data from: Global phenology maps reveal the drivers and effects of seasonal asynchrony

Terasaki Hart et al. 2025, "Global phenology ..." <p>This archive contains all data and results generated by this study, as well as some input data derived from publicly accessible resources. All contents are covered by the CC-BY-NC-SA license (in short: please use, but with attribution, without commercial gain, and only if you share too).</p> <h2>Contents</h2> <p>NOTE: see Filename abbreviation glossary at bottom to decipher shorthand in filenames.</p> <ul> <li><strong><em>./rasters/</em></strong>: <ul> <li><strong><em>./main/</em></strong>: main rasters resulting from harmonic regressions and asynchrony calculations, including:<br> <ul> <li><strong><em>*_coeffs.tif</em></strong>: harmonic regression coefficients (5 coefficient bands: intercept, sine and cosine of annual cyclical time in radians, then sine and cosine of semiannual cyclical time in radians) <em>(global, 0.05 degree, EPSG:4326)</em></li> <li><strong><em>*_harm_reg_R2.tif</em></strong>: harmonic regression R^2s (1 band) <em>(global, 0.05 degree, EPSG:4326)</em></li> <li><strong><em>*_asynch_*km.tif</em></strong>: asynchrony results (including maps in Fig. 2A and Extended Data Fig. 7B) for all three neighborhood radii (50, 100, and 150 km) (each raster has 4 bands: asynchrony value, then R^2 value, P-value, and n of the OLS regression from which the asynchrony value (i.e., slope) was derived) (NOTE: the *_STRICT* qualifier in *NIRv_STRICT* and *SIF_STRICT* indicates the additional land cover filtering that masked all agricultural land before calculating LSP asynchrony) <em>(global, 0.05 degree, EPSG:4326)</em></li> </ul> </li> <li>  <strong><em>./masks/</em></strong>: data for all 5 masking maps presented in Extended Data Fig. 1B (<em>lcMask_DEFAULT.tif</em> includes only the red masked areas in the land cover map in the figure, whereas <em>lcMask_STRICT.tif</em> includes also the black masked (agricultural) areas) <em>(global, 0.05 degree, EPSG:4326)</em></li> <li><strong><em>./drivers/</em></strong>: rasters specific to the LSP asynchrony drivers analysis, including: <ul> <li><strong><em>hansen_lulcc_pct_neigh_mean.tif</em></strong>: neighborhood (100 km radius) mean of land use and cover change, derived from <a href="https://glad.umd.edu/dataset/global-land-cover-land-use-v1">GLAD Lab Hansen et al. 2019 global harmonized land use and land cover change</a> <em>(global, 0.05 degree, EPSG:4326)</em></li> <li><strong><em>MODIS_fire_freq_mean.tif</em></strong>: neigborhood mean (100 km radius) burn frequency, derived from <a href="https://lpdaac.usgs.gov/products/mcd64a1v061/">MODIS Burned Area (MCD64A1.061)</a> <em>(global, 0.05 degree, EPSG:4326)</em></li> <li><strong><em>MODIS_IGBP_veg_entropy.tif</em></strong>: neighborhood (100 km radius) entropy of vegetation structure, derived from <a href="https://lpdaac.usgs.gov/products/mcd12c1v061/">MODIS annual land cover(MCD12C1.061)</a> <em>(global, 0.05 degree, EPSG:4326)</em></li> <li><strong><em>err_map_*COORDS_*_*km.tif</em></strong>: mapped prediction errors from all LSP asynchrony drivers models ('yCOORDS' or 'nCOORDS' = whether geographic coordinates were included or excluded as covariates; 'NIRv' of 'SIF' = whether NIRv- or SIF-based LSP asynchrony was modeled; '50km','100km', or '150km' = whether a 50, 100, or 150 km neighborhood radius was used) <em>(global, 0.05 degree, EPSG:4326)</em></li> <li><strong><em>SHAP_map_*COORDS_*_*_*km.tif</em></strong>: maps of SHAP values for all 11 covariates (if 'yCOORDS', in which case longitude ('x') and latitude ('y') are included as covariates) or 9 covariates (if 'nCOORDS') for all LSP asynchrony drivers models ('NIRv' of 'SIF' = NIRv- or SIF-based LSP asynchrony; '50km','100km', or '150km' = 50, 100, 150 km neighborhood radius used) <em>(global, 0.05 degree, EPSG:4326)</em></li> <li><strong><em>SHAP_predom_top.tif</em></strong>: map of normalized-difference of SHAP values of the top two LSP asynchrony drivers from the