This readme.txt file was generated on 2026-07-14 by Bill Nelson. GENERAL INFORMATION 1. Title of Dataset: Imaging and analysis data of short-term co-culture in a SoilChip 2. Principal Researcher: Name: Kirsten Hofmockel Institution: Pacific Northwest National Laboratory Email: kirsten.hofmockel@pnnl.gov 3. Additional Author Contact Information Name: Natalie Sadler Institution: Pacific Northwest National Laboratory Email: natalie.sadler@pnnl.gov 4. Information about funding sources supporting the data: This program is supported by the U. S. Department of Energy, Office of Science, Office of Biological and Environmental Research, through the Genomic Science Program, under FWP 70880. SHARING/ACCESS INFORMATION 1. Licenses/restrictions placed on the data: This work is marked with CC0 1.0: https://creativecommons.org/publicdomain/zero/1.0/. The authors do request that you appropriately cite the dataset when referencing or re-using the dataset. 2. Links to publications that cite or use the data: 3. Links to other publicly accessible locations of the data: NA 4. Links/relationships to ancillary data sets: NA 5. Was data derived from another source? No 6. Recommended citation for this dataset: Song Feng, Natalie Sadler, Arunima Bhattacharjee, Ryan McClure, Kirsten Hofmockel. (2026) Imaging and analysis data of short-term co-culture in a SoilChip. PNNL DataHub https://data.pnnl.gov/group/nodes/dataset/34902 https://doi.org/10.25584/3375157 7. Software used to analyze dataset: See https://github.com/PNNL-SoilSFA/soil1.3_biofilm DATA & FILE OVERVIEW 1. File List: See file-level-md.csv for a listing of all files and experimental metadata. README.txt - this file file-level-md.csv - metadata for all files in data package. See details below under DATA-SPECIFIC INFORMATION Supplemental_Table_S2_biomass_data_for_anova_table_2.csv – habitat and biomass area data supporting ANOVA analysis testing the main and interaction effects of growth substrate, species, and time on biomass across the whole imaging region Supplemental_Table_S3_random_windows_for_anova_table_3.csv – habitat and biomass area data supporting ANOVA analysis testing the main and interaction effects of species, growth substrate, and habitat size on species’ biomass using randomly sampled sliding windows One raw data folder for each replicate experiment: Activity_1.3_rep1 Activity_1.3_rep2 Activity_1.3_rep3 Each raw data folder has subfolders organized by treatment - NAG (N-acetyl-glucosamine), Chito5 (pentaacetyl-chitopentaose, also referred to as chitopentaose), and Chitin - and sampling date/time (mm.dd.yy_hh). SoilChips were inoculated with BFP-tagged Rhodococcus, GFP-tagged Sphingopyxis, and RFP-tagged Variovorax and imaged at 0, 11, 22, 37, 66, and 90 hours after inoculation. Imaging consisted of a collection of 8 overlapping tiles with 40 images taken along the Z plane at 0.5 µm depth. Imaging was performed using bright field (BF) and three fluorescence channels ([C]): 405 nm Ex, 460/50 m Em (‘DAPI’); 488 nm Ex, 525/50 Em (‘GFP’); and 561 nm Ex, 595/30 m Em (‘RFP’). In a few cases, the initial round of imaging was not sufficient, and a second round was performed. In the data folders are the following file types: *.nd - ND v2.0 image acquisition metadata file generated by MetaMorph software. (Further description below.) One file per imaging round. *_[C]_s[N].TIF - raw image file for tile [N] (valid values 1-8) visualized using channel [C] (valid values BF, DAPI, GFP, RFP), containing 40 images along the Z stack, except for bright field images which are a single layer. One file per round, channel, tile. *_[C]_stitched.TIF – image of the complete SoilChip made by stitching together the 8 tile images and using the max intensity Z-projection; one file for each channel [C]; *_w1SD_BF_stitched_MASK.tif - pillar mask; manually marked pillar regions (from the bright field stitched image) used to exclude pillar autofluorescence; one file. *_DAPI_s[N]_masked.TIF - masked image for tile [N]; pillar region mask applied to tile image; only for 'DAPI' channel; one file per tile image. *_[C]_s[N]_wo_bg.TIF - background-subtracted image for tile [N], channel [C] (valid values DAPI, GFP, RFP); rolling-ball background removal (and Otsu thresholding); one file per fluorescence channel (none for BF), tile. The results folder: results_20220408 contains a subfolder for each replicate experiment. In the folders are the following data types: *[C]_masked.TIF - masked single-channel image for channel [C] (valid values DAPI, GFP, RFP); fluorescence image with pillar region mask applied thresholding. One file per date, treatment, fluorescence channel. *[C]_processed.csv - quantification table for channel [C]; per-image extracted values used for ANOVA and ratio calculations. One file per date, treatment, fluorescence channel. *[C]_processed.TIF - processed single-channel image for channel [C] after Z-projection, stitching, background removal, and ratio calculations. One file per date, treatment, fluorescence channel. *[C]_sharp.TIF - sharp processed image for channel [C]. One file per date, treatment, fluorescence channel. *_combined_processed.pkl - python 'pickle' object; serialized processed image/quantification data (arrays or dataframes) for downstream Python analysis. One file per date, treatment. *_combined_processed.TIF - merged RGB composite; the three normalized fluorescence channels (DAPI/GFP/RFP) combined into single image. One file per date, treatment. *_combined_processed_sharp.TIF - exposure-adjusted versions of *_combined_processed.TIF files. One file per date, treatment. *_combined_processed_small.TIF - version of *_combined_processed_sharp.TIF that is smaller and includes scale markings in pixels. One file per date, treatment. *_combined_processed_1d.png - 1D biomass profile plot; sliding-window quantification of species biomass from left (inoculation port) to right. One file per date, treatment. 2. Relationship between files, if important: The three raw data folders are independent replicate experiments (biological replicates). The results folder contains three folders of independent replicate experiments (biological replicates) corresponding to the raw data folders. 3. Additional related data collected that was not included in the current data package: N/A 4. Are there multiple versions of the dataset? **No** DATA-SPECIFIC INFORMATION FOR: file-level-md.csv 1. Number of variables: 8 2. Number of cases/rows: 4557 (including header) 3. Variable List: filepath - path to file within data package directory structure filename - name of the file sample_datetime - date and hour data was obtained from sample; ISO 8601 timepoint – sampling timepoint; valid values 0, 11, 22, 37, 66, 90; N.B. this value might not map precisely to the sample_datetime, i.e. some timepoint 22 samples were only incubated for 21 hours, and some timepoint 37 samples were only incubated for 35 hours. treatment - carbon source provided to culture; valid values 'chitin', 'chitopentaose', 'N-acetylglucosamine', NULL round - all samples had a first round of imaging; in a few cases, initial imaging was poor and a second round of imaging was performed; valid values 1, 2, NULL ex_filter - excitation filter; valid values are 'bright field', '405 nm' (DAPI=BFP), '488 nm' (GFP), '561 nm' (RFP), NULL em_filter - emission filter; valid values are '460/50 m' (DAPI=BFP), '525/50 m' (GFP), '595/30 m' (RFP), NULL tile - indicates subsection of stage imaged; valid values 1-8, NULL 4. Missing data codes: NULL 5. Specialized formats or other abbreviations used: DATA-SPECIFIC INFORMATION FOR: *.nd 1. Variables (taken from https://support.moleculardevices.com/s/article/MDA-file-formats 2026-07-01): NDInfoFile - Required title and version; Version 1 uses the legacy file format (.tif and .stk); Version 2 uses the Meta Series Single/Multi-plane TIFF file format Description - Description DoTimelapse - Time lapse performed? TRUE/FALSE NTimePoints - Number of time points; required if time lapse performed DoStage - Multiple stage positions acquired? TRUE/FALSE NStagePositions - Number of stage positions; required if time lapse performed Stage[N] - Label for stage position [N] (integer) DoWave - Multiple wavelengths acquired? TRUE/FALSE NWavelengths - Number of wavelengths acquired WaveName[N] - Name for wavelength [N] (must match name used for image files) WaveDoZ[N] - Wavelength [N] acquired as Z series? TRUE/FALSE DoZSeries - Z series performed? TRUE/FALSE NZSteps - Number of Z steps performed ZStepSize - Distance between Z steps WaveInFileName - Is wavelength encoded in image name? Applies to multi-wavelength series only EndFile - Required end of file marker 2. Missing data codes: NULL DATA-SPECIFIC INFORMATION FOR: Supplemental Table S2 1. Number of variables: 7 2. Number of cases/rows: 145 (including header) 3. Variable List: sample – treatment replicate description treatment – carbon source provided to culture; valid values “N-acetylglucosamine“, “Chitopentaose”, “Chitin” timepoint – sampling time (hours); valid values 0, 11, 22, 37, 66, 90. replicate – replicate index; valid values 1, 2, 3 total_habitat_area – area of SoilChip available for growth (pillars masked) (µm^2) organism – name of organism; valid values “Variovorax”, “Rhodococcus”, “Sphingopyxis” biomass_area – area of image covered by organism (µm^2) 