Radiocarbon is a critical constraint on our estimates of
the timescales of soil carbon cycling that can aid in identifying mechanisms
of carbon stabilization and destabilization and improve the forecast of soil
carbon response to management or environmental change. Despite the wealth of
soil radiocarbon data that have been reported over the past 75 years, the
ability to apply these data to global-scale questions is limited by our
capacity to synthesize and compare measurements generated using a variety of
methods. Here, we present the International Soil Radiocarbon Database
(ISRaD;
The study of soil organic matter (SOM) dynamics is essential to an improved understanding of terrestrial ecosystem dynamics and the Earth's carbon cycle (Oades, 1988; Heimann and Reichstein, 2008). Current evaluations suggest that SOM accounts for up to 2770 Pg of organic carbon in the top 3 m of soil (Jackson et al., 2017; Le Quéré et al., 2018), which makes it one of the largest actively cycling terrestrial carbon reservoirs and an important modulator of climate change (Sulman et al., 2018). However, the lack of clarity about which fraction of that reservoir will respond to ongoing environmental changes (i.e., timescales of years to centuries) and which will respond only on millennial timescales (He et al., 2016) makes it imperative to improve our understanding of the controls on soil carbon cycling. Additionally, many studies and models focus on only the top 0.5 m of soil or less, despite deeper soils contributing a significant proportion of SOM storage by way of low carbon concentrations but large deep soil mass (Rumpel and Kögel-Knabner, 2010). There is an urgent need to synthesize a wide variety of soil data to model the role of soil in the climate system (Bradford et al., 2016), to develop more data-driven estimates of soil health (Harden et al., 2017), to inform policy and land management plans that preserve and enhance soil carbon storage (Minasny et al., 2017; Poulton et al., 2018), and to extend our detailed understanding of soil developed from observations made at the profile scale to both regional and global extents. Here we describe a new open-source database for the synthesis of soil data with a particular focus on soil radiocarbon data.
Radiocarbon (i.e.,
Two recent soil radiocarbon synthesis efforts demonstrate the utility of
these data for improving predictions of SOM dynamics (He et al., 2016;
Mathieu et al., 2015). Bulk soil radiocarbon measurements, if not part of
repeated time series, provide only an approximation of the time elapsed
since carbon in the soil was fixed from the atmosphere. In other words, soil
carbon age as measured by radiocarbon is defined as the age of carbon stored
in the soil, from the time it enters until a time of observation. However,
this mean value is not representative of how fast soil carbon will respond
to a change in inputs, as it has been repeatedly demonstrated that SOM is
not homogeneous and that carbon stabilized by different physical, chemical,
or biological mechanisms cycles at different rates. Models can be used to
explain time series of bulk radiocarbon or physically and chemically
separated SOM fractions, but this requires model structures with multiple
pools cycling on different timescales to simultaneously explain the rate of
bomb
Ongoing study of soils has led to shifting conceptual views of the controls on SOM dynamics (Blankinship et al., 2018; Golchin et al., 1994; Lehmann and Kleber, 2015; Oades, 1988; Schmidt et al., 2011). Current conceptual views that emphasize the protection of SOM from microbial decomposition via physical isolation or sorption to soil mineral surfaces (Lehmann and Kleber, 2015) and within anaerobic microsites (Keiluweit et al., 2016) have largely replaced earlier paradigms of humification, selective preservation, and progressive decomposition. Three of the fundamental questions currently driving SOM research are (1) what are the controls on the partitioning of organic inputs between soil reservoirs cycling over different timescales; (2) what factors determine rates at which SOM in each reservoir is lost, retained, or transferred within the soil; and (3) which mechanisms contribute to transformation of SOM to stabilized or more protected forms? To address these questions, researchers typically measure the concentration or mass content of organic carbon along with other properties, including molecular composition, isotopic ratios, and the distribution of SOM between conceptually or operationally defined pools (e.g., Basile-Doelsch et al., 2009) or a time series of samples collected over the course of decades (e.g., Baisden et al., 2002a).
