Data from: The role of migration in the change of Northern Hemisphere vegetation for the past 50,000 years
Data files
Mar 05, 2025 version files 3.67 GB
-
out_data.zip
1.14 GB
-
README.md
10.31 KB
-
simulations.zip
2.42 GB
-
T_ann_60_0_kya.zip
111.58 MB
Abstract
Our primary aim was to assess how dispersal limitation affected forest changes in the Northern Hemisphere during the last glacial cycle, and especially after the last deglaciation. We used the LPJ-GM 2.0 dynamic global vegetation model to simulate the vegetation dynamics of the Northern Hemisphere over the past 50,000 years. We compared two dispersal modes by allowing establishment as soon as suitable environmental conditions were realised (free dispersal), or by additionally constraining establishment based on potential migration speed and maturation time (dispersal limitation). For both dispersal modes, we inferred (mega-)biomes and calculated the fraction of forest cover, earliest post-glacial forest establishment, and the difference between dispersal modes (dispersal lag). To evaluate our results, we compared simulated tree cover with pollen-based reconstructions of European Holocene vegetation and of North American vegetation since the Last Glacial Maximum. We simulated multi-millennial dispersal lags in post-glacial establishment of temperate forests in Europe, boreal forests in Central Siberia, and scattered forest areas close to mountain chains (Rocky Mountains and north of Himalaya). Overall, the simulation with limited dispersal had lower forest extents compared to the free-dispersal simulation, especially after the Last Glacial Maximum, with greater mismatches following intervals of rapid warming, notably the onsets of GI-1 and the Holocene. Comparisons with pollen-based reconstructions showed that the simulation with limited dispersal better captured both the expansion of European temperate trees and the distribution of broadleaved and needleleaved trees in North America during the early Holocene. The lag in response of forests to rapid climate change has implications for past ecosystems and projections of future ecosystem services.
The dataset contains the input/output data and R code associated with the article "The role of migration in the change of Northern Hemisphere vegetation for the past 50,000 years".
Description of the data and file structure
The dataset contains:
- analysis.R:
the R script to perform all analyses and generate all output figures and tables (stored in /out_data.zip/) included in the manuscript;
the code contains the directory paths (and links) to the input data necessary to perform the analysis and informative comments; - T_ann_60_0_kya.txt:
post-processed annual temperature (Kelvin) used in the biomization procedure (unprocessed data is the original paleo-climate; see Sharing/Access information below);
"NA" (not applicable) refer to cells underwater; - /out_data.zip/:
contains all output tables generated in the analysis via analysis.R;
the output tables are comma-separated values (CSV):- BiomeInference_domPFT_50_0ka.csv: inference of biome, mega-biome, Land Cover Type (LCT) and dominant Plant Functional Type (PFT). Each row indicates a grid cell (lon-lat coordinate). Variables include: dispersal scenario of the simulation (either "Free Dispersal" or "Dispersal Limitation"); longitude (Lon) and latitude (Lat) in degrees; year of the simulation in years before present BP (Year); simulated leaf area index (LAI, unitless) of each PFT (from BNE until C4G) and their total (Total) extracted from /simulations/.../lai.out; boolean for the landmask, whether active (1, vegetated grid cell) or inactive (0, under ice or water, i.e. unvegetated); total carbon mass (Cmass, kg/m2) extracted from /simulations/.../cmass.out; aggregated PFTs used in the biome inference (B-tree and Tr-shrub); mean annual temperature used to distinguish Tundra and Desert (T_ann) extracted from T_ann_60_0_kya.txt; area of the grid cell in km2 (GridArea); threshold of carbon biomass used in the assignment of low-carbon biomes (CMass_thr, kg/m2); inferred biome (Biome); inferred mega-biome (Mega-Biome); inferred Land Cover Type (LCT); PFT with the highest LAI (Dominant PFT); "NA" (not applicable) in T_ann refer to cells underwater;
- BiomeInference_domPFT_slices_averNeig.csv: inference of biome, mega-biome, Land Cover Type (LCT) and dominant PFT after calculating the average LAI-PFT values in the Moore neighborhood. Variables as above.
