Regional pollen-based Holocene temperature and precipitation patterns depart from the Northern Hemisphere mean trends

Ulrike Herzschuh, Thomas Böhmer, Manuel Chevalier, Raphaël Hébert, Anne Dallmeyer, Chenzhi Li, Xianyong Cao, Odile Peyron, Larisa Nazarova, Елена Новенко, Jungjae Park, Natalia Rudaya, Frank Schlütz, Lyudmila Shumilovskikh, Pavel E. Tarasov, Yongbo Wang, Ruilin Wen, Qinghai Xu, Zhuo Zheng

Abstract

A mismatch between model- and proxy-based Holocene climate change, known as the “Holocene conundrum”, may partially originate from the poor spatial coverage of climate reconstructions in, for example, Asia, limiting the number of grid cells for model–data comparisons. Here we investigate hemispheric, latitudinal, and regional mean time series and time-slice anomaly maps of pollen-based reconstructions of mean annual temperature, mean July temperature, and annual precipitation from 1908 records in the Northern Hemisphere extratropics. Temperature trends show strong latitudinal patterns and differ between (sub-)continents. While the circum-Atlantic regions in Europe and eastern North America show a pronounced Middle Holocene temperature maximum, western North America shows only weak changes, and Asia mostly shows a continuous Holocene temperature increase. Likewise, precipitation trends show certain regional peculiarities such as the pronounced Middle Holocene precipitation maximum between 40 and 50° N in Asia and Holocene increasing trends in Europe and western North America, which can all be linked with Holocene changes in the regional circulation pattern responding to temperature change. Given a background of strong regional heterogeneity, we conclude that the calculation of global or hemispheric means, which initiated the Holocene conundrum debate, should focus more on understanding the spatiotemporal patterns and their regional drivers.

Introduction: The Holocene Conundrum

Paleoclimatologists have long been puzzled by the Holocene Conundrum: the discrepancy between climate models, which generally show a steady increase in global temperatures over the last 11,000 years, and "proxy" datasets, which often suggest temperatures peaked about 6,000 years ago (the Mid-Holocene Thermal Maximum) followed by cooling.

By synthesizing nearly 2,000 pollen-based records across the Northern Hemisphere, this research suggests that this mismatch is largely a product of over-averaging and spatial bias. Most high-quality proxy records historically came from the North Atlantic region (Europe and Eastern North America), which was heavily influenced by the melting Laurentide ice sheet. Large swaths of the planet, particularly the Asian interior, have been underrepresented in global averages. This study fills those gaps to determine if a more geographically balanced dataset changes the narrative.

Methods and Datasets

The researchers utilized pollen-based reconstructions to estimate past temperatures (Mean Annual and July) and precipitation. Pollen is an excellent proxy because different plant species have specific "climate envelopes"; fossilized pollen in lake sediments or peat allows scientists to infer past conditions.

The study integrated several massive databases, primarily LegacyClimate 1.0 (containing 2,593 reconstructions) and Temp12k. The authors standardized 1,908 records, ensuring consistent taxonomies and chronologies across the Northern Hemisphere.

Spatial and temporal distribution of the 1,908 pollen-based records used in the study, showing (a) record locations and (b) the number of records covering distinct time periods.

Technical methods included:

  • WA-PLS (Weighted-Averaging Partial Least-Squares): A regression technique relating modern pollen assemblages to climate variables.
  • MAT (Modern Analog Technique): Estimating past conditions by finding modern vegetation samples similar to fossil samples.
  • Zonal Averaging: Calculating means for 10° latitudinal bands and specific sub-continents to capture regional heterogeneity rather than a single "global" number.

Regional Heterogeneity and Key Results

The study reveals that the Mid-Holocene Thermal Maximum is not a global phenomenon but a "circum-Atlantic" pattern.

  • Europe and Eastern North America: These regions showed a pronounced July temperature maximum of +1.3 ± 0.4 °C around 5.7 ka (thousand years ago).
  • Asia: In contrast, Asia showed almost no mid-Holocene maximum, instead following a trend of continuous temperature increase.
  • Precipitation Decoupling: Precipitation trends varied significantly. Asia saw a moisture maximum between 7–5 ka (likely due to monsoon dynamics), while Europe and Western North America showed a steady increase in precipitation throughout the Holocene.
Density of pollen-based time series across the Northern Hemisphere, highlighting the spatial distribution and concentration of data points.

When Asia is properly weighted in the hemispheric average, the "cooling" trend of the late Holocene becomes much less pronounced. The revised Northern Hemisphere mean annual temperature shows a warming of only 0.08 ± 0.04 °C between the early Middle Holocene and the Late Holocene—a much flatter trend than previous North Atlantic-centric studies suggested.

Limitations and Future Directions

While this synthesis provides a more balanced view, several challenges remain:

  • Vegetation Lag: Vegetation can lag behind climate changes by centuries, potentially smoothing rapid temperature shifts.
  • Human Impact: Long-term agriculture and grazing in Europe and China may bias recent pollen-based reconstructions.
  • Seasonal Bias: Pollen may be more sensitive to summer temperatures than annual means.
  • Southern Hemisphere Gap: This study is strictly Northern Hemisphere. Solving the global Holocene Conundrum requires similar high-density syntheses for the Southern Hemisphere and the tropics.

Summary updated

Abstract

A mismatch between model- and proxy-based Holocene climate change, known as the “Holocene conundrum”, may partially originate from the poor spatial coverage of climate reconstructions in, for example, Asia, limiting the number of grid cells for model–data comparisons. Here we investigate hemispheric, latitudinal, and regional mean time series and time-slice anomaly maps of pollen-based reconstructions of mean annual temperature, mean July temperature, and annual precipitation from 1908 records in the Northern Hemisphere extratropics. Temperature trends show strong latitudinal patterns and differ between (sub-)continents. While the circum-Atlantic regions in Europe and eastern North America show a pronounced Middle Holocene temperature maximum, western North America shows only weak changes, and Asia mostly shows a continuous Holocene temperature increase. Likewise, precipitation trends show certain regional peculiarities such as the pronounced Middle Holocene precipitation maximum between 40 and 50∘ N in Asia and Holocene increasing trends in Europe and western North America, which can all be linked with Holocene changes in the regional circulation pattern responding to temperature change. Given a background of strong regional heterogeneity, we conclude that the calculation of global or hemispheric means, which initiated the Holocene conundrum debate, should focus more on understanding the spatiotemporal patterns and their regional drivers.
BibTeX
@inproceedings{herzschuh2023regional,
  author = {Ulrike Herzschuh and Thomas Böhmer and Manuel Chevalier and Raphaël Hébert and Anne Dallmeyer and Chenzhi Li and Xianyong Cao and Odile Peyron and Larisa Nazarova and Елена Новенко and Jungjae Park and Natalia Rudaya and Frank Schlütz and Lyudmila Shumilovskikh and Pavel E. Tarasov and Yongbo Wang and Ruilin Wen and Qinghai Xu and Zhuo Zheng},
  title = {{Regional pollen-based Holocene temperature and precipitation patterns depart from the Northern Hemisphere mean trends}},
  booktitle = {Climate},
  year = {2023},
  doi = {10.5194/cp-19-1481-2023},
}
Figure