Weighted Averaging Partial Least Squares Research Articles

A modern pollen dataset from lake surface sediments on the central and western Tibetan Plateau

Abstract. Modern pollen datasets are essential for pollen-based quantitative paleoclimate (e.g. precipitation) reconstructions, which can aid in better understanding recent climate change and its underlying forcing mechanisms. A modern pollen dataset based on surface sediments from 90 lakes in the shrub, meadow, steppe and desert regions of the central and western Tibetan Plateau (TP) was established to fill geographical gaps left by previous datasets. Ordination analyses of pollen data and climatic parameters revealed that annual precipitation is the dominant factor in modern pollen distribution on the central and western TP. A regional transfer function for annual precipitation was developed with weighted averaging partial least squares (WA-PLS), which suggests a good inference power of the modern pollen dataset for annual precipitation. A case study in which the transfer function was effectively applied to a fossil pollen record from Tangra Yumco on the central TP for paleoprecipitation reconstruction demonstrated the significance of the modern pollen dataset in a lower data region for paleoclimate change studies. Data from this study, including pollen data for each sample and information on the sampled sites (location, altitude and climate data), are openly available via the Zenodo portal (https://doi.org/10.5281/zenodo.8008474, Ma et al., 2023).

Modern pollen–climate relationships and their application for pollen-based quantitative climate reconstruction of the mid-Holocene on the southern Korean Peninsula

A modern pollen dataset is a prerequisite for reconstructing quantitative paleoclimate and paleovegetation cover using fossil pollen records. Although multiple modern pollen–climate datasets have been established covering a wide range of climate conditions, such datasets are exceedingly rare for the Korean Peninsula (KP). In this study, we acquired a modern pollen dataset from 198 surface soil samples collected on 37 mountains on the southern KP. Redundancy analysis (RDA) and variation partitioning results identified mean annual temperature (Tann) as the most important climate variable shaping pollen assemblages on the southern KP. Using the pollen–climate relationships inferred from the RDA, we applied the Huisman–Olff–Fresco model and determined that arboreal pollen taxa responded sensitively to the climatic gradient, whereas non-arboreal pollen taxa did not. We applied weighted averaging–partial least squares (WA-PLS) and the modern analog technique (MAT) to the pollen dataset, and a comparison of the results showed that MAT performed better than WA-PLS. A transfer function was applied to fossil pollen records from the areas covered by our dataset; the results confirmed that annual precipitation (Pann) and Tann were modulated by different mechanisms, with Pann strongly affected by El Niño–Southern Oscillation-driven typhoons during the Holocene, whereas Tann was mainly influenced by the Tsushima Warm Current from 7500 to 5100 cal yr BP depending on Kuroshio Current inflow intensity, and subsequently followed by the East Asian winter monsoon during 5100–3400 cal yr BP.

Wet mid–late Holocene in central Asia supported prehistoric intercontinental cultural communication: Clues from pollen data

Past precipitation patterns in arid central Asia can help us to understand better the situation of the Eurasian steppe-road of prehistoric intercontinental culture communication. This requires reliable quantitative pollen-based reconstructions to reveal the spatiotemporal pattern of past precipitation (pollen is the most feasible proxy for quantitative reconstruction and spatial comparison). In this study, modern pollen data (n = 1336) from arid central Asian sites with a mean annual precipitation (Pann) less than 400 mm were used to explore the influence of Pann on modern pollen assemblages using canonical correlation analysis (CCA) and to create a pollen-Pann calibration-set using Weighted-Averaging Partial Least Squares (WA-PLS) and Random Forest (RF). We then reconstructed past Pann variations for 10 Holocene pollen spectra from north-western China and Mongolia. The reconstructions based on the two approaches were compared with short-term reconstructions and meteorological records (past 55 years) to assess their performance, with statistical significance testing of the reconstructions and assessment of analogue quality. Our results show that RF has a slightly better model performance than WAPLS, and is also a suitable approach for past climate reconstruction using pollen data. Lacustrine pollen spectra reflect the overall pattern of multi-year climate changes rather than interannual climate variations. Pann reconstructions show that eastern arid central Asia had a relatively dry early Holocene (11.5–8 cal ka BP) and a stable humid mid–late Holocene (after 6 ka), which supported prehistoric intercontinental cultural communication.

