1 Introduction
The Mongolian Rangelands and Resilience (MOR2) database provides researchers with information for assessing socio-ecological aspects of climate change and nomadic herder adaptations in Mongolia. MOR2 conducted research on management activities, hydrology, ecology, and social conditions on Mongolian rangelands in places that adopt community-based rangeland management (CBRM) and those adhering to traditional herding practices (non-CBRM). As a result, the MOR2 database has multiple types of data, organized into different thematic data sets, gathered by an interdisciplinary and multi-cultural research team (e.g., hydrologists, ecologists, geographers, and social scientists from US and Mongolia) using different data collection methods, scales, units of analysis, and analytical techniques. A central component of the database is the multiple types of field and secondary data collected by sub-teams. Our research teams were fluid, with members participating in multiple aspects of the project, strengthening the interdisciplinarity of the research and enriching the database design.
The database evolved organically over the course of the five-year project (2010−2015). Database discussion and development was ongoing, adaptive, and reactive to data collection activities. Data quality was our top concern involving various strategies for addressing heterogeneity. To mitigate issues with heterogenous data, we conducted a detailed analysis of the characteristics of the data, developed procedures for data use and sharing based upon established protocols (i.e.,
FHA, 2010).
We discuss key challenges and lessons learned through a description of the data content, collection methods, and data sharing protocols. We highlight the creation of a comprehensive database built into a simple storage structure of multiple thematic folders. We describe the data repository for data archiving and long-term maintenance. We conclude with lessons learned in creating a database across cultures and disciplines.
2 Data collection and content
The MOR2 database is comprised of social, ecological and hydro-meteorological data, herder environmental observations, and geospatial data. Data were collected from 36 soums (counties) pairing community-based rangeland management (CBRM) and with traditional herder groups (non-CBRM). CRBM and non CBRM organization level social data were collected from 142 pastoral groups, household level social data collected from 706 households, ecological data collected from 143 winter camps at three different grazing distances. Hydro-meteorological data, satellite images, herder environmental observations and participatory maps are also included in the database.
Study Sites The MOR2 study sites span four ecological zones across Mongolia: mountain and forest steppe, steppe, eastern steppe, and desert steppe. In each zone, paired soums (counties) were selected with (18) and without (18) formal CRBM groups for a total of 36 soums. Within each soum, 1−9 herder groups (CBRM) or traditional neighborhoods (non-CBRM) were randomly selected for study (
Ulambayar, 2015;
Ulambayar et al., 2017;
Fernández-Giménez et al., 2018;
Jamsranjav et al., 2018). Geospatial data visualize sample study locations using a Geographic Information System (ESRI, ArcGIS Version, 10.2) (Table 1; Fig. S1). Extensive fieldwork took place over three field seasons (2011, 2112, 2013) where ecological, social, physical, and boundary data were collected. Each research team collected data using a mixture of tools including global positioning systems (GPS), photographs, digitally recorded interviews, paper survey questionnaires, focus group discussions, ecological and hydrological sampling protocols. In the field, data were collected on datasheets for both ecological and social information and entered into digital databases after the fieldwork.
Social Data: Social data were collected from 142 pastoral groups and 706 member households, government officials from 36 soums and NGO staff, and soum residents. These data include existing soum statistics, soum-level development and social capital information collected via interviews with soum leaders, a development questionnaire and soum-level focus group survey, CBRM and non-CBRM organizational and household-level questionnaires. Household and organizational data were collected in the field via household questionnaires, group leader interviews and group member focus groups. Household and organizational data were entered into an MS Access database, queried, extracted and analyzed in SPSS (
Ulambayar, 2015;
Ulambayar et al., 2017;
Ulambayar and Fernández-Giménez, 2019). Soum statistics and focus group surveys were entered directly into Excel files.
