Access to this full-text is provided by Springer Nature.
Content available from Nature Communications
This content is subject to copyright. Terms and conditions apply.
Article https://doi.org/10.1038/s41467-024-47921-1
Early warning signals of the termination of
the African Humid Period(s)
Martin H. Trauth
1
, Asfawossen Asrat
2,3
,MarkusL.Fischer
1
,
Peter O. Hopcroft
4
, Verena Foerster
5
, Stefanie Kaboth-Bahr
6
,
Karin Kindermann
7
,HenryF.Lamb
8,9
,NorbertMarwan
10
, Mark A. Maslin
11
,
Frank Schaebitz
5
& Paul J. Valdes
12
The transition from a humid green Sahara to today’s hyperarid conditions in
northern Africa ~5.5 thousand years ago shows the dramatic environmental
change to which human societies were exposed and had to adapt to. In this
work, we show that in the 620,000-year environmental record from the Chew
Bahir basin in the southern Ethiopian Rift, with its decadal resolution, this one
thousand year long transition is particularly well documented, along with
20–80 year long droughts, recurring every ~160 years, as possible early
warnings. Together with events of extreme wetness at the end of the transi-
tion, these droughts form a pronounced climate “flickering”, which can be
simulated in climate models and is also present in earlier climate transitions in
the Chew Bahir environmental record, indicating that transitions with flicker-
ing are characteristic of this region.
The mid-Holocene climate transition from predominantly wet to dry
conditions in tropical and subtropical northern Africa is the most
dramatic example of a climate tipping point during the present inter-
glacial. Climate tipping points occur when small perturbations in the
forcing mechanism trigger a large, nonlinear response from the sys-
tem, moving climate to a different future state generally accom-
panied by dramatic consequences for the biosphere1–4.Themid-
Holocene climate transition underscores that the much-touted sta-
bility of the Holocene climate and its beneficial effects on the evo-
lution of human societies do not hold true, at least not for the low
latitudes. In fact, the transition from the African Humid Period (AHP),
and with it the so-called Green Sahara phase, 15–5 kiloyears before
present (kyr BP) to pronounced aridity after ~5 kyr BP led to sig-
nificant changes in human environments in large areas of northern
Africa3,5–7.
The mid-Holocene climate transition was caused by the decrease
in solar radiation in the Northern Hemisphere driven by the Earth’s
climate precession3,8. This change occurred slowly, as the sinusoidal
period of precession is ~23 kyrs, whereas the more rapid quasilinear
change in solar irradiance occurred during a quarter period of pre-
cession (~6 kyrs) between 9 and 3 kyr BP9. The response of the climate
system to the 7–8% decrease in solar irradiance within 6 kyrs wasmuch
more rapid, perhaps faster than ~200 years in western Africa, but up to
1000 years in eastern Africa3,7–9. The main reason suggested for the
rapid termination of the AHP is the positive feedback between the
monsoon and vegetation that amplifies the comparatively small
changes in external forcing5.
These rapid environmental changes had a strong impact on
humans in northern Africa, as their preferred habitats of grasslands,
open forests, and lakes disappeared10. They responded to increasing
Received: 4 October 2023
Accepted: 12 April 2024
Check for updates
1
University of Potsdam, Institute of Geosciences, Potsdam, Germany.
2
Botswana University of Science and Technology, Department of Mining and Geological
Engineering, Palapye, Botswana.
3
Addis Ababa University, School of Eart h Sciences, Addis Ababa, Ethiopia.
4
University of Birmingham, School of Geography,
Earth & Environmental Sciences, Birmingham, United Kingdom.
5
University of Cologne, Institute of Geography Education, Cologne, Germany.
6
Freie Uni-
versität Berlin, Institute of Geological Sciences, Berlin, Germany.
7
University of Cologne, Institute of Prehistoric Archaeology, Cologne, Germany.
8
Aber-
ystwyth University, Department of Geography and EarthSciences, Aberystwyth, UK.
9
Trinity College Dublin, Botany Department, S chool of Natural Sciences,
Dublin, Ireland.
10
Potsdam Institute for Climate Impact Research, Member of the Leibniz Association, Potsdam, Germany.
11
University College London,
Geography Department, London, UK.
12
University of Bristol, Bristol Research Initiative for the Dynamic Global Environment, School of Geographical Sciences,
Bristol, UK. e-mail: trauth@uni-potsdam.de
Nature Communications | (2024) 15:3697 1
1234567890():,;
1234567890():,;
Content courtesy of Springer Nature, terms of use apply. Rights reserved
aridity by retreating to regions with better water availability, such
as mountain refuges, oases and the Nile valley11,12. Only a few
settlements dated to the AHP have been recorded in the Nile valley
so far11. Due to marshy areas with frequent, extensive flooding,
wild animals (e.g., crocodiles) and disease-bearing mosquitos,
the Nile valley may have been rather unattractive for settlement
during this period11–14. With the onset of drier conditions at the
end of the AHP, the Nile valley developed into a more suitable human
habitat, with favorable conditions for farmers and livestock
keepers, and for the subsequent development for a more complex
society11,12. In this respect, the end of the AHP is an example of both
the negative and positive effects of a climate tipping point on early
societies6,11.
The mid-Holocene climate tipping point in tropical and sub-
tropical northern Africa has been the subject of much research2–4.This
is because there is a comparatively simple but nonlinear relationship
between the cause (orbital forcing) and the accelerated response of
the monsoon system. It is also much studied because current human-
induced climate change may reverse this climate tipping point; mod-
eling results suggest that the green parts of the Sahel are spreading
northward3. Recent literature distinguishes two major types of tipping
points according to the nature of the cause and response of the climate
system3,15. Tipping points of the first type are characterized by critical
slowing down and a decreasing recovery from perturbations near the
transition16,17. When the noise level is high, as in the African Monsoon
System, however, we encounter a tipping point of a different type,
namely a tipping accompanied by flickering between two stable states
near the transition2–4,17.
The two types of tipping points differ in the nature of the early
warning signals, which are increasingly becoming the focus of research
as they are particularly important to understand in order to predict
possible future human-induced climate tipping points16,17.While
slowingdowninthefirst type of tipping point leads to a decrease in
variability, autocorrelation and skewness, flickering in the second type
leads to exactly the opposite, and, in case of doubt, to the impending
tipping point not being recognized16,17. It is therefore important that
the underlying processes are faithfully captured by models, decisively
influencing the credibility of model-based predictions of abrupt
changes under future climate forcing.
For the African Monsoon System, flickering prior to transition was
recently predicted by a modeling study using HadCM315, whereas
records of past climate changes are of too low resolution to compre-
hensively validate these model simulations16,17. These studies indicate a
strong positive feedback between vegetation cover and precipitation,
caused by both radiative and hydrological effects. First, the darker
vegetation enhances solar energy absorbed fueling monsoonal type
circulation onto land and hence precipitation, and second, vegetation
as opposed to bare soil enables more efficient moisture-recycling back
to the atmosphere, thus providing a pair of self-reinforcing feedbacks.
In HadCM3, the albedo effect is stronger than the moisture effect by
about a factor of three, and the flickering arises as a result of the
approach to a critical value of external forcing of the climate system, in
this case the gradually declining summer insolation during the Holo-
cene. As the threshold is approached, small perturbations induced by
the simulated variability in the model can cause the model to flicker
between states15.
