Livestock enclosures in drylands of Sub-Saharan Africa are overlooked hotspots of N2O emissions

Field N₂O flux measurements

At seven different semi-arid and arid savanna regions in Kenya (Fig. 1, Supplementary Fig. 9), N₂O measurements from soils at 46 boma sites and 22 adjacent reference sites (undisturbed savanna) were carried out. At each boma and control site, 3–7 plots were chosen randomly for flux measurements. Local members of the pastoral communities, including herders and/or community elders were interviewed for information on the time since abandonment of each boma.

N₂O fluxes from bomas were measured using the fast-box chamber method¹⁵, deploying an ultra-portable greenhouse gas analyzer of ABB-Los Gatos Research Inc. (Modell 909–0041). A gas-tight, vented chamber (0.3 × 0.2 × 0.15 m) was pressed against the ground on foam frames for 4–7 min, during which time sample air was pumped from the headspace of the chamber to the analyzer and returned to the chamber thereafter. In this way, changes in headspace N₂O concentrations were continuously measured over the sample period, with a running average of every 5 s. Linear regression over the sample period was used to calculate fluxes. The detection limit for N₂O fluxes was <1 µg N₂O-N m⁻² h⁻¹.

Soil and environmental parameters

Environmental variables including gravimetric and volumetric soil water content, soil and air temperatures, organic layer thickness, and aboveground plant biomass (if present) were measured in all plots. At selected boma sites, we also measured total N and C stocks for 0–1 m depth (see Supplementary Table 1 and Supplementary Fig. 9). We noted crusts only on bomas that were abandoned for <5 years. As nutrient concentrations diminish with time, vegetation starts to grow on abandoned bomas. Furthermore, owing to rainfall and trampling of wild animals and sometimes of livestock, crusts are fragile, fragmented and <5 mm in thickness.

Identifying drivers of N₂O fluxes

For regression analyses on controlling parameters of N₂O fluxes (Supplementary Tables 2–4; Supplementary Notes 2 and 3), we calculated an average flux per boma (or control) site for our analyses. We used Welch’s t test to compare N₂O fluxes from boma and adjacent savanna plots, as well as for comparing N₂O fluxes from vegetated plots with those from bare soil plots and from plots of high versus low organic layer depth (>0.3 m versus <0.3 m). A depth of 0.3 m was chosen as a threshold value to distinguish between bomas that were used for 1–2 years and those that were used consecutively for >2 years, as manure accumulates at a rate of 0.1 to 0.15 m yr⁻¹.

To determine the effect of potential driver variables on N₂O fluxes (log10), we explored a multiple regression model for gravimetric soil water content (including a quadratic term to allow a decrease at high soil water content) as well as soil temperature, organic layer N, organic layer C, vegetation cover, time since abandonment, and organic layer depth (Supplementary Table 3 and 4, multiple r² = 0.41, adjusted r² = 0.26). Only soil temperature significantly affected N₂O fluxes in the multiple linear regression model. In addition, a mixed modeling approach was used to assess whether random effects (e.g., on site) improved the model. This did not improve the model performance, indicated by a higher AIC (138 versus 91).

To estimate overall N₂O emissions of bomas over time, we built a simple regression model relating boma age (since abandonment) to average boma N₂O fluxes on a log-log scale (Supplementary Table 2, displayed in Fig. 2). Since the variability explained by this model is moderate (r² = 0.13), there are relatively wide confidence intervals (Fig. 2), but these reflect the full variability of our measured N₂O fluxes. From the model fit, we calculated the cumulative N₂O emission on a per-area basis over 40 years, i.e., including oldest abandoned bomas investigated.

Upscaling N₂O fluxes from bomas to SSA

The estimated N₂O flux over the 40 years period served as an input for upscaling measurements, thereby taking into account uncertainty of estimated N₂O flux as well as the uncertainty in further parameters used for spatial upscaling (Supplementary Fig. 7 and below).

Gridded estimates of livestock numbers were taken from Gilbert et al.⁶ (Fig. 4f) and restricted to semi-arid and arid environments of SSA (Supplementary Fig. 10). The calculation of total livestock units for pastoral systems (i.e., sum of cattle+goat+sheep) was based on livestock conversion factors as reported by Houerou and Hoste²³, which are for cattle in herd 0.7 and for sheep and goat 0.1. Information on the boma area per livestock unit was obtained from Okello et al.²⁴. Based on interviews with herders and local experts it became evident that several bomas are used at the same time as herders move their livestock across grazing areas in response to the availability of forage and water and that at least two bomas are simultaneously in use on average—one close to a semi-permanent homestead and one within seasonal grazing areas. The number of years that a boma is used varies considerably across sites, between 1 month to up to several years in this study and confirmed by several others^12,25, with 3.7 years being the central value. Information on manure extraction from bomas was not available. Based on interviews with herders we assumed for 90% of bomas manure is either not extracted or irregularly or only partially extracted. Manure use increases in regions with mixed farming, i.e., simultaneous crop and livestock production and short (<2–5 km) transport distances, and with increasing humidity²⁶, but is hardly existing in remote regions with livestock production only, which are in the focus of our study.

Total N₂O emissions from bomas in semi-arid and arid environments of SSA were calculated as follows:

Step 1: Calculation of total livestock numbers (TLN) by combing population data for cattle, goat and sheep and converting to a single livestock unit, based on conversions for semi-arid and arid environments²³.

Cattle, sheep, goat: total number of cattle, sheep, goat

Step 2: Calculating boma use intensity (BUI)

BUI = BAL * NB * FMB/YB

with

BAL: boma area per livestock, the work of Okello et al.²⁴ indicates that the space for one cattle in a boma is ~4–16 m², central value: 10

NB: number of bomas in use at the same time, interviews and expert knowledge, central value = 2.5

FMB: Fraction of bomas without use of manure, interviews and expert knowledge, central value = 0.9

YB: years of boma use, interviews and expert knowledge, central value = 3.7

Note: in 2015 the total area of newly abandoned bomas in semi-arid and arid SSA was estimated at 1792 km². The total area of semi-arid and arid SSA equals 15.16 Mio km². Thus, the area of newly abandoned bomas equals 0.11‰ of the total area of semi-arid and arid SSA.

Step 3: Calculating N₂O emission intensity (N₂O_int)

N ₂O_int = N ₂O x N ₂O_years * 4/28

N₂O: mean average annual N₂O flux from bomas, this study

N₂O_years = observed minimum number of years with N₂O fluxes significantly greater than at adjacent savanna sites = 40 years

44/28 = conversion of N₂O-N to N₂O

Step 4: Calculating total N₂O emissions from bomas in semi-arid and arid environments

Or (summarizing steps 1–3)

Supplementary Fig. 7 shows histograms, probability distribution functions, and cumulative distribution functions of the different variables used for assessing the uncertainty of the upscaling procedure. To provide statistics such as median, 25 and 75% percentile values, etc. for upscaled N₂O emissions we used the Latin hypercube sampling (LHS) method. The concept behind LHS is to divide the cumulative curve of each variable into n equally probable intervals and take a random sample at each probable interval for each variable. The n values obtained for each of the components were then paired with each other to recalculate the equation n times for assessing the uncertainty in a prediction equation.

Source: Ecology - nature.com