimated ages of stratigraphic layers, they can vary depending on the dating method used. In this paper, we tested the potential of paleoclimate simulations to address this issue and contribute to the description of the environmental and chronological context of archaeological/paleontological sites. We produced a set of paleoclimate simulations corresponding to the stratigraphy of a Late-Pleistocene Holocene site, El Harhoura 2 (Morocco), and compared the climatic sequence described by these simulations to environmental inferences made from isotopes and faunal assemblages. Our results showed that in the studied site combined US-ESR ages were much more congruent with paleoenvironmental inferences than OSL ages. In addition, climatic variations were found to be more consistent with isotopic studies than faunal assemblages, allowing us to discuss unresolved discrepancies to date. This study illustrates the strong potential of our approach to refine the paleoenvironmental and chronological context of archaeological and paleontological sites.
- Preprint
(3730 KB) - Metadata XML
-
Supplement
(1165 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on cp-2021-185', Anonymous Referee #1, 02 Feb 2022
This study attempts to clarify the question of the chronology of a geological-archaeological sequence located in the Atlantic coast of Morocco, based on a comparison of the climate from a climate model and from the succession in different periods (approximately between the mid-Holocene and 100ka).
My overall comment is that the approach and conclusions are not convincing because (1) as stated by the authors, the grid resolution of the climate model used is 157 km, which is a very coarse regional climate for comparison with a sequence, (2) there seems to be a serious problem with the dating (or chronology) of the geologic sequence, which may lead to incorrect comparisons with each time slice from the model, (3) there is no quantitative climate reconstruction from the geologic sequence that could make the data-model comparison accurate, (4) there is a serious problem with the temporal resolution of the geologic sequence, since there are only 7 (or 8?) samples over a period of about 100,000 years. In addition, the ages seem dubious when comparing the different dating methods, and (5) overall, it is difficult to see how the use of a climate model has helped "refine the paleoenvironmental and chronological context of archaeological and paleontological sites," as the authors explain. The authors discussed more problems with their data-model comparison than benefits their approach brought.
-
AC1: 'Reply on RC1', Léa Terray, 14 Apr 2022
We would like to thank the reviewer for the detailed suggestions in the manuscript. They allowed us to better identify the reasons for the lack of clarity in our article and some of the stong comments. We apologize for it, and hope that the proposed revision address the concerns.
(1) as stated by the authors, the grid resolution of the climate model used is 157 km, which is a very coarse regional climate for comparison with a sequence
The model grid we use is a standard grid for global paleoclimate simulations and climate change experiment. Compared to other models with higher horizonral resolution the vertical resolution higher. However it is true that the model and the cave sequence do not depict conditions at the same scale. However, in nature, when considering large changes in climate as we do here between different climatic periods, the global climate conditions at least partly determine the environment at local scale. Thus, the provided information is not independent. In addition, any paleoclimate regional simulations would require the input of a large scale paleoclimate simulation and thereby represent the same large scale climatic differences between the different climate. We also use a model version that has some skill in reproducing the climate in the region of interest. Our intention was not to consider the large scale climate and the specific climatic conditions at the El Harboura 2 site as providing equal information, but to test the consistency between them, and to deduce from it potential implications for the chronological and paleoenvironmental context of El Harhoura 2.
(2) there seems to be a serious problem with the dating (or chronology) of the geologic sequence, which may lead to incorrect comparisons with each time slice from the model
Date a stratigraphic layer is not an easy task. Dating discrepancies are common in archeological context when several dating methods are used, mainly because they do not rely on the same proxy. Each present different benefits and limits. The OSL method is based on the sediment (quartz grains). It is usually pretty reliable, but it's not a straightforward technique. The combined US-ESR method is based on fossil teeth. Conversely to OSL datings, it is a direct technique, but it can be of lower resolution. The AMS-14C method rely on radiocarbon measurements. It is highly reliable to date recent stratigraphic layers, but not at all when applied to older layers.
For these reasons, it is quite normal to observe discrepancies between methods. What remain to decide is which method (i.e. which proxy) provide the best dating estimation regarding our site. This is precisely one of our objective.
Moreover, several studies have dated the stratigraphic layers of our site, providing repeatability for the dating estimations, and ensuring that the dates we used here are not likely to be biased by bioturbation or other taphonomic factors (e.g., Ben Arous et al., 2020a, b; Janati-Idrissi et al., 2012; Jacobs et al., 2012; Marquer et al., in press; Nespoulet and El Hajraoui, 2012). Because the geochronological context has been well studied, our site is a good model to test this bi-disciplinary approach.
(3) there is no quantitative climate reconstruction from the geologic sequence that could make the data-model comparison accurate
One of our objective is precisely to test the consistency between the climatology describe by the climate model and paleoenvironmental indicators from our site over the sequence. In other words, to test the coherence between atmospheric conditions at large geographic scale and the landscape and environnemental conditions at local scale.
The paleoenvironmental indicators we used are from the stratigraphic sequence itself. The THI (Taxonomic Habitat Index) is inferred from de microfaunic assemblage of the site, and the isotope survey rely on the study of micromammals teeth from the site. These indicators give quantified information about the environnemental variation over the sequence. They are usual approaches in archeology to quantitatively and qualitatively reconstruct paleoenvironments.
(4) there is a serious problem with the temporal resolution of the geologic sequence, since there are only 7 (or 8?) samples over a period of about 100,000 years. In addition, the ages seem dubious when comparing the different dating methods
The stratigraphic resolution is unfortunatly conditionned by the stratigraphic record. Moreover, a site with eight well defined, studied and dated levels are quite rare. First, because archeological sites as rich as EH2 are not common at all, and secondly because it requires an enormous amount of work to set the taxonomic, chronological and envrionmental context of a site (Ben Arous, 2019; Jeffrey, 2016; Stoetzel, 2009).
