Dataset Information

Machine-learning based reconstructions of primary and secondary climate variables from North American and European fossil pollen data.

ABSTRACT: We test several quantitative algorithms as palaeoclimate reconstruction tools for North American and European fossil pollen data, using both classical methods and newer machine-learning approaches based on regression tree ensembles and artificial neural networks. We focus on the reconstruction of secondary climate variables (here, January temperature and annual water balance), as their comparatively small ecological influence compared to the primary variable (July temperature) presents special challenges to palaeo-reconstructions. We test the pollen-climate models using a novel and comprehensive cross-validation approach, running a series of h-block cross-validations using h values of 100-1500?km. Our study illustrates major benefits of this variable h-block cross-validation scheme, as the effect of spatial autocorrelation is minimized, while the cross-validations with increasing h values can reveal instabilities in the calibration model and approximate challenges faced in palaeo-reconstructions with poor modern analogues. We achieve well-performing calibration models for both primary and secondary climate variables, with boosted regression trees providing the overall most robust performance, while the palaeoclimate reconstructions from fossil datasets show major independent features for the primary and secondary variables. Our results suggest that with careful variable selection and consideration of ecological processes, robust reconstruction of both primary and secondary climate variables is possible.

SUBMITTER: Salonen JS

PROVIDER: S-EPMC6825136 | biostudies-literature | 2019 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine-learning based reconstructions of primary and secondary climate variables from North American and European fossil pollen data.

Salonen J Sakari JS Korpela Mikko M Williams John W JW Luoto Miska M

Scientific reports 20191101 1

We test several quantitative algorithms as palaeoclimate reconstruction tools for North American and European fossil pollen data, using both classical methods and newer machine-learning approaches based on regression tree ensembles and artificial neural networks. We focus on the reconstruction of secondary climate variables (here, January temperature and annual water balance), as their comparatively small ecological influence compared to the primary variable (July temperature) presents special c ...[more]

PMID: 31676769

Similar Datasets

Project description:Elderberries are being increasingly produced and consumed in North America for their edible and medicinal flowers and fruits. The American elderberry (Sambucus nigra subsp. canadensis) is native to, and most often cultivated in North America. The European elderberry (S. nigra subsp. nigra) has been developed into an economically-important horticultural crop in Europe, but most European cultivars do not perform well in the midwestern USA. The genotype S. nigra subsp. nigra 'Marge' is an open-pollinated seedling of S. nigra subsp. nigra 'Haschberg', which is one of the most popular elderberry cultivars grown in Europe. In a four-year study (one establishment year followed by 3 production years; 2008-2011) at three Missouri (USA) locations, 'Marge' significantly out-performed and out-yielded eight American elderberry genotypes within the same replicated field plots. Across 3 production years at all three sites, 'Marge' achieved budbreak later, flowered earlier, suffered less Eriophyid mite damage, was taller, produced larger berries, and yielded significantly greater amounts of fruit compared with all eight American elderberry genotypes in the study. At one site, 'Marge' produced three times the yield (1.89 kg/plant) compared with the next highest-producing American elderberry genotype (0.65 kg/plant). It is an exceptionally robust and drought-resistant elderberry. The phenotypic attributes of 'Marge' are similar to that of European elderberry except that it performs exceptionally well in the midwestern USA. DNA marker results, along with phenological and morphological characteristics, indicate that 'Marge' is a European elderberry (S. nigra subsp. nigra). As with most European genotypes, 'Marge' does not fruit on first-year wood, and will therefore require a different pruning regimen compared with American elderberry for success in North American production. We do not yet know how 'Marge' will perform outside the midwestern USA, but it is so productive, unique, and mite resistant, that it merits introduction as a cultivar.

Project description:The inference of past temperatures from a sedimentary pollen record depends upon the stationarity of the pollen-climate relationship. However, humans have altered vegetation independent of changes to climate, and consequently modern pollen deposition is a product of landscape disturbance and climate, which is different from the dominance of climate-derived processes in the past. This problem could cause serious signal distortion in pollen-based reconstructions. In the north-central United States, direct human impacts have strongly altered the modern vegetation and hence the pollen rain since Euro-American settlement in the mid-19th century. Using instrumental temperature data from the early 1800 s from Fort Snelling (Minnesota), we assessed the signal distortion and bias introduced by using the conventional method of inferring temperature from pollen assemblages in comparison to a calibration set from pre-settlement pollen assemblages and the earliest instrumental climate data. The early post-settlement calibration set provides more accurate reconstructions of the 19th century instrumental record, with less bias, than the modern set does. When both modern and pre-industrial calibration sets are used to reconstruct past temperatures since AD 1116 from pollen counts from a varve-dated record from Lake Mina, Minnesota, the conventional inference method produces significant low-frequency (centennial-scale) signal attenuation and positive bias of 0.8-1.7 °C, resulting in an overestimation of Little Ice Age temperature and likely an underestimation of the extent and rate of anthropogenic warming in this region. However, high-frequency (annual-scale) signal attenuation exists with both methods. Hence, we conclude that any past pollen spectra from before Euro-American settlement in this region should be interpreted using a pre-Euro-American settlement pollen set, paired to the earliest instrumental climate records. It remains to be explored how widespread this problem is when conventional pollen-based inference methods are used, and consequently how seriously regional manifestations of global warming have been underestimated with traditional pollen-based techniques.

Dataset Information

Machine-learning based reconstructions of primary and secondary climate variables from North American and European fossil pollen data.

Publications

Machine-learning based reconstructions of primary and secondary climate variables from North American and European fossil pollen data.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets