SR-SIM superresolution structured illumination microscope images of fossil and modern pollen of Poaceae

Title: SR-SIM superresolution structured illumination microscope images of fossil and modern pollen of Poaceae Authors: Surangi W. Punyasena, Michael A. Urban, Marc-Élie Adaimé, Ingrid C. Romero Subject: Biological sciences Keywords: Grasses; superresolution microscopy; Rutundu; CZI; visual analysis; Africa; Lake Ellis; Lake Nkunga. Funder: NSF-DBI Advances in Bioinformatics (NSF-DBI-1262561) Corresponding Creators: Surangi W. Punyasena and Marc-Élie Adaimé This dataset includes a total of 2,689 images of Poaceae pollen from modern specimens, as well as fossil and surface sediment samples (counts correspond to the original CZI files; files containing more than one grain were cropped so that each image shows a single grain, and some grains were discarded after visual inspection, so the final image set may differ across the three datasets). The modern pollen grains are from herbarium collections (1,189), and the fossils are from a lake sediment core (1,200) and lake surface sediments (300). These images were taken using SR-SIM superresolution microscopy at 630× magnification (63×/NA 1.4 oil DIC). The images are in the CZI file format. They can be opened using Zeiss proprietary software (Zen, Zen lite) or in ImageJ. More information on how to open CZI files can be found here: [https://www.zeiss.com/microscopy/us/products/microscope-software/zen/czi.html#microscope---image-data] The main goal of this dataset is to capture the diversity of pollen morphology across the Poaceae (grass) family and to create an image database for training image classification and computer vision models. Please cite this dataset and the associated papers when using these images. The images are organized into three main collections: 1) Modern grasses: This group contains all modern specimens, organized by species. Each modern species zip file had the prefix "modern_" and contains all the images corresponding to that species. There are 60 modern species in total. 2) Rutundu: This group contains images of fossil specimens collected from the Lake Rutundu core. Each zip file begins with the prefix "rutundu_" and ends with the core depth. There are 24 depths in total. 3) Lake surface samples: This zip file contains all samples collected from lake surface sediments. It is divided into six subfolders, each corresponding to a locality (lake). This includes Lake Ellis, Lake Nkunga, Rumiku Swamp, Simba Tarn, Small Hall Tarn, and the surface sediments of Lake Rutundu. Inside each subfolder are the images of the Poaceae pollen grains found in the surface sediments of that lake. Note: "Lake_surface_samples.zip" may also contain a "__MACOSX" folder and ".DS_Store" files (automatically generated by macOS). These are not part of the image dataset and can be safely ignored. The accompanying metadata includes an Excel file, "Metadata_Poaceae_list_specimens.xlsx", which contains detailed information about the folders, files, and specimens. The file contains three sheets: 1) info_images_modern This sheet includes subfamily, genus, species, number of pollen grains imaged, herbarium/source for the pollen sample, and the name of the zip file. The herbarium/source column indicates the specimen origin (e.g., MOBOT, Harvard, STRI). Where a folder name contains two names (e.g., Brachiaria_brizantha_Urochloa_brizantha), the first is the original name provided with the specimen and the second is the currently accepted name following taxonomic verification against Plants of the World Online (POWO, Kew). The subfamily, genus, and species columns follow the revised name. 2) info_images_Lake_Rutundu This sheet includes family, sample depth, sample age, number of pollen grains imaged, and the name of the zip file for that depth. All images are of grass pollen, so the identification is "Poaceae" for all specimens. 3) info_images_lake_surface This sheet includes locality, number of pollen grains imaged, and the name of the zip file for the collection, and the folder name for the lake within the zip file. The "locality" column records the lake from which surface sediments were collected, including Lake Rutundu.