In this work, we present a detailed statistical analysis of several
wet–dry transitions in the 620,000-year environmental record from
the Chew Bahir basin in the southern Ethiopian Rift. With its decadal
resolution, the 1000-year-long termination of the AHP is particularly
well documented, along with 20–80-year-long droughts, recurring
every ~160 years, as possible early warnings. Together with events of
extreme wetness at the end of the transition, these droughts form a
pronounced climate “flickering”, which can be simulated in climate
models and is also present in earlier climate transitions in the Chew
Bahir environmental record, indicating that transitions with flickering
are characteristic of this region.
Results
The Chew Bahir record of tipping points with precursors
The paleohumidity record from the Chew Bahir basin in the southern
Ethiopian Rift documents the climate history of eastern Africa for the
past ~620 kyrs18 (Fig. 1). The climate in this part of Africa is determined
by an interaction of several air streams and convergence zones, which
—in combination with the varied topography, large lakes, and the
nearby Indian Ocean—result in a complex spatiotemporal distribution
of precipitation19. The dual passage of the tropical rain belt results in a
bimodal distribution of rainfall in the Chew Bahir region today,
whereby moisture reaching the Ethiopian highlands comes from the
Mediterranean and Red Sea (55%), and from the Indian Ocean (31%)20.
On annual to decadal time scales, the intensity of rainfall is related to
the east–west adjustments in the Walker circulation associated with
the Indian Ocean Dipole19. During the African Humid Period (15–5kyr
BP), the Chew Bahir basin was filled with a large freshwater lake up to
the ~45 m overflow level to the Turkana basin18.
In the years from 2009 to 2014, we recovered sediment cores, up
to ~278 m long, from the southern part of the Chew Bahir basin9,18,21,22
(Supplementary Table 1). High potassium (K) content in the sediment,
determined by micro X-ray fluorescence (μXRF) scanning, was pre-
viously shown to be a reliable proxy for aridity in the Chew Bahir
basin9,21,23,24. The major control on K concentrations is the hydro-
chemistry of the paleolake and porewater. These exert a direct control
on the degree of authigenicmineral formation in thesediment, such as
low-temperature illitization of smectites and analcime formation dur-
ing episodes of higher alkalinity and salinity in the closed-basin lake
resulting from a drier climate23–26. The progressive authigenic
K-fixation in smectites with increasing evaporative conditions was
found to be further enhanced by the excess in the octahedral layer
charge that is caused by Al-to-Mg substitutions in clay minerals with
more alkaline and saline lake conditions23,24.
The decadal resolution of this proxy record provides an oppor-
tunity to examine the termination of the AHP and possible early
warning signals18,23. Based on six well-dated short sediment cores
(9–19 m, <47kyr BP) and two long cores (292.87 m, <620 kyr BP) we
studied the climate transition at ~5.5 kyr BP in detail, and similar
transitions, including possible early warning signals before the first
known occurrence of Homo sapiens on the African continent at
~318 kyr BP27. The Chew Bahir record is also particularly attractive
because it is located close to the Ethiopian Plateau where rainfall feeds
the sources of the Nile and other large rivers. The Chew Bahir record
thus provides a high-resolution reconstruction of hydrological fluc-
tuations that controlled living conditions along the Nile, and may
provide insights into cultural innovation in this region21.
At the end of the AHP in the short cores from Chew Bahir, we
observe at least fourteen dry events, each 20–80 years long and
recurring at 160 ± 40 years intervals (Fig. 2a). These dry events, inter-
preted as possible precursors of an imminent tipping point, would
have allowed a prediction of climate change in the Chew Bahir basin at
that time. Defining the actual tipping point in the r ecord is difficult due
to the presence of noise, but perhaps not crucial in assessing the
transition and possible early warning signals (see “Methods”). Later in
the transition, after ~6 kyr BP, seven wet events occur in addition to the
dry events, with similar duration and recurrence rate. These high-
frequency wet–dry extreme events represent a pronounced flickering
in line with recent modeling results15. Interestingly, the older sediment
record at Chew Bahir shows several transitions that are very similar to
the termination of the AHP (Supplementary Fig. 3 and Supplementary
Table 2). For example, the transition between ~382 and 376 kyr BP
seems in its evolution extremely similar to the termination of the AHP
and possibly points towards similar dynamics (Fig. 2b).
Article https://doi.org/10.1038/s41467-024-47921-1
Nature Communications | (2024) 15:3697 2
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Validation of the tipping points in recurrence plots
The use of nonlinear methods, such as recurrence plots (RPs) and
recurrence quantification analysis (RQA) reveals clear similarities in
the dynamics contained in the time series (Fig. 2c–f and Supplemen-
tary Fig. 6). RPs are graphical displays of recurring states of a system,
calculated from the (e.g., Euclidean) distance between all pairs of
observations, (if required) within a cutoff limit (or threshold). A thre-
sholded RP displays recurring states with a distance below the
threshold value as black dots, while states above the critical distance
are displayed as white dots. An unthresholded RP, not designated as
such in some fields, is simply a pseudocolor plot of the distances
between pairs of observations, therefore simply a graphical repre-
sentation of the distance matrix. RQA uses measures of complexity for
a quantitative evaluation of the RP’s small-scale structures28,29.Among
these, the recurrence rate (RR), indicated by the density of black dots
in the RP, describes the propensity of the system to recur in a parti-
cular time period28–30. The ratio of the recurrence points that form
diagonal structures (of a minimum length) is a measure for deter-
minism (DET)ofthesystem
28–30.
ExploringtheRP/RQAresultsoftheKrecordbetween9and3kyr
BP reveals two major square blocky features in the RP, connected at
~5.5 kyr BP which coincides with the inflection point in the K curve
(Fig. 2c). The two blocks differ significantly in their internal structure.
The block between 5.5 and 3kyr BP shows a very high density of
recurrence points, which is reflected in high values of RR and DET.In
contrast, the block between 9 and 5.5 kyr BP is, except for its older part
between 9 and 8 kyr BP, characterized by both vertical and horizontal
lines, representing episodes of stability (both wet and dry) interrupted
by a series of extremely dry events, indicated by white vertical lines in
the RP. The interval between 8 and 5.5 kyr BP also shows remarkable
diagonal features suggesting a cyclic recurrence of droughts in the
Chew Bahir basin within a period that had a generally wet climate. The
oscillating climatic conditions are reflected in higher RR and DET
values indicating a relatively high predictability of climate, but much
lower than before 8 kyr BP and after 5kyr BP, both being episodes of
relative stability and predictability, with extremely high RR and DET
values close to one.
Examining the RP/RQA results of the modeled precipitation
from a transient climate model simulation of the Holocene covering
10–0 kyr BP reveals very similar structures despite the differences
due to the different character of the climate variable (i.e., sedimen-
tary K concentrations versus simulated precipitation) (see “Methods”
for more information about modeling) (Supplementary Fig. 6). There
is also a blocky structure in the RP, but it looks different in detail
because of the different course of the K curve and the modeled
curve. The most striking similarity to the K curve from the Chew
Bahir, apart from the rapid transition between 7 and 4 kyr BP, is the
occurrence of diagonal structures in the range of the transition.
