Stalagmite-Inferred Climate in the Western Mediterranean during the Roman Warm Period

Abstract

The circum-Mediterranean region is the cradle of ancient civilizations that had their roots in the Holocene. Climate change has been considered a key element that contributed to their rise or fall. The Roman Warm Period (RWP), 200 B.C. to 400 A.D., was the warmest period in Europe during the last two thousand years. Hydroclimatic change at the end of the RWP has been suggested as a possible influence on the stability of the Roman political regime and the eventual collapse of the Roman Empire in 476 A.D. A lack of precise proxy records hampers our understanding of hydroclimatic variability over the RWP. Here we present a stalagmite-based climate record from 550 ± 10 B.C. to 950 ± 7 A.D. (2σ) from northern Italy, which reveals a climatic trend of warming and increased humidity throughout the RWP. By comparison with other proxy records in Europe and the circum-Mediterranean region, we argue that the warm, humid climate in southern Europe could be linked to the multi-centennial warming of the Mediterranean Sea. Our record further suggests a century-long rapid drying trend from the early-4th to early-5th century, followed by a 100-year-long drought event, which could have influenced the fall of the Roman Empire.

1. Introduction

Climate change has widely affected society, ecosystems, and agriculture [1]. Today, the circum-Mediterranean region, with over 500 million inhabitants, has been threatened by severe drought and extreme climate, including floods, heat waves, and cold spells [2,3,4]. Such climate changes have been proposed as significant factors in the stability of ancient civilizations [5,6,7]. For example, the well-known 4.2-k dry event (4.2 thousand years ago) may have induced the collapse of the Old Kingdom and Akkadian Empire [8,9]. Increasing climate variability at ~0 A.D. has been suggested as a trigger of the Celtic Expansion and Roman Conquest [6]. Human societies have been forced to adapt to severe and unstable hydroclimatic conditions, with consequent climate-driven famine, warfare, diseases and/or political collapse [10]. Complete and detailed paleoclimate records provide a means to investigate human responses to past climate change.

The Roman warm period (RWP), from ~200 B.C. to 400 A.D, is one of the warmest periods in the late Holocene [11,12]. This period was characterized by a warm, humid climate across Europe [13,14,15,16,17,18]. One recent study [19] even suggested that the average temperature in the RWP was 2 °C higher than the average temperature over the last century in the western Mediterranean. Although the RWP could be a regional warming event and limited in Europe and the North Atlantic region, to understand the RWP is essential for ancient European societal changes. Hampered by a lack of precisely dated records, the detailed climate variability in the circum-Mediterranean realm in the RWP remains unclear.

Here we present a well-dated stalagmite-based record of climate from the Bàsura cave of northern Italy, from 2500 to 500 years ago (yr BP, relative to 1950 A.D.). We used stalagmite carbon isotopes as a climate proxy and built an absolute age model with U-Th dating techniques. Our records show that precipitation in this region progressively increased over the RWP. A century-long shift to a dryer climate at the end of the RWP possibly contributed to the collapse of the Roman Empire in 476 A.D.

2. Material and Methods

2.1. Cave and Regional Settings

Bàsura cave (44°08′ N, 8°12′ E, Figure 1) is located in Toirano, northern Italy, 186 m above sea level [20] and 4.3 km from the coast of the Ligurian Sea. The cave has one natural entrance and another artificial tunnel connected to Santa Lucia cave. It features an 890 m-long passage with 10–20 m thick cave ceiling. Genoa (44°42′ N, 8°85′ E), 60 km northeast of the cave, has a large regional meteorological station. The station is only several km far from the coast and could share a similar climate feature with the cave, supported by high degrees of similarity of Genoa precipitation data and regional precipitation patterns (Supplementary Materials, Figure S1). Local observatory temperature and precipitation data used in this study were from this station. The region features a Mediterranean-type climate, with hot–dry summers and wet–temperate winters. Observational data of the Global Network of Isotopes in Precipitation-International Atomic Energy Agency (GNIP-IAEA) during 1961–1995 A.D. from the Genoa station show an annual temperature of 14.5 ± 1.2 °C (n = 29, 1σ) and an annual rainfall of 886 ± 106 mm (n = 35, 1σ). More than 70% of the precipitation is contributed during the rainy season between September and February. The average temperature in the Bàsura cave is 15.6 °C, and the relative humidity is 97–100% [21].