Ecological Data: Ecological data were collected from 143 winter pastures (428 plots) in four ecological zones (
Fernández-Giménez et al., 2018;
Jamsranjav et al., 2018). Ecological field data includes soil pit descriptions, soil surface data, site environmental data (i.e. metadata), vegetation data including plant biomass by functional groups and plant cover by species, as well as plot photos. Data were collected from 428 50 m×50 m plots along grazing gradients at distances of 100, 500, and 1000 m from winter shelters (
Jamsranjav et al., 2018). Ecological data are entered into the Database for Inventory, Monitoring, and Assessment (DIMA) developed by the Agricultural Research Service (ARS), at the Jornada Experimental Range, New Mexico. DIMA provides automated analysis routines for vegetation, biomass, and soils indicators. The ecological data are stored in both DIMA and exported into Excel files in the MOR2 database and imported into SAS or SPSS for analysis. Quality assurance and control (QA/QC) procedures were established for transferring data from hard copy forms to DIMA and from DIMA to excel spreadsheets. All data are documented with metadata and README files.
Hydro-meteorological Data: Measurements were collected in selected study sites to examine sub-watershed scale stream dynamics. Precipitation, temperature and hydrologic data collected from global and Mongolian governmental databases were used to derive national level interpolated climate data sets. Tree cores were collected from two sites for comparison and analysis with existing data sets stored in the International Tree-Ring Data Bank. Spatial climate data were accessed from information hosted by various international groups (e.g., Global Precipitation Climatology Centre, Climate Research Unit, University of Delaware, WorldClim (1950−2000), Climate Prediction Center (1979−2012), and Aphrodite (1961−2007)). Gridded meteorological data at a monthly timestep over the past 50 years were analyzed at several spatiotemporal scales using R statistical software.
Herders’ Environmental Observations: Herder data were collected using in-person interviews, a structured closed-ended questionnaire and several open-ended interview questions about environmental change and grassland health from 2011−2013 (
Bruegger et al., 2014;
Fernández-Giménez et al., 2015;
Jamsranjav et al., 2019). The original surveys were scanned, stored electronically, translated and data entered into an Excel spreadsheet
Remote Sensing Data: Remotely sensed satellite data include multiple resolutions Advanced Very High Resolution Radiometer (AVHRR, 1 km) (1983−2015) and Moderate Resolution Imaging Spectroradiometer (MODIS, 250 m) (2001−2016) data that span Mongolia. Due to enormous file sizes, the remote sensing data resides on an external hard drive along with metadata that can be accessed by contacting the CSU repository.
Participatory Mapping and GIS Data: Participatory mapping resulted in hard copy maps, digital maps, global positioning system (GPS) points, and interview data used to analyze social, ecological, physical and administrative boundaries. The participatory mapping data contain information about social- ecological boundaries including herder territories, toponyms, cultural places, hydrological, and physical features shown in the hand-drawn maps. These maps and summary reports were scanned and stored in the CSU data repository.
3 Data integration
Data discussions throughout the project created a flexible and adaptive structure to manage and store the data. Using a numbered coding system created linkages and relationships between the distinct ecological, social and physical databases, establishing an unambiguous label for each administrative unit (soum or county and aimag or province), organization, household, winter camp and ecological plot. The coding system enables cross-referencing and merging of data across the various databases, using a spatially explicit hierarchy, to facilitate integrated analysis of social, ecological and physical data (
Driscoll et al., 2007). The coding system is used in each separate database. The code is the “key” that links the social (both household and organizational), ecological, and spatial data (Table S1). The data are organized so that social data are easily aggregated from the household to the group or the soum level. Similarly, ecological data can be analyzed at the plot, pasture/group or soum scale. To conduct integrated analysis of ecological and social data, the databases were merged on key fields, facilitating powerful analyses of this integrated social-ecological database (
Fernández-Giménez et al., 2018).
4 Data protocols
Formal written protocols for all data collection activities ensure consistency. Protocols for the database include the development of internal metadata standards. The metadata for the database was created through a series of dynamic README files that are located within each folder. Team members may access the different data in the folders, but any changes or updates are recorded in the README file within that folder. This creates a distributed, living data dictionary. However, it is incumbent upon team members to maintain these README files. Additionally, as new data are added and transferred to the latest version of the database, an overall spreadsheet is maintained to track these data changes. Each thematic database has a data steward to oversee these processes.