Furthermore, it is striking that the spacing of the diagonals is also
similar at about 150 years, which corresponds to the recurrence of
droughts in the Chew Bahir. Thus, despite the differences in the
characteristics of the climate variables, it looks like the model has
captured a key feature of the tipping point at the termination of the
AHP, including flickering.
Chew Bahir
Lake Tana
Lake Turkana
Indian Ocean
Red Sea
Nile
Gulf of Aden
Area >1,000 m
above sea level
Tectonic fault
Med Sea
0°N
10°N
20°N
30°N
40°N 50°E40°E30°E
Ethiopian Plateau
02550 km12.5
Quaternary
Alluvial, fluviatile and lacustrine sediments
Neogene
Basalts, rhyolites, trachytes, phonolites, ignimbrites
Paleogene
Basalts, rhyolites, trachytes, tuffs, ignimbrites
Paleozoic and Precambrian
Granites, syenites, diorites and gabbros
Biotite and hornblende granites and diorites
Metasedimentary gneisses
Muscovite-biotite granitoid gneisses
Layered gneisses and amphibolites
Hornblende-quartz-feldspar gneisses
Magnetite-quartz-feldspar gneisses
Ashkare
Aluma
Hammar
Range
Omo
Turkana
Basin
Kino Sogo
Fault Zone
Teltele
Plateau
Bala
Chamo
Basin
Chew
Bahir
ETHIOPIA
KENYA
6°N
5°N
4°N
37°E 38°E
CB02
CB03
CB01
CB04 CB05
CB06
CHB14-2
CHB14-1
ba
Fig. 1 | Study area and localities of the lake cores. a Map of northeastern Africa
and adjacent areas showing the Ethiopian Plateau (in gray), the Ethiopian rift
(marked with thin black lines), the Chew Bahir basin (4°45'40.55''N 36°46'0.85''E,
~500 m above sea level) and the river Nile with its two tributaries. Coastline and
river polygons from the Global Self-consistent, Hierarchical, High-resolution Geo-
graphy Database (GSHHG)68.Topography from the1 arc-minute globalrelief model
of the Earth’ssurface(ETOPO1)
69.bGeological map of the Chew Bahir basin,
showing the four generalized rock types: Quaternary rift sediments, Neogene
and Paleogene rift volcanics, and Paleozoic–Proterozoic basement, and the loca-
tion of the short cores CB01–06, the intermediate core CHB14-1 and the longcores
CHB14-2Aand 2B. Compilation based on Omo River Project Map41, Geological map
of the Sabarei Area70, Geological map of the Yabello Area71, and Geological map of
the Agere Maryam Area72.
Article https://doi.org/10.1038/s41467-024-47921-1
Nature Communications | (2024) 15:3697 3
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Comparing the RP/RQA results of the K record between 9 and
3 kyr BP with those of older transitions, the one between
~382 and 376 kyr BP looks like an almost perfect replica of the termi-
nation of the AHP (Fig. 2). The RP of the transition at ~382–376 kyr BP
shows the same blocky features, connected at ~379 kyr BP, and the
diagonal lines,although not quite as clearly as in the RP of the K curve
between 9 and 3 kyr BP, due to the cyclic recurrenceof droughts in the
range of the transition (Fig. 2c, d). Also, the curves of the RR and DET
are very similar, with higher RR and DET values indicating a relatively
high predictability of climate, but much lower than before 381.5kyr BP
and after 378 kyr BP, both being episodes of relative stability and
predictability, with extreme RR and DET values (Fig. 2e, f). This is
noteworthy considering thatthe older transition at 202.86 meters core
depth (mcd) is documented in a core obtained with a different coring
technique, and whose K concentration was measured with a different
µXRF scanner. It also confirms K concentration as a very reliable cli-
mate proxy, even 379 kyr BP before today and, considering older
similar transitions, before that (Fig. 2, Supplementary Fig. 3, and
Supplementary Table 2). The similarity of the two transitions in
duration and structure further falsifies Wright’shypothesis
31 that
humans, through pastoralism and agriculture, influenced the rate and
structure of the climate transition in the mid-Holocene, at least for the
region around the Chew Bahir. The transition at ~382–376 kyr BP is
quasi-natural (without anthropogenic overprints) because larger
populationsofhumans(namelyH. sapiens)wereabsentfromthe
region, in contrast to the transition at 9–3kyrBP.
Discussion
The flickering wet–dry events prior to major climate transitions, which
seem to recur throughout the ~620kyr Chew Bahir record, combined
with the HadCM3 modeling results presented here, confirm the exis-
tence of precursor events prior to a tipping point, previously predicted
only in theory. Although the Chew Bahir record offers what appears to
be the clearest example of flickering, there are other examples, at least
at the end of the AHP. For example, log(Ti/Ca) data from core MD04-
2726 from the Nile delta deep sea fan, interpreted as a record of Nile
flood events, show extreme droughts with similar duration and
recurrence but about 1550 years earlier than at Chew Bahir32.Other
0
0.2
0.4
0.6
0.8
1
Recurrence Rate
0
0.2
0.4
0.6
0.8
1
Recurrence Rate
Determinism
Recurrence rate
Determinism
Recurrence rate
Age (kyr BP)
Age (kyr BP)
Age (kyr BP)
Age (kyr BP)
0.5
3
3
4
4
5
5
6
6
7
7
8
8
9
9
376
376
377
377
378
378
379
379
380
380
381
381
382
382
Age (kyr BP) Age (kyr BP)
3 4 5 6 7 8 9 376 377 378 379 380 381 382
0.6
0.7
0.8
0.9
1
Determinism
-2
-1
0
1
2
Potassium (standardized)
0.5
0.6
0.7
0.8
0.9
1
Determinism
-2
-1
0
1
Potassium (standardized)
ba
cd
ef
Transition with Flickering Transition with FlickeringStable dry
Pyramids
of Giza
Stable wet Stable dry Stable wet
7.7 ka
drought
~14+2 droughts
~7 wet events
8.2 ka
drought
~13+2 droughts
drought
drought
~10 wet events
Diagonal lines
depicting deterministic
system dynamics
Diagonal lines
depicting deterministic
system dynamics
Fig. 2 | Wet–dry transitions in the Chew Bahir during the past ~620 kyrs,
recurrence plots, and recu rrence quantification analysisresults. a,bRecords of
relative aridity in the Chew Bahir basin, southern Ethiopia, between (a)9–3kyrBP
and (b)382–376kyr BP interval. During the past ~620 kyrs, climate in northeastern
Africa passed multiple tipping points, for example, at ~7kyr BP and ~380 kyr BP,
respectively. After passing the tipping points, climate entered ~0.9–1.5kyr long
transitions from stable wet to stable dry climate, as described by nonlinear least-
squares fitting a ramp function(dotted purpleline) to the K curve from ChewBahir.