2.2. Samples, U-Th Dating and Isotope Measurements

One candle-shaped aragonitic stalagmite, BA14-1, was collected from Bàsura cave. This stalagmite is 160 mm in length with no visible hiatuses. Detailed sample figures (Supplementary Materials, Figure S2A) and scanning electron microscope images are presented in Hu et al. (2022) [21]. A depth interval at 6.5–80.5 mm from the top was used in this study. About 0.02–0.05 g subsamples from thirty-five layers were U-Th dated. Subsamples were drilled for U-Th chemistry, which was performed on class-100 benches in the class-10000 clean room at the High-Precision Mass Spectrometry and Environment Change Laboratory (HISPEC), Department of Geosciences, National Taiwan University (NTU) [29,30]. U and Th isotopic compositions were measured on a multiple collector inductively coupled plasma mass spectrometer, Thermo-Fisher Neptune, linked with a dry sample introduction system, Cetec Aridus [31]. An assumed initial atomic ²³⁰Th/²³²Th ratio of 4 ± 2 × 10⁻⁶ was used for correcting U-Th ages. Reported uncertainties in the U-Th isotope data and calculated ages are given at a two-sigma (2σ) uncertainty level, or two standard deviations from the mean (2σ_m). Based on the StalAge depth-age model (Supplementary Materials, Figure S2B) [21,32], this segment covers a time period from 550 ± 10 B.C. to 950 ± 7 A.D., including the RWP from 200 B.C. to 400 A.D.

A total of 286 powdered subsamples, 50–100 μg each, were drilled from BA14-1 in intervals of 6.5–80.5 mm from the top for stable isotope analyses. Measurements of stable carbon isotopes were performed on a Micromass IsoPrime IRMS at the Stable Isotope Laboratory, Department of Earth Sciences, National Taiwan Normal University. The stable isotope data are reported in per mil (‰), relative to Vienna Pee Dee Belemnite (VPDB) and calibrated with two international standards, the National Bureau of Standards 19 (NBS-19) and IAEA-CO-1. The one-sigma external precision (1σ) for the δ¹³C analyses is ± 0.06‰ based on long-term measurements of NBS-19, IAEA-CO-1, and one in-house marble standard. The measured value for IAEA-CO-1 is 2.45 ± 0.06‰, consistent with the certificated value of 2.49 ± 0.03‰. The measured value for NBS-19 is 1.94 ± 0.06‰, consistent with the certificated value of 1.95‰.

3. Results and Discussion

3.1. Carbon Isotope Results

The δ¹³C results from stalagmite BA14-1 are given in Supplementary Materials, Table S1. This record (Figure 2B), from 550 ± 10 B.C. to 950 ± 7 A.D. (2500 to 1000 yr BP), varies from −11 to −5‰ (Figure 2B). It features stable δ¹³C values hovering around −9 to −7‰ at 550–200 B.C. and a multi-centennial decreasing trend from −8 to −10‰, between the onset of the RWP in 200 B.C. and its end in 400 A.D. A 100 yr interval featuring a rapid δ¹³C increase, from −10 to −8‰, occurred from 341 ± 14 to 432 ± 17 A.D. This was followed by a high-δ¹³C period that persisted until 526 ± 4 A.D., and subsequent centennial-scale δ¹³C oscillations ranging from −6 to −9‰ were observed after that (Figure 2B).

3.2. Interpretation of Carbon Isotope Variations

The Hendy test [34] on BA14-1 shows two-sigma variability of ±0.08–0.21‰ in δ¹⁸O and no correlation between δ¹⁸O and δ¹³C for coeval subsamples of the same growth layers, suggesting negligible kinetic isotope fractionation [21]. Further duplication tests [35] also indicated that this stalagmite was deposited with insignificant kinetic isotope effects so that the δ¹³C values reflect environmental factors [21]. When stalagmites form under or close to isotopic equilibrium, the carbonate δ¹³C mainly reflects the isotopic composition of soil CO₂, which is inherited from the atmosphere and through soil respiration [36]. The variation of stalagmite δ¹³C could hence be controlled by bedrock, vegetation productivity, soil activity, and cave pCO₂ [37,38,39]. During carbonate precipitation, temperature-controlled water-calcite fractionation may also affect the δ¹³C values preserved in the stalagmites [37].