The queries about dating methods have been addressed in (2).
(5) overall, it is difficult to see how the use of a climate model has helped "refine the paleoenvironmental and chronological context of archaeological and paleontological sites," as the authors explain. The authors discussed more problems with their data-model comparison than benefits their approach brought.
It is true that, out of caution, we have extensively discussed the limits of our approach. Nevertheless, there is two major findings. First, the climate sequence describe by simulation is clearly more consistent with the paleoenvironmental indicators when we rely on datings performed with combined US-ESR than OSL. This provides substantial support for combined US-ESR datings, thus refining the chronological context of our site. This is an important result, knowing that, until now, there was no method capable of discriminating between these dating techniques on our site. Second, the climate sequence allowed us to discuss the few environmental inconsistencies existing between the paleoenvironmental indicators (THI and isotopes) by informing us about the climate trends at large scale, thus refining environmental inferences.
We hope that these precisions can clarify the objective of our article, and support the relevance of our approach. Following the suggestions of both reviewers, we propose a rearrangement of the plan of the manuscript in order to improve its clarity:
- Introduction
We will more clearly state our objective, the different issues we aim to address and the approach choosen to achieve it. We will also better present the limits that are usual and inevitable in archeological context.
- Material and Methods
This part will be recentered on the newly produced simulations using LMDZOR. Several figures currently presented in the results part will be placed here to support the clarity of the reasonning. Pre-existing simulations will be less detailed. We’ll present how they were run, because using the same forcings for the LMDZOR simulations, but we’ll do not named them and will just refer to the period they represent We will focus on the fact that they have been running using from different model versions, that present different systematic biases. This will be illustrated by comparing the sea surface temperature of historical simulations of each model with observations (Fig2). Fig 3 and 4, illustrating the impact of these biases on the climatology will be passed in supplementary material. We will then present the bias correction we realized in order to reduce this systematic bias between model versions. As a reference model version, we choose CM6A which is the best model version for our region of interest, as illustrated by FigA2 that will be placed in the main text.
- Results
In this part, we will directly start by presenting the results of the new simulations (Fig 5 and 6). We will add a paragraph and a figure explaining the regional and global response to the forcings behind the climate variation found on our region of interest. We’ll add a figure illustrating the dynamical changes inspired from Fig 1rep attach to this response to better stress how temperature, moisture and wind lead to the changes over the NW Moroccan region.
- Discussion
We will focus on the significant results of the consistency of the different proxies, and less on the different biases that have already been presented in the introduction. Further discussion about the interpretational meaning of the correlations between climate variables and paleoenvironmental proxies will be included. We also propose to more explicitly state the major results on which our conclusions are based, and present the perspectives they offer for future works.
We will also to seriousely consider all the constructive comments and suggestions the reviwer has made in the pdf manuscript and submit the paper to English proof-reading.
Please note that substantial changes have already been made in the manuscript since its submission, notably english improvement and clarification of the material and methods.
Best regards,
Léa Terray, on behalf all authors
-
AC1: 'Reply on RC1', Léa Terray, 14 Apr 2022
-
RC2: 'Comment on cp-2021-185', Anonymous Referee #2, 03 Feb 2022
Summary
The authors use a suite of climate simulations to discern between two alternative chronologies of paleoclimate levels in a record covering most of the last glacial cycle from North Africa. The simulations seem to favour the isotope-derived timeseries.Main comments:
The authors address an important issue in paleoclimate and make seemingly innovative steps in combining climate modelling with uncertain paleoclimate chronologies. That said it is not clear that this method would in general produce robust results. Although, this paper demonstrates one example, it does not take account of uncertainties or biases in the climate model simulations or in the records (other than treating the two as equally likely). Something more formalised based on for example Bayesian methods would seem to be more robust, though it may not exist yet. The present method also does not address the underlying causes of interpretational issues with the paleoclimate data. For example, would it not be possible that the improved correlation between climate and isotopes could be because the more biologically-derived proxies are a more complex function of physical drivers than the water isotope signal? It is not clear how the combining (and perhaps weighting) across different climate model variables does or could give a different outcome. Overall, this is a difficult problem to tackle, but it seems that here a clearer demonstration that this method can deal effectively with these issues is needed.
---------------------
Minor comments:For the description of the climate simulations it would make more sense to re-write this with a focus on the new simulations that you have performed rather than on the older ones on which you based the SST fields. Then, whilst I appreciate the effort in evaluating the biases in the different model versions, I would recommend that the discussion of this, along with figures 2-4 are placed in an appendix. To me this would improve the readability of the paper.
The wording used to describe DH1 and 2 is sometimes confusing. DH1/2 are first introduced as two alternative chronologies from the paleoclimate recrods, but later in the text they are used to describe a chronology of the simulations (e.g. line 367). Please could you go through and ensure that it is clear in each instance which is meant?
Abstract: 1st sentence isn't clear. How about something like? : "Reconstructing the paleoenvironmental and chronological context of archaeological/paleontological sites is a key step for understanding the evolutionary history of organisms."In the abstract please define "US-ESR" and "OSL", or you could just say two different dating methods?
Line 11: Please elaborate what you mean typically by "drastic discrepancies".Line 28: "whose microvertebrate assemblages have been extensively studied" - please provide a few example references here.
Figure 1: not clear how the dates in panel F relate to the other variables.Figures 5/6: can you add the times in kyr BP to the grey boxes/legend to match the text (e.g. at line 257).