Lake surface sediment pollen dataset for the alpine meadow vegetation type from the eastern Tibetan Plateau and its potential in past climate reconstructions

Abstract. A modern pollen dataset with an even distribution of sites is essential for pollen-based past vegetation and climate estimations. As there were geographical gaps in previous datasets covering the central and eastern Tibetan Plateau, lake surface sediment samples (n=117) were collected from the alpine meadow region on the Tibetan Plateau between elevations of 3720 and 5170 m a.s.l. Pollen identification and counting were based on standard approaches, and modern climate data were interpolated from a robust modern meteorological dataset. A series of numerical analyses revealed that precipitation is the main climatic determinant of pollen spatial distribution: Cyperaceae, Ranunculaceae, Rosaceae, and Salix indicate wet climatic conditions, while Poaceae, Artemisia, and Chenopodiaceae represent drought. Model performance of both weighted-averaging partial least squares (WA-PLS) and the random forest (RF) algorithm suggest that this modern pollen dataset has good predictive power in estimating the past precipitation from pollen spectra from the eastern Tibetan Plateau. In addition, a comprehensive modern pollen dataset can be established by combining our modern pollen dataset with previous datasets, which will be essential for the reconstruction of vegetation and climatic signals for fossil pollen spectra on the Tibetan Plateau. Pollen datasets including both pollen counts and percentages for each sample, together with their site location and climatic data, are available at the National Tibetan Plateau Data Center (TPDC; Cao et al., 2021; https://doi.org/10.11888/Paleoenv.tpdc.271191).

Comparison of boosted regression trees vs WA-PLS regression on diatom-inferred glacial-interglacial climate reconstruction in Lake Tiancai (southwest China)

Quantitative paleoclimate reconstructions based on biological fossils over glacial-interglacial timescales are a major source of information on long-term climate variability. However, such reconstructions can present major methodological challenges, as calibration methods based on modern climatic and biological patterns may yield biased results, especially when data are particularly sensitive to the influence of secondary environmental variables in the training set or fossil assemblage. A machine-learning technique, boosted regression tree (BRT), is compared with a weighted averaging-partial least squares (WA-PLS) regression in a diatom-based summer (mean July) temperature transfer function with a calibration set of 273 samples collected from China. Using these calibration methods and a fossil diatom sequence from Lake Tiancai (southwest China; 3898 m a.s.l.), the summer temperatures covering the last glaciation and entire Holocene (18000 cal yr BP to present) are reconstructed. The record is validated by applying several statistical reconstruction diagnostics and compared with pollen and diatom records from the same sediment core. The results suggest that the BRT-based calibration model, which produced a coefficient of determination (r2 = 0.85) and root-mean-square error of prediction (RMSEP = 1.65 °C) offers substantial improvements over WA-PLS. The record shows that the summer temperature varies by ~2.5 °C across the entire period. The temperature evolution inferred by diatom-based BRT model is characterized by an early cool phase (with a MJT of 9.3 °C) prior to 10000 cal yr BP, followed by a clear mid-Holocene thermal maximum (~1.1 °C above the present-day temperature) from 6000 to 3500 cal yr BP and a subsequent cooling that ended at 1500 cal yr BP. This contrasts with the chironomid-based temperature transfer function in the same sequence showing a warmer period between 8500 and 6000 cal yr BP. The overall pattern of diatom-based summer temperature reconstructions broadly matches the climate change inferred from pollen records from Tiancai Lake and other records from the surrounding regions, suggesting that summer temperatures respond to insolation forcing and the variability of the southwest monsoon. We argue that the BRT-based climate model potentially outperforms the other models in southwest China when using a large and non-linear calibration set of diatoms for reconstructing long-term temperature.

