Evolution of tropical land temperature across the last glacial termination

The study addresses the magnitude and timing of tropical land temperature change from the Last Glacial Maximum (LGM) to the Holocene, a period marked by a large rise in atmospheric CO2 (~200 to ~280 ppmv). While deglacial changes in CO2, ice sheets, and ocean circulation have been studied extensively, precise and accurate temperature records from low-latitude land regions remain scarce, limiting understanding of regional climate sensitivity to greenhouse forcing. In the tropics, proxy-based estimates of deglacial warming span 1–6 °C and climate models suggest 1.6–3.5 °C, with sea surface temperature (SST) reconstructions showing divergent evolution even within the tropical West Pacific. This ambiguity, and the lack of robust terrestrial thermometers, complicates assessment of whether tropical temperatures primarily respond to Northern Hemisphere (NH) abrupt events, Southern Hemisphere (SH) changes, greenhouse gases, or their combination. This study aims to provide a precise terrestrial temperature reconstruction from Northern Borneo to clarify the magnitude and phasing of tropical warming across the last deglaciation.

Prior work indicates wide-ranging estimates of tropical deglacial warming: proxy compilations suggest 1–6 °C while climate models estimate 1.6–3.5 °C. SST reconstructions within the tropical West Pacific show differing structures of warming, with some sites indicating step-like SH-paced warming (increase during HS1, slight ACR cooling, further increase during YD), while others (e.g., Indonesian Throughflow region) suggest steady warming across the deglaciation. Earlier SST compilations estimated ~1–2 °C low-latitude warming, whereas noble gas measurements in groundwater implied ~6 °C land warming. Recent proxy-model syntheses for tropical land suggest ~3.4 °C warming (15°S–15°N). Hydroclimate reconstructions from speleothem δ18Occ across the Indo-Pacific Warm Pool document sensitivity to NH abrupt events via δ18O shifts interpreted as rainfall changes, and changes in Atlantic meridional overturning circulation have been tied to tropical rainfall shifts and ITCZ migrations.

Study site and context: Stalagmite SC02 was collected in 2006 from Secret Chamber, Clearwater Cave system, Gunung Mulu National Park, Northern Borneo (4°N, 115°E), in the tropical West Pacific. The cave experiences year-round deep tropical convection under the ITCZ with ~5000 mm annual rainfall and weak seasonality; ENSO modulates interannual precipitation. Present-day cave temperature is 24.0 °C (23.9–24.1 °C) based on 2018–2019 logger data. Inner cave temperatures are slightly cooler (~0.4 °C) than cave entrance, likely due to evaporative cooling assumed stable over time.

Sample and preparation: SC02 is ~580 mm long; the studied 204 mm section covers ~5.4–23 ka with ~11 µm/yr growth and columnar calcite fabric. Fluid inclusions are intercrystalline, mainly monophase liquid, elongated parallel to growth. Two-phase inclusions were excluded. Inclusion sizes range 10^2–10^5 µm³. A 13 mm thick slab previously used for U/Th dating and stable isotope analyses was subsampled. Sections (~300 µm) were prepared, cut into 5 mm-wide strips, and fragments used for microthermometry. Layers were traced to the center to tie analyses to exact ages and δ18Occ.

Fluid inclusion (FI) microthermometry: The paleothermometer relies on the density of former drip water trapped in microscopic inclusions formed during calcite growth. Because stalagmite inclusions are typically monophase and metastable against vapor nucleation, single ultra-short laser pulses were used to induce vapor bubble nucleation, creating a stable liquid-vapor equilibrium. Upon heating, inclusions homogenize to liquid at Th(obs); Th was corrected for surface tension effects (volume-dependent) to yield the homogenization temperature of an infinitely large inclusion, equivalent to formation temperature Tf if original water density is preserved. Bubble radii were measured from images at known temperature to perform the correction.

Experimental setup: Measurements used a Linkam THMSG600 heating/cooling stage on an Olympus BX51 microscope with LMPLFLN 100x/0.8 objective and Leica DFC350 FX camera. Stage accuracy was ±0.1 °C based on synthetic H2O and H2O–CO2 standards. An amplified Ti:Sapphire laser (CPA-2101) provided femtosecond pulses for vapor nucleation.

Data density and uncertainties: Each temperature level represents ~30–40 individual FI measurements within a growth layer, yielding near-Gaussian distributions with SD 0.5–1.8 °C (avg 1.0 °C). Averaged temperature uncertainties are reported as two standard errors of the mean (2 SEM), typically 0.2–0.4 °C. Each point integrates 50–250 years depending on layer width and growth rate. U/Th ages (with 95% CIs) from prior work provide the age model.

Corrections and calculated parameters: To isolate climatic warming, measured cave temperatures were corrected for glacial sea-level-induced altitude changes using a regional lapse rate of 0.6 °C/100 m and a global sea-level reconstruction (Lambeck et al., 2014). Hydroclimate was assessed by combining δ18Occ (10–20 ka from prior work; Holocene from this study) with cave temperature to compute δ18O of drip water (δ18Odw), correcting for global seawater δ18O changes due to ice volume (~1.0‰ at LGM). Present-day relationships in Borneo link δ18Odw closely to rainwater δ18O and precipitation amounts.