Lines 297-303: It's not clear to me how the principal component analysis is calculated e.g. from which variables and times? Please could you add a paragraph here to explain this?
It's not clear that the approach for MIS5d in DH2 (L5-8). Have you replicated the climate simulation for 115k in each of these? Is this reasonable given the variations in curves B-E of Figure 1?
Line 366: do you mean overestimated as in too old, or that the uncertainties are over/under-estimated?
Line 425: "interglacial"->"interstadial"? There was no interglacial in MIS3?
Discussion:Since your results come down in favour of the isotope-based record, it would be worth discussing how the inclusion of water and/or carbon isotopes in the climate model could better refine future work?
I also wonder if you might speculate on how applicable this approach is going to be? It require lots of climate simulations, and is presumably only applicable where the chronological/interpretational uncertainties are large.
---------------------
Technical corrections;Abstract sentence 2: Commonly used method .. -> Methods commonly used ...
Line 3: Reconstruct->Reconstructing
Line 5: method -> methods
line 29: describe->described
line 71 and throughout: precipitations -> precipitationTable 1 lig155k -> lig115K
Table 1 and throughout : Gaz-> GasLine 148: forced-> prescribed
Line 169: "In global the model reach equilibrium ...". Do you mean "The globally averaged quantitites show the model has reached equilibrium by eight years"?Line 422: "goes until 11 ka on L8", do you mean "is as large as 11 ka for L8"?
Line 445: "reasonnalbe"->"reasonable"
Overall, the language could do with further proof-reading/editing for English.
Citation: https://doi.org/10.5194/cp-2021-185-RC2 -
AC2: 'Reply on RC2', Léa Terray, 14 Apr 2022
We thank the reviewer for constructive comments and suggestions.
Main comments:
The authors address an important issue in paleoclimate and make seemingly innovative steps in combining climate modelling with uncertain paleoclimate chronologies. That said it is not clear that this method would in general produce robust results. Although, this paper demonstrates one example, it does not take account of uncertainties or biases in the climate model simulations or in the records (other than treating the two as equally likely). Something more formalised based on for example Bayesian methods would seem to be more robust, though it may not exist yet.
The study we present in this article is indeed completely exploratory. Our aim was first to test which information we could deduce from a “simple” consistency approach between global paleoclimate simulations and paleoenvironmental inferences from an archeological site. As the reviewer suggests, this may pave the way for the development of more robust methods combining both fields (paleoclimatoloy and archeology/paleontology). So far, our approach may indeed concern a limited number of well studied sites. However, with more powerful statistical tools, it could be extended to other sites whose context is less referenced, and perhaps even help establish it. The limit is that the sampling is limited so that rigorous uncertainties ‘or uncertainty matrix’ might be difficult to properly estimate
The present method also does not address the underlying causes of interpretational issues with the paleoclimate data. For example, would it not be possible that the improved correlation between climate and isotopes could be because the more biologically-derived proxies are a more complex function of physical drivers than the water isotope signal?
It is true that the paleoenvironmental proxies we used are not equally related to climate variables. Isotopes have been collected from rodents teeth, meaning that the isotope fractions is related to the dietary preferences of organisms, and depend on temperature and vegetation. Thus, isotopes are related to the magnitude of seasonal variation in insolation, water stress, temperature and diurnal temperature range. These variables condition the presence of essential elements for plants survival (e.g. sunlight, water in the soil), and thus may participate to determine the type of vegetation. On the other hand, the taxonomy habitat index (THI) rely on ecological preferences of species. Altogether, the variables of the THI give information about the proportion of biomes (e.g. forest, bush, steppe), and thus the spatial distribution and density of the vegetation. Consequently, its relationship to climate is more indirect than for isotopes. Moreover, the THI is linked to the overall (mean and seasonal) climate variability, while isotopes are related to the magnitude of seasonal variation only. Isotopes somehow better reflects the large climatic changes in terms of temperature and precipitation that have occurred in the past. The THI seems indeed to be a more complex function of climate variables than isotopes. However, these two paleoenvironmental proxies can be considered as complementary since they do seem not reflect the same climate parameters. They thus offer a way to infer the overall consistency between climate and the cave environment. We added these comments in the discussion of the manuscript.
It is not clear how the combining (and perhaps weighting) across different climate model variables does or could give a different outcome.
The choice of climate variables was based on their potential direct or indirect impact on organisms, as explained in the manuscript. However, at first, we selected many more variables. Then, we tested the correlations between the selected climate variables. We excluded redundant variables and chose to use a reduce number of uncorrelated variable in order to reduce data dimensionality.
Minor comments:
For the description of the climate simulations it would make more sense to re-write this with a focus on the new simulations that you have performed rather than on the older ones on which you based the SST fields. Then, whilst I appreciate the effort in evaluating the biases in the different model versions, I would recommend that the discussion of this, along with figures 2-4 are placed in an appendix. To me this would improve the readability of the paper.
Indeed, it is true that the emphasis on model biases make the paper heavy to read. We will rearrange the manuscript and pass this part in appendix following the reviewer suggestion.
The wording used to describe DH1 and 2 is sometimes confusing. DH1/2 are first introduced as two alternative chronologies from the paleoclimate recrods, but later in the text they are used to describe a chronology of the simulations (e.g. line 367). Please could you go through and ensure that it is clear in each instance which is meant?
We take good note of it, these parts will be rephrased to avoid confusion.
Abstract: 1st sentence isn't clear. How about something like? : "Reconstructing the paleoenvironmental and chronological context of archaeological/paleontological sites is a key step for understanding the evolutionary history of organisms."
In the abstract please define "US-ESR" and "OSL", or you could just say two different dating methods?
