1. Introduction

In the era of big data, the process of data integration often involves extracting large amounts of data from diverse and wide-ranging sources. In response to differences in data sources and types, various data integration methods have emerged (Lin et al., 2019). Three methods of data integration are common. One of the most widely adopted data integration methods utilizes Tag Image File Format (TIFF) raster data. Many studies have released data sets in raster format, e.g., TIFF files containing actual or extrapolated data on global biome patterns over the past 140 thousand years (Allen et al., 2020), as well as raster data sets for terrestrial ecology and regionalization (Olson et al., 2001). TIFF raster files encompass essential geographic variables like cartographic projection, datum level, and ground pixel size within their structure (Ritter and Ruth, 1997). Users can directly open and extract detailed data from TIFF raster files using ArcGIS software or utilize R or Python packages ('raster' package or 'Pillow' library) for data extraction. Furthermore, this method of data integration exhibits a small storage footprint and excels in space manipulation performance. However, a raster data file can only represent a single type of data set. To overcome this limitation, researchers have proposed methods for overlaying multiple tiff raster layers. For instance, Fischer et al. (2022) superimposed 31 distinct layers of terrestrial biome and land-cover classifications into a unified RasterStack. Users can access specific layers through the R codes provided by Fischer et al. (2022), all spatial data manipulations and displays were performed with the R software environment for statistical computation utilizing the 'raster', 'rgdal', 'maptools', 'viridis', and 'terra' packages, but this process may require significant parsing time and impose demanding requirements on the device's GPU performance. Second, researchers use the form of tabular data for data integration, i.e., the main body of this database is in the form of a table. For example, Sun et al. (2023) compiled 438 measurements of global grassland net primary productivity (NPP) from 1957 to 2018 and published it as the NPP (BNPP) database. This approach allows for clear visualization of the data. However, organizing tables from different sources and formats together is cumbersome and lacks a systematic structure. Another approach to data integration involves the utilization of enterprise servers for data standardization and publication. A prominent example is the database server offered by Esri Corporation in ArcGIS, which enables users to access and download data, though specific authorizations might be necessary (Oberle, 2004). Data integration has been inefficient due to the presence of diverse data types and variations in data source resolutions, which impeded studies on plant diversity.

Here, we propose a new data integration method called the gridded vector data integration method. The gridded vector data integration method combines grid-based techniques with vectorization, allowing for the integration of diverse data types from multiple sources into grids of the same scale, with each grid cell accommodating multiple types of data. Grids have gained prominence in biodiversity studies and researchers commonly employ 1° × 1° grid cells for their investigations (see Hempson et al., 2015; Rangel et al., 2018; Zhang et al., 2018). Vector data, which is one of two types of GIS data file, alongside raster data, is stored in the shapefile (.shp) format, which integrates the database, graph, coordinate, and table files into a single file. Notably, .shp files have the advantage of accommodating an unlimited amount of data in a single file (Tresch et al., 2019).

To demonstrate the utility of the gridded vector data integration method, we have selected western China as a case study. Previous studies have explored various aspects of plant biology in western China, including the pattern and formation of plant diversity (Yu et al., 2019, 2020, 2024; Ding et al., 2022), conservation efforts in protected areas (An et al., 2021), the impact of ancient climates on vegetation changes (Li et al., 2023a), the use of remote sensing data to calculate vegetation growth rates (Dong et al., 2022), and climate change (Li et al., 2023b). However, these studies commonly collect, organized, and analyze data independently, rarely sharing data. This lack of collaboration hampers research efforts in western China. We reasoned that establishing a comprehensive data system and integrating diverse data sets that include plant distribution and various environmental factors such as precipitation, historical climate, topography, hydrology, and socio-economic data, will assist botanists, ecologists, conservation biologists, and biogeographers in conducting more in-depth exploration of plant distribution patterns and dynamic changes in western China.

In this study, we implemented our new gridded vector data integration method using western China as a case study. We created a comprehensive 1° × 1° database, named Expeditionplus, that includes plant distribution information and environmental factor data. We hope that this database will stimulate research on plant diversity and provide useful information for scientific expedition.