Various vegetation types modulate soil CO₂ isotopic compositions via different photosynthesis pathways [40]. For example, previous studies suggest that C3 and C4 plants may lead to δ¹³C values in cave calcite in the range of −6 to −14‰ and +2 to −6‰, respectively [39,40,41]. Considering the reported calcite-aragonite offsets of 1.1–2.3‰ [38], the values of aragonitic BA14-1 δ¹³C ranging from −6 to −11‰, are equivalent to calcitic carbon isotopic compositions of ~−7 to −13‰, well within the predicted C3-type carbon δ¹³C range (−6 to −14‰). This result agrees with previous studies, which suggest that C3-type plants dominate the vegetation surrounding Bàsura cave in modern times [42]. However, vegetation changes are not likely to control BA14-1 δ¹³C variations during the late Holocene since this region has been persistently dominated by C3-type plants, even in glacial times [21,43,44].

Under warm or humid conditions, dense vegetation coverage atop the cave and enhanced soil biological activity could lead to negative values in soil δ¹³C [45]. Meanwhile, increased precipitation could enhance the drip water and infiltration rates, with a reduction in drip water CO₂ degassing and more negative speleothem δ¹³C values [37,39,42], and vice versa. Prior carbonate precipitation (PCP) [46,47] is an additional factor affecting stalagmite δ¹³C values. PCP is defined as the carbonate precipitation from seepage water before it drips into the cave. The process involves CO₂ degassing along the percolation flow path, which favors carbonate precipitation with light carbon isotopes in the cave ceiling, leading to relatively high δ¹³C values in the remaining water body [39,46,47]. Under a dry (wet) condition, long (short) infiltration residence times enable (depress) PCP, enriching (depleting) the aqueous solution and stalagmite in ¹³C.

For the reasons above, a warm and humid climate could promote soil activity and vegetation cover, reduce PCP in the epikarst, and decrease stalagmite δ¹³C values in the cave. The negative (positive) BA14-1 δ¹³C value could hence register a humid and warm (cold and dry) climate in this region. This argument can be supported by the covariation of BA14-1 δ¹⁸O and δ¹³C time series (Figure 2B). Stalagmite δ¹⁸O at the Bàsura cave is dominated by both the amount effect and the source effect [21]. The amount effect describes the negative correlation between δ¹⁸O values and precipitation amount. The source effect means that more (less) moisture sources from the remote Atlantic Ocean with low δ¹⁸O values, compared to the regional Mediterranean Sea, could reduce (elevate) stalagmite δ¹⁸O. Hence, a negative δ¹⁸O shift reflects strong precipitation or dominant Atlantic-sourced moisture carried by the westerly winds. In the Mediterranean region, the control of δ¹³C and δ¹⁸O covariation has been suggested to be mainly hydroclimatic, e.g., [42,44,48] since high precipitation can decrease δ¹³C by promoting soil activity and vegetation respiration, and δ¹⁸O by the amount effect. A degree of similarity (r = 0.53, n = 286, p < 0.05) between BA14-1 δ¹³C and δ¹⁸O data supports that stalagmite δ¹³C variations document precipitation changes, with high (low) precipitation corresponding to low (high) δ¹³C values.

3.3. Hydroclimate Changes during the Roman Warm Period

From the onset of the RWP at 200 B.C., both BA14-1 δ¹³C and δ¹⁸O express a multi-centennial decreasing trend until 350 A.D. This trend indicates increasing precipitation during this 550 yr interval in northern Italy (Figure 2B). It is similar to the reduced growth rates (bandwidth) recorded in Scotland stalagmites (Figure 2A) [22], which suggests a shift towards a more humid climate. Flood event records from Lake Savine (Figure 2C; [23]), northern Italy, display two intervals with over six flooding events per 100 years centered at 50 B.C. and 350 A.D. (Figure 2C) [23]. The two periods with intense flooding events are in accord with high stalagmite-based precipitation index values in Romania (Figure 2D) [24], which also shows periods with relatively high precipitation at 50 B.C. and 350 A.D. While at the first half of the RWP, from 200 B.C. to 100 A.D., BA14-1 δ¹³C and δ¹⁸O do not suggest strong precipitation as the records from Lake Savine (Figure 2C) [23] and Romania (Figure 2D) [24] do, at the second half of the RWP at 100–400 A.D., all records (Figure 2B–D) reveal that southern Europe could have experienced a wet climate, especially at the late stage of the RWP centered at 350 A.D. The overall 550 yr humid pattern from 200 B.C. to 350 A.D. can also be supported by records from central Italy (Supplementary Materials, Figure S3B,C) [16,18] and the Dead Sea (Figure S3D) [17]. The stalagmite δ¹³C from central Italy [16] shows a negative excursion during the whole RWP (Figure S3B), suggesting a warm and humid climate. The composite lake level records (Figure S3C) reveal a highstand at the second-half of the RWP at ~350 A.D. [18]. The lake level record from the Dead Sea suggested relatively high lake levels during the RWP compared to those in pre-RWP. Combined with our record (Figure S3A), this evidence suggests that the RWP featured a humid climate in the Mediterranean realm.