Line 11: Please elaborate what you mean typically by "drastic discrepancies".
Line 28: "whose microvertebrate assemblages have been extensively studied" - please provide a few example references here.
We thank the reviewer for these suggestions. We will further detail our terms in the abstract and the introduction.
Figure 1: not clear how the dates in panel F relate to the other variables.
The dates and the variables are related through the stratigraphical layers (analyses have been performed on the same layers). We will clarify the figure.
Figures 5/6: can you add the times in kyr BP to the grey boxes/legend to match the text (e.g. at line 257).
Sure, we will add times to the figures.
Lines 297-303: It's not clear to me how the principal component analysis is calculated e.g. from which variables and times? Please could you add a paragraph here to explain this?
The principal components analyses (PCA) have been performed on climate variables presented in Table 2, and grouped per stratigraphical layers. For each climate variable, we considered both the annual mean and seasonal variation (standard deviation). Thus, the PCAs present the climate proximity/differences between the stratigraphical layers of El Harhoura 2. A detailed explaination will be added to the manuscript.
It's not clear that the approach for MIS5d in DH2 (L5-8). Have you replicated the climate simulation for 115k in each of these? Is this reasonable given the variations in curves B-E of Figure 1?
Yes, the climate simulation for 115k has been replicated for layers L8-L5. Indeed, paleoenvironmental indicators support different conditions between L8 and L5. However, the dating resolution does not allow us to distinguish the ages to these layers. This is illustrated by the inconsistency between absolute and relative datings: L8 is dated to ~106k, while L6 is dated to ~116k, while L8 is below L6, and therefore necessarily older. That is why we choose to group those layers.
Line 366: do you mean overestimated as in too old, or that the uncertainties are over/under-estimated?
It is overestimated as in “too old”, this precision will be added in the manuscript.
Line 425: "interglacial"->"interstadial"? There was no interglacial in MIS3?
Indeed, there is no interglacial in MIS3. We will correct this spelling error in the manuscript.
Discussion:
Since your results come down in favour of the isotope-based record, it would be worth discussing how the inclusion of water and/or carbon isotopes in the climate model could better refine future work?
It is indeed a very interesting lead for perspective. Isotopes could be used to refine, or at least cross verify, the simulation of seasonal variation in future works. This would indeed add a more direct consistency test to such a study. However, it is important to keep in mind that there is a difference with isotop records that are usually used in climatology: here they have been collected from rodents teeth, impliying that they are also impacted by the dietary preferences of these rodents. Note also that we have tried to have a “minimum computing” time approach, and that isotope unable models are still quite expensive in computing time, and a full coupled model version with isotopes is underconstruction and was not available for this study.
I also wonder if you might speculate on how applicable this approach is going to be? It require lots of climate simulations, and is presumably only applicable where the chronological/interpretational uncertainties are large.
Such uncertainties are unfortunatly common in archeology when different methods are applied, since the biases discussed depend on the methods and not especially on the site. However, if the results of this approach are promising in the case of our site, El Harhoura 2, it needs indeed the site to have been well studied. For this approach to be applicable on other sites depends of the following requirements :
- Data about the chronology of the sequence must be available from different methods
- Paleoenvironmental inferences need also to be available, and preferentially from different sources
- The sequence must be composed of several levels
- Fully coupled climate simulations of the periods of interest must be available, otherwise their complete production would represent a considerably larger amount of work. In our case, we were able to use an ensemble of multi-period opportunity ran with different versions of the IPSL model over a period of 10 years. However, such simulations are becoming more common and distributed. It opens new avenues on the way to use them to test the consistency of paleoclimate or paleoenvironement reconstructions in different regions.
However, as the reviewer suggested, with more powerful statistical tools, this approach could be extended to other sites whose context is less referenced.
All technical corrections will be included in the manuscript.
We hope that these precisions can support the relevance of our approach. Following the suggestions of both reviewers, we propose a rearrangement of the plan of the manuscript in order to improve its clarity:
- Introduction
We will more clearly state our objective, the different issues we aim to address and the approach choosen to achieve it. We will also better present the limits that are usual and inevitable in archeological context.
- Material and Methods
This part will be recentered on the newly produced simulations using LMDZOR. Several figures currently presented in the results part will be placed here to support the clarity of the reasonning. Pre-existing simulations will be less detailed. We’ll present how they were run, because using the same forcings for the LMDZOR simulations, but we’ll do not named them and will just refer to the period they represent We will focus on the fact that they have been running using from different model versions, that present different systematic biases. This will be illustrated by comparing the sea surface temperature of historical simulations of each model with observations (Fig2). Fig 3 and 4, illustrating the impact of these biases on the climatology will be passed in supplementary material. We will then present the bias correction we realized in order to reduce this systematic bias between model versions. As a reference model version, we choose CM6A which is the best model version for our region of interest, as illustrated by FigA2 that will be placed in the main text.
- Results
In this part, we will directly start by presenting the results of the new simulations (Fig 5 and 6). We will add a paragraph and a figure explaining the regional and global response to the forcings behind the climate variation found on our region of interest. We’ll add a figure illustrating the dynamical changes inspired from Fig 1rep attach to this response to better stress how temperature, moisture and wind lead to the changes over the NW Moroccan region.
- Discussion
We will focus on the significant results of the consistency of the different proxies, and less on the different biases that have already been presented in the introduction. Further discussion about the interpretational meaning of the correlations between climate variables and paleoenvironmental proxies will be included. We also propose to more explicitly state the major results on which our conclusions are based, and present the perspectives they offer for future works.