Anthropogenic modification of soil communities in northern China for at least two millennia: Evidence from a quantitative mollusk approach

Reconstructing and quantifying human impacts is an important step in understanding how, when, and to what extent humans have changed terrestrial ecosystems via agricultural land use during the Holocene. However, the use of soil-faunal indicators and an associated quantitative model of land use intensity to study anthropogenic disturbance is relatively poorly developed. In this study we collected 139 samples of soil-dwelling snails from urban land and human settlements, cultivated land, modified and natural vegetation mosaics, and natural habitats in northern China, in order to characterize the snail assemblages in human-modified habitats. The results show that cultivated land and planted forest are mainly dominated by snails of Cathaica fasiola, Bradybaena ravida, Vallonia tenera, etc., while natural habitats are characterized by an even distribution of members of the families, such as Ariophantidae, Cochlicopidae, Pupillidae, Subulinidae. Canonical correspondence analysis indicates that human activity (quantified by the human influence index, HII) is one of the most important factors shaping the composition of snail fauna in man-modified habitats. There is a significant relationship between modern snail data and HII, and the transfer function established by weighted-averaging partial least squares (WA-PLS) model for HII exhibits a good statistical performance. We then applied the calibration model to a fossil snail record spanning the last 12 kyr from the Chinese Loess Plateau (CLP). The reconstructed HII increases slightly at ∼5 ka and abruptly at ∼2 ka, coinciding with increases in the human population of the CLP, local cultural development, and human-induced changes in vegetation cover. Our quantitative reconstruction indicates that human activities, via agricultural activity, may have permanently altered natural ecosystems and the soil fauna for at least the last two millennia.

The impact of methodological decisions on climate reconstructions using WA-PLS

AbstractMost techniques for pollen-based quantitative climate reconstruction use modern assemblages as a reference data set. We examine the implication of methodological choices in the selection and treatment of the reference data set for climate reconstructions using Weighted Averaging Partial Least Squares (WA-PLS) regression and records of the last glacial period from Europe. We show that the training data set used is important because it determines the climate space sampled. The range and continuity of sampling along the climate gradient is more important than sampling density. Reconstruction uncertainties are generally reduced when more taxa are included, but combining related taxa that are poorly sampled in the data set to a higher taxonomic level provides more stable reconstructions. Excluding taxa that are climatically insensitive, or systematically overrepresented in fossil pollen assemblages because of known biases in pollen production or transport, makes no significant difference to the reconstructions. However, the exclusion of taxa overrepresented because of preservation issues does produce an improvement. These findings are relevant not only for WA-PLS reconstructions but also for similar approaches using modern assemblage reference data. There is no universal solution to these issues, but we propose a number of checks to evaluate the robustness of pollen-based reconstructions.

Modern pollen rain, vegetation and climate along elevation gradients in the upper-middle Yellow River: numerical approaches for quantitative environmental reconstruction

A total of 245 modern pollen samples available in the East Asian Pollen Database and our unpublished pollen data (n = 204 and 41, respectively) were compiled to determine the correlation between modern pollen rain and vegetation in the watershed regions along the upper and middle Yellow River (UMYR) using principle component analysis (PCA). Despite considerable geographical differences and an uneven spatial vegetation distribution, pollen assemblages could differentiate major vegetation and climate zones. The PCA result reveals a high correlation between pollen taxa and key environmental factors, suggesting the high potential of this database for quantitative climate reconstruction. After screening, 199 samples were finally used to quantitatively reconstruct climate parameters with three methods namely, the weighted averaging partial least squares (WA-PLS) method, locally weighed weighted averaging (LWWA) method, and modern analogue technique (MAT).The comparison of multiple reconstructions indicates that WA-PLS provides quite accurate climate reconstructions from limited pollen data, whereas LWWA and MAT appear more suitable for large pollen datasets associated with a broad geographic range. We also discovered that the selection of sub-divided watershed areas yielded significant reduction in prediction errors and, in turn, improved accuracy of the quantitative climate reconstruction. The statistical results indicate that the transfer functions are reliable for both climate variables and elevation that is controlled mainly by the lapse rate of the altitudinal temperature. The current study provides evidence that modern pollen data in the UMYR can be used as a reliable reference dataset for pollen-based quantitative climate reconstructions.

