Dynamic and thermodynamic influences on precipitation in Northeast Mexico on orbital to millennial timescales

Northern Mexico is projected by many climate models to dry in the future, yet the spatial pattern and magnitude of drying remain uncertain due to limited model agreement and sparse local proxy data. Severe droughts historically have had major social and economic impacts in the region, emphasizing the need for improved projections. Paleoclimate records can help constrain dynamical drivers of precipitation variability, evaluate models, and provide data for proxy–model comparisons. Despite extensive karst in Northeast (NE) Mexico, few high-resolution, long-duration records exist. Key open questions include whether orbital-scale precipitation is paced by seasonal insolation (e.g., via ITCZ migration and CLLJ strength), and on millennial scales whether changes are driven by dynamics of the Caribbean Low-Level Jet (CLLJ) or by thermodynamic forcing from Atlantic–Pacific sea-surface temperature (SST) gradients. This study presents a new, high-resolution speleothem record from Tamaulipas spanning 62.5–5.1 ka to test these hypotheses and uses an isotope-enabled climate model to separate dynamic versus thermodynamic controls.

Prior work in NE Mexico offers mixed evidence for insolation control: some lake records correlate runoff with Northern Hemisphere summer insolation (NHSI), while others implicate autumn or spring insolation as more relevant. In contrast, NW and Southern Mexico show stronger NHSI–precipitation links via the North American Monsoon and ITCZ shifts. On millennial timescales, interpretations have linked dry conditions during Heinrich Stadials (HS) to a weakened CLLJ from a southward ITCZ, though this conflicts with modern seasonal dynamics where the CLLJ strengthens when the ITCZ is south. Numerous reconstructions and modeling studies across Mesoamerica indicate strong precipitation sensitivity to both Atlantic and Pacific SSTs on interannual to centennial scales, including a north–south dipole during the Common Era. However, the role of SSTs on orbital and millennial scales in NE Mexico remains poorly constrained due to sparse records and age-model uncertainties. Some NE Mexico lake records link extreme rainfall and erosion to warm Gulf of Mexico SSTs (mid-Holocene, Bølling–Allerød), but associations with mean precipitation are weak or inconsistent. Collectively, the literature highlights uncertainty about the dominance of insolation, CLLJ dynamics, and SSTs in controlling NE Mexico hydroclimate.

Speleothem sampling and chronology: A 78 cm candle-shaped stalagmite (CB2) was collected from Cueva Bonita (23°N, 99°W; 1071 m asl) in the Sierra Madre Oriental, Tamaulipas. The cave environment is stable; dripwater–calcite equilibrium and minimal kinetic effects were evaluated. Thirty-three 230Th–234U ages were measured at ~2.5 cm intervals (U = 18–63 ng/g) using MC-ICP-MS (Nu Plasma II-ES with Aridus 2). Ages were corrected using an initial 230Th/232Th = 10.5 ± 5.3 ppm, constrained by stratigraphic consistency tests and a modern speleothem calibration (CB4). All dates are stratigraphically ordered within 2σ. An age–depth model was generated using COPRA with 2000 Monte Carlo realizations; the record spans 62.5–5.1 ka with average growth ~14 µm/yr and mean temporal resolution ~36 years. Potential minor growth-rate changes occur near 11–6 ka and 30–37 ka, but no fabric changes were observed; Holocene data are used only for glacial–interglacial comparison due to a dating gap. Modern isotope monitoring and moisture source: A local precipitation station (Alta Cima, ~1 km from cave) collected 48 samples (June 2018–May 2019) using evaporation-limiting collectors; δ18O and δD were analyzed on a Picarro L2130i (σ ≈ 0.11‰ for δ18O). Back-trajectory analysis (HYSPLIT) for 2005–2018 (rain-bearing only; 3600 trajectories; launched every 6 h at 1500 m, 72 h back) indicates moisture is consistently sourced from the Gulf of Mexico and Caribbean. Observed monthly precipitation δ18O correlates strongly and negatively with precipitation amount (p < 0.01; r = 0.88; slope ≈ −2‰ per 100 mm). Isotope-enabled GCM simulations over the last 40 years also support δ18O primarily reflecting precipitation amount. Proxy measurements and interpretation: CB2 was micromilled at 500 µm increments (to 1 mm depth), yielding 1578 stable isotope samples (δ18O, δ13C) analyzed with a Kiel IV + Thermo Delta V IRMS (precision: 0.08‰ δ18O; 0.05‰ δ13C). Ice-volume corrections were applied to δ18O. Trace elements (Mg/Ca) were measured on 789 aliquots (every other sample) via Nu Attom HR-ICP-MS with bracketing standards and Ge internal standard. Multi-proxy interpretation: δ18O of calcite is taken to reflect dripwater/precipitation δ18O and thus regional precipitation amount, given equilibrium conditions and modern calibration. δ13C and Mg/Ca inform local water balance via prior calcite precipitation (PCP) and epikarst processes; higher δ13C and Mg/Ca indicate enhanced PCP during drier conditions. Temperature effects on Mg partitioning are also considered, particularly across the deglaciation. Climate model experiments: The isotope-enabled Community Earth System Model (iCESM1; CAM5/CLM4/POP2/CICE4; nominal 1–2.5°) was configured for 21 ka (PMIP4 forcings; ICE-6G ice sheets). Ocean initial δ18O from GISS dataset with +1‰ global enrichment; other ocean states from prior LGM runs. The LGM control was integrated 500 years; analysis from final 50 years. A Heinrich-like freshwater forcing experiment added 0.50 Sv with δ18O = −30‰ into 50–70°N North Atlantic for 100 years, then off for 50 years; analysis from final 50 years. Moisture budget decomposition followed Seager & Henderson to separate thermodynamic (specific humidity) and dynamic (winds) contributions; residual includes transient and nonlinear terms. Model–data comparisons and auxiliary datasets: Pre-industrial iCESM precipitation was compared with GPCP merged products; known overestimation in the E. Tropical Pacific is attributed to orography resolution, with better fidelity at subtropical latitudes. Additional comparisons include Greenland ice core δ18O, atmospheric pCO2, and regional SST reconstructions.