Soil fractionation is the operationally defined separation of soils into distinct pools or “fractions” through a variety of physical, chemical, and biological approaches. Soil fractionation is generally intended to isolate soil fractions that reflect SOM in different physico-chemical states or mechanisms of SOM protection (Trumbore and Zheng, 1996); these mechanisms may operate on distinct temporal scales (e.g., Khomo et al., 2017). For example, density fractionation of SOM is a commonly applied technique (Golchin et al., 1994, 1995; Crow et al., 2007; Sollins et al., 2006, 2009; Swanston et al., 2005). The “light” fraction of soil material that floats in a dense solution (e.g., sodium polytungstate) or gets picked up by electrostatic attraction (Kaiser et al., 2009) is sometimes used as a proxy for rapidly cycling SOM, as this material is generally observed to have a shorter mean residence time compared with the bulk-soil average, while the “heavy” or dense material is used as a proxy for mineral-associated SOM, which is assumed to cycle more slowly (e.g., Sollins et al., 2009). In some cases, sonication of the suspension may be used to further isolate occluded SOM, i.e., organic material in soil aggregates (Golchin et al., 1994; Kaiser and Berhe, 2014). Other methods for isolating SOM with different cycling rates in the soil include, but are not limited to, physical separation of aggregates by size and water stability (Jastrow et al., 2006; Plante et al., 2006; Six and Paustian, 2014) or of different-sized soil particles (Desjardins et al., 1994), biological incubation of soils (Torn et al., 2005; Trumbore, 2000; Paul et al., 2001), and chemical extractions (Heckman et al., 2018; Masiello et al., 2004).
Comparing the mass and radiocarbon signature of the carbon leaving or entering the soil system (fluxes) with those of specific soil fractions provides insight into the rates of transfer between pools and provides a means for differentiating between various measures of dynamics, ranging from mean age to the transit time of carbon for the whole soil, a given depth increment, or a given SOM pool (Gaudinski et al., 2000; Baisden et al., 2002a, b; 2003; Sierra et al., 2014; Ohno et al., 2017; Ziegler et al., 2017; Szymanski et al., 2019). Similarly, measurements of interstitial soil carbon (i.e., in soil water or gases collected from within an intact soil profile) and its isotopic signature provide key information about the dynamics of the carbon present in the soil solution (Sanderman et al., 2008). Soluble carbon is believed to be the dominant pathway for vertical transport of organic carbon (Kaiser and Kalbitz, 2012; Angst et al., 2016) and also an intermediate stage through which carbon exchanges from being vulnerable to microbial decomposition to being stabilized on mineral surfaces (Jackson et al., 2017; Leinemann et al., 2018).
Measurements of bulk soils as well as soil fractions are evaluated in the context of other soil properties to better understand the controls on SOM preservation. However, the diversity of soil fractionation methods makes it difficult to compare measurements across soils or to evaluate best practices (e.g., Trumbore and Zheng, 1996). Combining radiocarbon measurements of soil carbon fractions, time series, incubations, interstitial observations, and fluxes has proven useful in resolving the contribution of different soil carbon persistence mechanisms in a site-specific modeling context (Braakhekke et al., 2015), but the application of this approach beyond the site scale has thus far been limited due to the lack of globally synthesized data.
With a changing paradigm for SOM dynamics and ever-evolving SOM models, it
is more important now than ever that we synthesize existing soil radiocarbon
measurements
Here, we present a flexible database spanning broad spatial scales and capturing a range of data types, including diverse soil fractionation methods, incubations, fluxes, and interstitial measurements and spanning a range of spatial scales. Our goal is to provide an open-access data resource that will encourage the scientific community to apply the database for a variety of synthesis studies or metaanalyses and also contribute data to the repository.
The International Soil Radiocarbon Database (ISRaD) is designed to be an
open-source platform that (1) provides a repository for soil radiocarbon and
associated measurements, (2) is able to accommodate data collected from a
large variety of soil radiocarbon studies, including the diversity of
fractionation techniques applied to soils as well as repeated bulk
measurements made over spatial or temporal gradients, and (3) is flexible
and adaptable enough to accommodate new variables and data types. Although
ISRaD was specifically developed with soil radiocarbon measurements in mind,
it is well suited for synthesizing other soil measurements, including stable
carbon and nitrogen isotopes. Importantly, we currently focus only on
natural abundance isotopic measurements and therefore exclude data from
isotopic tracer studies. The ISRaD v1.0 data are archived and freely
available at
Conceptual diagram of an entry in the database. Each box represents a table in an entry; the horizontal bars distinguish the hierarchical levels of the database. Arrows show the hierarchical relationship between and among levels of the database. Time is considered at the profile level, as this is the coarsest spatial scale for which observational data are reported. Every time a profile is sampled, a unique profile identifier must be generated, consisting of the profile name combined with the profile observation date, which is then linked to all measurements made at or below the profile level of the hierarchy.