North Atlantic Oscillation (NAO) variation has been widely used to understand precipitation changes over Europe on different geological time scales, e.g., [28,49]. At a positive (negative) phase of the NAO, increased (decreased) pressure contrast between the Azores High and the Icelandic Low results in poleward (equatorial) migrations of the westerlies over the North Atlantic, bringing in strong precipitation in northern Europe (southern Europe and the Mediterranean region) [49]. However, the reconstructed NAO index [22,33] suggests that the NAO shifted from a negative state to a positive condition during the RWP (Figure 1A). The reconstructed NAO index increased from −3 at the beginning of the RWP at 200 B.C., to 0 at 100 B.C.

From 100 B.C. to the end of the RWP at 400 A.D., the reconstructed NAO index, fluctuating between 0 and 2, features two peaks of 1 and 2 at 0 and 300 A.D., respectively (Figure 2A). Most of the RWP was dominated by an NAO-positive phase. A positive NAO-state, associated with enhanced Azores High and Icelandic Low, should force poleward shifts of the westerlies and result in drought conditions in southern Europe. Hence, the humid conditions reflected by Bàsura stalagmite records and other archives in southern Europe (Figure 2B–D) cannot be attributed to the NAO phase.

Bàsura stalagmite-inferred precipitation increases in southern Europe during the RWP (Figure 3A) coincide with multi-centennial warming registered in sea surface temperature (SST) records in the Mediterranean Sea (Figure 3B,C). Sediment SST records from the Alboran Sea in the western Mediterranean (TTR17-434G, Figure 3B; [25]), the Sicily Channel of the central Mediterranean (SW104-ND11; Figure 3B; [19]) and the Minorca Basin from the northwestern Mediterranean (MINMC06 and HER-MC-MR3; Figure 3C; [27]) all suggest a clear warming trend from 200 B.C. to 200 A.D. and a plateau at 200–400 A.D. This 200-yr high SST in the Mediterranean Sea could increase the land-sea temperature contrast in autumn and winter, and enhance frontal rains along the Mediterranean borderland [50,51]. The warm Mediterranean Sea could also promote Mediterranean-origin cyclogenesis [19,52], which brings strong precipitation over northern Italy. Even in northern Africa, the stalagmite-based rainfall index from Morocco shows that regional precipitation increased from 0 to 400 A.D. (Figure 2E; [28), consistent with the observations at Bàsura (Figure 2B). Accordingly, centennial-scale humid conditions in southern Europe could be attributed to warm Mediterranean seawater, instead of the NAO state.

Rapid positive δ¹³C and δ¹⁸O shifts in the precisely dated Bàsura records (Figure 2B) reveal a dramatic drop from 341 ± 14 to 432 ± 17 A.D. This 100-yr precipitation decrease is also documented at Lake Savine (Figure 2C) and in Romania (Figure 2D). The agreement indicates a large-scale century-long hydroclimate shift at the end of the RWP. Bàsura isotopic records also show a centennial-scale drought lasting from 432 ± 17 to 526 ± 4 A.D. The drying trend from 341 ± 14 to 432 ± 17 A.D. and the following 100-yr drought might be a factor that contributed to the collapse of the Roman Empire in 476 A.D [6,13].

4. Conclusions

We present a 1500 yr stalagmite proxy hydroclimate record from Bàsura cave, northern Italy from 550 B.C. to 950 A.D., covering the RWP from 200 B.C. to 400 A.D. Our results show that during the RWP, southern Europe experienced a warm and humid climate. A concurrent warming trend observed in Mediterranean SST records suggests that an increased land-sea temperature contrast and increasing cyclones could be possible factors contributing to multi-centennial wet conditions in southern Europe during the RWP. A succession involving a rapid 100-yr drying shift from 341 ± 14 to 432 ± 17 A.D. and a 100-yr drought event from 432 ± 17 to 526 ± 4 A.D. are observed in the Bàsura stalagmite proxy record and other natural archives. A dramatic hydroclimate change from wet to dry conditions from 341 ± 14 to 432 ± 17 A.D. may have influenced the collapse of the Roman Empire in 476 A.D.