4. Missing data codes: NULL 5. Specialized formats or other abbreviations used: NA DATA-SPECIFIC INFORMATION FOR: Supplemental Table S3 1. Number of variables: 8 2. Number of cases/rows: 66199 (including header) 3. Variable List: analysis_specific_index – analysis-specific index; N.B. these values are not unique timepoint – sampling time (hours); valid values 0, 11, 22, 37, 66, 90. treatment – carbon substrate provided to culture; valid values “Chitin”, “Chitopentaose”, “N-acetylglucosamine” habitat_category - the area within the sliding window available for biomass to occupy was calculated (i.e., the habitat feature size) and divided by the total size of the window to yield the relative habitat size; relative habitat feature sizes were categorized into three intervals: areas exceeding 0.8 were classified as ‘pore_bodies’, areas smaller than 0.4 were classified as ‘pore_throats’, and the remainder as ‘intermediate’. Only pore_throats and pore_bodies were used in this analysis. window_position – distance from left edge of image to window (µm) habitat_to_total_area_fraction – fraction of image area comprising habitat (µm^2) organism – organism for which data was collected biomass_area - area of image covered by organism (µm^2) 4. Missing data codes: NULL 5. Specialized formats or other abbreviations used: METHODOLOGICAL INFORMATION 1. Description of methods used for collection/generation of data: SoilChip design and fabrication SoilChips were generated as described previously [1]. The SoilChip devices are optically clear and have low autofluorescence and a thin base to allow for high resolution microscopy and imaging. They have a non-reactive and hydrophilic surface to allow culturing. A landscape of pillars provides a porous and structured environment functionally similar to soil. We used a biocompatible thiol-ene based resin (NOA 81) [2] to fabricate a habitat composed of a series of pillars, 5 ports, and 4 fiducial markers within a habitat space measuring 8 mm. The pillars have a 100 µm diameter with 20 µm pore throats between pillars and 60-120 µm pore bodies. We chose this design to mimic the porosity between soil particles. [3,4]. While this does not represent the spatial heterogeneity found in soil [5], we opted for a repeating pattern pillar design to achieve more controlled diffusion and chemical gradients that could affect early bacterial migration. The porous device layers were covered with a transparent PDMS membrane with port holes punched in after assembly. Since our experimental designs include low volume static cultures incubated over days, we incorporated a chamber that enables maintaining moisture levels. To do this, we attached the structured culture devices to the glass coverslip bottoms of culture dishes. SoilChips were treated with N-acetyl glucosamine (soluble monomer), pentaacetyl-chitopentaose (soluble polymer, also referred to as chitopentose), or chitin (insoluble polymer) as growth substrates and inoculated with all three fluorescently tagged species (Rhodococcus, Variovorax, and Sphingopyxis). To each assembled SoilChip, 1 mL of 0.4% agar was dispensed around the internal perimeter of the SoilChip petri dish chamber to prevent the SoilChips from drying out and allowed for maintaining 100% relative humidity. Additionally, the large headspace-to-SoilChip habitat ratio and the highly air-permeable PDMS cover ensured that SoilChips did not become anoxic. A total of 20 µL of growth substrate-specific media (10 mM N-acetylglucosamine; C8H15NO6 (Sigma), 2 mM pentaacetyl-chitopentaose (chitopentose); C40H67N5O26 (NeoGen) was added to the central SoilChip ports. No additional growth substrate was added to SoilChips used for chitin experiments since it was deposited in the device during SoilChip assembly. Tagged Rhodococcus was streaked on LB agar plates supplemented with 30 mg L-1 apramycin, while tagged Sphingopyxis, and Variovorax were streaked on Modified M9 agar supplemented with 10 mM glucose and appropriate antibiotics (50 mg L-1 kanamycin, 15 mg L-1 tetracycline, and 15 mg L-1 chloramphenicol). Single colonies were transferred to 14 mL polystyrene culture tubes with 3 mL of R2A and incubated overnight at 25°C and shaking at 200 rpm. Cultures were centrifuged 5000 x g for 5 min at 25°C to pellet cells. Cell pellets were then resuspended in carbon-free Modified M9 and pelleted once more. The washed cell pellets were then resuspended in enough Modified M9 to achieve an OD600 of 1.0. Next, 100 µL of each species were combined and centrifuged (5000 x g for 5 min) to pellet cells. Supernatants were removed and replaced with 400 µL carbon free modified M9 thickened with 0.1% agar. The 