- CoverIncrease_maxima_H_BA.csv: rate of increase in forest cover (Increase, in percentage) and time-point of stable forest cover across the Holocene (Peak, in years BP) for each scenario (Dispersal scenario: "Free Dispersal", or "Dispersal Limitation"), forest type (LCT: "Boreal Forest", "Temperate Forest") and area (Area: "Northern Hemisphere", "Asia", "Europe", "North America"), for two relevant time periods of rapid climate warming (Period: Holocene "H", or Bolling-Allerod "BA"); forest cover is calculated from Cmass and GridArea from BiomeInference_domPFT_50_0ka.csv;
- ForestThermalChange_LGM_LH.csv: average and standard deviation of forest cover in percentage (cover_mean_LGM, cover_sd_LGM; cover_mean_LH, cover_sd_LH) and average and standard deviation temperature in degC (T_mean_LGM, T_sd_LGM; T_mean_LH, T_sd_LH) for the Last Glacial Maximum (LGM) and the Late Holocene (LH), and forest change (cover_change) and thermal difference (T_change) between the LGM and the LH, for each region (Area: "Northern Hemisphere", "Asia", "Europe", "North America"), forest type (Forest type: "Forest","Boreal Forest","Temperate Forest") and dispersal scenario ("Free Dispersal", or "Dispersal Limitation"); variables of forest cover are derived from CoverIncrease_maxima_H_BA.csv; temperature variables are calculated from the input paleo-climate (see Sharing/Access information below); "LGM" = Last Glacial Maximum; "LH" = Late Holocene;
- LCT_FractionBiomes_TemperatureAnomaly_NH_AS_EU_Am_500yrs.csv: fraction of grid cell area (FracArea) and fraction of grid cells (FracGrid_nr) occupied by forest type (LCT and Mega-biomes) across different areas (Area: "Northern Hemisphere", "Asia", "Europe", "North America"), and dispersal scenario ("Free Dispersal", or "Dispersal Limitation") each 500 years time-slice (Year). Added variables used in the calculation of fractions: total carbon mass (Cmass, kg/m2), area of the grid cell in km2 (GridArea), number of grid cells (P.n), are extracted from BiomeInference_domPFT_50_0ka.csv. Temperature variables in degC: mean temperature (Temperature (Mean)), standard deviation (Temperature (S.D.)), temperature anomaly (Temperature Anomaly), are extracted/calculated from TemperatureAnomaly_NH_AS_EU_Am_100yrs.csv;
- MinimumEstablishment_LCT_MegaBiomeForests.csv: first year of forest establishment (in years BP) after the LGM for boreal forest (Minimum Boreal Forest), temperate forests (Minimum Temperate Forest), and forests in general (Minimum Forest) for the two dispersal scenarios (Dispersal scenario: "Free Dispersal", or "Dispersal Limitation"). Each row indicates a grid cell (Lon / Lat); post-LGM establishment values are calculated from BiomeInference_domPFT_50_0ka.csv; "NA" values in Minimum correspond to uncolonized cells;
- REVEALS_Models_LCTs.csv: cover comparison (x.diff) as difference between simulated values (x.perc) and REVEALS estimates (x.perc_REVEAL) of for different Land-Cover Types (x: OL=open land; ET=evergreen trees; ST=summergreen trees; BF=boreal forests; TF=temperate forests; Forest=all kinds of forests). Simulated values of cover percentage (x.perc) are calculated by summing the Fractional Plant Cover (FPC) of each LCT-equivalent PFT, where the calculation is based on LAI values from /simulations/.../lai.out for the two dispersal scenarios (Dispersal scenario: "Free Dispersal", or "Dispersal Limitation") and averaged for each REVEAL time-window (mid-point: Year_mid, in range: Year_lower-Year_upper). The REVEALS estimates (x.perc_REVEAL) overlapping the simulated grid cells (Lon / Lat) are extracted from the REVEALS product (see Sharing/Access information below);