Both transitions are marked by pronounced flickering between the two extremes,
wet (blue arrows) and dry (red arrows). c,dRecurrence plots (RPs) showing
remarkable diagonalfeatures suggesting a pronounced flickering in the Chew Bahir
basin after the tipping points, and e,frecurrence quantification analysis (RQA)
measures recurrence rate (RR)anddeterminism (DET) of the Chew Bahir records
with higher DET values indicating a relatively high predictability of climate, but
much lower than before and after the transitions, both being episodes of relative
stability and predictability, with higher DET values. See the supplementary infor-
mation for more examp les of wet–dry transitions with pronounced flickeringin the
~620 kyr long climaterecord of the Chew Bahir, as wellas for a detailed description
and an interpretation of the RPs and RQA results.
Article https://doi.org/10.1038/s41467-024-47921-1
Nature Communications | (2024) 15:3697 4
Content courtesy of Springer Nature, terms of use apply. Rights reserved
data sets showing excursions at the termination and flickering during
the AHP include a lake record from Lake Dendi33, a lake record from
Lake Abiyata34, and a record from Congo stalagmites35.
The discovery of flickering prior to and during the transition from
wet to dry conditions in Africa has significant implications for inter-
preting the relationship between climate change, cultural develop-
ments, and human migrations in those regions. After the hyperarid
period coinciding with the high-latitude Younger Dryas (12.9–11.7 kyr
BP) stadial, increased precipitation turned northeast Africa into a
savannah-like environment with mean annual rainfall of about
50–150 mm36. Recolonization of this formerly desert area by small
hunter-gatherer groups was probably a very slow process taking place
over more than 1000 years when foraging ranges were gradually
expanded37. During their seasonal rounds, ephemeral waterholes were
one of the most important resources for these highly mobile groups,
other than areas with permanent water access38. Living conditions
remained harsh and must be assumed still to have been arid with high
climatic variability, and patchy, unpredictable access to surface water.
Therefore, people had to be able to adapt to changing and adverse
living conditions, for instance, by flexible seasonal resource manage-
ment, variable mobility patterns, and food sharing as risk minimization
strategies12.
By 7.3 kyr BP, the northeast African summer rain began to retreat
southward in what was probably a gradual, rather than a stepwise
shift9,11. During this period, hunter-gatherer groups began to leave
areas far from permanent water availability. During the main flickering
period with about fourteen short droughts, areas outside the Nile
valley, the oases, and the few mountain refugia were already
uninhabitable12. At ~7 kyr BP, many areas were abandoned due to cli-
matic instability. We suggest that environmental instability caused by
the climate flickering led to the abandonment of ancestral habitats, as
conditions became too unpredictable to allow either nomadic or
sedentary lifestyles. The discussion has too long centered on how fast
these types of transitions have occurred (abrupt vs. gradual) and what
stressthis uni-directionalchange exertedon humans9.Theflickering of
climate, however, would have had a much more dramatic impact than
the slow climate transitions spanning tens of generations. Confirma-
tion of the existence of flickering events as precursors several times in
the past, as shown by our data from Chew Bahir, may also provide
insights into possible early warning signals for future large-scale cli-
mate tipping points.
Methods
Chew Bahir setting, materials and climate proxy
Chew Bahir (4.1–6.3°N, 36.5–38.1°E; WGS 84; Fig. 1) is a tectonic basin
in the southern Ethiopian Rift with a ~32,400 km2catchment area. In
the west, it is separated from the Omo Basin by the Hammar Range,
which has become known through fossil findings of Homo sapiens39,40.
In the western catchment, mainly Precambrian and Paleozoic granites,
syenite, diorites, and gabbros occur, while in the easterncatchment, in
addition to these rocks, Paleogene and Neogene basalts, trachytes and
other volcanic rocks from the early phase of the formation of the rift
dominate41,42. During the rainy seasons, the Chew Bahir mudflat is
covered by water, mostly from the Weyto and Segen rivers entering the
basin fromthe north. Forming a terminal sink for eroded material from
thecatchmentsincetheNeogeneformationoftheChewBahir,the
basin today contains more than 5 km of sediment18.
In the years from 2009 to 2014, we recovered sediment cores of
different lengths from the southern part of the Chew Bahir mudflat. In
2009 and 2010, we took six up to 19-m long cores CB01–06 along a
NW–SE transect across the southwestern part of the basin9,18,21,22 (Fig. 1
and Supplementary Table 1). Of these, CB01 and CB02 were recovered
from the margin of an alluvial fan, an area that receives sediment pri-
marily from the Hammar Range and the Weyto River. CB04–06, on the
other hand, were collected further into the center of the Chew Bahir
and are supplied primarily by the Segen River from the north9,21. CB03
from an intermediate location along the transect, contains a fluvio-
lacustrine mix of Weyto and Segen sediments9,21. In mid 2014, we col-
lected the ~40 m CHB14-1 core in the central part of the basin, not far
from CB0522. In late 2014, we collected the longest coresCHB14-2A and
2B, reaching down to 278.58 and 266.38 m depth, respectively18.
The short cores CB01–06andCHB14-1wereshippedtotheU
Cologne for further analysis and storage9,18,21,22.Itisveryimportant
when evaluating micro X-ray fluorescence (µXRF) data and discussing
possible measurement artifacts that the elemental content of CB01
were measured with a molybdenum (Mo) tube as radiation source in
the µXRF scanning with an ITRAX core scanner, whereas cores
CB02–06 were measured with a chromium (Cr) tube which has slightly
different sensitivities towards elemental settings9,18. Due to the higher
sedimentation rate at the basin margin, CB01 has a higher temporal
resolution of a few years at the same spatial resolution of 0.5 cm,which
is why short-term fluctuations such as the 20–80 year droughts are
better represented in this core than in cores CB02–06 and CHB14-19.
The long cores CHB14-2A and 2B were shipped to the US Continental
Scientific Drilling Facility (LacCore) for sampling, analysis and storage,
except for µXRF measurements that were performed at the LacCore-
associated Large Lakes Observatory (LLO)18.Theshortercores
CB01–06 were dated by the radiocarbon method, recalibrated using
IntCal2043 for this work, and a composite age model was developed by
linear and spline interpolation9,18. We used exclusively data from core
CB01 for the analysis of the mid-Holocene tipping point to avoid
artifacts in the analysis from stitching the composite of multiple short
cores9,29. For this reason, the analysis is limited to the 9–3 kyr BP time
interval, which does not show major fluctuations in the spacing of the
data points, including gaps9,29.
Core CHB14-1 was dated by radiocarbon (14C) dating and thermo-
luminescence/optically stimulated luminescence (TL/OSL) age deter-
minations; the age model was found using Bayesian age–depth
modeling22. The long cores CHB14-2A and 2B were dated using 14C,
OSL, and 40Ar/39Ar ages, and tephrochronological data, again using
Bayesian age–depth modeling to develop the RRMarch2021 age
model9,44. Cores CHB14-2A and 2B were spliced together to a common
depth, forming composite core CHB14-2, using visual and physical
sediment properties, resulting in 292.87 m long composite core with a
core recovery of ~90%. The μXRF data sets were subjected to intensive
quality control, which included outlier elimination, correction of off-
sets, and cleanup of duplicate values9,18.
