the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 May 2026
How abruptly did the Holocene Green Sahara end?
Abstract. Discussion about abrupt changes in the Sahara at the end of the African Holocene Humid Period has generally remained qualitative. Here we use a quantitative approach to characterize abruptness of vegetation changes in a transient simulation of the MPI (Max Planck Institute) Earth System model and proxy records (dust flux and pollen data) for the last 8000 years. In the simulations, we find the strongest long-term increase in simulated bare surface in a north-west to south-east oriented strip of some 400 km to 800 km in north-south direction between approximately 20°N in the west and 15°N in the east. Grid boxes in this region reveal abrupt, step-like transitions with changes in simulated bare surface, or conversely, in the total vegetation cover, which occur on average some 4 times faster than the approximately linear trend in the orbital forcing. The transitions in simulated grass and tropical tree plant functional types, i.e., in the type of the vegetation cover, appear to be up to twice as abrupt and up to twice as fast as the transition in the total vegetation cover. The reconstructed dust flux into the Atlantic shelf area at around 20°N, which we interpret as spatially aggregated changes in the openness of the landscape in the western Sahara, reveal a step-like transition with a much stronger abruptness than the simulated transition in bare surface in this region. The pollen records, however, indicate a rather gradual increase in desert-like conditions (with only one exception), but in many records, abrupt and fast transitions in Guinean, Sudanian and Sahelian phytogeographical groups occur. Our quantitative analysis thus confirms earlier propositions that there was no “collapse” of the green Sahara and that gradual changes in some proxy records and abrupt shifts in others are not necessarily contradictory. Our analysis also suggests that a gradual expansion of the Sahara during the mid-Holocene may have been accompanied by abrupt changes within the ecosystems.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Climate of the Past.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(4046 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 16 Jul 2026)
- RC1: 'Comment on egusphere-2026-2587', Anonymous Referee #1, 19 Jun 2026 reply
-
RC2: 'Valuable and comprehensive manuscript; could use a little streamlining', Jed Kaplan, 29 Jun 2026
reply
Review of
How abruptly did the Holocene Green Sahara end?
By Martin Claussen et al.
This manuscript describes a study to investigate the end of the “Green Sahara” period in the middle Holocene. First noticed by explorers of the Sahara nearly a century ago, research has demonstrated that much of the area of northern Africa that is currently occupied by arid and hyper-arid deserts was occupied by grasslands, shrublands, and savannas, and lakes and wetlands during the mid-Holocene. This transition from vegetated and watered surfaces to extreme desert represents one of the greatest land cover changes over the preindustrial Holocene and in terms of areal extent would be second only to the waxing and waning of the ice sheets on Quaternary timescales. While it is well understood that the ultimate driver of land cover change in the Sahara is orbital forcing, mainly the precessional cycles, that slow and relatively linear effect does not correspond well with paleoenvironmental reconstructions that suggested land cover change in the Sahara may have been abrupt. An abrupt end to the Green Sahara period was initially posited by analyses of the dust record in Atlantic Ocean marine cores off the coast of West Africa. In the more than 25 years since these records were published, a wide range of studies have attempted to understand the timing and extent of the end of the Green Sahara, not least in part because it is one of the best examples we have of a rapid change in land cover in response to slow forcing, which could contain lessons for the future about the stability of terrestrial ecosystems in the context of the buildup of anthropogenic greenhouse gases in the atmosphere.
In this manuscript, Claussen et al. describe the most comprehensive and quantitative study to-date to understand the spatial and temporal signature of the end of the Green Sahara period based on paleoenvironmental archives and, using an earth system model, investigate the mechanisms that caused that dramatic land cover change. Overall, this is a fascinating study and a detailed, and well-presented manuscript that will certainly be of interest and value to the community. There are a few things that could be done to make this manuscript easier to read and therefore more impactful, but overall it should be suitable for publication with minor revisions.