We will also seriousely consider all the constructive comments and suggestions the reviewer has made in the pdf manuscript and submit the paper to English proof-reading.
Please note that substantial changes have already been made in the manuscript since its submission, notably english improvement and clarification of the material and methods.
Best regards,
Léa Terray, on behalf all authors
-
AC2: 'Reply on RC2', Léa Terray, 14 Apr 2022
Status: closed
-
RC1: 'Comment on cp-2021-185', Anonymous Referee #1, 02 Feb 2022
This study attempts to clarify the question of the chronology of a geological-archaeological sequence located in the Atlantic coast of Morocco, based on a comparison of the climate from a climate model and from the succession in different periods (approximately between the mid-Holocene and 100ka).
My overall comment is that the approach and conclusions are not convincing because (1) as stated by the authors, the grid resolution of the climate model used is 157 km, which is a very coarse regional climate for comparison with a sequence, (2) there seems to be a serious problem with the dating (or chronology) of the geologic sequence, which may lead to incorrect comparisons with each time slice from the model, (3) there is no quantitative climate reconstruction from the geologic sequence that could make the data-model comparison accurate, (4) there is a serious problem with the temporal resolution of the geologic sequence, since there are only 7 (or 8?) samples over a period of about 100,000 years. In addition, the ages seem dubious when comparing the different dating methods, and (5) overall, it is difficult to see how the use of a climate model has helped "refine the paleoenvironmental and chronological context of archaeological and paleontological sites," as the authors explain. The authors discussed more problems with their data-model comparison than benefits their approach brought.
-
AC1: 'Reply on RC1', Léa Terray, 14 Apr 2022
We would like to thank the reviewer for the detailed suggestions in the manuscript. They allowed us to better identify the reasons for the lack of clarity in our article and some of the stong comments. We apologize for it, and hope that the proposed revision address the concerns.
(1) as stated by the authors, the grid resolution of the climate model used is 157 km, which is a very coarse regional climate for comparison with a sequence
The model grid we use is a standard grid for global paleoclimate simulations and climate change experiment. Compared to other models with higher horizonral resolution the vertical resolution higher. However it is true that the model and the cave sequence do not depict conditions at the same scale. However, in nature, when considering large changes in climate as we do here between different climatic periods, the global climate conditions at least partly determine the environment at local scale. Thus, the provided information is not independent. In addition, any paleoclimate regional simulations would require the input of a large scale paleoclimate simulation and thereby represent the same large scale climatic differences between the different climate. We also use a model version that has some skill in reproducing the climate in the region of interest. Our intention was not to consider the large scale climate and the specific climatic conditions at the El Harboura 2 site as providing equal information, but to test the consistency between them, and to deduce from it potential implications for the chronological and paleoenvironmental context of El Harhoura 2.
(2) there seems to be a serious problem with the dating (or chronology) of the geologic sequence, which may lead to incorrect comparisons with each time slice from the model
Date a stratigraphic layer is not an easy task. Dating discrepancies are common in archeological context when several dating methods are used, mainly because they do not rely on the same proxy. Each present different benefits and limits. The OSL method is based on the sediment (quartz grains). It is usually pretty reliable, but it's not a straightforward technique. The combined US-ESR method is based on fossil teeth. Conversely to OSL datings, it is a direct technique, but it can be of lower resolution. The AMS-14C method rely on radiocarbon measurements. It is highly reliable to date recent stratigraphic layers, but not at all when applied to older layers.
For these reasons, it is quite normal to observe discrepancies between methods. What remain to decide is which method (i.e. which proxy) provide the best dating estimation regarding our site. This is precisely one of our objective.
Moreover, several studies have dated the stratigraphic layers of our site, providing repeatability for the dating estimations, and ensuring that the dates we used here are not likely to be biased by bioturbation or other taphonomic factors (e.g., Ben Arous et al., 2020a, b; Janati-Idrissi et al., 2012; Jacobs et al., 2012; Marquer et al., in press; Nespoulet and El Hajraoui, 2012). Because the geochronological context has been well studied, our site is a good model to test this bi-disciplinary approach.
(3) there is no quantitative climate reconstruction from the geologic sequence that could make the data-model comparison accurate
One of our objective is precisely to test the consistency between the climatology describe by the climate model and paleoenvironmental indicators from our site over the sequence. In other words, to test the coherence between atmospheric conditions at large geographic scale and the landscape and environnemental conditions at local scale.
The paleoenvironmental indicators we used are from the stratigraphic sequence itself. The THI (Taxonomic Habitat Index) is inferred from de microfaunic assemblage of the site, and the isotope survey rely on the study of micromammals teeth from the site. These indicators give quantified information about the environnemental variation over the sequence. They are usual approaches in archeology to quantitatively and qualitatively reconstruct paleoenvironments.
(4) there is a serious problem with the temporal resolution of the geologic sequence, since there are only 7 (or 8?) samples over a period of about 100,000 years. In addition, the ages seem dubious when comparing the different dating methods
The stratigraphic resolution is unfortunatly conditionned by the stratigraphic record. Moreover, a site with eight well defined, studied and dated levels are quite rare. First, because archeological sites as rich as EH2 are not common at all, and secondly because it requires an enormous amount of work to set the taxonomic, chronological and envrionmental context of a site (Ben Arous, 2019; Jeffrey, 2016; Stoetzel, 2009).
The queries about dating methods have been addressed in (2).
(5) overall, it is difficult to see how the use of a climate model has helped "refine the paleoenvironmental and chronological context of archaeological and paleontological sites," as the authors explain. The authors discussed more problems with their data-model comparison than benefits their approach brought.