Expanding the Swiss‐Norwegian chironomid training set with Polish data

The Swiss‐Norwegian training model established in 2011 has been used across Europe to infer past mean July air temperatures using Chironomidae subfossils. It combines reference sites from Norway and the Swiss Alps but lacks representative localities from the central European plains belt. This gap complicates reconstructions from temperate European latitudes, and thus, a Polish training set (TS) was developed. Data from 102 lakes, mainly from Pomerania and the Masurian Lakelands, as well as the Polish Plain, the Sudetes, and the Carpathian Mountains were included. After selecting and merging suitable chironomid taxa, the Swiss‐Norwegian‐Polish TS includes 357 lakes and 134 chironomid morphotypes, and it has a mean July air temperature range of 3.5 to 20.0 °C. The weighted averaging‐partial least squares (WA‐PLS) and artificial neural network (ANN) transfer function based on the extended TS performs well: R2jack = 0.91/0.95 and root mean squared error of prediction (RMSEP) = 1.39 °C/1.34 °C, respectively. The transfer function applied to the Żabieniec Lake sequence (central Poland) reveals higher summer temperatures (2.5–4.0 °C) for the Lateglacial and for the Holocene (1.0 °C) compared to the Swiss‐Norwegian model. The reconstruction based on the Swiss‐Norwegian TS shows an increasing trend from the Bølling‐Allerød Interstadial to the Meghalayan period, whereas the new model reconstruction shows no increasing trend when using WA‐PLS and a slightly decreasing trend when using ANN. Summer temperatures in the Lateglacial in Żabieniec inferred using the Swiss‐Norwegian TS are consistent with other temperature reconstructions for 50–55°N in Europe, while those based on the Swiss‐Norwegian‐Polish TS (SNP TS) show closer affinity to more western sites than the eastern European ones. Despite being uncommon, the application of ANN can be a good equivalent of the WA‐PLS technique especially in situations where the available database is limited to fewer than a few hundred variables.

Ecology of peatland testate amoebae in the Alaskan continuous permafrost zone

Arctic peatlands represent a major global carbon store, but rapid warming poses a threat to their long-term stability. Testate amoebae are sensitive hydrological indicators that offer insight into Holocene environmental change in peatlands. However, in contrast to temperate peatlands, there have only been a few studies into the ecology of testate amoebae and their efficacy as environmental indicators in permafrost peatlands. We present the first study of testate amoeba ecology from peatlands in the continuous permafrost zone, based on samples from across the Alaskan North Slope. Multivariate statistical analyses show that pore water electrical conductivity (EC), a proxy for nutrient status along the ombrotrophic-minerotrophic gradient, is the dominant control on testate amoeba distribution. Water-table depth (WTD) is also a significant control on testate amoeba distribution, but is secondary to EC. We present two new testate amoeba-based transfer functions to reconstruct both EC (TFEC) and WTD (TFWTD), the first for peatlands in the continuous permafrost zone. The transfer functions are based on Weighted Averaging Partial Least Squares (WAPLS) regression and were assessed using leave-one-out (LOO) cross-validation. We find that both transfer functions have good predictive power. TFWTD is the best performing model (R2JACK = 0.84, RMSEPJACK = 6.66 cm), but TFEC also performs well (R2JACK = 0.76, RMSEPJACK = 146 μS cm−1). Our findings are similar to those conducted in peatlands in discontinuous permafrost regions. The new transfer functions open the opportunity for reconstructing the Holocene dynamics of peatlands of the continuous permafrost zone in Alaska, which represent rapidly changing ecosystems.

