Images, data, and statistical analysis scripts for review article on cover crop roots

Images, data, and statistical analysis scripts for review article on cover crop roots. Optimization of root traits to provide enhanced ecosystem services in agricultural systems: a focus on cover crops - [https://doi.org/10.1111/pce.14247] Research site, planting, and growth 10/2020 - 04/26/2021 cover crop field trial. DDPSC FRS at Planthaven Farm, O'Fallon, MO 63366 (latitude 38.848240°, longitude -90.686640°). The field was tilled before sowing of cover crops. Seed for each cover crop were spread in using a push seed spreader and were lightly irrigated. Alfalfa (Medicago sativa), dundale pea (Pisum sativum), milkvetch (Astragalus canadensis, Astragalus bisulcatus), crimson clover (Trifolium incarnatum), hairy vetch (Vicia villosa), mustard (Brassica juncea var Mighty Mustard, var Kodiak), barley (Hordeum vulgare), wheat (Triticum aestivum, winter, spring), winter rye (Secale cereale), and triticale (× Triticosecale Wittmack). Field harvest measurements Four canopy images were taken across each cover crop row using a Canon 5DS R camera. Images were taken from above each plot at 5ft height manually. Green color was thresholded from the canopy images in batch using OpenCV python script and the percent green cover calculated (Jupiter notebook). Five soil monoliths were excavated using a "shovelomics" approach with an average monolith size of 25.4cm x 25.4cm x 20 cm. The remaining four soil monoliths were destructively analyzed. One soil monolith was imaged using a Canon 50D DLSR camera in a photogrammetry shed. All photogrammetric analysis was conducted using Pix4D mapper software (Pix4D S.A. Prilly, Switzerland), and point cloud cleaning was conducted in CloudCompare V2. 10.2. Cover crop shoots from the remaining soil monoliths were cut and placed into a paper bag for dry biomass determination (60oC for 5 days). A cover crop shoot count was conducted for each monolith with each tiller considered as a shoot for the grasses (barley, wheat, triticale). After cover crop shoot harvesting, a photo was then taken of each soil monolith with remaining weed biomass. A weed score was assigned to each image by one trained researcher with a score 1 low weeds to 5 high weed presence. Soil monoliths were the soaked briefly in water and then the soil washed using a hose keeping the roots. Roots were then scanned on an Epson Expression 12000XL Photo Scanner with transparency unit. Images labeled with "_part" were samples with too many roots for scanning and so were separately weighed. Dry root biomass was taken for the scanned and unscanned roots separately. Root length was determined from images using software RhizoVision Explorer (https://doi.org/10.5281/zenodo.4095629), total root length was estimated using scanned root length and scanned dry biomass with unscanned root biomass. Along each cover crop plot a 10ft trench was dug using a Yanmar Excavator Vi020-6 perpendicular to the row with each trench fully bisecting the plot. Trench was one bucket wide (19 inches) and approximately 36 inches deep in the middle of the row. The five deepest roots that could be observed in the trench wall was measured manually with a tape measure for each cover crop. A garden trowel