main (NIRv-based, 100 km neighborhood, with geographic coordinates included) drivers model (asynchrony of minimum temperature and precipitation seasonality) (plotted in Fig. 2B) <em>(global, 0.05 degree, EPSG:4326)</em></li> <li><strong><em>SHAP_predom_all.tif</em></strong>: map of the index of the predominant driver of LSP asynchrony at each pixel, according to SHAP values for the main (NIRv-based, 100 km neighborhood, with geographic coordinates included) model (0: ppt.asy, 1: tmp.min.asy, 2: brn.frq.mea, 3: tmp.max.masy, 4: def.asy, 5: cld.asy, 6: veg.ent, 7: vrm.med, 8: luc.pct.mea) (plotted in Extended Data Fig. 9C) <em>(global, 0.05 degree, EPSG:4326)</em></li> </ul> </li> <li><strong><em>./etc/</em></strong>: other rasters generated as part of analysis, including: <ul> <li><strong><em>NIRv_4_EOFs_sqrt_coswts_standts.tif</em></strong>: 4 EOFs calculated from the global map of fitted annual average NIRv LSP phenocycles (calculated using square-root-cosine weighting on latitude and using standardized phenocycle time series) <em>(global, 0.05 degree,EPSG:4326)</em></li> <li><strong><em>NIRv_4_EOFs_sqrt_coswts_standts.tif</em></strong>: the same EOFs as the previously listed filed, but transformed for RGB visual display in Fig. 1 (scaled to the [0, 1] interval, 'folded' over the ITCZ using a latitudinally varying weighted sum of EOF and 1-EOF, and reprojected to the Greenwich-centered Equal Earth projection) <em>(global, EPSG:8857)</em></li> <li><strong><em>NIRv_LSP_modality_EPSG8857.tif</em></strong>: LSP modality (from perfect annual to perfect semiannual), plotted in Extended Data Fig. 6A <em>(global, EPSG:8857)</em></li> <li><strong><em>NIRv_SIF_phen_R2s.tif</em></strong>: R^2s between all NIRv and SIF LSP phenocycles, plotted in Extended Data Fig. 6B <em>(global, 0.05 degree, EPSG:4326)</em></li> </ul> </li> </ul> </li> <li><strong><em>./tables/</em></strong>: <ul> <li><strong><em>./LSP_fitting_examples/</em></strong>: GeoJSONs of raw data extracted at the example sites plotted in Extended Data Fig. 2A-D</li> <li><strong><em>./itcz/</em></strong>: Shapefiles of the boreal summer and winter average ITCZ lines, digitized from <a href="https://doi.org/10.1146/annurev-earth-060313-054623">Zhisheng et al.</a> and used to calculate the annual-mean ITCZ map that was used in visualization in Fig. 1 and Extended Data Fig. 4</li> <li><em><strong>./isoclim/</strong></em>: Shapefile of results of the isoclimatic phenological asynchrony analysis (Fig. 3)</li> <li><strong><em>./phen/inat/</em></strong>: data and results for iNaturalist flowering phenology analysis, including: <ul> <li><strong><em>TID_*.json</em></strong>: GeoJSONs of the flowering observation data used to plot the two examples shown in Fig. 4A</li> <li><strong><em>all_inat_plant_phen_taxa.csv</em></strong>: all iNat taxa that had available plant phenology data, as of date of download</li> <li><strong><em>inat_hex_results.json</em></strong>: GeoJSON of hextile-summarized results of flowering modality (Extended Data Fig. 10)</li> <li><strong><em>iNat_MMRR_results_ALL.csv</em></strong>: unformatted version of <em>SUPP_TAB_4_iNat_MMRR_results.csv</em>, containing results for all tested taxa (Extended Data Table 4 shows only taxa with significant results for the LSP-distance coefficient)</li> </ul> </li> <li><strong><em>./phen/coffea_arabica/</em></strong>: four CSVs of sampling points digitized within polygons shown in the [Bacca et al.](https://doi.org/10.19053/20275137.3200) Fig. 2 version of the Colombian coffee harvest map, for analysis of harvest asynchrony analysis (analyzed in Fig. 4C)</li> <li><em><strong>./gen/rhinella_granulosa/</strong></em>: tables of geographic sampling locations and pairwise genetic distances (derived from <a href="https://datadryad.org/stash/dataset/doi:10.5061/dryad.pc866t1p4">Thomé et al. data</a>), for analysis of genetic isolation by phenological asynchrony in <em>Rhinella granulosa</em> (Fig. 4B)</li> <li><em><strong>./gen/xiphorhnychus_fuscus/</strong></em>: tables of geographic