In its most general form, ISRaD is an implicitly relational database. It
consists of a linked hierarchical list of tables that contain soil
measurements, i.e., variables (Fig. 1). The fundamental unit of organization
in ISRaD is the
Transparency and traceability are fundamental tenants of ISRaD. Accordingly, each entry, whether ingested individually or as a compilation, must have a DOI. For data from published studies, the DOI of the publication is acceptable. Data from unpublished studies must be registered for a DOI through a DOI registration agency (e.g., Zenodo, Pangaea, etc.) prior to ingestion into ISRaD. As it is equally important to be able to reconstruct prior data compilations, e.g., synthesis studies, the specific references for individual datasets making up a synthesis are ingested as part of the synthesis entry, and the entry is flagged within the database with an additional reference to the synthesis study itself. For example, several of the major data sources added to ISRaD were synthesis studies (e.g., He et al., 2016; Mathieu et al., 2015), and users can generate reports of data from these prior syntheses by constructing a query that utilizes this synthesis flag.
Each ISRaD release will be available in two forms: (1) a raw version of data
(
An entity relationship diagram for the International Soil
Radiocarbon Database (ISRaD). A short description of the required variables
for each entity is shown along with the field name used in the database and
the variable data type. Crow's foot connections with a straight line
indicate mandatory daughter entities (one or more), whereas a crow's foot
with an open circle indicates optional (zero or more) daughter
entities. The “*” indicates entries indicate keys, or linking variables,
which are repeated at each successive level of the ISRaD hierarchy. The
“
The ISRaD data hierarchy consists of eight levels of information (Fig. 2). The top level of the data hierarchy is the metadata table (1), which includes information describing the source of data for a particular entry. The remainder of the hierarchical levels can be defined by the spatial extent of the information included in each table. The site (2), profile (3), layer (4), and fraction (5) tables represent information captured from decreasing spatial extents: from the scale of the study area to individual mass fractions isolated from a single soil sample. Special cases of the last three spatial extents further accommodate the temporal context of repeated measurements: (6) fluxes, (7) interstitial, and (8) incubations. In the sub-sections below, we provide overviews and examples of the types of information reported at each level and for each of the tables that occupy these levels (Fig. 2).
The data hierarchy is maintained across tables through the use of unique
keys, or linking variables (noted with a “*” in the following
descriptions), that are required in each record (row) of data in each table.
In addition to the table-specific key, each subordinate table in the
hierarchy must also contain the key variables of the above tables. For
example, in addition to a unique
ISRaD provides basic quality assurance and quality control (QA/QC) protocols
(described below) and expert review that are applied prior to ingesting
entries. These protocols are used to ensure that required variables are complete,
that the key variables match across levels of the hierarchy (more detail
below), and that data entered match the specified data type and range for a given
variable. Variables that are not designated as required need only be
completed if those data are available. The ISRaD template and a detailed
description containing the full list of variables along with instructions
for populating the template can be downloaded or viewed from the
“Contribute” page of the web interface (
For all variables across all hierarchical levels, it is important to observe the acceptable data types (character and numeric) and units. Variable names, descriptions, and reporting conventions are given in the heading columns of the ISRaD template file (ISRaD_Template.xlsx), and more detailed information is provided in the data dictionary (ISRaD_Template_Info.xlsx). Allowed values include unrestricted text, controlled text, or numeric variables with or without defined ranges. Unrestricted text is generally limited to naming and note data fields, while controlled text fields are implemented for certain variables in an attempt to standardize the data and simplify data analysis. In the event that desired variables are not included in the current version of ISRaD, users may submit a request to add new variables. This process is initiated by posting an issue at the ISRaD GitHub repository and is described in more detail in Sect. 3.4.
The metadata table provides information for the characterization of the
entry itself. Required metadata include the entry name (i.e.,
Site-level data are limited to the geospatial details defining the coarsest
scale of the study area(s) included in each entry. We define a site as a
spatially defined location that includes one or more soil profiles. By
convention, we define a site as having a
Profile-level data include details pertaining to specific sampling locations. If available, profile-scale spatial coordinates should be provided in addition to site-scale coordinates.