General comments
While I really appreciated this manuscript, I found it hard to read in many places and was distracted by the extensive use of symbology, dense text especially in the methods and results sections, and some figures that were initially confusing. These issues can be easily addressed with minor edits. Anything that could be done to streamline the text in sections 2 and 3, and avoid the use of symbols and acronyms, would make the manuscript text easier to read. See my specific comments below for further details.
One of the reasons why this manuscript is dense and difficult to get through is because of the use of many symbols. Maybe it would be more intuitive and easier to read quickly if these variables were given descriptive names. For example, the authors could simply use the label “vegetation change” or “land cover change” in the main text instead of the symbol ΔXf. Alternatively, or as well, all the symbols along with their names, units, and a short description could be included in a small table instead of buried in the manuscript text, perhaps as part of section 2.5.
Furthermore, the A, O, and R values are critical to the presentation of the manuscript. Do these variables have units associated with them or can they be interpreted in terms of physical units? This should be noted.
Specific comments
Line 80
Following radiocarbon convention of BP time being indexed to 1950 CE, 8000 BP is only approximately 6000 BCE and so the word “approximately” should be added here. 8ka is actually equivalent to 6051 BCE.
Furthermore, the use of negative values for time (-8 ky and later in the paper -4 ky) is unconventional and unnecessary. The abbreviation “ka” means “years ago” and so it would be easier for readers to immediately grasp your message if throughout the text that convention was followed, e.g., “8 ka” or “4 ka”.
Line 96
Although it may not make a big difference for the current study, I find it curious that “a delayed development of bare surface is assumed with a time scale of 50 years”. In ecosystems dominated by herbaceous vegetation, bare surface can develop seasonally, and it certainly takes much less than 50 years for a surface to transition from a vegetated to unvegetated state. Given that the timescale for the end of the Green Sahara is on a few centuries (see discussion around lines 520-530), it would actually make sense that, around the time of the transition to drier conditions, land cover “flickered” between years with herbaceous vegetation cover and years that were mainly bare ground during the transition. The authors should acknowledge the limitations of including this unrealistic boundary condition in the vegetation model and in the discussion on uncertainties comment on how this limitation could affect the interpretation of the model results and the timescale for the simulated abruptness to the end of the Green Sahara.
Line 187
Please include a scientific name for “Tiger bush”.
Figure 2
It would be easier to interpret, particularly if someone extracts this figure, e.g., to include in a presentation, if units were provided for each of the scale bars. There is a lot of information in this figure, and I didn’t immediately notice that the color bar in panel (a) is inverted compared to panels (b) and (c). Furthermore, it’s not immediately apparent what the 0-6 and 1-7 scales represent. I would almost suggest trying to break this figure into three separate figures, to get a little more space and put the common color bars at the bottom and make the key results easier to immediately appreciate.
Line 328-330
It is not surprising that the change in simulated bare surface fraction is not as abrupt as the dust flux reconstructions because most of the dust comes, even at present, from playas that are not explicitly simulated by the model. This fact is noted in the discussion, but it could also be mentioned in the introduction and methods that the model does not include lake and wetland dynamics.
Lines 383-385
Again, it is not surprising that less change is observed in pollen records compared to dust flux records. This may of course be because non-vegetated surfaces produce no pollen at all and so unless one considered pollen accumulation rate as opposed to taxonomic fractions, one might not expect an increase in bare ground fraction to make a difference in the pollen record that is immediately visible in the pollen record.
Line 450
It’s also not surprising that dust fluxes increase before bare surface fraction since, as noted above, most of the dust is coming from playas, which are not included in the modeling experiments. We might expect the dynamics of shallow playas such as those in the Lake Chad-Bodélé Basin to have been much more sensitive to climate variability than the native grasses, shrubs, and trees of the Sahel and Sahara that are well adapted to seasonal and internannual drought.
Figure 7
I found this figure very nice and easy to understand, but as noted above, I would prefer to see standard “ka” notation instead of using negative numbers for time.