Phytolith-inferred transfer function for paleohydrological reconstruction of Dajiuhu peatland, central China

Depth to water table (DWT, the depth from the water surface to the top of the peat surface) is one of the most important environmental variables related to the habitat types and distribution of vegetation within a subalpine peatland. The distribution of phytolith assemblages and basic environmental data from 43 surface soil samples with significant ecological and hydrological gradients were investigated to generate transfer functions for quantitative reconstruction of paleoenvironmental changes in Dajiuhu peatland, central China. Detrended correspondence analysis (DCA) and redundancy analysis (RDA) were employed to explore the relationship between main environmental variables and phytolith morphotypes and distributions. Our results indicate that the spatial distribution of phytoliths was significantly correlated with the DWT (25% variance), total organic carbon (TOC, 10% variance) and magnetic susceptibility (χ, 7% variance). We established the transfer functions for the significant variables based on modern analogue technique (MAT), weighted averaging techniques (WA) and weighted averaging partial least squares (WA-PLS), and model performance was assessed using bootstrap cross-validation. The high correspondence of the predicted DWT results based on phytolith-environment calibration data with observed data reflects that the phytolith-based WA-PLS is a reliable effective calibration method for the quantitative DWT reconstruction of ombrotrophic (rain-fed) subalpine peatland.

Evidence of Holocene water level changes inferred from diatoms and the evolution of the Honghe Peatland on the Sanjiang Plain of Northeast China

Diatom-inferred changes in past water levels and the evolution of peatlands during the Holocene period are investigated in the Honghe National Natural Reserve (HNNR), which is located on the Sanjiang Plain of Northeast China. This investigation began with analyses of fossil diatoms and the corresponding water environments from the surface sediments of peatlands. The detrended correspondence analysis (DCA) and redundancy analysis (RDA) methods show that the water level is one of the major factors that drives the distribution of diatom assemblages in the peatlands of the HNNR. A diatom-based inference model is developed, and the Holocene water levels are reconstructed based on the inference model. Applied to a one-component weighted averaging partial least squares (WAPLS) model, the jackknife statistical test yields a relatively high correlation coefficient and a low root-mean-squared error of prediction (R2jack = 0.220; RMSEPjack = 0.2339) compared with a series of models. The peatland water level was relatively high from 6.5 to 2.8 ka (1 ka = 1000 cal yr BP), and species that prefer deep water, such as Aulacoseira crenulata (Ehrenberg) Krammer and Navicula cuspidata Kützing, were dominant at that time. The water level also increased markedly from 2.8 to 2.5 ka and subsequently declined from 2.5 ka to the present. A clear reciprocal shift is noted between an assemblage of deep-water species that includes A. crenulata and species that prefer sphagnum bogs, including Nitzschia perminuta (Grunow) Peragallo. This evidence suggests that the water level of the HNNR peatland became shallower, implying the presence of a paleolake in the early or middle Holocene that was followed by the development of the peatland, which is a result of paludification, up to the present time. These results strengthen the potential for paleo-reconstructions of the hydrology of the peatlands of Northeast China using multiple proxies.

From micromammals to paleoenvironments

Micromammal assemblages are often used in paleoecological reconstructions. However, in studies to date, the environmental variables reconstructed were not shown to drive micromammal community structure before the reconstruction was attempted. Furthermore, the relative abundance data is too often dismissed despite its potential paleoecological value. Here we chose to reconstruct the botanical succession stage known to impact micromammal communities in the Mediterranean zone of Israel. We used weighted averaging partial least squares (WAPLS) regression to model the connection between botanical succession phase and the micromammal faunas as reflected in Tyto alba pellets. In addition to cross-validation, we validated our model using historical T. alba pellet data and contemporary maps. Moreover, we tested the repercussions of misidentifying the owl species in archaeological micromammal assemblages by applying our model to Recent assemblages collected by other owl species. Our WAPLS model differentiated forested environments from garrigue or grassland dominated ones in both the cross-validation and in the reconstructed historical environments. However, urban environments were not well distinguished from grasslands and garrigues, and misidentifying the owl species severely compromised model performance. Our results stress the potential insight deducible from micromammal relative abundance data, given its reliance on empirically tested relationships between fauna, environment, and taphonomic agent (predator).

