Organic matter cycling along geochemical , geomorphic and disturbance gradients in forests and 1 cropland of the African Tropics-Project TropSOC Database Version 1 . 0 2 3

3 Sebastian Doetterl, Rodrigue K. Asifiwe, Geert Baert, Fernando Bamba, Marijn Bauters, 4 Pascal Boeckx, Benjamin Bukombe, Georg Cadisch, Matthew Cooper, Landry N. Cizungu, 5 Alison Hoyt, Clovis Kabaseke, Karsten Kalbitz, Laurent Kidinda, Annina Maier, Moritz 6 Mainka, Julia Mayrock, Daniel Muhindo, Basile B. Mujinya, Serge M. Mukotanyi, Leon 7 Nabahungu, Mario Reichenbach, Boris Rewald, Johan Six, Anna Stegmann, Laura 8 Summerauer, Robin Unseld, Bernard Vanlauwe, Kristof Van Oost, Kris Verheyen, Cordula 9 Vogel, Florian Wilken, Peter Fiener 10

C cycling and C stabilization in soils complement the dataset to deliver one of the first landscape 48 scale datasets to study the linkages and feedbacks between geology, geomorphology and 49 pedogenesis as controls on biogeochemical cycles in a variety of natural and managed systems 50 in the African Tropics. 51 The hierarchical and interdisciplinary structure of the TropSOC database allows for linking a wide 52 range of parameters and observations on soil and vegetation dynamics along with other 53 supporting information that may also be measured at one or more levels of the hierarchy.

Tropical soils responding to disturbance 105
With the expansion of cropland into forested landscapes soil erosion rates are expected to 106 continue to increase. Soil erosion will undoubtedly impact biogeochemical cycles and change the 107 input, storage and exchange of C between soils and atmosphere as well as the flux of nutrients 108 between plants and soils in tropical systems in the region. To understand how tropical soils and 109 ecosystems respond to erosional disturbance, it is necessary to consider the combined effects of 110 climate, geology, topography, soil formation, biological processes and human disturbance. To 111 date, no study on the interrelationship of these controls on biogeochemical cycles has been 112 carried out in tropical ecosystems. However, studies carried out in other regions have shown that 113 controls on soil C dynamics, for example, are highly interlinked (Doetterl et al., 2015a;Hobley and 114 Wilson, 2016; Nadeu et al., 2015). 115 Soil redistribution as a consequence of erosion also changes the functionality of landscape units. 116 For example, soil degradation on hillslopes is matched by a rapid buildup of sediment deposits in 117 valley bottoms, where C and nutrient rich soil is rapidly buried in subsoils under new sediments. 118 While this consequence of deforestation can lead to an increase in the residence time of C due 119 to slower microbial C turnover in buried soil (Doetterl et al., 2012;Alcantara et al., 2017), important 120 nutrients are now lost to plants leading to biomass productivity (Veldkamp et al. 2020) and 121 degraded tropical forests generally negative for microbial processes in soils (Sahani & Behera, 122 2001). Soil redistribution is also known to change the temporal and spatial patterns of soil 123 weathering and affects C stabilization. In agricultural systems, the effects of this pressure can be 124 Feedbacks on biogeochemical cycles between soil weathering, erosion will differ significantly not 129 only between natural and disturbed systems, but also between systems with differing soil mineral 130 reactivity. Recent advances have shown that mineral reactivity, constrained predominantly by soil 131 weathering and the mineralogy of the soil parent material, has direct control over soil organic 132 carbon, with climate exerting only indirect control through its impact on biogeochemical processes 133 and matter fluxes (Doetterl et al., 2015a;Tang and Riley, 2015). However, the exact effects of 134 mineralogy on the temperature sensitivity of microbial decomposer communities and the primary 135 productivity of ecosystems have, to date, not been constrained (Hahm et al., 2014;Tang and 136 Riley, 2015). 137

Importance and outlook of research on the future of tropical biogeochemical cycles 138
Tropical Africa is expected to experience great changes to both soil biogeochemical cycling and 139 ecosystem level carbon (C) fluxes between soil, plants and the atmosphere, with unknown 140 consequences for biogeochemical cycles. Despite decades of recognizing their importance, 141 tropical soils remain among the least studied in the world (Mohr and van Baren, 1954;Mohr et 142 al., 1972;Ssali et al., 1986;Juo and Franzluebbers, 2003). Although a more complete 143 understanding on soil-plant coupling in tropical environments is critical, most of our process 144 understanding on biogeochemical cycling between plant and soil is still derived from temperate 145 regions. However, due to differences in their environmental setting and soil forming history, many 146 tropical soil systems will likely react very differently to soil disturbance and land conversion than 147 temperate soil systems. For example, temperate ecosystems can differ fundamentally in the way 148 nutrients cycle and in the dominating and limiting factors for plant growth (Du et al., 2020). In 149 contrast to soils in the temperate zone, long lasting chemical weathering has led to a massive 150 depletion of mineral nutrients from soils in many tropical systems, although the remaining 151 available nutrients are very efficiently re-cycled in natural tropical biospheres (Walker and Syers, 152 1976;Vitousek, 1984). Hence, any loss of nutrients is therefore a critical disturbance with direct