0.1% agar was supplied to slow minimize immediate cell diffusion away from the inoculation port during inoculation. To inoculate, 0.5 µL of the cell suspension was injected into the central port of each SoilChip. The culture dish lids were placed on top of each dish and then wrapped in parafilm. Species were co-cultured in this manner for 4 days, and images were collected as described below at 0, 11, 22, 37, 66, and 90 hours after inoculation, with time points selected based on our prior work with these species [6,7]. Between timepoints, SoilChips were incubated in the dark at 25°C. Experiments were repeated 2 additional times for a total of 3 replicates per growth substrate. In total, we examined data from 9 separate SoilChips (3 different growth substrates, N-acetylglucosamine/chitopentose/chitin, with 3 biological replicate SoilChips per growth substrate). Image analysis was conducted for the total biomass of each species within the entire imaging region, and the species occupancy within specific habitat features (e.g. pore throats versus pore bodies). An inverted confocal microscope (Dmi6000b, Leica Microsystems, Wetzlar, Germany) equipped with a confocal scanning unit CSU 10 (Yokogawa Corporation of America, Sugar Land, TX) was used for this work. Laser lines used included 405 nm Ex; 460/50 m Em for BFP, 488 nm Ex; 525/50 Em for GFP, and 561 nm Ex; and 595/30 m Em for RFP. Laser power was set to 40% for each laser line and 300 ms exposure time for all laser lines. The Instrument is fitted with a Leica Plan APO 20.07 objective that uses Coolsnap HQ2 (Photometrics, Tucson, AZ, USA) controlled by MetaMorph version 7.7.8.0 (Molecular Devices, Sunnyvale, CA, USA) software. SoilChips were imaged using a 63x oil objective. The sample stage was set to capture 8 neighboring tiles (2 x 4 panel) covering the device from the left side (next to the inoculation port) to the right (Fig. 1A). These 8 tiles comprise the whole imaging region. At each of the 8 tile positions, a series of 0.5 µm increment Z stacks were captured for the 3 laser lines at each Z plane before moving to the next plane. Bright field images were also collected, but only for the bottom most Z plane. Additionally, 1 pixel is 0.125 µm and the frame size for each image tile is 174 X 130 µm. 1. Lukowski, J. K. et al. Expanding Molecular Coverage in Mass Spectrometry Imaging of Microbial Systems Using Metal-Assisted Laser Desorption/Ionization. Microbiology Spectrum 9, 10.1128/spectrum. 00520-00521 (2021). 2. Bartolo, D., Degré, G., Nghe, P. & Studer, V. Microfluidic stickers. Lab on a Chip 8, 274-279 (2008). 3. Smercina, D. N., Bailey, V. L. & Hofmockel, K. S. Micro on a macroscale: relating microbial-scale soil processes to global ecosystem function. FEMS Microbiol Ecol 97 (2021). https://doi.org:10.1093/femsec/fiab091 4. Lipiec, J. et al. The effect of aggregate size on water retention and pore structure of two silt loam soils of different genesis. Soil and Tillage Research 97, 239-246 (2007). 5. Deng, J. et al. Synergistic effects of soil microstructure and bacterial EPS on drying rate in emulated soil micromodels. Soil Biology and Biochemistry 83, 116-124 (2015). 6. McClure, R. et al. Development and analysis of a stable, reduced complexity model soil microbiome. Frontiers in Microbiology 11, 1987 (2020). 7. McClure, R. et al. Interaction Networks Are Driven by Community-Responsive Phenotypes in a Chitin-Degrading Consortium of Soil Microbes. Msystems, e00372-00322 (2022). 2. Methods for processing the data: For each image taken, four image types were collected, namely bright field, BFP (Rhodococcus), GFP (Sphingopyxis), and RFP (Variovorax). The raw images in each channel consist of 8 tiles, each tile in each fluorescence channels (BFP, GFP, and RFP) is a TIFF file with 40 images (pages) along the Z stack. The bright field images are of a single layer. We first stitched together all the bright field images, accounting for the movement of the lens based on the scanning parameters, the overlap between two tiles is 29.125 µm (i.e., 233 pixels equivalently). The stitched image is processed with ImageJ [1] to mark the region of pillars manually. These marked pillar regions represent masks that we then apply to fluorescence images to remove pillars from the images. The SoilChip pillars exhibited autofluorescence signal in the blue channel, so to process and denoise the images in a consistent way, we removed these pillar regions from the images via masking. In each of the fluorescence channels, we processed the image by projecting all z stacks into a 2D image by taking the maximum intensity at each pixel along the z-axis. Then the images were stitched