Many variables that may initially appear to belong at the site level are
instead included at the profile level to facilitate accurate representation
of spatial heterogeneity at a finer scale than the site level (e.g., for
multiple profiles observed at the same site). Examples include local mean
annual temperature and precipitation, soil taxonomic classification,
vegetation type, land cover, depth to bedrock, and parent material
composition. Other than the entry name and site name, the only additional
required variable at the profile level is the profile name
(
Soil flux data present a special case of observations that correspond to the
profile level of the database hierarchy. Flux-level data allow for
reporting temporally explicit measurements of mass or energy transfer
occurring at the profile scale. Both gas and liquid analytes (e.g.,
Layer-level data correspond to measurements made for a specific depth
increment collected from a soil profile. The required variables at the layer
level include layer name (
The interstitial level is a special case of layer-level data. Specifically, interstitial data refer to measurements made on material occupying the interstices of the soil structure. In most cases, this material can be thought of as being mobile relative to the rest of the soil matter. Some common examples include gases, liquids, and colloids. Like flux data, the interstitial data table accommodates repeated measurements of these properties through time, and as such, the observation date must be recorded for each record in the interstitial table. Because interstitial records may not correspond to the same depth increments defined for solid phase analyses, separate depth reporting is used in the interstitial table distinct from what is reported in the layer table. Both sampling methodology as well as the properties of interstitial samples are reported in the interstitial table.
Compared with most other soil databases, the fraction data table of ISRaD is
unique. The fraction data fields are designed to accommodate and allow for
fair comparison of the wide-ranging methodologies utilized to partition
soils into discrete fractions. As such, there are more required fields for
the fraction level compared with the other hierarchical levels. These
required fields include fraction name (
For example (see Fig. 3), most soil density fraction (
Flux rates and isotopic signatures of laboratory-incubated samples are reported in the incubation table. Sample processing data (e.g., whether or not roots have been removed from samples prior to incubation) are recorded as well as incubation conditions (e.g., temperature, moisture, and duration). Repeat measurements, such as incubation time series, can also be recorded. Incubation records must be linked either to a layer or both a fraction and a layer, e.g., roots isolated from a specific bulk layer sample.
Radiocarbon measurements of environmental samples have a long history, much
of which is reviewed in Trumbore (2009), including common units. Radiocarbon
data ingested to ISRaD are required to adhere to some basic reporting
conventions. First, measurements of radiocarbon may be reported in units of
either fraction modern (FM) or
New data entries are added, or ingested, into ISRaD through a user-initiated
process. The most common means of ingesting entries is via the template
provided on the ISRaD website (
Data templates that have passed QA/QC should be submitted via email to ISRaD at info.israd@gmail.com. These templates are then distributed to ISRaD expert reviewers, who inspect template files to ensure proper completion of the more complex aspects of the template, such as classification of soil fractionation methods. If problems are identified with a submitted dataset during the expert review process, reviewers will work with the data curator to ensure that these problems are corrected. Once the expert reviewer signs off on a submitted template, it will be ingested into the database.
One key feature of the ISRaD structure
is the ability to classify and categorize data generated from diverse
methods for fractionating soils. The ISRaD approach requires specification
of the fractionation scheme applied, which may include, but is not limited
to, density
The current version, v1.2.3, of ISRaD includes a total of 263 individual data entries and 730 sites spanning the globe (Fig. 4). The current distribution of data across the various levels of the database hierarchy is shown in Table 1, and a full list of data entry references is provided in Table S1 in the Supplement.
The number of data points currently included at each hierarchical level in ISRaD v1.2.3.2019-12-20
Details of some core calculations included with
ISRaD_extra. Additional variables will be regularly added to
ISRaD_extra, and an up-to-date list can be found at
Users may access ISRaD and its supporting information three ways: (1) the website, (2) the ISRaD-R package, and (3) the GitHub repository. Each of these access points is described in more detail below.
Most simply, users can access ISRaD data and associated resources by way of
the website (
ISRaD has been designed for ease of use in the R computing environment (R Foundation for Statistical Computing, Vienna, Austria) in order for users to be able to take advantage of the full suite of R capabilities and functionality to manipulate and analyze ISRaD data. Many of the basic functionalities, such as loading current versions of the ISRaD data objects, can be performed in the R environment without installation of the ISRaD-R package. A number of vignettes including R scripts for some commonly used data manipulations or plotting are given on the website.
