Globally, peatlands cover approximately 420 million ha, equivalent to 2.8% of the world land area (1). Of this area, approximately 44-170 million ha are found in the tropics (2–4), with an estimated tropical peatland carbon (C) pool of 81.7-91.9 Gt C (4). The C rich soils in peatlands are formed from long-term accumulation of dead plant material, preserved under anaerobic conditions due to high water table levels. Land use change in peatlands results in significant C losses to the atmosphere, principally due to drainage leading to microbial decomposition of the soil C, aquatic export of dissolved and particulate organic C, and methane emissions from drainage channels (5). Of the total area of tropical peatlands, it is estimated that approximately 24.2 Mha are degraded, with a loss of 1.48 Gt CO₂ per year (3).

While the Congo Basin in central Africa has the most extensive peatland complex on the continent [and the most extensive in the tropics (6)], neighbouring Uganda is the largest contributor to emissions from degrading peat in Africa, where at least 0.5 Mha of Uganda’s estimated 1.3 Mha of peatland has been degraded, resulting in CO₂ emissions of up to 20 Mt yr^-1 (7). However, there is limited research into the use of peat soils in Uganda, and the impacts of peatland degradation on soil C and greenhouse gas (GHG) dynamics (8).

Across Uganda wetland types vary, and can be found fringing lakes, in lowland plains and in valley bottoms. Vegetation found in these wetlands are diverse and include grassland, Cyperus papyrus, woodlands, and palms, with plants either anchored into the soil or floating on the water’s surface (9, 10). These wetland areas are where Uganda’s peat soils are found. While the extent of peat soils across the country is still to be fully documented, they are often found underpinning papyrus-dominated wetland areas, in particular in the SW of the country (11). The valley bottom wetlands of the Kigezi highlands close to the border with Rwanda have some of the deepest reported peat deposits of all of Uganda (11), but the most widespread wetland drainage and land use change occurs here as well (12). Wetland conversion is largely due to agricultural drivers, with cultivation of potatoes on these peat soils estimated to provide income of US$ 3,000 ± 1,000 ha^-1 yr^-1 to smallholder farmers (12).

While the key environmental drivers affecting GHG emissions from soils have been documented in many studies and indicate strong relationships between temperature, water table depth (WTD) and soil respiration flux (13–18), there is limited understanding of these relationships for the diversity of conditions found across tropical peatlands. It is important to highlight the need to differentiate the CO₂ emissions from heterotrophic respiration during microbial decomposition of soil organic matter (R_h) and the emissions from roots and autotrophs which is not a result of soil carbon loss (19), The majority of tropical peatland emissions research has been conducted in SE Asia (20, 21), and to a lesser extent in the Amazon basin (22, 23), but there are considerable gaps when it comes to understanding the range and biogeochemistry of C losses experienced by these drained peatlands under the diversity of land use changes and wetland conditions occurring in Sub-Saharan Africa. At an ecosystem level the work of Saunders et al. (8) identified reductions in ecosystem C due to agricultural conversion in wetlands in Uganda based on measurements of photosynthetic CO₂ assimilation, but how much soils contribute to overall carbon loss from these systems is unknown.

For these reasons, this study set out to understand the biogeochemistry and rates of C losses from peat soils under land use change, through measurement of heterotrophic CO₂ fluxes from a potato cultivated peatland in a deep peat wetland system in SW Uganda. The key objectives of this were to quantify the potential range of soil CO₂ fluxes associated with this land use change on this soil, identify if there is temporal variability in these CO₂ fluxes, and identify any relationships between soil CO₂ fluxes and key environmental variables.

This study was implemented in the catchment of the Ruhuma wetland system in Muko Sub-county, Kabale District, in the Kigezi Highlands of south western Uganda (-1.208752° latitude, 29.830968° longitude, approximately 1,900 m above sea level). The landscape comprises of steep mineral soil hillsides which slope into a minerotrophic wetland valley bottom. The entire landscape is a mosaic of small, privately cultivated plots, with those in the wetland managed under farmer cooperatives, and individual plots distributed between farmers, most of whom manage multiple plots.

The Ruhuma wetland system is part of a network of peat dominated, valley bottom wetlands in the area, located 35 km SW of Kabale town, where the river Ruhuma flows north out of Lake Bunyonyi. The wetland covers approximately 5.47 km², at an altitude of 1,930 m above sea level. Wetland edge cultivation began in the 1980s and by the late 1990s the majority of the wetland was drained and converted for cultivation (personal communication). Peat depths in this wetland are significant, with measurements in the site indicating depths of over 10 m (personal data), and other studies measuring peat depths of over 20 m in neighbouring wetlands (11). The sapric peat soils found in this wetland are highly organic, with average organic matter contents measured at 87.22 ± 3.9% [± SD, n=198 (24)] and bulk densities averaging 0.125 ± 0.04 g cm³ [± SD, n=835 (24)].

