The persistent degradation of ecosystems around the world has created an unprecedented need for restoring biodiversity and ecosystem function. However, the science of restoring ecosystems is still in its infancy, and restoration practice often results in widely variable and unpredictable outcomes. As a scientific and applied field, we are still building our understanding of the processes that cause spatiotemporal variation in restoration success and failure. An example is seed-based restoration in drylands, where the increasingly harsh and variable climate means that an individual seed faces multiple barriers to successfully germinating, emerging, and persisting. Despite this, seeding is one of the common techniques in dryland restoration; though there are many success stories to drive its continued use as a restoration tool. To improve seeding outcomes generally, restoration practitioners must learn collectively from our successes and failures. Thus, a team of us set out to amalgamate datasets on seeding experiments in dryland ecosystems. By 2019, we had collected data from across 174 restoration sites on six continents.

The work for this paper started in January 2018 when Dr. Nancy Shackelford started a post-doc position with Dr. Katharine Suding at the University of Colorado in Boulder. Nancy began a collaborative databasing effort referred to as the Global Arid Zone Project (GAZP). As an ecologist and data scientist, Nancy was well placed to lead the effort and leaned on her existing dryland collaborations to create a core steering committee to guide project development.

Months of work was spent reaching out to scientists and land managers with known datasets in dryland systems. This was accomplished through endless phone calls, emails, follow-up phone calls and emails, networking events, presentations at meetings and conferences, call-outs to relevant society groups, literature reviews, and the development of communication tools like the website and a quarterly newsletter. Although the data collection is ongoing and the database continues to grow, data submitted before August 2019 were used to test the global status and trends of dryland seeding efforts for GAZP’s first project paper.

While data from six continents were contributed, the number of contributions were highest for high GDP (gross domestic product) countries including the United States and Australia. In particular, we received few contributions from Asia and Africa. While some researchers from countries on these continents were contacted, the distance, difference in restoration methods and data collection efforts, and regional isolation in some places are still clear barriers to developing new collaborations in these areas. We are continually seeking a more global representation in the database, and welcome connections with restoration researchers from any region.

One of the major challenges with collecting distinct datasets from a range of experts was cleaning the data and finding a consistent form of measurement. Some projects measured vegetation density, cover, frequency, or all of the above. Some seeding was performed on its own, others as part of a complex restoration design. Some projects were designed experiments, others included monitoring data from restoration practice. Nancy spent months cleaning the data, aligning the vocabulary, taxonomy, and structure of the data, and quality checking it. She then sent a message to all contributors for participation in writing the first paper from this database.

Two group meetings with interested collaborators were held. We identified four major questions that we wanted to ask using the database and split into four writing teams. Nancy and Dr. Gustavo Paterno sought out additional databases, such as the TRY Plant Trait Database that houses species-level trait data, which were used to compliment the GAZP database for the analyses. Then, Nancy and Gustavo conducted the analyses. Though we know measures of density and cover are more aligned with the goals of restoration practitioners, trying to analyze across data sets meant that we focused on the most universal measure—the presence or absence of a seeded species in a given sampling location. Each team wrote up the results section, and then Nancy, Dr. Elizabeth Leger, Dr. Daniel Winkler and Dr. Lauren Svejcar wrote an introductory section and edited the paper to improve flow and clarity. The Global Arid Zone Project has now become the Global Restore Project through a strong partnership with iDiv in Germany. We have doubled the database size and include grassland data from around the world. We continue to expand our efforts and scope, and hope to soon launch a publicly available tool for exploring and using the data. For people interested in contributing data, please visit the GAZP website and follow the instructions on how to contribute and participate.