2.1 Study site
The research was carried out in the Guder watershed (10°59′30″–11°1′0″N, 36°54′0″–36°56′0″E, 2498–2857 m a.s.l., average 2590 m a.s.l.), which is located in the UBNB, Ethiopia (Fig. 1), in the highlands of the subtropical agro-ecological region. The soil in experimental plot was a Malabon silty clay loam (Pachic Ultic Argixerolls) ( Soil Survey Staff, 2014).
<< Insert Fig. 1 Approximately here >>
Temperature and precipitation data were obtained from Dangila and Enjibara meteorological stations, respectively. At Guder, the average yearly rainfall was 2394 mm, with monthly temperature averaging 16–20°C. Lowest temperatures ranged from 4.94°C in January to 12.88°C in June. Highest temperatures ranged from 21.98°C in August to 28.72°C in March (Fig. 2). Highest rainfall occurs in summer (June–Aug), and lowest rainfall occurs in winter (Dec–Mar). Autumn (Sept–Nov), when the soil is still moist, is the grain harvest season. In spring (Mar–May) the weather transitions from the dry season to the rainy season, with rains usually beginning in May. The research site’s main crops are potato (Solanum tuberosum), teff (Eragrostis tef), barley (Hordeum vulgare), and wheat (Triticum aestivum). Natural forests (3.8%), grazing land (9.5%), bush land (17.5%), acacia plantations (30.1%), and farmland (39.1%) are the predominant land use types (Sultan et al. 2017).
<< Insert Fig. 2 Approximately here >>
2.2 Experimental plots
Soil bunds (SB), a common land management (LM) practice in the study watershed, were established in 2015 in an area with an average slope of 15%. The spacing between bunds was 5.5 m. The numbers of bunds in treated plots were three bunds. The area coverage by bund was 2133 m2 ha− 1. We planted teff, the main crop in Guder (Mulualem et al. 2021) and the major food crop in Ethiopia (Asargew et al. 2021), during two experimental years [from September 2020 to August 2021 for seasonal soil respiration, and 16–29 September 2021 for diurnal soil respiration].
Polyvinyl chloride collars (diameter 0.19 m, length 0.11 m, insertion depth 0.05 m) were used to measure soil respiration. To reduce the soil disturbance impact, all collars were installed at least one week before the first flux measurements, and not moved during the course of the experiment. Six PVC collars (three replicates for control and soil bund treatments) were used to examine seasonal variation. In these same experimental plots, eighteen fixed PVC collars were established in the second year (nine replicates × two treatments (control and soil bund) (Fig. 3).
<< Insert Fig. 3 approximately here >>
2.3 Data collection
Soil respiration was monitored using a LI-8100A soil gas flux system (LI-COR, Lincoln, NE, USA). To assess seasonal fluctuations, soil respiration was assessed monthly (Liang et al., 2019) in SB and control plots. Seasonal soil respiration data were only collected on rainless mornings, between 9 am and 12 am (Sheng et al., 2010). We took soil respiration measurements in the middle of every month from September 2020 to August 2021. Each soil respiration reading lasted for 90 seconds and the average value was calculated for each PVC collar.
To examine diurnal fluctuations in soil respiration, we collected data four times each day at six hours interval from 18 PVC collars [9 SB and 9 control plots] for two weeks (16–29 September 2021, at 11 a.m., 5 p.m., 11 p.m., and 5 a.m.) (Fig. 3).
Using linear models of CO2 accumulation vs. time, we evaluated the linearity of CO2 accumulation for each collar on each sample day. Observations that did not meet the R2 \(\geqslant\) 0.9 threshold were discarded (Savage et al. 2008). During short enclosure periods (1–3 minutes), this technique yields more consistent soil respiration results (Kandel et al. 2016).
<< Insert Fig. 4 Approximately here >>
Soil moisture and temperature were measured concomitantly with both diurnal and seasonal soil respiration. We took 200-g soil samples for soil moisture analysis in each experimental plot while collecting data on soil respiration (Fig. 4C and D). Soil samples were dried in an oven for 24 h at 105°C. The weight difference between wet and dry soil samples was divided by that of the dry sample, and then multiplied by 100 to yield the gravimetric soil moisture content (%). Soil temperature was measured using a soil temperature Omega probe 6000-09TC (Li-Cor, Lincoln, NE, USA) at the same depth (0.05 m) as the soil respiration measurements.
2.4 Soil sample analysis
Soil samples for physical and chemical analysis were collected from every plot. Soil samples were crushed and sieved with a 2-mm mesh sieve after air drying at room temperature. A pH meter was used to test the pH of the soil in the supernatant suspension of a 1:2.5 soil and water combination (Peech 1965). A C/N analyzer (Macro Corder JM1000CN, J-Science Lab, Kyoto, Japan) was used to analyze soil total nitrogen and carbon. The Olsen method was used to determine the amount of available phosphorus in the soil (Olsen et al., 1954). Oven-dried soil mass was divided by its volume to determine bulk density. Soil texture analysis was performed by using hydrometer (Bouyoucos 1962).
2.5 Statistical analysis
The mean differences of seasonal soil respiration data between SB and control plots were checked for normality and the mean difference was not normal. Thus, to test whether soil respiration differed significantly between plots with and without soil bunds, we used the Wilcoxon signed-rank test.
Diurnal soil respiration data were not normally distributed and were transformed by using the square root data transformation; the Kolmogorov–Smirnov normality test was applied at p < 0.05. The homogeneity of variances for diurnal soil respiration and temperature was tested using Mauchly’s sphericity test. One-way repeated measure analysis of variance (RMA) was used to examine the significance of differences between plots. After determining the significance of differences between mean values, the least significant difference test (LSD) was used to determine mean separation at p < 0.05. Diurnal soil temperature data were normally distributed (Kolmogorov–Smirnov normality test at p < 0.05) and were analyzed using a one way-repeated measure ANOVA. A paired t-test analysis was employed to evaluate differences in soil moisture between SB and control plots.
Pearson’s correlation coefficient was used to assess the association between soil respiration and soil moisture and temperature. SPSS Statistics software was used to conduct the statistical analysis (v.26 for Windows, IBM Corp., Armonk, NY, USA). OriginPro 2019b (Origin Lab. Corp., Northampton, Mass., USA) software was used for generating graphs.
Soil CO2 flux was calculated by using the following equation (LI-COR, 2015) (Eq. 1).
$$Rs=\frac{{V{P_o}}}{{RS({T_o}+273.15)}}\frac{{dC}}{{dt}}$$
1
,
where Rs is soil respiration rate (µmol m− 2 s− 1), V is chamber volume (cm3), Po is initial pressure (kPa), R is the universal gas constant (0.008314 cm3 kPa mol− 1 K− 1), S is soil surface area (cm2), To is initial air temperature (°C) and dC/dt is the initial rate of CO2 mole fraction change (µmol mol− 1 s− 1).
An exponential function was used to simulate the relationship between soil respiration and temperature (Lloyd and Taylor 1994) (Eq. 2):
,
where; Rs is CO2 flux (µmol m− 2 s− 1), T is the soil temperature (°C); a and b are the model coefficients, and e, base of the natural logarithm (~ 2.718281828459).
A polynomial function was used to estimate the link between soil moisture and soil respiration (Peng et al. 2015) (Eq. 3) :
,
where Rs is as defined above; m is soil moisture (%); a, b, and c are the model coefficients.
To determine how soil temperature and moisture interact to affect soil respiration, we used multiple regression (Cui et al., 2020) (Eq. 4):
,
where all variables are as defined in the above equations.