5 Data sharing
A Data Ownership and Use Protocol provides guidance for database use and encourages early career and student researchers to develop their research skills. Each researcher must apply to use the data and explain their research questions, methods, and analysis. The protocol provides for and ensures oversight of analytical approaches, scientific peer review, and appropriate referencing of credit for data development. The process is meant to be flexible and to promote the adoption of sound scientific approaches in using a unique database. As data are processed, stored, and shared, procedures to ensure security, trust and privacy are essential (
Nica et al., 2011). However, multiple challenges with implementing this protocol occurred due to differences in scientific training and standards, disciplinary approaches, and languages (
Fernández-Giménez et al., 2019). (Refer to Discussion and Conclusions).
6 Data storage and access
The long-term stewardship of databases for maintenance, access and research requires resources and support (
FHA, 2010). The CSU Morgan Library provides data curation services to researchers including data set assessment by data experts, advice on how to increase the usability of the data by others over time, and a repository to share data publicly (172 unique data sets) The repository provides unique identifiers (DOIs and handles) for data sets that allow direct citation with a permanent URL (available at CSU Mountain Scholar website
). This repository enhances accessibility and discoverability of the MOR2 database including each of separate data sets (Table S2).
In general, the working data format includes Excel files with multiple tabs, often one for metadata and one for data. To make the data FAIR (Findable, Accessible, Interoperable, and Reusable;
Wilkinson et al., 2016), several curation processes were performed. Library staff used R programming language to read the data, largely in Excel format, reformat it to conform to tidy data table standards (
Wickham, 2014), and export the data into non-proprietary, archival formats such as .csv files. Using the Open Science Framework (OSF), a research data management platform, the Mongolia Project (
Magle et al., 2020) was created to provide a high-level overview of the data curation process that can be tracke. README files were created, containing an inventory of the data files, descriptions of their contents, and any other contextual information needed for reusability. This README file acts as a new user’s first introduction to the contents of the data set.
Some of the MOR2 data set cannot be made publicly available and require controlled access. These data include MOR2 team interviews where individuals could be identified, breaching IRB rules on confidentiality. Library staff created an authorized user group within the repository’s Mountain Scholar system (
Magle et al., 2017). Only individuals who belong to that group can download data sets, but the discovery metadata and README files are still accessible to the public. If a researcher wants to access the data, they must contact the repository manager through Mountain Scholar’s “request a copy” form.
7 Data analysis
The research teams are conducting numerous types of analyses using these compiled MOR2 data sets. GIS layers and physical landscape data provide the basis for analysis to create derived data products. For example, flow analysis, stream networks, and digital elevation models were created for Khangai Mountain region river basins using the suite of ArcGIS Spatial Analyst tools (ESRI, ArcGIS 10.2) (
Venable et al., 2012,
2015;
Fassnacht et al., 2015,
2018;
Wolf and Venable, 2015;
Venable, 2017;
Tumenjargal et al., 2020). Household-level data are aggregated and combined with organizational profile data to analyze social outcomes of community-based management (
Ulambayar et al., 2017;
Ulambayar and Fernández-Giménez, 2019). Ecological field data are analyzed to assess effects of grazing gradients and community- based management and combined with remote sensing to compare patterns in ground and remotely sensed data (
Jamsranjav et al., 2018). Ecological and social data are combined into one integrated database to assess the relationship between household pasture management practices and pasture conditions (
Fernández-Giménez et al., 2018). Ecological data and herder observations are combined to examine whether herder observations can serve as a proxy for ecological monitoring (
Jamsranjav et al., 2019). Participatory maps are coded using visual grounded theory to identify a typology of boundary types associated with herder management practices. Next steps include examining the ecological outcomes of CBRM institutions, rules, land tenure, and specific management processes – the research is on-going, diverse, complex, and exciting.