We compare the Chew Bahir data with simulated precipitation
from a transient climate model simulation of Holocene covering
10–0kyrBP
15. This 10,000-year simulation employs the Hadley Centre
coupled model version 3 (HadCM3)45,46. HadCM3 is a coupled three-
dimensional atmosphere–ocean general circulation model with
schemes for sea ice, and dynamic vegetation which is represented
using a tiling of nine different land-surface covers (MOSES 2.1 with
dynamic vegetation represented with TRIFFID, M2.1d)47.Thehor-
izontal resolution of the atmospheric model is 3.75° × 2.5° in
longitude–latitude with 19 unequally spaced vertical levels. In the
ocean the resolution is 1.25° × 1.25° with 20 unequally spaced vertical
levels.
The University of Bristol configuration of this model (HadCM3B-
M2.1d) is comparable to other more recent models in terms of skill in
simulating present-day climatology46.Inrecentworkthreenewcon-
figurations of this were developed, applying mid-Holocene-
constrained parameter updates to the model’s atmospheric convec-
tion (+CONV), vegetation (+VMS) or both (+CONV+VMS). The first
three configurations15: standard (STD-equivalent to the Bristol config-
uration), +CONV, +VMS are unable to reproduce the mid-Holocene
green Sahara convincingly, which is similar to the majority of other
coupled climate models48. In this study we use the fourth configuration
+CONV+VMS (or HadCM3BB-M2.1d)15 because it shows a convincing
Article https://doi.org/10.1038/s41467-024-47921-1
Nature Communications | (2024) 15:3697 5
Content courtesy of Springer Nature, terms of use apply. Rights reserved
greening of the Sahara under mid-Holocene forcing and, when run
transiently across the Holocene, it shows excellent agreement with
precipitation reconstructions from northwestern Africa49 against
which it was not optimized. This version is the only configuration to
display flickering and has recently convincingly reproduced the timing
of multiple gr een Sahara pha ses back to 800 ka BP50.
For the Holocene, the HadCM3BB-M2.1d was forced with changes
in Earth’sorbit
51, the time-varying distribution of ice sheets, land area
and sea level from ICE-6G52,53 and mixing ratios of carbon dioxide,
methane and nitrous oxide as reconstructed from ice cores54–56. Var-
iations in tropospheric aerosols are not included, but volcanic erup-
tions, the total solar irradiance and land use have been analyzed
separately and do not markedly alter the Holocene simulation over
northern Africa15. Volcanic stratospheric aerosol optical depth is pre-
scribed in the model based on the results from HolVol v1 bipolar
reconstruction57 and total solar irradiance has been reconstructed
using cosmogenic isotopes58. In this study, we analyze the climate
simulation forced with Holocene variations in orbit, ice sheets,
greenhouse gases, solar irradiance and volcanic eruptions and using
the paleo-conditioned configuration, HadCM3BB-M2.1d15,59.
Validation of the tipping points in statistical parameters
The K curve of Chew Bahir core CB01 was analyzed section by section
(within ~6-kyr windows) using methods from nonlinear dynamics such
as recurrence plots (RPs) and recurrence quantification analysis (RQA)
(Fig. 2and Supplementary Figs. 4–6). These methods require evenly-
spaced time series, which is why we first converted the data from
composite depth to age using a previously published linear age
model60.Subsequently,thedatawereinterpolatedtoanevenly-spaced
time axis running from 45.358 to 0kyr BP at 10-year intervals using a
shape-preserving piecewise cubic interpolation61 implemented in the
MATLAB function pchip.
Recurrence plots (RPs) are graphical displays of recurring states
of a system, calculated from the (e.g., Euclidean) distance between all
pairs ofobservations, (if required) withina cutoff limit28–30,62.Thevisual
inspection of RPs is often complemented by a recurrence quantifica-
tion analysis (RQA), which uses measures of complexity for a quanti-
tative evaluation of the RP’s small-scale structures28–30. Among these,
the recurrence rate (RR) is measuring the density of black dots in the
RP, describingthe propensity of the system to recur in a particular time
period28–30. Diagonal lines in RPs are diagnostic of predictablebehavior
in time series and can be used to predict future conditions from the
present and past. The ratio of the recurrencepoints that form diagonal
structures (of a minimum length) to all recurrence points is a measure
of determinism (DET)ofthesystem
28–30.
For the analysis of the 9–3 kyr BP record, the K record from the
short cores was embedded in a phase space with a dimension of m=3
and temporal distances of τ= 4 (Supplementary Fig. 4), equivalent to
4 × 10 years =40 years, where 10 years is the resolution of the time
series following a piecewise cubic Hermite polynomial interpolation63.
We use the window size w= 50 and the step size ws = 5 data points of
the moving window to calculate the RQA measures. The size wof the
window corresponds to 50 ×10 years = 500 years and the step size is
5 × 10 years = 50 years. We use a minimum length of 10 points to
compute DET. To compare the RP/RQA-based dynamics in the Chew
Bahir record of aridity and the modeled precipitation record, we
interpolated the modeled record to the same time axis, used the same
embeddingparameterstocreatetheRPsandusedthesamewindow
size to calculate the RQAmeasures (Supplementary Fig.6). Similarities
in the texture of the recurrence plots of both proxy records show that
the embedding provides comparable results with these values.
Determining the duration and start of tipping
We statistically analyzed the K curve from the Chew Bahir for the
amplitude, duration, and midpoint of a climate transitions between 9
and3kyrBPandbetween382and376kyrBPusingthreedifferent
methods: sigmoid fit, ramp fit and change point detection9,64 (Sup-
plementary Fig. 7). First, we applied a change point search algorithm61
to the standardized K record, i.e., the mean K values were subtracted
from the individual K values and then divided by the standard devia-
tion. The algorithm, which has been included in MATLAB as the
findchangepts function, detects change points by minimizing a cost
function over all possible numbers and locations of change points.The
findchangepts function yields the number of significant changes in
the mean, the standard deviation, and the trend of a time series (not
exceeding a maximum number of permissible changes defined by the
user) that minimize the sum of the residual error and an internal fixed
penalty for eachchange. Second, we fitted a sigmoidfunction with four
parameters a,b,c,andd
xfitðtÞ=a+b
1+e
dðtcÞ,for1<t<+1ð1Þ
to the records xiwith 1 < i<nand ndata points using nonlinear least-
squares fitting. The sigmoid function (in its normalized representa-
tion) is a monotonic s-shaped curve, often referred to as a smooth
version of a step function. The sigmoid function is bounded by two
horizontal asymptotes xfit tðÞ!0and1ast!1 and + 1,