Objectives and framework 164
In the following we aim at providing an overview on the data collected by project TropSOC which 165 is now available to the research community as an open access database. We give a brief 166 description of the project's design before elaborating the structure of the database and its content. 167 Note that beyond the overview information presented here, more details to methods and sampling 168 designs for each assessed parameter is explained in great detail in the supplementary metadata 169 files accompanying the database. 170 The main objective of project TropSOC was to develop a mechanistic understanding of plant and 171 microbial process responses to changing soil properties in the African Tropics exemplified along 172 land use, erosional and soil geochemical gradients studied in the Congo and the Albertine Rift. 173 Trying to understand biogeochemical cycling affected by human activities in tropical (agro-174 )ecosystems as a whole, TropSOC had two main foci: 175 (i) investigate how nutrient fluxes and organic matter allocation between tropical soils, plants differ 176 in relation to the controlling factors geochemistry, topography and land use. 177 (ii) investigate how the geochemistry of soils and their parent material control, interact with or 178 mediate the severity of erosional disturbance on C cycling in tropical soils. 179 In order to address these objectives, project TropSOC investigates effects on tropical soil 180 biogeochemical cycling and biological responses to variation in soil and environmental properties 181 along three main vectors (

Study area -Geochemistry and soil types 221
Within the study area three regions each representing a geochemical differing parent material for 222 soil formation were determined. The first region (Figure 2a) is predominantly situated on mafic 223 magmatic rocks, typically mafic alkali-basalts ranging in age between 9-13 Ma (Schlüter 2006  publication. In order to cover potentially stable, eroding and depositional landforms, topographic 287 positions of plots ranged from plateaus (slope < 5%), over two slope positions (slopes between 9 288 and 60%) to valley positions (slopes < 5%) (Table 1). 289 Table 1. Topographic information of TropSOC plots across different geochemical regions and land use. Slope and altitude are displayed as minimum and maximum values. Each topographic position per geochemical region contains the range between 3-7 field replicate plots.

Forest plot installation 291
Sampling in forests followed a strict catena approach and plots were established following an three replicate plots each in three geochemical regions). Note that three plots in the mafic region 299 had to be relocated due to safety reasons after the sampling period. For an overview on forest 300 plot sampling design see Figure 5a. 301

Sampling mineral and organic soil layers 302
At the time of plot installation, four replicate soil cores per plot (one in each subplot) were taken 303 in a depth-explicit way in 10 cm increments up to 1 m soil depth, and combined as composites 304 per plot. In addition, one soil profile pit was dug to a depth of 100 cm in the center of one of three 305 replicate plots (

Forest inventory and aboveground standing biomass 318
In 2018, full inventories of the forest tree species and standing aboveground biomass (AGB) were 319 conducted on all forest plots. The forest inventory followed an international, standardized protocol 320 for tropical regions (Matthews et al., 2012). First, we identified the species of all living trees with 321 a diameter at breast height (DBH, measured at 1.3 m above ground) greater than 10 cm in each 322 plot. Second, these identified trees were classified into the following empirical DBH classes: 10 -323 20 cm, 20 -30 cm, 30 -50 cm and > 50 cm. Third, to estimate the above-ground biomass (AGB), 324 we constructed stand-specific height diameter (H-D) allometric relationships using a 325 representative subset of the plot-specific trees (Méchain et al., 2017). For this, 20% of all 326 measured, specific trees were selected for height measurement, across the DBH range that was 327 recorded per plot. Depending on the tree abundance of each DBH class, the height of three to 328 five individual trees were then measured using a hypsometer (Nikon Laser Rangefinder Forestry 329 Pro II, Nikon, Japan). AGB for each individual tree was then estimated using the allometric 330 equation as described by Chave et al. (2014) for moist tropical forests. To estimate wood density 331 data, we used species averages from the DRYAD global wood density database (Zanne et al., 332 2009). To extrapolate this information for the entire plot for all our sites, we applied a stand-333 specific height-diameter regression model; modelHD, available within the R package BIOMASS 334 (Méchain et al., 2017). In a last step, aboveground standing biomass carbon stock was estimated 335 assuming that that all samples standing biomass has a 50 wt.% share of C (Chave et al., 2005). 336 A re-census was carried out in 2020, in order to detect changes in above-ground standing 337 biomass and to determine tree mortality. Tree mortality rate (λ) at each plot was assessed 338 following Lewis et al. (2004), using inventories conducted in 2018 and 2020. Tree mortality rate 339 was calculated for all tree stems with DBH>10cm in every plot. 340

