| 07 Aug 2026
Early to Middle Holocene aridification trend in southeastern Iberia and the termination of the last North African Humid Period revealed by a multi-proxy speleothem record
Abstract. During the Early Holocene, relatively humid conditions prevailed across northern Africa, transitioning to a drier climate by the end of the mid-Holocene. While the timing and nature of this climatic shift are well-established in the Maghreb region of Africa, uncertainties remain regarding the pace of the wet-to-dry transition in the southern Iberian Peninsula. We present a multi-proxy stalagmite record from Simarrón II cave, spanning 9.7 to 4.4 ka BP. The dataset combines stable carbon and oxygen isotopes in calcite (δ18Ocarb, δ13C), Mg/Ca ratios, petrographic observations, fluid inclusion isotopes (δ18Ofi, δ2H) and TEX₈₆-derived paleotemperatures to reconstruct paleohydrological and climatic changes. From 9.7 to 4.4 ka, progressively higher δ¹⁸Ofi and δ²H values, coupled with decreasing deuterium excess, are consistent with a gradual shift in rainfall seasonality and/or moisture sources (i.e., frontal Atlantic-type precipitation regime during the early Holocene to a more convective Mediterranean regime). This trend could be overprinted by enhanced soil and in-cave water evaporation during the late Mid-Holocene because of climate aridification. Lower δ18Ocarb, δ13C, and Mg/Ca values between 9.7 and 6.0 ka also indicate relatively wet conditions during the early Holocene and part of the middle Holocene. From ~6.0 ka onward, these proxies show a progressive increase, stabilizing after ~5.5 ka, consistent with a climate shift toward more arid conditions that occurred within ~500 yrs. This is interpreted as reflecting enhanced prior calcite precipitation rates and kinetic isotope effects associated with intensified in-cave CO₂ degassing, due to the dry climate during the termination of the last North African Humid Period. Paired δ18Ocarb– δ18Ofi paleotemperature estimates yield unrealistically high values (20–30 °C) for most of the investigated period, further supporting non-equilibrium fractionation processes and likely in-cave water evaporation. In contrast, TEX86-based temperatures (13–17 °C) suggest that early–mid Holocene cave temperatures were higher than present-day values in the cave (11.7 °C).
This manuscript presents a new multi-proxy speleothem record from Simarrón II cave in southeastern Iberia, covering an important interval of Early-to-Middle Holocene hydroclimatic change. The combination of carbonate stable isotopes, fluid-inclusion water isotopes, Mg/Ca ratios, petrography, U–Th dating, and TEX₈₆ measurements is potentially valuable. The high-elevation setting and comparison with records from Iberia and northern Africa are also of considerable interest to Climate of the Past.
The manuscript is generally well structured and the dataset has the potential to make a useful contribution. However, several of the main conclusions;particularly those concerning fluid-inclusion isotope trends, quantitative temperature changes, the timing of North African Humid Period termination, and the role of high-elevation areas as climatic refugia; are currently stated with greater confidence than the available evidence permits. The chronology and analytical uncertainties also need to be documented more completely.
The chronology needs substantially more documentation. Eleven U–Th ages are reported, but the supplement contains only one petrographic figure and no table of dating results. A complete table should report sample depth, 238U and 232Th concentrations, isotope activity ratios, initial 234U/238U, uncorrected and detrital-corrected ages, correction assumptions, and analytical uncertainties. There also appears to be an inconsistency in lines 240–244. The youngest age is said to be 4,621 ± 12 yr BP at 70 mm from the top, whereas Figure 2 appears to place this date much closer to the top. This should be checked carefully. Furthermore, the manuscript states that the record spans 9.7–4.4 ka, although the directly dated interval is reported as 9.244–4.621 ka. The authors should explain how the end ages were extrapolated and report their model uncertainties. The possible hiatus at approximately 30 mm must be explicitly incorporated into the age modelling. Interpolating a continuous COPRA model across a physical hiatus could produce misleading ages and growth rates. The authors should clarify whether the sections above and below the hiatus were modelled separately, quantify the possible duration of the hiatus, and show the complete age-model uncertainty envelope. Growth-rate uncertainties should also be presented.
The interpretation of the fluid-inclusion isotope record is presently ambiguous. The measured relationship, δ²H = 4.2 × δ¹⁸O − 15.8, has a slope considerably lower than either the GMWL or the reported Almería LMWL. This strongly indicates evaporation and/or analytical fractionation. It is therefore difficult to reconcile this result with the statement that most samples plot along the GMWL. More importantly, the uncorrected δ¹⁸O, δ²H, and d-excess values are subsequently used to infer changes in moisture source and rainfall seasonality. If evaporation during crushing, soil/epikarst evaporation, and in-cave evaporation are all plausible, the present data cannot uniquely distinguish among these mechanisms.