Users who need to locally compile a version of ISRaD (e.g., using their own
templates) or who want access to the full suite of reporting functions can
access these features by installing the ISRaD-R package (also called
ISRaD, which is available at the Comprehensive
R Archive Network – CRAN – repository,
The source code for the ISRaD-R package is hosted under version control on
the GitHub repository ISRaD
(
For students or non-experts interested in learning more about the science
behind the data, we have developed the Soil Organic Carbon Information Hub
(SOC-Hub). The SOC-Hub (
Individual data entries (i.e., completed templates or templates output from ingested compilations) that have passed QA/QC and the expert review process are hosted in the “ISRaD_data_files” folder of the GitHub repository. Users may download these entries in order to add new data or to make corrections to existing data if problems are discovered. Corrected files can be resubmitted to ISRaD once they pass QA/QC by emailing the updated template and a text file of the QA/QC report to the ISRaD editor (info.israd@gmail.com) and will be reingested after passing the expert review process. This process of user-initiated revision of existing data entries is particularly useful when large data compilations are ingested into ISRaD from previously published syntheses (e.g., He et al., 2016; Mathieu et al., 2015) or when publications report treatment means. Depending on the scope of the synthesis efforts, entries ingested into ISRaD this way may omit data available from the original studies, and the entry modification process allows those data to be added or corrected as needed.
Access to the source code underlying the ISRaD database compilation and calculations allows for users to check for errors and contribute to the functionality of ISRaD. Users with a registered GitHub account are invited to write code that adds to or improves upon the existing database tools. Using standard GitHub tools, users will submit a “pull request”, and following code testing and evaluation of utility to the ISRaD community, user-submitted code will be incorporated into the ISRaD-R package.
One of the most important tools available to ISRaD users is the ability to post questions, report issues, or make suggestions, including requests to incorporate new variables. We use issue tracking tools provided by GitHub to track and categorize user input including: suggestions for improvements; problems or errors with the website, the R package, code, or any other aspects of ISRaD; requests for new variables or issues related to existing variables (e.g., incorrect acceptable ranges used in QA/QC); or asking questions related to template entry or any other aspect of ISRaD. While the GitHub issue-reporting functionality is the preferred means for reporting questions or issues with the database or process, it does require that users register a GitHub account. Users who do not wish to or are not able to register with GitHub can also submit issues or questions via an email to the ISRaD editor (info.israd@gmail.com); however, the response time may be slower.
Geographic location of sites currently included in ISRaD v1.0. Circles that appear darker in color indicate multiple overlapping sites at the resolution of the map.
Official releases of ISRaD data will be issued periodically, following major structural changes to the database or after the ingestion of a substantial amount of new data. Versioning is tracked using a three-tiered sequential numeric identifier, i.e., the version number (in the form of “major.minor.patch”) in addition to the date of the most recent change. The major version (i.e. the first numeral of the version number) is used to track official releases. Following each official release a DOI will be issued and the data will be archived by Zenodo (
Users citing ISRaD should cite this publication as well as the most recent official data release at the time that they accessed the data. In their citation of the official release, users should also reference the version of the data they used (e.g., v1.2.3.2019-12-20).
ISRaD is not the only soil database available to the international research community (Malhotra et al., 2019). The primary niche of ISRaD is the ability to synthesize soil radiocarbon data and provide a framework for comparing soil carbon fraction data. For other purposes, there may be other soil databases that are more applicable. However, as a benefit of adding data to ISRaD, we facilitate sharing of data ingested into ISRaD with other databases developed by the soil science community. At present, ISRaD has a reciprocal agreement with the International Soil Carbon Network (ISCN), which is focused on soil carbon content and related variables from bulk soils (i.e., no isotope or soil fraction information). As per this agreement, the ISCN retrieves bulk-soil data from ISRaD and is responsible for filtering duplicate entries and incorporating any new data into the ISCN database.
A simplified depiction of the ISRaD governance pyramid, where the scientific steering committee is responsible for approving major management decisions and data maintainers are responsible for implementing broad changes, but data contributors and users are the primary drivers of the evolution of the data product.