and shovel were used to excavate and confirm roots in trench wall. Data was analyzed using R Statistics script and raw data used for data processing and figure generation (2021PlantHavenCovercrop_dataprocessing.R). PCA analysis was conducted using the “FactoMineR” package (Husson et al. 2019) to explore the relationships between the traits within the dataset and clustered by family. Individual ZIP file contents: 2021PlantHavenCovercrop_CanopyImages.zip – Raw canopy images, processed percent green cover images, and Jupiter notebook python script (2021PlantHavenCovercrop_ImageBatchColorThreshold.ipynb). 2021PlantHavenCovercrop_RootFlatbedImages.zip – Raw flatbed root scans of cover crops and processed images using RhizoVision Explorer. 2021PlantHavenCovercrop_SoilMonolithWeedImages.zip – Images of soil monoliths after cover crop shoot biomass was removed. 2021PlantHavenCovercrop_dataprocessing.zip – R Statistics script and raw data used for data processing and figure generation (2021PlantHavenCovercrop_dataprocessing.R). 2021PlantHavenCovercrop_ShootPhotogrammetry.zip – 3D models of cover crop shoots from excavated soil monoliths. The .bin files can be opened using CloudCompare app.本数据集包含用于撰写覆盖作物根系综述文章的图像、数据及统计分析脚本。 《优化根系性状以提升农业系统生态系统服务功能：聚焦覆盖作物》——DOI: 10.1111/pce.14247 研究地点、种植与生长情况 2020年10月至2021年4月26日开展覆盖作物田间试验，试验地点位于美国密苏里州奥法伦市Planthaven农场DDPSC FRS（北纬38.848240°，西经90.686640°，邮编63366）。 试验田在播种覆盖作物前已进行翻耕。各覆盖作物种子采用手推式播种机撒播，并辅以轻度灌溉。 供试覆盖作物包括：紫花苜蓿（Medicago sativa）、邓代尔豌豆（Pisum sativum）、加拿大紫云英（Astragalus canadensis）、二裂紫云英（Astragalus bisulcatus）、绛三叶（Trifolium incarnatum）、长柔毛野豌豆（Vicia villosa）、芥菜（Brassica juncea 品种‘Mighty Mustard’、‘Kodiak’）、大麦（Hordeum vulgare）、小麦（Triticum aestivum，分冬小麦、春小麦）、冬黑麦（Secale cereale）以及小黑麦（× Triticosecale Wittmack）。 田间收获测定项目 采用佳能5DS R相机，在每条覆盖作物种植行上采集4幅冠层图像。图像通过手动方式在每个样地上方5英尺高度处拍摄。使用OpenCV Python脚本对冠层图像进行批量绿色阈值分割，并通过Jupyter Notebook计算绿色覆盖百分比。 采用‘铲式组学（shovelomics）’方法挖掘5个土壤原状柱样品，单柱平均尺寸为25.4cm×25.4cm×20cm。剩余4个土壤原状柱样品用于破坏性分析。 选取1个土壤原状柱样品，在摄影测量棚内使用佳能50D数码单反相机进行成像。所有摄影测量分析均采用Pix4D Mapper软件（瑞士普里利Pix4D公司）完成，点云清理工作在CloudCompare V2.10.2中进行。 将剩余土壤原状柱样品中的覆盖作物地上部分剪下，装入纸袋以测定干生物量（60℃烘干5天）。对每个原状柱样品的覆盖作物茎秆数进行统计，对于禾本科作物（大麦、小麦、小黑麦），每个分蘖计为1茎。完成覆盖作物地上部分收获后，拍摄每个原状柱样品及残留杂草生物量的照片。由1名经过培训的研究人员对每张照片进行杂草评分，评分范围为1（杂草极少）至5（杂草极多）。 将土壤原状柱样品短暂浸泡于水中，随后用水管冲洗土壤以保留根系。随后使用爱普生Expression 12000XL照片扫描仪（配备透射扫描单元）对根系进行扫描。标注有‘_part’的图像对应根系过多无法一次性扫描的样品，此类样品将单独称重。分别对已扫描根系和未扫描根系进行干生物量测定。利用RhizoVision Explorer软件（DOI: 10.5281/zenodo.4095629）从扫描图像中计算根长，结合已扫描根系的根长、干生物量以及未扫描根系的干生物量，估算总根长。 在每个覆盖作物样地中，沿垂直于种植行的方向开挖一条长10英尺的沟渠，沟渠完全贯穿样地。沟渠宽度为1铲斗宽（19英寸），样地中部沟渠深度约为36英寸。针对每个覆盖作物，手动使用卷尺测量沟渠壁上可见的5条最深根系。使用园艺小铲和铁锹挖掘并确认沟渠壁内的根系。 数据分析采用R语言统计脚本完成，原始数据用于数据处理与图表生成（2021PlantHavenCovercrop_dataprocessing.R）。使用‘FactoMineR’包（Husson等，2019）开展主成分分析（PCA），以探究数据集内各性状间的关联，并按科进行聚类。 各ZIP压缩包内容如下： 2021PlantHavenCovercrop_CanopyImages.zip：原始冠层图像、处理后的绿色覆盖百分比图像，以及Jupyter Notebook Python脚本（2021PlantHavenCovercrop_ImageBatchColorThreshold.ipynb）。 2021PlantHavenCovercrop_RootFlatbedImages.zip：覆盖作物根系平板扫描原始图像，以及使用RhizoVision Explorer处理后的图像。 2021PlantHavenCovercrop_SoilMonolithWeedImages.zip：覆盖作物地上生物量收获后土壤原状柱样品的图像。 2021PlantHavenCovercrop_dataprocessing.zip：用于数据处理与图表生成的R语言统计脚本及原始数据（2021PlantHavenCovercrop_dataprocessing.R）。 2021PlantHavenCovercrop_ShootPhotogrammetry.zip：挖掘获得的覆盖作物地上部分的三维模型，.bin文件可通过CloudCompare应用程序打开。