Impacts of the spatial extent of pollen-climate calibration-set on the absolute values, range and trends of reconstructed Holocene precipitation

Pollen-based quantitative reconstructions of past climate variables is a standard palaeoclimatic approach. Despite knowing that the spatial extent of the calibration-set affects the reconstruction result, guidance is lacking as to how to determine a suitable spatial extent of the pollen-climate calibration-set. In this study, past mean annual precipitation (Pann) during the Holocene (since 11.5 cal ka BP) is reconstructed repeatedly for pollen records from Qinghai Lake (36.7°N, 100.5°E; north-east Tibetan Plateau), Gonghai Lake (38.9°N, 112.2°E; north China) and Sihailongwan Lake (42.3°N, 126.6°E; north-east China) using calibration-sets of varying spatial extents extracted from the modern pollen dataset of China and Mongolia (2559 sampling sites and 168 pollen taxa in total). Results indicate that the spatial extent of the calibration-set has a strong impact on model performance, analogue quality and reconstruction diagnostics (absolute value, range, trend, optimum). Generally, these effects are stronger with the modern analogue technique (MAT) than with weighted averaging partial least squares (WA-PLS). With respect to fossil spectra from northern China, the spatial extent of calibration-sets should be restricted to radii between ca. 1000 and 1500 km because small-scale calibration-sets (<800 km radius) will likely fail to include enough spatial variation in the modern pollen assemblages to reflect the temporal range shifts during the Holocene, while too broad a scale calibration-set (>1500 km radius) will include taxa with very different pollen-climate relationships.

Examining bias in pollen-based quantitative climate reconstructions induced by human impact on vegetation in China

Abstract. Human impact is a well-known confounder in pollen-based quantitative climate reconstructions as most terrestrial ecosystems have been artificially affected to varying degrees. In this paper, we use a human-induced pollen dataset (H-set) and a corresponding natural pollen dataset (N-set) to establish pollen–climate calibration sets for temperate eastern China (TEC). The two calibration sets, taking a weighted averaging partial least squares (WA-PLS) approach, are used to reconstruct past climate variables from a fossil record, which is located at the margin of the East Asian summer monsoon in north-central China and covers the late glacial Holocene from 14.7 ka BP (thousands of years before AD 1950). Ordination results suggest that mean annual precipitation (Pann) is the main explanatory variable of both pollen composition and percentage distributions in both datasets. The Pann reconstructions, based on the two calibration sets, demonstrate consistently similar patterns and general trends, suggesting a relatively strong climate impact on the regional vegetation and pollen spectra. However, our results also indicate that the human impact may obscure climate signals derived from fossil pollen assemblages. In a test with modern climate and pollen data, the Pann influence on pollen distribution decreases in the H-set, while the human influence index (HII) rises. Moreover, the relatively strong human impact reduces woody pollen taxa abundances, particularly in the subhumid forested areas. Consequently, this shifts their model-inferred Pann optima to the arid end of the gradient compared to Pann tolerances in the natural dataset and further produces distinct deviations when the total tree pollen percentages are high (i.e. about 40 % for the Gonghai area) in the fossil sequence. In summary, the calibration set with human impact used in our experiment can produce a reliable general pattern of past climate, but the human impact on vegetation affects the pollen–climate relationship and biases the pollen-based climate reconstruction. The extent of human-induced bias may be rather small for the entire late glacial and early Holocene interval when we use a reference set called natural. Nevertheless, this potential bias should be kept in mind when conducting quantitative reconstructions, especially for the recent 2 or 3 millennia.

