Data from: Biological processes underpin the persistence of dryland productivity following extreme wet years

Global warming has induced more years of above-average rainfall,
significantly affecting the interannual variability of the terrestrial
global carbon cycle. An extreme wet year can cause changes to vegetation
structure and function that persist beyond itself, referred to as “legacy
effects”. The physical and biological mechanisms underlying these effects
are poorly understood, introducing uncertainty into climate–carbon models
to accurately represent post–wet year vegetation dynamics. Here we used
multi-source satellite-derived vegetation productivity metrics, as well as
eddy covariance (EC) measurements, to investigate the legacy effects of
extreme wet years on the productivity of Australia’s drylands. We found
that the impact of the 2010–2011 extreme wet year extended beyond
generating a record-breaking carbon uptake, which exceeded the 40-year
mean by more than 1.5 standard deviations. It also resulted in a
widespread positive legacy effect in the following year. Specifically, up
to 56% of the vegetated areas that experienced anomalous wetness showed
significant legacy effect after one year, with impact size contributing up
to 40 percent of total productivity in those regions. Biological memory in
wet years, representing a potential process for carbon storage and
subsequent remobilization, was shown to dominate the legacy effect. Random
forest analysis identified key ecogeographic controls on biological
memory, such as resource-conservative strategies associated with drier
climates and relatively fertile soils. Comparisons with Dynamic Global
Vegetation Models (DGVMs) further revealed that current models may
underestimate this biological memory by up to 70%, in part due to limited
representation of carbon storage dynamics. Our results contribute to more
accurate modelling of dryland carbon cycle and provide a framework to
better account for post-wet-year legacy effects by incorporating the
influence of wet-year productivity.

# Data from: Biological processes underpin the persistence of dryland
productivity following extreme wet years Dataset DOI:
[10.5061/dryad.51c59zwnb](https://doi.org/10.5061/dryad.51c59zwnb) ##
Description of the data and file structure This repository provides the
codes used in the analysis for the manuscript: ***Biological Processes
Underpin the Persistence of Dryland Productivity Following Extreme Wet
Years***. * **Figure 1 and Figure 2** were produced using **MATLAB**
scripts. * **Figure 3** was produced using **R** scripts. All codes are
provided for reproducibility and transparency of the analysis. * The
MATLAB scripts include procedures for detrending, normalization,
regression, and legacy effect quantification. * The R scripts include
procedures for recursive feature elimination (RFE), random forest
modeling, and SHAP value analysis. The codes are generalized: * Input data
are expected in raster format for gridded spatial data, including
vegetation indices and meteorological variables. Detailed specifications
are provided in the code annotations and in the associated manuscript: *
**For Figures 1 and 2** * **Vegetation index**: The PKU GIMMS NDVI
product \ Available at:
[https://doi.org/10.5281/zenodo.8253971](https://doi.org/10.5281/zenodo.8253971) * **Meteorological data**: SPEI dataset.\ Available at: [https://digital.csic.es/handle/10261/332007](https://digital.csic.es/handle/10261/332007) - **For Figure 3** * **Soil and landscape data**: Soil and Landscape Grid of Australia.\ Available at: [https://www.csiro.au/en/research/natural-environment/land/soil-and-landscape-grid-of-australia](https://www.csiro.au/en/research/natural-environment/land/soil-and-landscape-grid-of-australia) - Spatial and temporal dimensions are determined by the input data. - Output files (tables, models, figures) are saved in the user-defined **target folder**. ## Access information Other publicly accessible locations of the data: * n/a Data was derived from the following sources: * n/a