Line 506-508
Finally, it is explained here that lake dynamics are undoubtably of importance for understanding the dust record of the Sahara. This information should be included in the introduction and signposted to the reader that these lake dynamics are not included in the model simulations.
Line 545-546
I understand that the red dots were used in the analysis and not the gray ones, but I had to re-read the figure caption several times because I thought these different colors represented terrestrial and marine records, respectively. Consider re-wording for clarity.
Figure A4
Again, it should be noted that the color scale is inverted for bare surface compared to the other panels. Consider using a different color scheme for these panels to make it more immediately obvious.
Appendix B
I really liked this deep-dive into the vegetation model behavior and results.
Reviewed by Jed O. Kaplan
Citation: https://doi.org/10.5194/egusphere-2026-2587-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 221 | 103 | 15 | 339 | 9 | 13 |
- HTML: 221
- PDF: 103
- XML: 15
- Total: 339
- BibTeX: 9
- EndNote: 13
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of “How abruptly did the Holocene Green Sahara end?” by Claussen et al.
This manuscript provides a highly rigorous quantitative framework to evaluate the termination of the African Holocene Humid Period (AHHP). The authors successfully move the long-standing debate from a qualitative discussion of "collapse vs. gradual decline" to a standardized, metric-driven assessment. The integration of transient simulations from the MPI-ESM with carefully aggregated pollen groups and marine dust flux records is solid. By applying a changepoint detection method to linear regressions, the authors effectively demonstrate that the end of the Green Sahara was characterized by a complex spatial and temporal mosaic: a gradual expansion of desert areas on a macro scale, accompanied by abrupt, step-like transitions within specific PFTs and regional ecosystems. The discussion of the limitations of the single model study (such as the challenge of representing savanna biomes in JSBACH), and the uneven distribution of proxy data availability provides a clear path forward. The manuscript is a pleasure to read, and my comments below are offered as minor (and mostly technical) suggestions to further polish an already outstanding paper.
1. As the authors noted, the complete omission of the Sahelian group from the model comparison—due to JSBACH's known inability to accurately simulate savanna ecosystems—is a significant limitation. Given that the Sahelian transition is central to the aridification of the Sahara, the authors should explicitly outline how the lack of a savanna biome representation might skew the modeled rapidity of the bare-surface expansion. If grass and trees cannot properly co-exist as savanna in the model, does the transition to bare soil appear artificially abrupt?
2. Orbital forcing vs. internal variability. The definition of abruptness relies on changes occurring faster than the linear trend in orbital forcing. While this is a solid baseline, it would be beneficial to elaborate on the potential role of unforced, centennial-scale internal climate variability. Does the MPI-ESM transient simulation show periods where internal variability can coincidentally push the vegetation over a bioclimatic threshold or beyond the defined abruptness threshold, accelerating the transition independently of the orbital decline?
3. To restrict the changepoint detection a priori to a single transition in linear regression is a logical choice to isolate the primary step-like termination of the AHHP. However, some of the highly variable proxy records (and perhaps some regional grid boxes in the transient simulation) might naturally exhibit multi-stage transitions or "flickering" before settling into a desert state. The authors might consider adding brief discussion on whether releasing this single-changepoint constraint in future studies might reveal secondary and smaller transitional phases.
4. To ensure consistency between the high-resolution simulation output and the proxy data, the authors interpolate pollen and dust records to annual values before applying a 500-year Butterworth filter. Because the temporal resolution of the proxy records varies considerably (e.g., from roughly 50 years to nearly 400 years for pollen, and up to 990 years for the GC49 marine core), the authors might briefly clarify if the interpolation process introduces any minor spectral artifacts or aliasing, particularly for the coarsest records. A minor note assuring the reader that the changepoint detection is insensitive to this specific interpolation step would be a helpful technical addition.
Typos:
L46. adaption -> adaptation
L101. On the northern hemisphere... -> In the …
L133. contritbutors -> contributors
L146. 8000 year -> years
L152. …even more as… -> even more so as
L195. "mineral dust update". Do you mean “uptake”?