- New decadal-resolution speleothem record (CB2) from NE Mexico spans 62.5–5.1 ka with ~36-year mean resolution and 33 U–Th ages; highest-resolution continuous Mexican record over this interval. - δ18O reflects precipitation amount; modern station data show δ18O–precipitation amount correlation (p < 0.01; r = 0.88; slope ≈ −2‰ per 100 mm). Moisture source is Gulf of Mexico/Caribbean. - Multi-proxy coherence indicates PCP-driven responses to local water balance: δ13C co-varies with δ18O (r = 0.53, p < 0.01). During glacial periods, δ13C–Mg/Ca correlations are significant (r = 0.51–0.77, p ≤ 0.01), supporting PCP as a key control under drier conditions. - CB2 δ18O lacks a clear precessional (insolation) signal across ~2.5 precession cycles; correlations with seasonal insolation are weak over ~57 ka. A glacial–interglacial shift of ~1‰ in δ18O can be explained by cave temperature changes; early Holocene wetting observed regionally is muted in CB2 after temperature correction. - Strong millennial-scale variability: pronounced positive δ18O excursions (drier) during HS 5 (50–47 ka), HS 4 (43–42 ka), HS 3 (31–28 ka), HS 1 (18–15 ka), and the Younger Dryas (12–10 ka); HS 2 (~24 ka) is an exception with increased local water balance. - δ18O values (n = 1578) span −1.29‰ to +6.30‰ (VPDB); largest enrichment during HS1 (~16.8 ka). δ13C decreases ~3‰ across the deglaciation; millennial δ13C increases up to +3.94‰ (HS5) and ~+1.04‰ (HS3). Mg/Ca average glacial value ≈ 27 mmol/mol; rises to 37 (YD), 68 (HS1), 34 (HS3), 37 (HS4), 35 (HS5), and 33 (HS6) mmol/mol. Divergent deglacial Mg/Ca increase likely reflects temperature-dependent partitioning rather than PCP. - On orbital scales, CB2 δ18O more closely tracks global/regional temperature and atmospheric pCO2 than insolation (e.g., δ18O–pCO2 r = −0.61, p < 0.05 over full record, mainly driven by deglaciation; insignificant over 20–62.5 ka: r = −0.41, p = 0.11). - iCESM Heinrich-like hosing on LGM background produces strong N. Atlantic cooling, with tropical Atlantic cooling ~−6°C vs. tropical Pacific ~−1°C, intensifying an Atlantic–Pacific SST and SLP gradient. Strengthened tropical easterlies and enhanced Bermuda High reduce precipitable water and precipitation over Mesoamerica, shifting moisture convergence south. - Modeled NE Mexico precipitation δ18O increases are consistent with speleothem δ18O enrichments (~+1.5‰ during HS1). Bølling–Allerød shows opposite signals: δ18O −2‰, δ13C −3‰, Mg/Ca −15 mmol/mol, consistent with wetter conditions under strengthened AMOC and warmer SSTs. - Comparison with regional records (Cuba, Costa Rica speleothems; Cariaco Basin; Guatemala lake proxies) indicates broadly in-phase, spatially coherent drying across Mesoamerica and the Caribbean during HS events, driven by inter-basin SST/SLP gradients and dynamics rather than local insolation forcing. - HS2/LGM anomaly: CB2 indicates increased local water balance during HS2 near the LGM, potentially due to (i) a weaker AMOC reduction and mitigated SST cooling, and/or (ii) reduced evapotranspiration under colder temperatures increasing P−ET; PMIP3 simulations support higher soil moisture despite uncertain precipitation changes.