sampling locations and pairwise genetic distances (derived from <a href="http://zenodo.org/records/5012226">Quintero et al. data</a>), for analysis of genetic isolation by phenological asynchrony in <em>Xiphorhynchus fuscus</em> (Fig. 4B)</li> <li><strong><em>./SI/SUPP_TAB_*</em></strong>: tables from the Supplementary Information</li> <li><em><strong>./ED/TAB_*</strong></em>: tables containing content that is visualized in the Extended Data figures</li> <li><em><strong>./etc/</strong></em>: additional tabular results that are reported but are not presented in full within the paper, including: <ul> <li><em><strong>EXTRA_TAB_landgen_MMRR_results.csv</strong></em>: a table of the full landscape genetic MMRR results</li> <li><em><strong>EXTRA_TAB_drivers_model_tuning_results_subset_frac_*_NIRv_100km.csv</strong></em>: tables of the hyperparameter tuning results for the phenological asynchrony drivers model, for both of the subsetting fractions of the full raster dataset that were tested (0.05, and 0.005); tuning was done using a 100 km neighborhood and the NIRv-based LSP asynchrony map</li> </ul> </li> </ul> </li> <li>.<em><strong>/figures/</strong></em>: <ul> <li><em><strong>./main/FIG_*</strong></em>: the main figures</li> <li><em><strong>./ED/ED_FIG_*</strong></em>: the Extended Data figures</li> </ul> </li> <li><em><strong>./videos/</strong></em>: <ul> <li><em><strong>./SI/SUPP_VID_1_normalized_NIRv_LSP_300dpi.mp4</strong></em>: Supplementary Information Video 1, animating the average annual LSP phenocycles for all global pixels</li> </ul> </li> <li><em><strong>./logs/</strong></em>: files logging additional metrics and results that are printed to STDOUT by some analyses, including: <ul> <li><em><strong>NIRv_SIF_LSP_R2_median_stats_result.log.png</strong></em>: NIRv-SIF LSP correlation results (Extended Data Fig. 6B)</li> <li><em><strong>cheatgrass_stats_results.log.png</strong></em>: the cheatgrass analysis (Fig. 2B)</li> <li><em><strong>./drivers/</strong></em>: short summary info for all phenological asynchrony random forest models (Extended Data Fig. 9A)</li> <li><em><strong>isoclim_stats_results.log.png</strong></em>: isoclimatic phenology asynchrony analysis (Fig. 3)</li> <li><em><strong>inat_phen_MMRR.log</strong></em>: iNat phenology MMRR analysis (Fig. 4A)</li> <li><em><strong>plot_flowphen_landgen_cafe_results.log</strong></em>: log produced by the script that runs all analyses plotted in Fig. 4</li> </ul> </li> </ul> <h2><br>Filename abbreviation glossary</h2> <ul> <li><em><strong>*</strong></em>: 'wildcard' (indicates that multiple files exist with different filename patterns in this position) </li> <li><strong><em>NIRv</em></strong>: NIRv-derived LSP</li> <li><strong><em>SIF</em></strong>: SIF-derived LSP</li> <li><strong><em>tmmn</em></strong> and <strong><em>tmmx</em></strong>: minimum and maximum temperature</li> <li><strong><em>tmp.min.asy</em></strong> and <strong><em>tmp.max.asy</em></strong>: asynchrony in seasonality of minimum and maximum temperature</li> <li><strong><em>pr</em></strong>: precipitation</li> <li><em><strong>ppt.asy</strong></em>: asynchrony in seasonality of precipitation</li> <li><em><strong>def</strong></em>: climate water deficit</li> <li><em><strong>def.asy</strong></em>: asynchrony in seasonality of climate water deficit</li> <li><em><strong>cloud</strong></em>: fractional cloud cover</li> <li><em><strong>cld.asy</strong></em>: asynchrony in seasonality of fractional cloud cover</li> <li><em><strong>brn.frq.mea</strong></em>: 100 km neighborhood mean burn frequency</li> <li><em><strong>veg.ent</strong></em>: 100 km neighborhood entropy in vegetation structure</li> <li><em><strong>vrm.med</strong></em>: 100 km median vector ruggedness metric</li> <li><em><strong>luc.pct.mea</strong></em>: 100 km neighborhood mean percent land use and land cover change</li> <li><em><strong>EOF</strong></em>: empirical orthogonal functions</li> </ul> <h3><br>Questions? Please reach out! *drew DOT terasaki DOT hart AT gmail DOT com*</h3>