It is true that, out of caution, we have extensively discussed the limits of our approach. Nevertheless, there is two major findings. First, the climate sequence describe by simulation is clearly more consistent with the paleoenvironmental indicators when we rely on datings performed with combined US-ESR than OSL. This provides substantial support for combined US-ESR datings, thus refining the chronological context of our site. This is an important result, knowing that, until now, there was no method capable of discriminating between these dating techniques on our site. Second, the climate sequence allowed us to discuss the few environmental inconsistencies existing between the paleoenvironmental indicators (THI and isotopes) by informing us about the climate trends at large scale, thus refining environmental inferences.
We hope that these precisions can clarify the objective of our article, and support the relevance of our approach. Following the suggestions of both reviewers, we propose a rearrangement of the plan of the manuscript in order to improve its clarity:
- Introduction
We will more clearly state our objective, the different issues we aim to address and the approach choosen to achieve it. We will also better present the limits that are usual and inevitable in archeological context.
- Material and Methods
This part will be recentered on the newly produced simulations using LMDZOR. Several figures currently presented in the results part will be placed here to support the clarity of the reasonning. Pre-existing simulations will be less detailed. We’ll present how they were run, because using the same forcings for the LMDZOR simulations, but we’ll do not named them and will just refer to the period they represent We will focus on the fact that they have been running using from different model versions, that present different systematic biases. This will be illustrated by comparing the sea surface temperature of historical simulations of each model with observations (Fig2). Fig 3 and 4, illustrating the impact of these biases on the climatology will be passed in supplementary material. We will then present the bias correction we realized in order to reduce this systematic bias between model versions. As a reference model version, we choose CM6A which is the best model version for our region of interest, as illustrated by FigA2 that will be placed in the main text.
- Results
In this part, we will directly start by presenting the results of the new simulations (Fig 5 and 6). We will add a paragraph and a figure explaining the regional and global response to the forcings behind the climate variation found on our region of interest. We’ll add a figure illustrating the dynamical changes inspired from Fig 1rep attach to this response to better stress how temperature, moisture and wind lead to the changes over the NW Moroccan region.
- Discussion
We will focus on the significant results of the consistency of the different proxies, and less on the different biases that have already been presented in the introduction. Further discussion about the interpretational meaning of the correlations between climate variables and paleoenvironmental proxies will be included. We also propose to more explicitly state the major results on which our conclusions are based, and present the perspectives they offer for future works.
We will also to seriousely consider all the constructive comments and suggestions the reviwer has made in the pdf manuscript and submit the paper to English proof-reading.
Please note that substantial changes have already been made in the manuscript since its submission, notably english improvement and clarification of the material and methods.
Best regards,
Léa Terray, on behalf all authors
-
AC1: 'Reply on RC1', Léa Terray, 14 Apr 2022
-
RC2: 'Comment on cp-2021-185', Anonymous Referee #2, 03 Feb 2022
Summary
The authors use a suite of climate simulations to discern between two alternative chronologies of paleoclimate levels in a record covering most of the last glacial cycle from North Africa. The simulations seem to favour the isotope-derived timeseries.Main comments:
The authors address an important issue in paleoclimate and make seemingly innovative steps in combining climate modelling with uncertain paleoclimate chronologies. That said it is not clear that this method would in general produce robust results. Although, this paper demonstrates one example, it does not take account of uncertainties or biases in the climate model simulations or in the records (other than treating the two as equally likely). Something more formalised based on for example Bayesian methods would seem to be more robust, though it may not exist yet. The present method also does not address the underlying causes of interpretational issues with the paleoclimate data. For example, would it not be possible that the improved correlation between climate and isotopes could be because the more biologically-derived proxies are a more complex function of physical drivers than the water isotope signal? It is not clear how the combining (and perhaps weighting) across different climate model variables does or could give a different outcome. Overall, this is a difficult problem to tackle, but it seems that here a clearer demonstration that this method can deal effectively with these issues is needed.
---------------------
Minor comments:For the description of the climate simulations it would make more sense to re-write this with a focus on the new simulations that you have performed rather than on the older ones on which you based the SST fields. Then, whilst I appreciate the effort in evaluating the biases in the different model versions, I would recommend that the discussion of this, along with figures 2-4 are placed in an appendix. To me this would improve the readability of the paper.
The wording used to describe DH1 and 2 is sometimes confusing. DH1/2 are first introduced as two alternative chronologies from the paleoclimate recrods, but later in the text they are used to describe a chronology of the simulations (e.g. line 367). Please could you go through and ensure that it is clear in each instance which is meant?
Abstract: 1st sentence isn't clear. How about something like? : "Reconstructing the paleoenvironmental and chronological context of archaeological/paleontological sites is a key step for understanding the evolutionary history of organisms."In the abstract please define "US-ESR" and "OSL", or you could just say two different dating methods?
Line 11: Please elaborate what you mean typically by "drastic discrepancies".Line 28: "whose microvertebrate assemblages have been extensively studied" - please provide a few example references here.
Figure 1: not clear how the dates in panel F relate to the other variables.Figures 5/6: can you add the times in kyr BP to the grey boxes/legend to match the text (e.g. at line 257).
Lines 297-303: It's not clear to me how the principal component analysis is calculated e.g. from which variables and times? Please could you add a paragraph here to explain this?
It's not clear that the approach for MIS5d in DH2 (L5-8). Have you replicated the climate simulation for 115k in each of these? Is this reasonable given the variations in curves B-E of Figure 1?
Line 366: do you mean overestimated as in too old, or that the uncertainties are over/under-estimated?
Line 425: "interglacial"->"interstadial"? There was no interglacial in MIS3?