Diatom (Bacillariophyceae) assemblages in salt marshes of south‐central Chile: Relations with tidal inundation time and salinity

SUMMARYTo reconstruct sea‐level history from changes in tidal environments using diatom assemblages, we need to better understand the relations among brackish diatom assemblages and changing environments along elevational gradients from diverse coastal sites. Our statistical analysis reveals relations between environmental variables and brackish benthic diatom assemblages in the little studied region of south‐central Chile. Along four transects across salt marshes at two sites, we identified 224 diatom taxa in 112 samples. Detrended canonical correspondence analysis showed that tidal exposure time index and salinity were appropriately regressed against the abundance of diatom species using unimodal‐based methods. Our tests of classical and inverse regressions of weighted average and weighted averaging partial least squares (WA‐PLS) showed that WA‐PLS resulted in the highest coefficient of determination and the lowest root‐mean square of the error of prediction. Our regression will be useful in reconstructing environmental variables from fossil diatom assemblages in Chile.

Impact of eustatic and tectonic processes on the southeastern Mediterranean shelf during the last one million years: Quantitative reconstructions using a foraminiferal transfer function

Benthic foraminifera-based transfer function models provide accurate sea-level reconstructions for shelf and coastal environments around the world. This study presents for the first time results of a foraminiferal transfer function aiming at the reconstruction of relative sea-level in the southeastern Mediterranean shelf during the past one million years. A Weighted Averaging-Partial Least Squares (WA-PLS) regression method was used for the development of the transfer function based on an extensive set of surface samples (modern training data set) from the study area. The transfer function was applied to three shelf sediment successions retrieved from Haifa Bay, Israel. The Modern Analog Technique (MAT) and the random TF test were used to investigate similarity between modern and fossil foraminiferal faunas, testing the reliability and significance of our model for paleowater depth reconstruction. The performance of the WA-PLS transfer functions suggests a strong relationship between the observed and estimated water depths in the modern data set with relatively precise reconstructions of 7.0m (Root Mean Squared error of Prediction) over a water depth range of 3 to 80m. The deepest paleo water depths were reconstructed for the earliest sea-level highstands recorded in the sediment records (MIS 27? and MIS 29?) in the down-faulted boreholes due to tectonic movements, whereas the younger interglacial sedimentary units (MIS 25?, 13, 11, 7, 5.5 and 1) were mostly deposited at <15m water depths or slightly deeper during MIS 7. The relative sea-level history of the southeastern Mediterranean shelf reveals deviations from the global sea-level trend during MIS 5.5, which may have resulted from minor tectonic movements in the study area and/or un-representation of the phase of maximum sea-level in the sediment succession. Our paleowater depth estimates demonstrate a shallowing trend during the Holocene, generally matching other regional Mediterranean sea-level records.

East Asian summer monsoon precipitation variations in China over the last 9500 years: A comparison of pollen-based reconstructions and model simulations

To better understand the long-term changes of the East Asian summer monsoon precipitation ( Pjja), quantitative reconstructions and model simulations are needed. Here, we develop continental-scale pollen-based transfer functions for Pjja with weighted averaging–partial least squares (WA-PLS) regression and a Bayesian multinomial regression method. We apply these transfer functions to a set of fossil pollen data from monsoonal China for quantitatively reconstructing the Pjja changes over the last 9500 years. We compare the reconstructions with Pjja simulations from a coupled atmosphere–ocean–sea ice general circulation model (the Kiel Climate Model, KCM). The results of cross-validation tests for the transfer functions show that both the WA-PLS model ( r2 = 0.83, root mean square error of prediction (RMSEP) = 112.11 mm) and the Bayesian model ( r2 = 0.86, RMSEP = 107.67 mm) exhibit good predictive performance. We stack all Pjja reconstructions from northern China to a summary curve. The stacked record reveals that Pjja increased since 9500 cal. yr BP, attained its highest level during the Holocene summer monsoon maximum (HSMM) at ~7000–4000 cal. yr BP and declined to present. The KCM output and the reconstructions differ in the early-Holocene (~9500–7000 cal. yr BP) where the model suggests higher Pjja than the reconstructions. Moreover, during the HSMM, the amplitude of the Pjja changes (~20–60 mm above present) in simulations is lower than the reconstructed changes (~70–110 mm above present). The rising (declining) Pjja patterns in reconstructions before (after) the HSMM are more pronounced and fluctuating than in simulations. Other palaeohydrological data such as lake-level reconstructions indicate substantial monsoon precipitation changes throughout the Holocene. Our results therefore show that the KCM underestimates the overall amplitude of the Holocene monsoon precipitation changes.