2. Methods 2.1. Climate, topographical, hydrological, and socio-economic data

Western China covers the Qinghai-Tibet Plateau, Tianshan Pamir Plateau, Tarim Basin and Hengduan Mountain region (Fig. 1). The climate data for this area were obtained from the WorldClim website (v.2.1; http://worldclim.org/). Global terrestrial biome data and landcover classifications were obtained from Fischer et al. (2022). Hydrological data were obtained from the Hydrologic data center website (https://www.theia-land.fr/en/hydroweb/). Road network data were obtained from the Roads website (https://www.naturalearthdata.com/downloads/10m-cultural-vectors/roads/). Nighttime light time series data were obtained from the National Centers for Environmental Information website (https://ngdc.noaa.gov/eog/dmsp/downloadV4composites.html). Soil data were obtained from the China soil database website (http://www.vdb3.soil.csdb.cn). NDVI data were obtained from the NASA website (https://www.nasa.gov/). POI data were obtained from Amap information base (https://www.amap.com/). Leaf area index data were obtained from Liu and Liu (2013). Plant transpiration data were obtained from the Science Data Bank website (https://www.scidb.cn/en/detail?dataSetld=846627369274834944). Historical precipitation data were obtained from the China Meteorological Administration website (http://data.cma.cn). Plant (forest) carbon density data over China from 1982 to 2020 were obtained from https://www.scidb.cn/en/detail?dataSetld=906108877987119104. Spatial distribution data of population density was obtained from the National Bureau of Statistics of China website (Seventh census data; http://www.stats.gov.cn/). Esa Mosaic data was processed by Google Image Maps (https://www.google.com/maps/). Carbon emission data was obtained from the National Bureau of Statistics of China website (http://www.stats.gov.cn/). Data for plant specimen distribution, climate, topography, hydrology, and socio-economic status are shown in Table 1.

2.2. Plant specimen distribution data

The distribution data of plant specimens were obtained from multiple sources, including the Global Biodiversity Information Facility (https://www.gbif.org/, accessed: 04 June 2023, https://doi.org/10.15468/dl.engrf4), the Chinese Virtual Herbarium database (http://www.cvh.ac.cn), the Herbarium of the Kunming Institute of Botany (KUN) (http://www.kun.ac.cn/), and the e-Expedition website (http://ekk.ac.cn/). These sources encompass a comprehensive collection of specimen records gathered during our fieldwork in western China over the past decade. We selected records that included identification information, latitude and longitude coordinates, and excluded irrelevant data.

2.3. Coordinate system

We use WGS-84 coordinate systems instead of the Mercator projection for the following two reasons. First, the use of the Mercator projection can lead to significant distortion in the grid cells, particularly in western China near the edge of the elliptical shape of the Mercator projection. Second, Wang et al. (2022) indicated that the deformation error of the WGS-84 coordinate system is only 0.36 m on the x-axis and 1.22 m on the y-axis when establishing a 300 × 300 km grid. As the 1° × 1° grid cells are only a third of the size of the 300 × 300 km grid, the error value is expected to be even smaller.

2.4. Grid system construction

The grid cells were generated using the fishnet creation tool available in the ArcMap software toolbox. First, we imported the western China layer in the WGS-84 coordinate system. Next, we accessed the "Sampling" subtoolbox within the Data Management Tool and selected the "Create Fishnet" tool. We then inputted the western China range layer as the template extent and proceeded to specify a pixel width and height of 1° while leaving the other parameters as default. To ensure the integrity of the data, we implemented a unique numbering algorithm following the approach proposed by Yildirim et al. (2006). This algorithm facilitates intelligent sorting and assigns distinct numbers to each grid cell (Fig. 2).

2.5. Standardization for point data

First, using the WGS-84 coordinate system, we converted the table containing point data into a point file within the ArcMap software 10.2. Next, we established a mapping relationship between the distribution points and the grid cells. Then, we retrieved the information of points from the attribute table of each grid point and exported the resulting layer (Fig. 3a).

2.6. Standardization for surface data

In general, surface data in a table or database without latitude and longitude information cannot be directly matched to grid cells. For example, we obtained the GDP data table for each county from the website of the National Bureau of Statistics (http://www.stats.gov.cn). First, we checked and matched the county names with the corresponding counties in the.shp layer. Then, we assigned the GDP values of each county to the attribute table of each layer. Finally, the attribute value of each grid could be calculated based on the proportion of grid area occupied by the county area (Fig. 3b).

2.7. Standardization for TIFF data

The processing of TIFF data requires converting raster layers into surface layers. First, since TIFF raster data often cover a large area, we first used Spatial Analyst Tools to extract western China area. Then, the raster-to-plane layer conversion was performed using the partition statistical tool. Next, depending on the requirement, our method calculates either the mean or modal value. Note that when a pixel value represents a specific attribute, such as altitude, the average value is extracted for each grid. However, when a pixel value represents a specific category, such as a numerical value indicating a land type, the value with meaning attached is extracted to assign the category to the grid. Finally, the mask extraction tool was used to convert the conversion of tiff raster data to gridded vector data (Fig. 3c).

2.8. Data integration

For the same type of data, such as paleoclimate raster data, we can combine hundreds of layers. Generally, we use the "Link property sheet" function in GIS tool to link the property sheets of multiple layers and export a new table, and finally add the new table to the base layer to get the integrated layer. For the above operations, we can use the 'rgdal' package to implement in the R (v.4.2.1).