ISRaD is a community effort with multiple contributors operating at
different levels. Governance of ISRaD is required in order to ensure
continuity of services and to plan for the future evolution of this data
repository. The governance structure of the ISRaD is pyramid-shaped (Fig. 5). The ISRaD
Database
Data
Finally, ISRaD
Although the structure of the ISRaD governance pyramid is oriented around individual users, the nature of scientific research is often more group-focused. For example, teams of researchers generally work together to seek out funding and to conduct research. Thus, in some cases a group or team of individuals may seek to utilize or modify ISRaD for their purposes. Such groups can petition the scientific steering committee to be formally designated as an ISRaD organization. This process should be followed when groups seek to leverage the ISRaD resources beyond the scope of a basic user or contributor. The steering committee will consider the scope of the work proposed by the group and, when appropriate, provide a letter of support for funding proposals. Approved organizations should nominate a member to serve on the steering committee and, in the case of organizations making large changes or additions to ISRaD, a data maintainer to coordinate the technical aspects of that work.
As detailed above, ISRaD is an open-source project that provides several
ways for participation. ISRaD v1.0 data (Lawrence et al., 2019) are archived
and freely available at When utilizing the resources provided by ISRaD, including the complete dataset, individually curated entries, or value-added calculations included in the R-package, users should cite this publication and reference the version of ISRaD that was used for their work (see Sect. 3.6 above). Additionally, if users leverage individual data entries from the database, they should also cite the original source dataset and/or paper. When users interpret their own data in the context of data accessed from ISRaD, they should submit those new data for inclusion in ISRaD after they have published their results and/or obtained a DOI for their dataset.
ISRaD is an interactive open-source data repository specializing in
radiocarbon data associated with measurements of soils spanning a broad
range of spatial scales. The ISRaD dataset is unique in that it includes not
only measurements of bulk soils but also measurements of soil water, gases,
and the wide diversity of soil pools isolated through different
fractionation methodologies. Most of the studies summarized in ISRaD were
conducted with a goal of understanding the factors controlling timescales of
carbon cycling in specific sites, regions, or biomes. ISRaD is an attempt to
gather the data from these individual studies in one place and in the same
format to facilitate comparisons and synthesis activities. There are three
ways through which potential users can access ISRaD: (1) the web interface
enables users to download the most recently compiled report formatted as
a .csv file, (2) the ISRaD-R package provides access to the compiled reports
as well as visualization tools and R-based querying tools, or (3) the GitHub
repository provides direct access to the source code for the ISRaD-R
package as well as data from individual entries and the compiled database.
Currently, the ISRaD dataset contains
The supplement related to this article is available online at:
The creation of ISRaD was a community effort. The initial effort to build ISRaD started with the USGS Powell Center working group on Soil Carbon Storage and Feedbacks but benefited greatly from early efforts of the International Soil Carbon Network and other individual efforts to compile soil fraction or radiocarbon data. Scientists from the Max Planck Institute for Biogeochemistry joined forces with the Powell Center group to greatly expand the scope and technical complexity of ISRaD. CRL, JBM, AMH, GM, CAS, SS, KH, JCB, SEC, GMc, and ST designed and built ISRaD and led the preparation of the paper. PL, OV, KTB, CR, CEHP, CAS, KM, and SD provided technical contributions, including coding, to the creation of the database and assisted with the ingestion of data. CH, YH, CT, JH, MT, and CEA provided large datasets or data compilations. AAB, MK, AK, EMS, AFP, AT, JPS, LV, SFvF, and RW contributed to the conceptual framing of ISRaD and assisted with data ingestion. All authors read and commented on the paper. The USGS Powell Center working-group participants are CRL, JBM, AMH, GM, CAS, KH, JCB, SEC, ST, CR, CEHP, CS, AAB, MK, EMS, AFP, AT, JPS, and RW.
The authors declare that they have no conflict of interest.
We gratefully acknowledge our funding sources and the ESSD journal reviewers. We also thank Lisamarie Windham-Myers and Sanjay Advani, who provided thoughtful comments on an early version of this paper. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
This research has been supported by the US Geological Survey Powell Center for the working group on Soil Carbon Storage and Feedbacks, the Max Planck Institute for Biogeochemistry, the European Research Council (Horizon 2020 Research and Innovation Programme, grant agreement 695101), the USGS Land Change Science mission area, and the US Department of Agriculture (Soil Carbon Working Group award 2018-67003-27935).
This paper was edited by Giulio G. R. Iovine and reviewed by Jonathan Sanderman, Troy Baisden, and three anonymous referees.