respectively. It has a bell-shaped first derivative curve and exactly
one inflection point (parameter c), which can be interpreted as the
midpoint of the transition. In our analysis, we used the function fit
together with fitoptions and fittype included in the Curve Fitting
Toolbox of MATLAB to fit a sigmoid function to the data (Supplemen-
tary Fig. 7). Third, we statistically re-analyzed these records, fitting a
ramp function again with four parameters x1,x2,t1,andt2
xfit tðÞ=
x1,fort≤t1
x1+tt1
ðÞ
x2x1
ðÞ
t2t1,fort1<t≤t2
x2,fort>t2
8
>
>
<
>
>
:
ð2Þ
to the ndata points. The monotonic ramp-shaped curve has two
horizontal pieces and an inclined piece, connected by two abrupt
changes of direction and with a discontinuous first derivative. Both the
sigmoid and ramp functions are widely used to describe transitions in
climatic and environmental conditions as well as the response of the
biosphere61,65. Again, we used the function fittogether with fitop-
tions and fittype included in the Curve Fitting Toolbox of MATLAB
to fit a ramp function to the data (Supplementary Fig. 7). Using the
methods do describe the transitions, we found the following measures
for the transition between 9 and 3kyr BP
changepoints =
5.6950
sigmoidfit=
General model:
sigmoidfit(x) =a+b*(1./(1+exp(-d*(x-c))))
Coefficients (with 95% confidence bounds):
a=0.4223 (0.3727, 0.4719)
b=-2.053 (-2.127, -1.979)
c=5.722 (5.66, 5.784)
d=4.178 (3.238, 5.118)
rampfit=
General model:
rampfit(t) =rampfunction(t,t1,t2,x1,x2)
Coefficients (with 95% confidence bounds):
t1 =5.064 (4.951, 5.177)
t2 =6.37 (6.256, 6.483)
x1 =0.4175 (0.3691, 0.4659)
x2 =-1.623 (-1.674, -1.572)
Article https://doi.org/10.1038/s41467-024-47921-1
Nature Communications | (2024) 15:3697 6
Content courtesy of Springer Nature, terms of use apply. Rights reserved
resulting in the lower and upper knick points of the ramp at
t1= 5.1 kyr BP (corresponding to ~3.71 mcd) and t2=6.4kyr BP (corre-
sponding to ~4.30 mcd) (using a previously published linear age
model60, but with radiocarbon ages recalibrated to IntCal2043). For the
transition between 382 and 376 kyr BP, we found
changepoints =
379.0600
sigmoidfit=
General model:
sigmoidfit(x) =a+b*(1./(1+exp(-d*(x-c))))
Coefficients (with 95% confidence bounds):
a=0.8967 (0.8415, 0.952)
b=-1.9 (-1.985, -1.816)
c=379.2 (379.1, 379.3)
d=3.186 (2.481, 3.89)
rampfit=
General model:
rampfit(t) =rampfunction(t,t1,t2,x1,x2)
Coefficients (with 95% confidence bounds):
t1 =378.4 (378.2, 378.5)
t2 =380.2 (380, 380.3)
x1 =0.8868 (0.8346, 0.939)
x2 =-1.013 (-1.07, -0.956)
resulting in the lower and upper knick points of the ramp at
~380.2 kyr BP (corresponding to ~203.5 mcd in the core) and 378.4 kyr
BP (corresponding to ~197.7 mcd) (using the age model
RRMarch202144). It is tempting to mark the upper knick point t2of the
ramp as the most likely time for the climate tipping. However, the
fitting of a ramp but also other methods for determining the tipping
point66,67 use only static approaches to define a knick point, but can
never determine the actual tipping. Furthermore, the flickering in our
type of tipping point, as well as the presence of noise, makes it difficult
to precisely define a tipping point in a proxy record. The visual
inspector might be tempted to mark the tipping point in the younger
transition at ~5.9 kyr BP, and perhaps at 379.2 kyr BP in the older
transition, i.e., at the last wetevents in each case with an amplitude that
reaches the level of K values in the stable wet phase before the tran-
sition. But perhaps the precise definition of the beginning of tipping is
not crucial. It is much more interesting to see that the climate starts to
flicker well before the actual transition (e.g., more than one thousand
years earlier in our examples) and hence the tipping elements sent out
clear early warning signals in the form of 20–80-year-long extreme
events with clear regularity in recurrence.
Data availability
The data that support the findings of this study are available from
Zenodo with the identifier https://doi.org/10.5281/zenodo.10624471
(https://zenodo.org/records/10624471).
Code availability
TheMATLABcodetorecreatethefigures are available from Zenodo
with the identifier https://doi.org/10.5281/zenodo.10624471 (https://
zenodo.org/records/10624471).
References
1. Lenton, T. M. et al. Tipping elements in the Earth’s climate system.
Proc. Natl. Acad. Sci. USA 105,1786–1793 (2008).
2. Drijfhout, S. et al. Catalogue of abrupt shifts in intergovernmental
panel on climate change climate models. Proc. Natl. Acad. Sci. USA
112, E5777–E5786 (2015).
3. Pausata, F. S. R. et al. The greening of the Sahara: past changes and
future implications. One Earth 2,235–250 (2020).
4. Brovkin, V. et al. Past abrupt changes, tipping points and cascading
impactsintheEarthsystem.Nat. Geosci. 14,550–558 (2021).
5. Claussen, M. Modeling bio-geophysical feedback in the African and
Indian monsoon region. Clim. Dyn. 13,247–257 (1997).
6. Brooks, N. Cultural responses to aridity in the Middle Holocene and
increased social complexity. Quat. Int. 151,29–49 (2006).
7. Shanahan,T.M.etal.Thetime-transgressiveterminationofthe
African Humid Period. Nat. Geosci. 8,140–144 (2015).
8. Kutzbach, J. E. & Street-Perrott, F. A. Milankovitch forcing of fluc-
tuations in the level of tropical lakes from 18 to 0 kyr BP. Nature 317,
130–134 (1985).
9. Trauth,M.H.etal.Abruptorgradual?changepointanalysisofthe
late Pleistocene–Holocene climate record from Chew Bahir,
southern Ethiopia. Quat. Res. 90,321–330 (2018).
10. Drake, N. A. et al. Sedimentary and geomorphic evidence of
Saharan megalakes: a synthesis. Quat.Sci.Rev.276, 107318 (2022).
11. Kuper, R. & Kröpelin, S. Climate-controlled Holocene occupation in
the Sahara: motor of Africa’s evolution. Science 313,803–807
(2006).
12. Riemer, H. & Kindermann, K. In Handbook of Ancient Nubia (ed.
Raue,R.),Ch.II.3(DeGruyter,2019).
13. Macklin,M.G.etal.Anewmodelofriverdynamics,hydroclimatic
change and human settlement in the Nile valley derived from meta-
analysis of the Holocene fluvial archive. Quat. Sci. Rev. 130,
109–123 (2015).
14. Zaki, A. S. et al. Did increased flooding during the Arican Humid
Period force migration of modern humans from the Nile valley?
Quat. Sci. Rev. 272, 107200 (2021).
15. Hopcroft, P. O. & Valdes, P. J. Paleoclimate-conditioning reveals a
North Africa land–atmosphere tipping point. Proc. Natl. Acad. Sci.
USA 118, e2108783118 (2021).
16. Scheffer, M. et al. Early-warning signals for critical transitions.
Nature 461,53–59 (2009).
17. Dakos,V.,VanNes,E.H.&Scheffer,M.Flickeringasanearly
warning signal. Theor. Ecol. 6,309–317 (2013).
18. Foerster, V. et al. 620,000 years of eastern African climate varia-
bility and hominin evolution. Nat. Geosci. 15,805–811 (2022).
19. Nicholson, S. E. Climate and climatic variability of rainfall over
eastern Africa. Rev. Geophys. 55,590–635 (2017).
20. Viste, E. & Sorteberg, A. The effect of moisture transport variability
on Ethiopian summer precipitation. Int. J. Climatol. 33,3106–3123
(2013).