Pollen-based temperature reconstructions from Jeju island, South Korea and its implication for coastal climate of East Asia during the late Pleistocene and early Holocene

We derived statistically robust, Weighted Averaging Partial Least Squares (WAPLS) transfer functions using modern surface pollen samples from Mt. Halla in Jeju Island, Korea. Pollen-based temperature reconstructions were performed by applying the best transfer function to lake sediments from Hanon maar paleolake in the southern part of Jeju Island. These quantitatively reconstructed temperatures were thereafter adjusted for the lack of warm analogue with the aid of Japanese and Russian transfer functions. Our temperature reconstructions demonstrated that annual mean temperatures during the Last Glacial Maximum (LGM) were ~8.5°C, i.e., about 8°C lower than the present, and that the Younger Dryas (YD) cooling amplitude was ~1.5°C between Bølling–Allerød (BA) (~11.3°C) and YD (~9.8°C). Spectral and wavelet analyses on detrended paleotemperatures identified significant periodicities at 1340years, primarily after 13,000calyr BP. The influence of Antarctic deglacial climate on the western North Pacific probably decreased with increasing northern latitude, as expected. Our results and Borneo oxygen isotope data showed very high correlation, indicating that Jeju Island was substantially influenced by the deglacial variabilities of the western Pacific warm pool via the Kuroshio Current. The climate of the study area seemed to have been mainly controlled by variation in the Atlantic Meridional Overturning Circulation (AMOC) before the beginning of the last deglaciation and subsequently by variation in the Western Tropical Pacific (WTP) Sea Surface Temperature (SST). We present significant information on the correlation of the late Pleistocene climate change with the Greenland oxygen isotope data and western North Pacific SST records. This correlation bridges the gap between terrestrial records, which mostly show AMOC variability, and oceanic records, which show WTP variability. These study results will increase our understanding of regional-scale climate change during the late Pleistocene and Early Holocene.

A modern pollen–climate dataset from China and Mongolia: Assessing its potential for climate reconstruction

A modern pollen dataset from China and Mongolia (18–52°N, 74–132°E) is investigated for its potential use in climate reconstructions. The dataset includes 2559 samples, 229 terrestrial pollen taxa and four climatic variables — mean annual precipitation (Pann): 35–2091mm, mean annual temperature (Tann): –12.1–25.8°C, mean temperature in the coldest month (Mtco): –33.8–21.7°C, and mean temperature in the warmest month (Mtwa): 0.3–29.8°C. Modern pollen–climate relationships are assessed using canonical correspondence analysis (CCA), Huisman–Olff–Fresco (HOF) models, the modern analogue technique (MAT), and weighted averaging partial least squares (WA-PLS). Results indicate that Pann is the most important climatic determinant of pollen distribution and the most promising climate variable for reconstructions, as assessed by the coefficient of determination between observed and predicted environmental values (r2) and root mean square error of prediction (RMSEP). Mtco and Mtwa may be reconstructed too, but with caution. Samples from different depositional environments influence the performance of cross-validation differently, with samples from lake sediment-surfaces and moss polsters having the best fit with the lowest RMSEP. The better model performances of MAT are most probably caused by spatial autocorrelation. Accordingly, the WA-PLS models of this dataset are deemed most suitable for reconstructing past climate quantitatively because of their more reliable predictive power.