- TemperatureAnomaly_NH_AS_EU_Am_100yrs.csv: mean (Mean) and standard (SD) deviation of temperature in degC across areas (Area: "Northern Hemisphere", "Asia", "Europe", "North America") at each 100 years time-slice (Year) for the two dispersal scenarios (Dispersal scenario: "Free Dispersal", or "Dispersal Limitation"). Mean and SD are calculated from the input paleo-climate (see Sharing/Access information below);
- /simulations.zip/:
formatted output of the LPJ-GM simulations for the free dispersal mode (/FREEDISP.zip/) and the dispersal-limited mode (/FIXSPEED.zip/) used as input data in the analysis.R;
the formatted output consists of leaf area index, unitless (lai.out) and carbon mass, kg/m2 (cmass.out) averaged each 50 years (model output setting) and covering the simulation domain at 0.5 degrees resolution; the output is organized in sub-folders at 5000-years intervals from the start of the simulation (slice_51000_45050, meaning from 51,000 until 45,050 years BP) until the end of the simulation (slice_5000_50; meaning from 5,000 until 50 years BP);- lai.out / cmass.out: each row indicates a grid cell (lon-lat coordinate). Variables include: longitude (Lon) and latitude (Lat) in degrees; year of the simulation in years before present BP (Year); simulated leaf area index (lai.out) / carbon mass (cmass.out) of each PFT (from BNE until C4G) and their total (Total).
Acronyms for Plant Functional Types (PFTs)
- BNE: boreal needle-leaved shade-tolerant evergreen tree
- BINE: boreal needle-leaved shade-intolerant evergreen tree
- BNS: boreal needle-leaved shade-intolerant summergreen tree
- BIBS: boreal broad-leaved shade-intolerant summergreen tree
- TeNE: temperate needle-leaved shade-intolerant evergreen tree
- TeBS: temperate broad-leaved shade-tolerant summergreen tree
- TeIBS: temperate broad-leaved shade-intolerant summergreen tree
- TeBE: temperate broad-leaved shade-tolerant evergreen tree
- BESh: boreal broad-leaved shade-intolerant evergreen shrub
- BSSh: boreal broad-leaved shade-intolerant summergreen shrub
- TeESh: temperate broad-leaved shade-intolerant evergreen shrub
- TeRSh: temperate broad-leaved shade-intolerant raingreen shrub
- TeSSh: temperate broad-leaved shade-intolerant summergreen shrub
- TrESh: tropical broad-leaved shade-intolerant evergreen shrub
- TrRSh: tropical broad-leaved shade-intolerant raingreen
- TrBE: tropical broad-leaved shade-tolerant evergreen tree
- TrIBE: tropical broad-leaved shade-intolerant evergreen tree
- TrBR: tropical broad-leaved shade-tolerant raingreen tree
- C3G: cool season grass
- C4G: warm season grass
Sharing/Access information
Links to the external sources (and associated publications) used in the analysis of analysis.R (see also initial comment in the code) to generate datasets in /out_data.zip/ (see above).
- https://catalogue.ceda.ac.uk/uuid/4ca242208e904efe830af45f1697f730: input paleo-climate and landscape data from "Armstrong, E., Hopcroft, P. O., & Valdes, P. J. (2019a). A simulated Northern Hemisphere terrestrial climate dataset for the past 60,000 years. Scientific Data, 6(1), 265."
- https://doi.org/10.1594/PANGAEA.937075: European pollen-based REVEALS land-cover reconstructions for the Holocene from: Githumbi, E., Fyfe, R., Gaillard, M. J., Trondman, A. K., Mazier, F., Nielsen, A. B., Poska, A., Sugita, S., Woodbridge, J., Azuara, J., Feurdean, A., Grindean, R., Lebreton, V., Marquer, L., Nebout-Combourieu, N., Stančikaite, M., Tanţǎu, I., Tonkov, S., Shumilovskikh, L., ... Zernitskaya, V. (2022). European pollen-based REVEALS land-cover reconstructions for the Holocene: Methodology, mapping and potentials. Earth System Science Data, 14(4), 1581-1619. https://doi.org/10.5194/essd-14-1581-2022
Corresponding author: Dr. Veiko Lehsten (https://orcid.org/0000-0002-0649-0042)