21. Foerster, V. et al. Climatic change recorded in the sediments of the
Chew Bahir basin, southern Ethiopia, during the last 45,000 years.
Quat. Int. 274,25–37 (2012).
22. Viehberg,F.A.etal.EnvironmentalchangeduringMIS4andMIS3
opened corridors in the horn of Africa for Homo sapiens expansion.
Quat. Sci. Rev. 202,139–153 (2018).
23. Foerster, V. et al. Towards an understanding of climate proxy for-
mation in the Chew Bahir basin, southern Ethiopian rift. Palaeo-
geogr. Palaeoclimatol. Palaeoecol. 501,111–123 (2018).
24. Gebregiorgis, D. et al. Modern sedimentation and authigenic
mineral formation in the Chew Bahir basin, southern Ethiopia:
Implications for interpretation of late Quaternary paleoclimate
records. Front. Earth Sci. 9, 607695 (2021).
25. Arnold, G. E. et al. Advanced hyperspectral analysis of sediment
core samples from the Chew Bahir basin, Ethiopian Rift, in the
spectral range from 0.25 to 17 µm. Front. Earth Sci. 9,606588
(2021).
26. McHenry, L. J., Foerster, V. & Gebregiorgis, D. Paleolakes of Eastern
Africa: zeolites, clay minerals, and climate. Elements 19,96–103
(2023).
27. Hublin, J.-J. et al. New fossils from Jebel Irhoud, Morocco and the
pan-African origin of Homo sapiens. Nature 546,289–292 (2017).
Article https://doi.org/10.1038/s41467-024-47921-1
Nature Communications | (2024) 15:3697 7
Content courtesy of Springer Nature, terms of use apply. Rights reserved
28. Marwan,N.,Romano,M.C.,Thiel,M.&Kurths,J.Recurrenceplots
for the analysis of complex systems. Phys. Rep. 438,237–329
(2007).
29. Trauth, M. H. et al. Classifying past climate change in the Chew
Bahir basin, southern Ethiopia, using recurrence quantification
analysis. Clim. Dyn. 53,2557–2572 (2019).
30. Trauth, M. H. et al. Recurring types of variability and transitions in
the 620 kyr record of climate change from the Chew Bahir basin,
southern Ethiopia. Quat. Sci. Rev. 266, 106777 (2021).
31. Wright, D. K. Humans as agents in the termination of the African
Humid Period. Front. Earth Sci. 5, 237134 (2017).
32. Mologni, C. et al. Frequency of exceptional Nile flood events as an
indicator of Holocene hydro-climatic changes in the Ethiopian
highlands. Quat. Sci. Rev. 247,106543(2020).
33. Wagner, B. et al. Holocene rainfall runoff in the centralEthiopian
highlands and evolution of the river Nile drainage system as
revealed from a sediment record from Lake Dendi. Glob. Planet.
Change 163,29–43 (2018).
34. Roubeix, V. & Chalié, F. New insights into the termination of the
African Humid Period (5.5 ka BP) in central Ethiopia from detailed
analysis of a diatom record. J. Paleolimnol. 61,99–110 (2019).
35. Dupont, L. A. et al. Episodic deposition of stalagmites in the
northeastern Democratic Republic of the Congo suggests Equa-
torial humid periods during insolation maxima. Quat. Sci. Rev. 286,
107552 (2022).
36. Neumann, K. Holocene vegetation of the eastern Sahara: charcoal
from prehistoric sites. Afr. Archaeol. Rev. 7,97–116 (1989).
37. Kindermann, K., Kuper, J., Bubenzer, O. & Henselowsky, F. In The
Journeys of Modern Humans from Africa to Europe (eds Litt, T. et al.)
Ch. 1.5 (Schweizerbart Science Publishers, 2021).
38. Bubenzer, O. & Riemer, H. Holocene climatic change and human
settlement between the central Sahara and the Nile valley:
archaeological and geomorphological results. Geoarchaeology 22,
607–620 (2007).
39. McDougall,I.,Brown,F.H.&Fleagle,J.G.Stratigraphicplacement
and age of modern humans from Kibish, Ethiopia. Nature 433,
733–736 (2005).
40. Vidal, C. M. et al. Age of the oldest known Homo sapiens from
eastern Africa. Nature 601,579–583 (2022).
41. Davidson, A. The Omo River Project: Reconnaissance Geology
and Geochemistry of Parts of Ilubabor, Kefa, Gemu Gofa and
Sidamo. Ethiopian Institute of Geological Surveys Bulletin no. 2
(1983).
42. Moore, J. & Davidson, A. Rift structure in southern Ethiopia. Tecto-
nophysics 46,159–173 (1978).
43. Reimer, P. J. et al. The IntCal20 northern hemisphere radiocarbon
age calibration curve (0–55 cal kBP). Radiocarbon 62,725–757
(2020).
44. Roberts, H. M. et al. Using multiple chronometers to establish a
long, directly-dated lacustrine record: Constraining ~600,000
years of environmental change at Chew Bahir, Ethiopia. Quat. Sci.
Rev. 266, 107025 (2021).
45. Gordon, C. et al. The simulation of SST, sea ice extents and ocean
heat transports in a version of the Hadley Centre coupled model
without flux adjustments. Clim. Dyn. 16,147–168 (2000).
46. Valdes, P. J. et al. The BRIDGE HadCM3 family of climate
models:HadCM3@Bristolv1.0.Geosci. Model Dev. 10,
3715–3743 (2017).
47. Cox, P. Description of the TRIFFID dynamic global vegetation model
Hadley Centre. Met Office Technical Note 24,1–16 (2001).
48. Brierley, C. M. et al. Large-scale features and evaluation of the
PMIP4-CMIP6 midHolocene simulations. Climate 16,1847–1872
(2020).
49. Tierney, J. E., Pausata, F. S. R. & deMenocal, P. B. Rainfall regimes of
the Green Sahara. Sci. Adv. 3, e1601503 (2017).
50. Armstrong, E., Tallavaara, M., Hopcroft, P. O. & Valdes, P. J. North
African humid periods over the past 800,000 years. Nat. Commun.
14, 5549 (2023).
51. Berger, A. Long-term variations of daily insolation and quaternary
climatic changes. J. Atmos. Sci. 35,2362–2367 (1978).
52. Argus, D. F., Peltier, W. R., Drummond, R. & Moore, A. W. The Ant-
arctica component of postglacial rebound model ICE-6G C
(VM5a) based on GPS positioning, exposure age dating of ice
thicknesses, and relative sea level histories. Geophys. J. Int. 198,
537–563 (2014).
53. Peltier,W.R.,Argus,D.F.&Drummond,R.Spacegeodesycon-
strains ice age terminal deglaciation: The global ICE-6G C (VM5a)
model: Global glacial isostatic adjustment. J. Geophys. Res. Solid
Earth 120,450–487 (2015).
54. Bereiter, B. et al. Revision of the EPICA dome C CO
2
record from
800 to 600 kyr before present: Analytical bias in the EDC CO2
record. Geophys. Res. Lett. 42,542–549 (2015).
55. Loulergue, L. et al. Orbital and millennial-scale features of atmo-
spheric CH4 over the past 800,000 years. Nature 453,383–386
(2008).