Discussion:Since your results come down in favour of the isotope-based record, it would be worth discussing how the inclusion of water and/or carbon isotopes in the climate model could better refine future work?
I also wonder if you might speculate on how applicable this approach is going to be? It require lots of climate simulations, and is presumably only applicable where the chronological/interpretational uncertainties are large.
---------------------
Technical corrections;Abstract sentence 2: Commonly used method .. -> Methods commonly used ...
Line 3: Reconstruct->Reconstructing
Line 5: method -> methods
line 29: describe->described
line 71 and throughout: precipitations -> precipitationTable 1 lig155k -> lig115K
Table 1 and throughout : Gaz-> GasLine 148: forced-> prescribed
Line 169: "In global the model reach equilibrium ...". Do you mean "The globally averaged quantitites show the model has reached equilibrium by eight years"?Line 422: "goes until 11 ka on L8", do you mean "is as large as 11 ka for L8"?
Line 445: "reasonnalbe"->"reasonable"
Overall, the language could do with further proof-reading/editing for English.
Citation: https://doi.org/10.5194/cp-2021-185-RC2 -
AC2: 'Reply on RC2', Léa Terray, 14 Apr 2022
We thank the reviewer for constructive comments and suggestions.
Main comments:
The authors address an important issue in paleoclimate and make seemingly innovative steps in combining climate modelling with uncertain paleoclimate chronologies. That said it is not clear that this method would in general produce robust results. Although, this paper demonstrates one example, it does not take account of uncertainties or biases in the climate model simulations or in the records (other than treating the two as equally likely). Something more formalised based on for example Bayesian methods would seem to be more robust, though it may not exist yet.
The study we present in this article is indeed completely exploratory. Our aim was first to test which information we could deduce from a “simple” consistency approach between global paleoclimate simulations and paleoenvironmental inferences from an archeological site. As the reviewer suggests, this may pave the way for the development of more robust methods combining both fields (paleoclimatoloy and archeology/paleontology). So far, our approach may indeed concern a limited number of well studied sites. However, with more powerful statistical tools, it could be extended to other sites whose context is less referenced, and perhaps even help establish it. The limit is that the sampling is limited so that rigorous uncertainties ‘or uncertainty matrix’ might be difficult to properly estimate
The present method also does not address the underlying causes of interpretational issues with the paleoclimate data. For example, would it not be possible that the improved correlation between climate and isotopes could be because the more biologically-derived proxies are a more complex function of physical drivers than the water isotope signal?
It is true that the paleoenvironmental proxies we used are not equally related to climate variables. Isotopes have been collected from rodents teeth, meaning that the isotope fractions is related to the dietary preferences of organisms, and depend on temperature and vegetation. Thus, isotopes are related to the magnitude of seasonal variation in insolation, water stress, temperature and diurnal temperature range. These variables condition the presence of essential elements for plants survival (e.g. sunlight, water in the soil), and thus may participate to determine the type of vegetation. On the other hand, the taxonomy habitat index (THI) rely on ecological preferences of species. Altogether, the variables of the THI give information about the proportion of biomes (e.g. forest, bush, steppe), and thus the spatial distribution and density of the vegetation. Consequently, its relationship to climate is more indirect than for isotopes. Moreover, the THI is linked to the overall (mean and seasonal) climate variability, while isotopes are related to the magnitude of seasonal variation only. Isotopes somehow better reflects the large climatic changes in terms of temperature and precipitation that have occurred in the past. The THI seems indeed to be a more complex function of climate variables than isotopes. However, these two paleoenvironmental proxies can be considered as complementary since they do seem not reflect the same climate parameters. They thus offer a way to infer the overall consistency between climate and the cave environment. We added these comments in the discussion of the manuscript.
It is not clear how the combining (and perhaps weighting) across different climate model variables does or could give a different outcome.
The choice of climate variables was based on their potential direct or indirect impact on organisms, as explained in the manuscript. However, at first, we selected many more variables. Then, we tested the correlations between the selected climate variables. We excluded redundant variables and chose to use a reduce number of uncorrelated variable in order to reduce data dimensionality.
Minor comments:
For the description of the climate simulations it would make more sense to re-write this with a focus on the new simulations that you have performed rather than on the older ones on which you based the SST fields. Then, whilst I appreciate the effort in evaluating the biases in the different model versions, I would recommend that the discussion of this, along with figures 2-4 are placed in an appendix. To me this would improve the readability of the paper.
Indeed, it is true that the emphasis on model biases make the paper heavy to read. We will rearrange the manuscript and pass this part in appendix following the reviewer suggestion.
The wording used to describe DH1 and 2 is sometimes confusing. DH1/2 are first introduced as two alternative chronologies from the paleoclimate recrods, but later in the text they are used to describe a chronology of the simulations (e.g. line 367). Please could you go through and ensure that it is clear in each instance which is meant?
We take good note of it, these parts will be rephrased to avoid confusion.
Abstract: 1st sentence isn't clear. How about something like? : "Reconstructing the paleoenvironmental and chronological context of archaeological/paleontological sites is a key step for understanding the evolutionary history of organisms."
In the abstract please define "US-ESR" and "OSL", or you could just say two different dating methods?
Line 11: Please elaborate what you mean typically by "drastic discrepancies".
Line 28: "whose microvertebrate assemblages have been extensively studied" - please provide a few example references here.
We thank the reviewer for these suggestions. We will further detail our terms in the abstract and the introduction.
Figure 1: not clear how the dates in panel F relate to the other variables.
The dates and the variables are related through the stratigraphical layers (analyses have been performed on the same layers). We will clarify the figure.
Figures 5/6: can you add the times in kyr BP to the grey boxes/legend to match the text (e.g. at line 257).