The study tests whether orbital to millennial hydroclimate variability in NE Mexico is controlled primarily by seasonal insolation/ITCZ–CLLJ dynamics or by thermodynamic forcing from SST gradients. The multiproxy stalagmite record shows negligible precessional pacing and strong millennial-scale drying during Heinrich events and the Younger Dryas, except HS2. Isotope-enabled modeling demonstrates that Atlantic cooling and a strengthened Atlantic–Pacific SST/SLP gradient intensify easterlies and shift moisture convergence southward, reducing precipitable water and convection over NE Mexico. Thus, millennial-scale droughts are governed by a combination of dynamic (wind) and thermodynamic (SST) influences, not a weakened CLLJ as previously inferred from seasonal analogs. The coherent response with southern Mesoamerica and Caribbean proxies indicates the importance of inter-basin gradients and AMOC variability for region-wide precipitation. The HS2/LGM exception underscores sensitivity to event magnitude and local energy balance (ET). Overall, the findings reinforce that paleoclimate SST gradients and AMOC state are key determinants of NE Mexico hydroclimate, providing critical constraints for evaluating model performance and future projections.

This work provides the first decadal-resolution, multiproxy speleothem record from NE Mexico spanning 62.5–5.1 ka, revealing: (1) weak to absent direct insolation control on precipitation; (2) strong millennial-scale drying during the YD and HS 1, 3–6; (3) glacial–interglacial variations closely linked to Atlantic–Pacific SST changes; and (4) HS2/LGM conditions with increased local water balance likely due to weaker event forcing and reduced ET. Isotope-enabled model experiments attribute Heinrich drying to cooler Atlantic SSTs and strengthened easterlies establishing a strong inter-basin SST/SLP gradient and southward-shifted moisture convergence. These mechanisms yield a spatially broad, coherent hydroclimate response across Mesoamerica and the Caribbean under weakened AMOC. The CB2 record offers stringent targets for model validation of inter-basin gradients across key intervals (LGM, HS events, deglaciation) and suggests that improved representation of Atlantic–Pacific SST gradients will enhance reliability of future rainfall projections. Future research should refine event-specific dynamics (e.g., HS2), quantify ET and cloud/radiative impacts on local water balance, and incorporate cyclone-resolving models and higher-resolution topography to reduce regional biases.

- Age-model uncertainties increase in intervals with fewer dates (e.g., 11–6 ka gap; 30–37 ka slower growth), though stratigraphy and multiple dates support continuity; Holocene component not interpreted in detail. - Initial 230Th/232Th correction is higher than traditional assumptions, potentially inflating U–Th age uncertainties, though stratigraphic tests and a modern speleothem constrain the value. - Proxy interpretations may include non-unique influences: δ18O can reflect source/seasonality/convective structure; δ13C and Mg/Ca influenced by vegetation, soil respiration, temperature, and water–rock interaction; deglacial Mg/Ca likely includes temperature effects. - Climate model limitations include coarse topographic resolution and lack of explicit tropical cyclone simulation; known biases in Eastern Tropical Pacific precipitation magnitude. Thus, modeled precipitation amounts are emphasized qualitatively via mechanisms rather than exact values. - The correlation between δ18O and pCO2 is chiefly deglacial; relationships are weaker over much of the glacial period, indicating non-stationarity in forcings. - HS2 interpretation remains uncertain due to debated AMOC weakening magnitude; multiple mechanisms (weaker forcing, ET changes) may contribute.