56. Schilt, A. et al. Glacial–interglacial and millennial-scale variations in
the atmospheric nitrous oxide concentration during the last
800,000 years. Quat. Sci. Rev. 29,182–192 (2010).
57. Sigl, M., Toohey, M., McConnell, J. R., Cole-Dai, J. & Severi, M.
Volcanic stratospheric sulfur injections and aerosol optical depth
during the Holocene (past 11 500 years) from a bipolar ice-core
array. Earth Syst. Sci. Data 14,3167–3196 (2022).
58. Vieira,L.E.A.,Solanki,S.K.,Krivova,N.A.&Usoskin,I.Evolutionof
the solar irradiance during the Holocene. Astron. Astrophys. 531,
A6 (2011).
59. Hopcroft, P. O. & Valdes, P. J. Green Sahara tipping points in tran-
sient climate model simulations of the Holocene. Environ. Res. Lett.
17, 085001 (2022).
60. Trauth, M. H. et al. Episodes of environmental stability versus
instability in late Cenozoic lake records of eastern Africa. J. Hum.
Evol. 87,21–31 (2015).
61. Killick, R., Fearnhead, P. & Eckley, I. A. Optimal detection of chan-
gepoints with a linear computational cost. J. Am. Stat. Assoc. 107,
1590–1598 (2012).
62. Eckmann,J.-P.,Kamphorst,S.O. & Ruelle, D. Recurrence plots of
dynamical systems. EPL 4,973–977 (1987).
63. Fritsch,F.N.&Carlson,R.E.Monotonepiecewisecubicinterpola-
tion. SIAM J. Numer. Anal. 17,238–246 (1980).
64. Trauth, M. H. et al. Northern Hemisphere Glaciation, African climate
and human evolution. Quat. Sci. Rev. 268, 107095 (2021).
65. Mudelsee, M. & Raymo, M. E. Slow dynamics of the Northern
Hemisphere Glaciation. Paleoceanography 20,PA4022(2005).
66. Livina, V. N., Ditlevsen, P. D. & Lenton, T. M. An independent test of
methods of detecting system states and bifurcations in time-series
data. Phys. A 391,485–496 (2012).
67. Marwan, N., Donges, J. F., Donner, R. V. & Eroglu, D. Nonlinear time
series analysis of palaeoclimate proxy records. Quat.Sci.Rev.274,
107245 (2021).
68. Wessel,P.&Smith,W.H.F.Aglobal,self-consistent,hierarchical,
high-resolution shoreline database. J. Geophys. Res. 101,
8741–8743 (1996).
69. Amante, C. & Eakins, B. W. ETOPO1 1 arc-Minute Global Relief Model:
Procedures, Data Sources and Analysis (National Geophysical Data
Center, NOAA, 2009).
70. Key, R. M. Geological Map of the Sabarei Area 1:250 000 (Ministry of
Environment and Natural Resources, Mines and Geology, 1988).
71. Hailemeskel,A.&Fekadu,H.Geological Map of the Yabello Area
1:250 000 (Geological Survey of Ethiopia, 2004).
72. Hassen, N., Yemane, T. & Genzebu, W. Geological Map of the Agere
Maryam Area 1:250 000 (Geological Survey of Ethiopia, 1997).
Article https://doi.org/10.1038/s41467-024-47921-1
Nature Communications | (2024) 15:3697 8
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Acknowledgements
Support for the Hominin Sites and Paleolakes Drilling Project (HSPDP)
led by Andrew S. Cohen has been provided by the National Science
Foundation (NSF) grants and the International Continental Drilling Pro-
gram (ICDP). Support for Chew Bahir Drilling Project (CBDP) led by A.A.,
H.L., F.S. and M.H.T has been provided by Germany Research Founda-
tion (DFG) through the Priority Program SPP 1006 ICDP (SCHA 472/13
and /18, TR 419/8, /10 and /16), the CRC 806 Research Project “Our way
to Europe”Project Number 57444011 and the UK Natural Environment
Research Council (NERC, NE/K014560/1, IP/1623/0516). We also thank
the Ethiopian permitting authorities to issue permits for drilling in the
Chew Bahir basin. We also thank the Hammar people for the local
assistance during drilling operations. We thank DOSECC Exploration
Services for drilling supervision and Ethio Der Pvt. Ltd. Co. for providing
logistical support during drilling. Initial core processing and sampling
were conducted at the US National Lacustrine Core Facility (LacCore) at
the University of Minnesota. We thank Christopher Bronk Ramsey,
Melissa S. Chapot, Alan L. Deino, Christine S. Lane, Helen M. Roberts,
and Céline M. Vidal for discussions on geochronology and age model-
ing. S.K.B. has received further financial support from the University of
Potsdam Open Topic Postdoc Program. This is publication #55 of the
Hominin Sites and Paleolakes Drilling Project.
Author contributions
M.H.T., A.A., H.F.L. and F.S. designed the Chew Bahir Drilling Project.
A.A., H.F.L., F.S. and V.F. led the drilling campaign. V.F., H.F.L., A.A. and
F.S. measured and sampled the cores. M.H.T. and N.M. designed and ran
the time series analysis experiments. P.O.H. and P.J.V. performed
modeling experiments. M.H.T. led the writing of the paper and wrote the
first draft of the text, including the paragraphs on time series analysis
and the determination of the duration and start of the tipping. V.F. and
M.H.T wrote the paragraph on validating climate tipping in the sediment
in the methods section. M.H.T., M.L.F. and N.M. wrote the paragraph on
recurrence plots and recurrence quantification analysis in the methods
section. K.K. and M.H.T. wrote the paragraphs on the implications of
climate tipping for humans in the main text. M.H.T., A.A., H.L., F.S., S.K.B.,
M.L.F., M.A.M. and V.F. edited the Introduction of the main text. M.H.T.
designed all the figures in the paper. M.H.T., A.A., M.L.F., P.O.H., V.F.,
S.K.B., K.K., H.F.L., N.M., M.A.M., F.S. and P.J.V. discussed the results and
contributed input to the manuscript.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary information The online version contains
supplementary material available at
https://doi.org/10.1038/s41467-024-47921-1.
Correspondence and requests for materials should be addressed to
Martin H. Trauth.
Peer review information Nature Communications thanks the anon-
ymous reviewers for their contribution to the peer review of this work. A
peer review file is available.
Reprints and permissions information is available at
http://www.nature.com/reprints
Publisher’s note Springer Nature remains neutral with regard to jur-
isdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as
long as you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons licence, and indicate if
changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless
indicated otherwise in a credit line to the material. If material is not
included in the article’s Creative Commons licence and your intended
use is not permitted by statutory regulation or exceeds the permitted
use, you will need to obtain permission directly from the copyright
holder. To view a copy of this licence, visit http://creativecommons.org/
licenses/by/4.0/.
© The Author(s) 2024
Article https://doi.org/10.1038/s41467-024-47921-1
Nature Communications | (2024) 15:3697 9
Content courtesy of Springer Nature, terms of use apply. Rights reserved
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com
Content uploaded by Asfawossen Asrat
Author content
All content in this area was uploaded by Asfawossen Asrat on May 07, 2024
Content may be subject to copyright.