The authors should report the absolute water yield for every analysis, not only water content per gram, provide the five replicate results individually, test δ¹⁸O, δ²H, and d-excess against water yield and sample mass, state the validated water-volume range and any data-exclusion criteria for the analytical system, quantify the temporal trends with slopes, confidence intervals, and significance tests, apply and discuss the evaporation correction consistently, not only for paleotemperature calculations. Unless these tests provide stronger discrimination, the shift from Atlantic frontal precipitation toward Mediterranean convective precipitation should be presented as one hypothesis among several, rather than as a principal conclusion.
The interpretation of higher δ¹³C and Mg/Ca as evidence of increased prior calcite precipitation and aridity is plausible, but not unique. Mg/Ca may also respond to water–rock interaction, residence time, source-rock variability, and temperature-dependent partitioning. Similarly, δ¹³C can be influenced by vegetation, soil productivity, cave-air CO₂, degassing, and PCP. The proposed relationship between dry conditions and faster stalagmite growth also requires more careful justification. Reduced infiltration may increase degassing and PCP, but it may simultaneously reduce drip supply and stalagmite growth. Growth rate should not be treated as independent confirmation of aridity without a more explicit process-based argument. The reported δ¹⁸O–δ¹³C correlation over the 240–140 mm interval may partly reflect two variables sharing the same temporal trend. The authors should test correlations after detrending or using first differences and account for serial autocorrelation. Correlations among Mg/Ca, δ¹³C, δ¹⁸O, growth rate, and water content should be reported quantitatively. If kinetic fractionation is central to the interpretation, along-layer measurements or another independent kinetic test would be valuable. If such analyses are not possible, the limitations should be stated explicitly.
Only six TEX₈₆ measurements are available, yet the manuscript uses them to infer a 2.5 °C warming, rapid climate oscillations, and warming associated with termination of the humid period. These conclusions are too strong for the current sampling density and uncertainty treatment. The authors should provide the individual GDGT abundances, TEX₈₆ values, analytical replicates, error estimates, and relevant quality-control indices. Potential changes in GDGT source or microbial community should also be discussed, because hydroclimatic changes may influence TEX₈₆ independently of cave temperature. The statement that calibration uncertainty is approximately ±1 °C at 2 SD needs to be reconciled with the uncertainties reported for the intercept and slope of the calibration equation. Repeatability of the in-house standard is analytical precision and should not be presented as total paleotemperature accuracy. Errors should be propagated and displayed in Figures 5 and 6. Until this is done, the inferred 2.5 °C variation cannot be evaluated robustly. The TEX₈₆ results should probably be presented as tentative evidence for temperature variability rather than a quantitative demonstration of rapid warming.
The regional comparison is useful, but Figure 7 combines records with different temporal resolutions, age uncertainties, and proxy sensitivities. The authors should explain the common interval used for z-score calculation, whether the records were resampled, and how chronological uncertainty was considered. Positive δ¹⁸O z-scores do not necessarily have an identical hydroclimatic meaning at every site. The conclusion that aridification was delayed at high elevations; and that mountain areas acted as climatic refugia; is based primarily on an approximately 500-year difference between Simarrón II and Cueva Victoria. This difference could also result from age-model uncertainty, cave hydrology, proxy-specific responses, or local climatic variability. The refugium interpretation should therefore be presented as a hypothesis unless it can be supported by additional high and low elevation records. Similarly, hydroclimatic change in southeastern Iberia should not automatically be equated with termination of the African Humid Period. The authors should distinguish between the African monsoon phenomenon itself and its possible western Mediterranean expression or teleconnection.
The final archaeological paragraph is not adequately connected to the preceding analyses and appears to contain a chronological mismatch. The mid-sixth millennium cal BC corresponds to approximately 7.5 ka BP, whereas the main aridification discussed in the manuscript culminates around 5.8–5.3 ka BP, approximately 3850–3350 BC. These events are therefore separated by well over a millennium. The claimed “close temporal correspondence” between the Mesolithic–Neolithic transition and the reconstructed aridification is not supported as currently presented. This paragraph should either be removed or developed into a properly referenced discussion using consistent BP/BC chronologies and a clearly defined archaeological transition.
Specific comments