The research was conducted at Endehua Kebele in the Fageta Lekoma District, Northwestern Ethiopia ( Figure 1). It is situated at 11°01’12” N latitude and 36°55’23” E longitude with 2761 meter elevation. The area belongs to the moist subtropical agro-climatic zone ( Bireda, 2015) having about 16.8°C of mean annual temperature and 1,702 mm of mean annual rainfall between 2010 and 2019 based on the Addis Kedam weather station in Northwestern Ethiopia Metrological Center ( NWMSC, 2020).

Cropland (52.3%), forest land (25. 6%), grassland (21.3%), and villages (0.8%) are the major land uses with high land conversion rates to A. decurrens woodlots ( Menale & Wolde 2018). According to the Food and Agriculture Organization ( FAO 2015b) classification, Acrisol and Nitosol are the predominant soil types in the area and known with low productivity ( Bireda 2015). The district’s total population was projected to be about 151,220 in 2017 with a density of approximately 226 inhabitants km ^-2 ( Central Statistics Agency (CSA), 2013). The major sources of income for community's livelihood are crop and livestock production and tree products mainly A. decurrens ( Yigez et al., 2018). Recently, about 90% of the population has been involved in A. decurrens charcoal production chains ( Bireda 2015).

The experimental A. decurrens stands were established in July 2015 using four month-old seedlings. The land use history of the site was only cropland. A randomized complete block design (RCBD) was used having five planting density as treatments (Treatment 1 (T ₁): 0.5 m x 0.5 m; Treatment 2 (T ₂): 0.75 m x 0.75 m; Treatment 3 (T ₃): 1 m x 1 m; Treatment 4 (T ₄): 1.25 m x 1.25 m; and Treatment 5 (T ₅):1.5 m x 1.5 m) with four replications. The number of seedlings planted initially per plot was about 196 for T ₁, 100 for T ₂, 49 for T ₃, 36 for T ₄, and 25 for T ₅. Each plot has a surface area of 49 m ² with 2 and 3 m distance between plots and blocks, respectively.

All stands were neither fertilized nor irrigated but hand weeding was applied every two months during the first half-year. Data were collected when the stands were 4.5 years old, within the ranges of stand rotation period (4–6 years) in the area ( Takele, 2019). The stand characteristics for each treatment during data collection are illustrated in Table 1.

Fine root sampling and processing

During December 2019, fine roots (<2 mm in diameter) were collected using a soil core method, because such methods are relatively easy to apply in the field, cost-effective, and accurate while having rather limited disturbances of the site ( Ravindranath & Ostwald , 2008; Addo-danso et al., 2016). Before coring, sample points were selected randomly at each plot and their litter layer of surface soil were removed. About 100 root and soil core samples were taken by a stainless corer (6.67 cm diameter) up to 50 cm depths at all selected points (5 spacing x 4 replicates x 5 soil depth class = 100). Sampling was done down to 50 cm because most species likely extend their root system down to 50 cm due to competition for soil nutrients and water and reach the bedrock. All individual core samples were separated into five depth classes (0–10, 10–20, 20–30, 30–40, and 40–50 cm) and then composites were made per plot and depth class and labeled in a separate plastic bag.

Fine root samples were sieved with 0.15 mm mesh size to separate roots from the soil. The roots were collected and then the sieve was rinsed with tap water to remove the remaining soil particles attached to the roots; then the roots were picked up and the soil handled for further analysis ( Ravindranath & Ostwald 2008). All fine roots were separated as live and dead by visual observation ( Dessie et al., 2017). This was based on the principle that live roots have usually brighter color, are elastic and have unbroken root tips, while dead roots are dark brown or black, broken root tips, are easily breakable, often shriveled ( Duncker et al., 2012; Jiang et al., 2018). The roots were further grouped as tree or herbaceous roots, whereby herbaceous roots possess a higher branching capacity than tree roots ( Jiang et al., 2018). All root samples were oven-dried at 70°C for 24 hours and then weighed using a sensitive balance (10 ^-4 g accuracy).

Coarse root sampling and processing

About 100 coarse root (diameter >2 mm) samples were also taken using the soil pit method that represents a compromise between sampling cost and efficiency ( Ravindranath & Ostwald 2008; Addo-danso et al., 2016). Two pits with a surface area of 900 cm ² (30 x 30 cm) were established down to a depth of 50 cm in each plot. One soil pit was dug in the center of the plot and the second one at a distance of 1 m further east to permit to take into consideration the sampling design. Then, all soil samples within each pit were taken separately with 10 cm intervals and composited per 10 cm depth interval. They were also passed through 2 mm sieve and handled by a separate plastic bag. The remaining sample processing was similar to that applied by fine roots.

Soil sampling and processing

About 100 Volumetric soil cores (5 cm diameter) were taken for soil bulk density determination. Thereafter, each sample was oven-dried at 105°C for 24 hours, weighed and then, sieved by 2 mm mesh size. Accordingly, the remaining soil fractions (>2 mm diameter) were also weighed to determine the coarse fraction content. The soil samples (taken on five depth classes of each plot) separated from the roots during root processing were also further analyzed for SOC content determination. For each depth class, soil samples were air-dried, ground and a further 20 g of soils were taken for SOC content laboratory analysis ( IPCC, 2003).

The oven-dried root biomass samples were standardized into gram per meter square (g m ^-2) for root biomass comparisons for each planting density and soil depth, then summed up for 0–50 cm depth as a total root biomass. The vertical distributions of fine and coarse root biomass at each depth were expressed as a percentage (in decimal) of the total root biomass (0–50 cm). Methods used for data estimations for each parameter are listed in Table 2.