Canopy leaves 341
To assess plant functional traits (leaf nitrogen, phosphorus, potassium, magnesium and calcium 342 content) of living canopy leaves (see section 2.7), we sampled, at the beginning of the weak dry 343 analyses, data quality and methodology. An overview of all datasets presented in this 526 database is given in Appendix Table A2. 527 In summary, TropSOC's first results demonstrate that even in deeply weathered tropical soils, 528 parent material has a long-lasting effect on soil chemistry that can influence and control microbial 529 activity, the size of subsoil C stocks, and the turnover of C in soil. Soil parent material and the 530 resulting soil chemistry need to be taken into account in understanding and predicting C 531 stabilization and turnover in tropical forest soils. Given the investigated rates of erosion on 532 cropland, our findings confirm the threat of large losses or organic matter leading to sharp decline 533 in soil fertility with little potential of soils to recover from nutrient losses naturally on decadal or 534

Basic information 562
The database comprises basic information of all plots and single point sampling positions where 563 data were collected during project TropSOC. An overview of the structure of the database is 564 presented in Appendix Table A2. The basic information of the database is structured in the 565 following way: 566 Part 1 -Location and basic background information for all plots and points where data were 567 collected. Data can be found in file 11_plots_points.csv, with description given in 568 11_plots_points.pdf. 569 Part 2 -Sample identifier for the database' internal connection between location of plots, points 570 and soil data from different soils depths as well as vegetation data. Data is stored in 571 12_sample_identifier.csv, with description given in 12_sample_identifier.pdf. 572 The key element to link all datatables for which data was collected and samples analyzed is the 573 plot ID and its derivative the sample ID. This identifier allows to link the results from sample 574 analysis with the locations given in 11_plots_points.csv. This results in a n:1 connection between 575 12_sample_identifier.csv and 11_plots_points.csv. See metadata file 11_plots_points.pdf for an 576 overview on the structure of the plots ID and 12_sample_identifier.pdf for an overview on the 577 structure of the sample ID. 578 579

Forest 580
TropSOC's forest data consists of seven parts (Table A2 for overview) structured as paired .csv / 581 .pdf files, containing the data (.csv) and accompanying metadata (.pdf) describing parameters 582 and methods. Additionally, an overview to all collected forest data is given in file 2_forest.pdf. 583

Cropland 602
TropSOC's cropland data consists of the following seven parts (Table A2 for overview) structured  603 as paired .csv / .pdf files, containing the data (.csv) and accompanying metadata (.pdf) describing 604 parameters and methods. Additionally, an overview to all collected cropland data is given in file 605 3_cropland.pdf. 606

Meteorological data 618
The meteorological data comprises 4 parts (Table A2 for  distribution of data points across the various levels of the database hierarchy is shown in Table  636 2. All individual data entries present in the database have passed quality control done by experts 637 that were involved in the creation of the data. Where applicable, reports on the quality assessment 638 of each parameter can be found in the metadata .pdf files accompanying the .csv files. 639 Table 2

Consecutive database versioning and archiving 674
Updated versions of the database will be periodically released following either substantial 675 changes or new peer-reviewed publications, leveraging the dataset. Versioning of these official 676 releases are tracked using an associated version number, e.g. TropSOC v1.0, and so on. These 677 official releases will be archived at ETH Zurich's Research collection via ETH's Soil Resources 678 Group (https://soilres.ethz.ch/) and the CBO data storage (https://www.congo-679 biogeochem.com/data) with a dataset DOI issued for each release via ETH Zurich so that users 680 may revert back to the earlier version if so required. These archived releases will be maintained 681 into perpetuity to facilitate reproduction of any analyses conducted using a past version of the 682 database. When accessing the dataset and using it for own research, users commit to cite the 683 original manuscript provided here in addition to the version number, DOI and any description 684 provided to future versions of the database (see section 6 for details). 685

Database governance and participation 686
TropSOC is a community effort with multiple contributors operating at different levels ( Figure 7). 687 Governance of TropSOC is required in order to ensure continuity of services and to plan for the 688 future evolution of this data repository. Studying the rapid environmental changes to the African 689 In addition, we strongly encourage TropSOC users to follow these simple guidelines for use: 746 (1) TropSOC users must agree not to manipulate the original source data without permission of 747 the TropSOC governance team described in section 5. This process should be followed in 748 particular when groups or individuals seek to use the TropSOC database beyond the scope 749 of its original objectives (see section 1.1). 750 (2) When utilizing TropSOC data, including the complete dataset, individually curated entries, or 751 value-added calculations, users should cite this publication and reference the version of 752 TropSOC that was used for their work under its specific DOI.

Appendix 804
Appendix Appendix Table A2. Structure of the TropSOC database. For each topic a .pdf file is given that 819 entails an overview for the available data on soil, vegetation and weather data collected for the 820 investigated forest and cropland plots. Each dataset then comprises a data-containing .csv file 821 and an additional metadata-containing .pdf file of the same name.