3. Results 3.1. Data table

We provide a total of 559 data files. Each file contains a unique grid identifier, allowing users to freely combine the provided data using the unique grid identifier. The first two folders contain information about grid location and unique grid identifiers. Files three to 244 contain information on the distribution of plant specimens in western China that includes specimen collection data at ten-year intervals and information on all plant specimens in each botanical family. The aforementioned folders all include SHP files and MXD files. Users can directly view and use the data in the ArcGIS software. The files numbered 250 to 490 contain current climate data and paleoclimate data for western China. Different folders represent climate factors for each grid from various databases, climate reconstruction models, and time periods. For example, "269.worldclimlastglacialccsm4bi1" represents bio1, i.e., annual mean temperature data, from the worldclim database, provided by the ccsm4 model for the Last Glacial Maximum. Each folder provides the mean, mode, maximum, minimum, range (difference between maximum and minimum values), and standard deviation of the corresponding climate factors for each grid. More information, including subfile name, each file format, and provided values, can be found in Table S1.

3.2. Use example

Our data integration system can directly utilize data sets in ArcGIS-related software. Fig. 4 illustrates how our system can be used to determine the top five plant families in terms of specimen quantity and the geographic distribution with the highest specimen count. Specifically, the number of specimens per grid revealed that species richness in two families (i.e., Rosaceae and Asteraceae) increased along a north-south gradient.

3.3. e-Expedition system

Based on the grid system described in this study, we have developed a software called "e-Expedition" that enables rapid photo-taking, information recording, and data uploading for field surveys of plants. The e-Expedition System is a comprehensive field research system for plant biologists consisting of an app terminal and a web terminal (http://www.ekk.ac.cn). Users can search and download the application from various app stores on mobile devices (Fig. 3d).

4. Discussion

In this study, we developed a standardized data system called e-Expedition system. To demonstrate the utility of our system, we created a database called named Expeditionplus and compiled data on the distribution of plants and environmental factors in western China. This system integrates attribute values (i.e., 558 in this study) based on a grid, allowing users to directly utilize data or view any grid in the field on a mobile app. The detailed information on species distribution in each grid and the varying environmental factors corresponding to each grid are invaluable for biogeographers studying plants or ecology.

The data set we used in this study has two limitations. Firstly, it is not exhaustive. Notably, our data set lacks scientific expedition records from regions with poor road accessibility, such as Kekexili (grid CHNR020 and adjacent grids), Qaidam (grid CHON021 and adjacent grids), and Tarim (grid CHNO011 and adjacent grids), despite satellite remote sensing data showing the existence of vegetation (Zhou et al., 2022). Incomplete data is a common issue in biogeographical studies, even in the best-researched regions globally (Bonardi et al., 2022). For field botanical expeditions, data may be particularly incomplete in uninhabited or difficult-to-access areas, presenting challenges in compiling biodiversity lists for these regions (Yang et al., 2014). Our database can help identify regions that require further investigation for future botanical field expeditions.

Secondly, the environmental data we used may be unreliable and is unevenly distributed. Ensuring the reliability and completeness of such data is crucial for its usability (Santos et al., 2010). Widely used ancient climate data often rely on means (Dimitradou and Nikolakopoulos, 2022). In contrast, our algorithm incorporates more parameters, including extreme and mode values, enhancing accuracy under specific conditions. Uneven spatial distribution results from interpolation at certain meteorological stations or values compiled by government agencies from administrative divisions. Generally, developed regions have larger and more accurate data sets, whereas sparsely populated areas suffer from severe data gaps.

Achieving more precise and comprehensive plant species distribution information depends on advancements in scientific technologies like remote sensing and the application of artificial intelligence in field exploration (Karina et al., 2021). Additionally, it relies on the development of computer AI technology and progress in plant taxonomy and phytogeography (Ledford, 2017). Expeditionplus will continue to be updated in the future, and the additional data will be accessible on the web terminal (http://www.ekk.ac.cn). We anticipate that more detailed and accurate information will enhance our understanding of the patterns and mechanisms of plant diversity in western China. Nonetheless, our study provides a promising framework for handling and applying integrated data.

Acknowledgements

This study was supported by the Second Tibetan Plateau Scientific Expedition and Research (STEP) program (2019QZKK0502), the National Natural Science Foundation of China (32322006), the Major Program for Basic Research Project of Yunnan Province (202103AF140005 and 202101BC070002), and the Practice Innovation Fund for Professional Degree Graduates of Yunnan University (ZC-22222401).

Data accessibility statement

The data used in this study have already been published. Details about the data of regions and plants for this study and the R-code used for analyses are available at http://www.ekk.ac.cn.

CRediT authorship contribution statement

Xinyuan Kuai: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft. Quansheng Fu: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – review & editing. Hang Sun: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Writing – review & editing. Tao Deng: Data curation, Formal analysis, Funding acquisition, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – review & editing.

Declaration of competing interest

The authors declare no competing interests.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2024.01.001.