A nutrient effect on the TEX₈₆ paleotemperature proxy

This is a repository for "A nutrient effect on the TEX₈₆ paleotemperature proxy" in Geophysical Research Letters.

Overview

The TEX₈₆ proxy is a widely used organic paleothermometer based on the ring distributions of isoprenoid glycerol dialkyl glycerol tetraethers (GDGTs) produced by marine ammonia-oxidizing archaea (AOA) (Nitrosospharales; formerly Thaumarchaeota). Because these archaea are also sensitive to nutrient availability — as shown in culture studies investigating growth rate/growth phase — TEX₈₆-based temperature reconstructions may carry a non-thermal nutrient bias.

This study demonstrates, using a global coretop compilation combined with World Ocean Atlas (WOA23) and Copernicus Marine (CMEMS) ocean products, that thermocline nitrate concentrations introduce a systematic, regionally variable offset in TEX₈₆ and can be used as a practical proxy to evaluate nutrient effect on lipid biosynthesis on marine AOA. Two paleoceanographic case studies (Arabian Sea and Tasman Sea) illustrate how this nutrient effect can affect downcore temperature reconstructions.

Citation

If you use this code or data, please cite:

Rattanasriampaipong, R., Tierney, J. E., Abell, J. T., & Gilmore, L. D. (2025). A nutrient effect on the TEX₈₆ paleotemperature proxy. Geophysical Research Letters.

Supplementary dataset (Zenodo)

Rattanasriampaipong, R., Tierney, J., Abell, J., & Gilmore, L. D. (2025). Supplementary Data for 'A nutrient effect on the TEX86 paleotemperature proxy' [Data set]. Zenodo. https://doi.org/10.5281/zenodo.14806962

Repository structure

nutrient-effect-on-TEX/
├── notebooks/
│   ├── SI_code1_PreProcessing_rev1.ipynb   # Stage 1: data assembly & gridding
│   └── SI_code2_DataAnalysis_rev1.ipynb    # Stage 2: analysis & all paper figures
├── spreadsheets/
│   ├── ds01_updated_global_coretop_tex.csv     # Global coretop TEX86 compilation
│   ├── ds02_manual_regionName_assignment.xlsx   # Regional assignments
│   ├── ds03_processed_coretop_tex.csv           # Processed coretop data (output of Stage 1)
│   ├── ds07_TasmanSea_paleorecords.xlsx         # Tasman Sea downcore records
│   ├── fitted_models/                           # Saved OLS model objects (.pkl)
│   └── published_data/                          # External datasets used in analysis
├── ncfiles/
│   ├── ds04_gridded_coretop_tex.nc              # Gridded TEX86 coretop data
│   ├── ds05_gridded_AOM_ds.nc                   # Gridded ammonia oxidation rate data
│   └── ds06_calculated_ocean_properties.nc      # Thermocline T and nitrate fields
├── figures/                                     # All paper figures (PDF/PNG/SVG)
├── bibtex/
│   └── references.bib                           # BibTeX references
├── environment.yml                              # Conda environment specification
└── conda-lock.yml                               # Locked dependency versions

Workflow

The analysis is split into two sequential Jupyter notebooks:

Stage 1 — SI_code1_PreProcessing_rev1.ipynb

Assembles and preprocesses all input data:

• Loads the global coretop TEX₈₆ compilation
• Extracts thermocline depth, thermocline-integrated temperature, and thermocline-integrated nitrate from WOA23 and CMEMS at each coretop site
• Regrids ocean properties to a common spatial grid
• Exports processed datasets (ds03, ds04, ds05, ds06) for use in Stage 2

Stage 2 — SI_code2_DataAnalysis_rev1.ipynb

Performs all statistical analyses and generates every paper figure:

• Fig. 1 — Global map of TEX₈₆ residuals relative to thermocline temperature
• Fig. 2 — Regional quantification of the nitrate effect
• Fig. 3 — Arabian Sea paleoclimate case study (δ¹⁵N proxy for denitrification)
• Fig. 4 — Tasman Sea paleoclimate case study (alkenone concentration proxy for productivity)
• Figs. S1–S11 — Supporting figures (ammonia oxidation rates, calibration comparisons, additional paleo records, age model)

Installation

Prerequisites

• Anaconda or Miniconda

Create the environment

# Using the locked environment (recommended for exact reproducibility)
conda-lock install --name texas-env conda-lock.yml

# Or using the environment.yml (may resolve to slightly different package versions)
conda env create -f environment.yml
conda activate texas-env

Launch JupyterLab

conda activate texas-env
jupyter lab

External data requirements

Some large data files are not stored in this repository and must be downloaded separately before running Stage 1:

Dataset	Source	Variable
WOA23 temperature (0.25°)	NCEI	Annual mean, 1991–2020 climatology
WOA23 nitrate (1°)	NCEI	Annual mean
CMEMS nitrate (0.25°)	Copernicus Marine	30-year monthly climatology (1993–2022)
NICOPP δ¹⁵N database	NOAA Paleoclimatology	Sedimentary bulk δ¹⁵N
Tang et al. (2023) nitrification database	ESSD	Global ocean nitrification rates

Stage 2 relies on the pre-processed .nc and .csv outputs from Stage 1, which are already included in this repository under ncfiles/ and spreadsheets/.

Key dependencies

Package	Purpose
xarray + dask	N-D array handling and lazy I/O for NetCDF ocean products
xesmf	Conservative regridding between ocean grids
cartopy / proplot	Map projections and publication-quality figures
statsmodels / scikit-learn	Ordinary least-squares regression and sliding-window analysis
baysplinepy / baysparpy	Bayesian alkenone (BAYSPLINE) and TEX86 (BAYSPAR) calibrations
cmdstanpy	Stan-based Bayesian modelling

License

Code is released under the MIT License. Please see the LICENSE file for details.