8 Discussion
Data management reflects the research process. Data bring research teams together in multiple ways. Field data collection builds research teams, comradery, shared experience and in MOR2, cross-cultural learning. Agreed upon data protocols ensure that researchers institute quality control for assurance of data integrity and dependability. MOR2 data were collected based upon research questions driven by disciplinary ground rules (i.e., ecological field data sheets, social science surveys, hydrologic flow data). Disciplinary vocabulary requires clear definitions to address both communication and conceptual (mis)understanding. The creation of the MOR2 database illuminates the cross-cultural challenges that include language barriers, differences in scientific approaches between the US and Mongolian researchers, communication barriers due to distance in time and space, different cultural norms and differing political and social contexts (
Laituri et al., 2015;
Fernández-Giménez et al., 2019). These issues are embedded in the praxis of trans-disciplinary and multicultural research.
Language barriers provide a unique set of disciplinary and cross cultural challenges. Working across different languages and alphabets – Mongolian and English – compound efforts to align and manage data. The project team did multiple back and forth translations of protocols, training materials, and field data sheets to ensure consistent vocabulary and definitions of scientific terminology. The Mongolian students and research partners were invaluable in bridging the cultural divide through translation, oversight, and participation in data and quality check assurances. However, these activities added time consuming tasks to the research project.
Research projects must meet the requirements of their institutions that include Institutional Review Boards, budget management practices, and the availability of resources for long-term data hosting and archiving. Partnering with other universities provides an opportunity to share insights into different management strategies to meet the needs of international, cross-cultural research to improve research outcomes. It is also important to recognize the serendipitous nature of the dynamics of a large, multi-cultural team where new ideas leads to unanticipated outcomes – publications, workshops, trainings, and conferences were all avenues that improved the outcomes of the MOR2 project.
Communication is essential to transdisciplinary and cross-cultural research. The MOR2 team benefitted from a bilingual project staff and strong leadership to provide consistent guidance. Regular meetings utilizing virtual meeting software (Skype, GoToMeeting) enabled face-to-face contact across the globe. These efforts built community and respect between project participants facilitating communication on research design, data development, and overcoming cross-cultural challenges. Additionally, the team sought to overcome the different social context of the academic environment where social relationships evolve differently between professors and students in the US and Mongolia. These discussions became important hubs for sharing approaches and demonstrating inclusion across disciplinary divides and academic standing.
9 Conclusions: lessons learned
Big data are the cornerstone of 21st century research. Data management has gained currency as a disciplinary field where sharing open data are a foundational tenet. The MOR2 database evolved during the rapid transformation of data management techniques, web-based data archiving, and the establishment of digital data repositories. Training and education on data collection are essential to high quality and consistent project data. In a multi-lingual environment, rigorous translation and back-translation of surveys, and field testing and revision before field use are essential to reduce ambiguity. A simple database design organized around research themes of the project provides a comprehensive, organizational framework. The coding scheme creates a spatially-explicit hierarchy to link the different thematic data sets, allowing for the creation of customized linked databases to address specific research questions.
The MOR2 database is not an integrated database, but integration is achieved through analytical approaches that link the social, ecological, and physical databases as well as through team-based approach to the research. The team dynamic has strong and long-term collaborative research relationships built on trust and respect. These are fundamental to the long-term use of the MOR2 database as access, curation, maintenance, and potential further development to be undertaken. This paper focuses on a description of the database and addresses the complex, underlying issues associated with building a transdisciplinary database in a cross-cultural environ-ment.
Since the completion of the MOR2 project there have been numerous advances and improvements in international collaboration and data management tools. The COVID pandemic fueled the use of platforms such as zoom (meetings and discussions) and slack (project management and sharing resources). Other advances include data management tools for hosting data and posting codes (e.g., GitHub). Practices for virtual research activity is much improved and promises exciting opportunities for the future. Cross-cultural research is challenging and issues related to language, custom, and equity remain. However, the MOR2 project demonstrated how many of challenges could be met. An examination of issues associated with large research teams, the challenges of international research, and designing approaches to complex problems are all avenues for future research.
PDF
(163KB)