and denoised by removing background noise estimated by a rolling ball algorithm [2]. We then applied masks generated from the bright field images to remove the pillar region to make subsequent analysis easier. After the masking of the 2D image, we filtered the noise from the image with a threshold value that is estimated based on Otsu’s method [2]. Because each fluorescence channel has different intensity levels, it is necessary to estimate and remove the background intensities separately. Finally, the images from different channels are normalized with their highest intensity values and then merged into an RGB image. The merged image intensities accurately reflect the occupied area by corresponding species. During image processing, a few poorly captured images from the total 54 images were identified due to focus issues, yielding no detectable signal. These included chitin at 21 hours in replicate 1, chitopentose at 0 hours, chitin at 11 hours, and all substrates at 90 hours in replicate 3. These images were treated as outliers and discarded. All remaining 48 images were retained for downstream quantification and analysis. For each image quantification, the biomass of each species in each image was calculated by summing the area occupied by the corresponding species (based on the fluorescent tag of that species), counted via Python (see https://github.com/PNNL-SoilSFA/soil1.3_biofilm). For each of the 48 images, biomass ratios were calculated by dividing the total area occupied by all three species by the areas occupied by each individual species. From the projected 2D image, we further generated a quantification of species biomass changes moving from left (the side of the inoculation port) to right. The species biomass change was calculated via Python by sliding a fixed width (500 pixels, equivalent to, 62.5 µm equivalently) rectangular window along the distance starting from inoculation port. The window width was chosen as half of the distance between two pillars in the diagonal direction (approximately 1000 pixels, or 125 µm). We then summed all the occupied pixels for each species in each of these windows. In this case, the summation is used to approximate the smoothed biomass curve of each species along the distance from inoculation port. Images with masked pillars were used to calculate the area within the sliding window that is available for biomass to occupy (i.e., the habitat feature size). Dividing this habitat feature size by the total size of the window provides the relative habitat size (ranging from 0 to 1). We categorized relative habitat feature sizes into three intervals corresponding to three habitat features: pore body, pore throat, and intermediate. Specifically, areas exceeding 80% of the maximal habitat size (1.625 X 104 µm2) were classified as pore bodies, while areas smaller than 40% of the maximum were classified as pore throats. The middle interval (40% ~ 80%), representing the transition between pore body and throat, were collectively treated as intermediates. From quantifying species biomass ratios, we can capture how these three species change along a spatial dimension. Images were taken from the second experimental replicate with three different times after inoculation for the co-culture to quantify community dynamics: 11 hours, 37 hours, and 90 hours. These early, middle, and late time points were chosen based on the growth rates of these species and community in our prior work [3, 4]. 1. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nature methods 9, 676-682 (2012). 2. Van der Walt, S. et al. scikit-image: image processing in Python. PeerJ 2, e453 (2014). 3. McClure, R. et al. Development and analysis of a stable, reduced complexity model soil microbiome. Frontiers in Microbiology 11, 1987 (2020). 4.McClure, R. et al. Interaction Networks Are Driven by Community-Responsive Phenotypes in a Chitin-Degrading Consortium of Soil Microbes. Msystems, e00372-00322 (2022). 3. Instrument- or software-specific information needed to interpret the data: .pkl files require the Python programming language and the pickle module. 4. Standards and calibration information, if appropriate: N/A 5. Environmental/experimental conditions: Three carbon sources, N-acetylglucosamine (NAG), pentaacetyl-chitopentaose (a NAG 5-mer) and chitin (a NAG polymer). Each carbon source was imaged at six timepoints: 0, 11, 22, 37, 66, and 90 hours after inoculation 6. Describe any quality-assurance procedures performed on the data: N/A 7. People involved with sample collection, processing, analysis and/or submission: Natalie Sadler, Song Feng, Bill Nelson, Ian Smith, Megan Oelgoetz, Kirsten Hofmockel