Research ArticlePALEOECOLOGY

Climate, CO2, and the history of North American grasses since the Last Glacial Maximum

See allHide authors and affiliations

Science Advances  25 Mar 2016:
Vol. 2, no. 3, e1501346
DOI: 10.1126/sciadv.1501346
  • Fig. 1 Location of bison and mammoth samples used in the CART analyses.

    White circles represent modern localities and red diamonds represent fossil localities. The references for these sites are listed in table S1. There are 632 individual specimens in total; 281 of these samples are identified as modern (dated to within the last 300 years), 256 specimens are identified as Holocene, and 95 specimens are dated to the LGM.

  • Fig. 2 Variable importance.

    Variable importance in determining bison and mammoth δ13C as inferred by the three different CART methods is shown. Because COT is a compound variable that incorporates temperature, precipitation, and atmospheric CO2, two analyses were executed to assess the overall importance of all variables and single climate variables. We only display the analysis excluding COT to show the importance of single climate variables. WP, proportion of precipitation falling in the winter; SP, proportion of precipitation falling in the summer; MSE, mean standard error.

  • Fig. 3 Partial dependence of individual variables.

    (A to D) Partial dependence of the COT variable (A), GSP (B), GST (C), and atmospheric CO2 (D) on the δ13C of bison and mammoth diets.

  • Fig. 4 Carbon isoscapes.

    Top: Predicted modern average δ13C of vegetation consumed by bison across western North America. The filled area represents the predicted δ13C values based on the mean of the three CART regression methods, and the dots are the differences between the measured and predicted δ13C of diet values at sample locations. (A, C, and E) Modern (A), mid-Holocene (C), and LGM conditions (E), respectively. Middle: Corresponding predicted modern relative abundance of C4 grasses converted from δ13C values based on typical C3 and C4 isotopic composition end members. (B, D, and F) Modern (B), mid-Holocene (D), and LGM conditions (F), respectively. Overlaying the modern C4 abundance prediction map (B) is the measured relative abundance of C4 vegetation for a number of sites across North America from a separate study (10). Gray squares in (E) and (F) represent areas covered by ice during the LGM. Bottom: Changes in the concentration of atmospheric CO2 from the LGM to the present (64). yBP, years before present.

  • Fig. 5 Change in the fractional abundance of C4 grasses.

    (A) From the mid-Holocene to the present. (B) From the LGM to the present. The colored areas represent the intersection of modern and past grass-dominated areas.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/2/3/e1501346/DC1

    Fig. S1. Carbon isotope composition of co-occurring contemporaneous bison and mammoth tissues.

    Fig. S2. Individual model predictions for the modern.

    Fig. S3. Individual model predictions for the mid-Holocene.

    Fig. S4. Individual model predictions for the LGM.

    Fig. S5. Model results incorporating ±1σ in mean CMIP5 paleoclimate model output for the mid-Holocene.

    Fig. S6. Model results incorporating ±1σ in mean CMIP5 paleoclimate model output for LGM.

    Fig. S7. Average summer temperature difference map from LGM to the modern (1901–1950).

    Fig. S8. SD maps for LGM COT, GSP, and GST.

    Fig. S9. Variability in importance rankings using ±1σ in mean CMIP5 paleoclimate model data.

    Table S1. Bison and mammoth isotope data and climate data used in CART analyses.

    Table S2. Comparison of hair, bone, and enamel of bison from the same locality.

    Table S3. Pearson correlation coefficients for climate variables used in CART analysis.

  • Supplementary Materials

    This PDF file includes:

    • Fig. S1. Carbon isotope composition of co-occurring contemporaneous bison and mammoth tissues.
    • Fig. S2. Individual model predictions for the modern.
    • Fig. S3. Individual model predictions for the mid-Holocene.
    • Fig. S4. Individual model predictions for the LGM.
    • Fig. S5. Model results incorporating ±1σ in mean CMIP5 paleoclimate model output for the mid-Holocene.
    • Fig. S6. Model results incorporating ±1σ in mean CMIP5 paleoclimate model output for LGM.
    • Fig. S7. Average summer temperature difference map from LGM to the modern (1901–1950).
    • Fig. S8. SD maps for LGM COT, GSP, and GST.
    • Fig. S9. Variability in importance rankings using ±1σ in mean CMIP5 paleoclimate model data.
    • Legends for tables S1 to S3

    Download PDF

    Other Supplementary Material for this manuscript includes the following:

    • Table S1 (Microsoft Excel format). Bison and mammoth isotope data and climate data used in CART analyses.
    • Table S2 (Microsoft Excel format). Comparison of hair, bone, and enamel of bison from the same locality.
    • Table S3 (Microsoft Excel format). Pearson correlation coefficients for climate variables used in CART analysis.

    Download Tables S1 to S3

    Files in this Data Supplement:

Navigate This Article