Comparative datasets: The Borneo temperature record was compared to Greenland NGRIP δ18O (NH temperature), Antarctic temperature anomalies, atmospheric CO2, mean global ocean temperature (ice-core noble gases), and North Atlantic 231Pa/230Th (AMOC strength), aligning on appropriate age scales (AICC2012, WD2014) for phasing assessment.

• Stalagmite SC02 cave temperatures indicate deglacial warming (ΔT) of 4.4 ± 0.3 °C (2 SEM) from the LGM (19.0 ± 0.2 °C) to the Holocene (23.4 ± 0.2 °C). After correcting for sea-level-induced altitude changes, ΔT is 3.6 ± 0.3 °C (2 SEM).
• Present-day cave temperature is 24.0 °C.
• The Borneo land temperature record exhibits step-like deglacial warming aligned with SH temperature and atmospheric CO2: early warming during HS1, slight cooling during ACR, renewed warming during YD, with accelerations coincident with rapid CO2 rises.
• No detectable cooling in Borneo land temperature during NH abrupt cooling events (HS1, YD), contrasting with NH signals in Greenland δ18O and AMOC slowdowns.
• Mean global ocean temperature (ice-core noble gas-based) shares the SH/CO2-aligned phasing, suggesting globally dominant timing similar to Borneo.
• The magnitude of tropical land warming (~3.6 °C corrected) agrees with recent proxy-model tropical estimates (~3.4 °C) and tropical African lake reconstructions, is larger than early SST compilations (~1–2 °C), and smaller than noble-gas groundwater land estimates (~6 °C).
• Regional context: Agreement with clumped isotope-based ~4 °C ASST; Mg/Ca SST records show 2–3 °C warming but potential biases (pH/salinity). The Borneo land record supports a step-like SH-paced evolution across the tropical West Pacific; deviations at archipelago sites likely reflect seasonality or sea level–driven circulation/salinity effects.
• Hydroclimate decouples from temperature: δ18Odw shows strong enrichment (interpreted as drying) during NH cooling events (HS1, YD), but little net difference between LGM (corrected δ18Odw ≈ −9.43‰; N=1) and Holocene (≈ −9.24‰; N=3), indicating minor LGM–Holocene hydroclimate change at the site.
• Sensitivity: Borneo land temperature increased 3.6 ± 0.3 °C in response to an ~80 ppm CO2 rise from LGM to Holocene, underscoring tropical temperature sensitivity to greenhouse forcing.

The findings resolve key uncertainties about tropical low-latitude land temperature evolution during the last deglaciation. The Borneo record demonstrates that tropical land temperature tracked atmospheric CO2 and SH temperature changes, not NH abrupt events, indicating primary control by greenhouse forcing with SH timing. This behavior contrasts with regional hydroclimate, which responded strongly to NH-driven millennial-scale perturbations (AMOC reductions) via ITCZ shifts and associated rainfall changes, as captured by δ18Odw. The decoupling between temperature and hydroclimate implies that the Indo-Pacific Warm Pool’s heat content (availability of heat and moisture) increased with SH/CO2 forcing, while the distribution (fate) of moisture was governed by interhemispheric temperature gradients and NH dynamics. Comparisons with global and regional datasets suggest potential NH over-representation in some global surface temperature reconstructions, while the step-like SH-paced pattern evident in Borneo and open-ocean West Pacific SST records likely reflects the broader regional signal. Methodological corrections (altitude/lapse-rate adjustment) refine the climatic warming magnitude, bringing it into agreement with recent tropical reconstructions and highlighting limitations in earlier SST-only or groundwater noble gas-based estimates.

This study provides a precise terrestrial tropical temperature reconstruction across the last deglaciation using stalagmite fluid inclusion microthermometry from Northern Borneo. After correcting for sea-level-induced altitude effects, Borneo land temperature warmed by 3.6 ± 0.3 °C from the LGM to the Holocene, with phasing closely aligned to atmospheric CO2 and SH temperature evolution. Hydroclimate, inferred from δ18Odw, was largely insensitive to greenhouse forcing but responded strongly to NH millennial-scale cooling via drying episodes, evidencing a decoupling of temperature and rainfall. These results underscore the sensitivity of tropical land temperatures to greenhouse forcing and clarify regional deglacial phasing in the tropical West Pacific. Potential future research could extend high-precision stalagmite FI temperature records across other tropical regions to test spatial consistency, refine lapse-rate and sea-level corrections with local paleoaltimetry constraints, integrate multi-proxy approaches (e.g., clumped isotopes, independent rainfall proxies) within the same archives, and improve chronological alignment across hemispheric records to further resolve phasing and mechanisms.

• The FI microthermometry approach assumes preservation of original inclusion water density and accurate correction of surface tension effects; undetected post-depositional alteration could bias temperatures.
• Cave temperatures may be slightly offset from outside air by evaporative cooling; this offset is assumed stable through time.
• The altitude correction relies on a modern lapse rate (0.6 °C/100 m) and a global sea-level reconstruction; spatial/temporal variability in lapse rates or local topographic effects could introduce uncertainty.
• Each temperature point averages 30–40 inclusions over 50–250 years, potentially smoothing short-lived events.
• δ18O-based hydroclimate interpretations assume precipitation amount is the dominant control on δ18Odw; changes in moisture source or rainout history could also influence δ18Odw.
• Limited number of LGM and Holocene δ18Odw points (N=1 for LGM; N=3 for Holocene) constrains statistical power for detecting small mean differences.