In order to determine the range and temporal variability in fluxes from peat soils under potato cultivation we measured CO₂ emissions at daily, weekly and monthly timesteps. To capture the weekly and monthly variability, we conducted four sets of repeated 5-day morning (between 9.00 and 11.00) and afternoon (between 15.00 and 17.00) measurements on thirty randomly located paired collars (15 on raised beds and 15 on adjacent trenches) within a potato field. In addition to this, on the 4^th day of each sampling period we added sampling at 18.00, 21.00 and 24.00, and 5.00 and 12.00 the following day, with the intention of capturing diurnal variability in temperatures and emissions. In order to determine the relationships between soil CO₂ fluxes and key environmental variables, soil CO₂ emissions, soil and air temperatures, WTD were all recorded in situ from each collar for every measurement. Measurements were all conducted from within the same farmer’s field in an area under management by the Rurigita Farmer’s Cooperative of the Ruhuma wetland.

When preparing the experimental design we were limited by the lack of available land use information, in particular around how the soils were used over the year. We targeted our sampling campaigns as best as possible towards capturing the potential difference in environmental and management conditions that the soils would experience over the year. Measurements were taken at the end of the rainy season in May 2014 when wetland soils were dug in preparation for planting of potatoes, in the dry season in August 2014 during the potato growing season, in the rainy season in November 2014 after the harvest at the start of the fallow period, and in February 2015 during the dry season and fallow period prior to the next season of planting. Farmers digging of the potato beds in May involved the clearing of fallow weeds into the old trenches between old, raised beds, and then piling the soil from the previous year’s beds into these trenches. Once the old bed had been transferred into the old trench, further digging down into fresher peat brought up previously unexposed soil which was piled on top of the new beds. After approximately 48 hours these new beds were planted with rows of potatoes. Potatoes grew for approximately 3.5 months, when they were harvested for market with the onset of the new rainy season in September. After harvesting, beds were left fallow until the following June, when the cycle of mining and mounding fresh peat to create new beds was repeated.

Collars made from PVC piping 10 cm in diameter and 5 cm in height were fully inserted into the soil. Care was taken to place collars away from any plants and they were checked on removal at the end of each sampling period to ensure no plant roots were present. These collars remained in situ for each 5-day set of samples and were then removed and randomly placed for the next month’s sampling. Soil CO₂ emissions were sampled using a PP Systems EGM-4 with a SRC-1 Soil Respiration Chamber placed onto the collars. Chamber measurements were set to run for up to 120 seconds. The WTD was recorded using a tape-measure and pre-prepared dip-well near to each pair of collars, and soil and air temperature were recorded during each flux measurement using digital thermometers with a 0.1°C resolution.

All data analysis was conducted in R Studio v 1.4.1106 using R v 4.0.4. A linear function was applied to the concentrations of CO₂ recorded for each measurement and data was then converted into flux based on the following equation:

where R_h was the rate of efflux (g m^-2 hr^-1), S was the slope of the concentration increase during the measurement (ppm s^-1), P was the pressure in the chamber (mBar, recorded by the EGM), T was the soil temperature (°C), V was the chamber volume (0.001171 m³), and A was the collar area (0.0078 m²). The ratio 40.01 22.41 accounts for the area occupied by one kg mol of CO₂ at standard temperature and pressure. Division by 1,000 and multiplication by 3,600 converts kg to g and seconds to hours.

Mean results are presented with either standard deviation (SD) or standard error (SE), as noted in the results. T-tests and analysis of variance (ANOVA) were used to compare results, along with Tukey’s Honestly Significant Difference (HSD). Linear models were developed based on log transformed R_h. We explored a series of linear models for R_h and WTD, soil temperature and soil management (raised beds, trenches and recent disturbance). Regression models were further improved with the log transformation of the R_h.

Over 1,300 soil flux measurements were taken as part of this study. We observed a significant difference in the rate of soil flux from the raised beds (referred to as raised beds, n=648), compared to the narrow drainage trenches between them (referred to as trenches, n=653, t-test, p<0.005, Table 1 ), and as such all results are split based on this micro-topographical difference. Across all samples, the mean heterotrophic respiration (R_h) emissions of CO₂ from the raised beds was 1.39 ± 0.6 g m² hr^-1 (± SD) and from trenches was 0.22 ± 0.1 g m² hr^-1.

Management of these peat soils for potato cultivation leads to significant R_h, in particular due to the mounding of fresh peats to ensure soils are raised above the water table. These raised beds had emissions nearly an order of magnitude higher than the trenches in between them.

While there were lower air temperatures at night, nocturnal soil temperature and R_h were not significantly affected and as such we conclude that there is no significant diurnal variability in emissions. Significant temporal variability was found over monthly time steps, rather than over hours or days, as we observed greater changes in key variables (soil temperature and WTD) from our measurements taken in the different months.

Soil management (the different microtopography from the raised beds and trenches, and disturbance from planting and harvesting), as well as the soil temperature and water table depth were the most significant variables that impacted the rate of R_h.

Our area weighed annual flux estimate of 98.79 ± 1.7 t CO₂ ha^-1 yr^-1 is the first attempt at providing an emissions factor for this land use on these peat soils in Uganda.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

JF designed the experimental work, conducted the field measurements and data analysis and lead the writing of the manuscript. CL supported the experimental design, contributed to the data analysis and provided input to the manuscript. JS supported the experimental design and provided input to the manuscript. All authors contributed to the article and approved the submitted version.

This research was conducted under the ALTER research project. ALTER (Grant No: NE-K010441-1) was funded with support from the Ecosystem Services for Poverty Alleviation (ESPA) programme. The ESPA programme is funded by the Department for International Development (DfID), the Economic and Social Research Council (ESRC) and the Natural Environment Research Council (NERC).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.