Sure, we will add times to the figures.
Lines 297-303: It's not clear to me how the principal component analysis is calculated e.g. from which variables and times? Please could you add a paragraph here to explain this?
The principal components analyses (PCA) have been performed on climate variables presented in Table 2, and grouped per stratigraphical layers. For each climate variable, we considered both the annual mean and seasonal variation (standard deviation). Thus, the PCAs present the climate proximity/differences between the stratigraphical layers of El Harhoura 2. A detailed explaination will be added to the manuscript.
It's not clear that the approach for MIS5d in DH2 (L5-8). Have you replicated the climate simulation for 115k in each of these? Is this reasonable given the variations in curves B-E of Figure 1?
Yes, the climate simulation for 115k has been replicated for layers L8-L5. Indeed, paleoenvironmental indicators support different conditions between L8 and L5. However, the dating resolution does not allow us to distinguish the ages to these layers. This is illustrated by the inconsistency between absolute and relative datings: L8 is dated to ~106k, while L6 is dated to ~116k, while L8 is below L6, and therefore necessarily older. That is why we choose to group those layers.
Line 366: do you mean overestimated as in too old, or that the uncertainties are over/under-estimated?
It is overestimated as in “too old”, this precision will be added in the manuscript.
Line 425: "interglacial"->"interstadial"? There was no interglacial in MIS3?
Indeed, there is no interglacial in MIS3. We will correct this spelling error in the manuscript.
Discussion:
Since your results come down in favour of the isotope-based record, it would be worth discussing how the inclusion of water and/or carbon isotopes in the climate model could better refine future work?
It is indeed a very interesting lead for perspective. Isotopes could be used to refine, or at least cross verify, the simulation of seasonal variation in future works. This would indeed add a more direct consistency test to such a study. However, it is important to keep in mind that there is a difference with isotop records that are usually used in climatology: here they have been collected from rodents teeth, impliying that they are also impacted by the dietary preferences of these rodents. Note also that we have tried to have a “minimum computing” time approach, and that isotope unable models are still quite expensive in computing time, and a full coupled model version with isotopes is underconstruction and was not available for this study.
I also wonder if you might speculate on how applicable this approach is going to be? It require lots of climate simulations, and is presumably only applicable where the chronological/interpretational uncertainties are large.
Such uncertainties are unfortunatly common in archeology when different methods are applied, since the biases discussed depend on the methods and not especially on the site. However, if the results of this approach are promising in the case of our site, El Harhoura 2, it needs indeed the site to have been well studied. For this approach to be applicable on other sites depends of the following requirements :
- Data about the chronology of the sequence must be available from different methods
- Paleoenvironmental inferences need also to be available, and preferentially from different sources
- The sequence must be composed of several levels
- Fully coupled climate simulations of the periods of interest must be available, otherwise their complete production would represent a considerably larger amount of work. In our case, we were able to use an ensemble of multi-period opportunity ran with different versions of the IPSL model over a period of 10 years. However, such simulations are becoming more common and distributed. It opens new avenues on the way to use them to test the consistency of paleoclimate or paleoenvironement reconstructions in different regions.
However, as the reviewer suggested, with more powerful statistical tools, this approach could be extended to other sites whose context is less referenced.
All technical corrections will be included in the manuscript.
We hope that these precisions can support the relevance of our approach. Following the suggestions of both reviewers, we propose a rearrangement of the plan of the manuscript in order to improve its clarity:
- Introduction
We will more clearly state our objective, the different issues we aim to address and the approach choosen to achieve it. We will also better present the limits that are usual and inevitable in archeological context.
- Material and Methods
This part will be recentered on the newly produced simulations using LMDZOR. Several figures currently presented in the results part will be placed here to support the clarity of the reasonning. Pre-existing simulations will be less detailed. We’ll present how they were run, because using the same forcings for the LMDZOR simulations, but we’ll do not named them and will just refer to the period they represent We will focus on the fact that they have been running using from different model versions, that present different systematic biases. This will be illustrated by comparing the sea surface temperature of historical simulations of each model with observations (Fig2). Fig 3 and 4, illustrating the impact of these biases on the climatology will be passed in supplementary material. We will then present the bias correction we realized in order to reduce this systematic bias between model versions. As a reference model version, we choose CM6A which is the best model version for our region of interest, as illustrated by FigA2 that will be placed in the main text.
- Results
In this part, we will directly start by presenting the results of the new simulations (Fig 5 and 6). We will add a paragraph and a figure explaining the regional and global response to the forcings behind the climate variation found on our region of interest. We’ll add a figure illustrating the dynamical changes inspired from Fig 1rep attach to this response to better stress how temperature, moisture and wind lead to the changes over the NW Moroccan region.
- Discussion
We will focus on the significant results of the consistency of the different proxies, and less on the different biases that have already been presented in the introduction. Further discussion about the interpretational meaning of the correlations between climate variables and paleoenvironmental proxies will be included. We also propose to more explicitly state the major results on which our conclusions are based, and present the perspectives they offer for future works.
We will also seriousely consider all the constructive comments and suggestions the reviewer has made in the pdf manuscript and submit the paper to English proof-reading.
Please note that substantial changes have already been made in the manuscript since its submission, notably english improvement and clarification of the material and methods.
Best regards,
Léa Terray, on behalf all authors
-
AC2: 'Reply on RC2', Léa Terray, 14 Apr 2022
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 781 | 239 | 50 | 1,070 | 97 | 40 | 46 |
- HTML: 781
- PDF: 239
- XML: 50
- Total: 1,070
- Supplement: 97
- BibTeX: 40
- EndNote: 46
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1