Impact of Slow-Forming Terraces on Erosion Control and Landscape Restoration in Central Africa’s Steep Slopes

Abstract

Large-scale land restoration projects require on-the-ground monitoring and evidence-based evaluation. This study, part of the World Bank Burundi Landscape Restoration and Resilience Project (in French: Projet de Restauration et de Résilience du Paysage du Burundi-PRRPB), examines the impact of slow-forming terraces on surface conditions and erosion in Isare (Mumirwa) and Buhinyuza (Eastern Depressions), Burundi. Slow-forming, or progressive, terraces were installed on 16 December 2022 (Isare) and 30 December 2022 (Buhinyuza), featuring ditches and soil bunds to enhance soil and water conservation. Twelve plots were established, with 132 measurement pins, of which 72 were in non-terraced plots (n_PT) and 60 were in terraced plots (PT). Monthly measurements, conducted until May 2023, assessed erosion reduction, surface conditions, roughness, and soil thickness. Terracing reduced soil loss by 54% in Isare and 9% in Buhinyuza, though sediment accumulation in ditches was excessive, especially in n_PT. Anti-erosion ditches improved surface stability by reducing slope length, lowering erosion and runoff. Covered Surface (CoS%) exceeded 95%, while Opened Surface (OS%) and Bare Surface (BS%) declined significantly. At Isare, OS% dropped from 97% to 80%, and BS% from 96% to 3% in PT. Similar trends appeared in Buhinyuza. Findings highlight PRRPB effectiveness in this short-term timeframe, and provide insights for soil conservation in steep-slope regions of Central Africa.

1. Introduction

Water erosion is a major environmental issue that involves the displacement and transport of soil particles by rain and surface runoff, leading to their deposition far from their original location [1]. This process significantly impacts soil functions such as food production, nutrient cycling, organic matter storage, and water retention [2].

Globally, approximately 75 billion tons of soil are lost annually due to erosion, affecting 12% of the Earth’s surface [3,4]. Agriculture contributes to about 75% of this soil loss [5].

In Africa, soil erosion is particularly severe, with the continent losing 280 million tons of agricultural yield annually, valued at approximately $127 billion [6]. In tropical Africa, erosion rates can reach 30–40 t/ha/year, significantly higher than the 17 t/ha/year in the U.S. and Europe, and well above the critical threshold of 12 t/ha/year [7,8].

The erosion problem in Africa is exacerbated by several factors, including poor agricultural practices, the cultivation of marginal lands, and rapid population growth [9]. Additionally, factors such as soil crusting [10] and intense rainfall [11], especially in steep-slope areas [12], contribute to the severity of erosion, particularly in Sub-Saharan Africa. Approximately 50% of African soils, including arenosols, leptosols, and ferralsols, have low fertility due to poor nutrient reserves and water retention capacity [13].

East Africa’s highlands are particularly vulnerable to land degradation, with water erosion affecting approximately 50% of arable land, reducing agricultural productivity [14]. These highlands, located above 1200 m a.s.l, span countries like Burundi, Ethiopia, Kenya, Rwanda, Tanzania, and Uganda. They are vital due to their agro-socio-ecological benefits, such as reliable rainfall, fertile soils, and low malaria rates [15,16]. However, high population densities, particularly in some areas with over 300 inhabitants per km², combined with unsustainable resource use, have exacerbated land degradation. Soil erosion depletes nutrients, leading to agricultural crises that threaten food security and sustainable development [17]. This issue has been documented in Burundi [18], Kenya [19], Rwanda [20], Tanzania [21], and Uganda [22]. In the highland regions of East African countries, Sustainable Land Management (SLM) practices, also referred to as Cross Slope Barrier Soil and Water Conservation (CSB-SWC) techniques, are implemented to mitigate soil erosion. In the context of this study, SLM refers to a set of integrated practices adopted by farmers to enhance agricultural productivity and land resilience.

Commonly adopted SLM measures include tree planting, construction of bench terraces, Fanya Juu terraces, contour grass strips, cut-off drains, application of livestock manure, improved agricultural practices (e.g., use of certified seeds and adherence to agricultural extension recommendations), mulching, crop rotation, compost application, livestock rearing (as a source of manure), and the use of chemical fertilizers to ensure adequate crop cover and yield.

Typically, farmers implement these technologies as part of a package comprising several of the aforementioned practices. However, survey data indicate that only approximately 5% of farmers had adopted such comprehensive SLM packages [17].

In Burundi, land degradation is particularly severe. According to the National Action Program for the Fight against Land Degradation, soil erosion is particularly problematic, with land losses estimated at 4 t/ha/year in the east, 18 t/ha/year in the central region, and over 100 t/ha/year in the mountainous areas of Mumirwa [23]. Addressing this issue requires adequate soil and water conservation (SWC) measures. Understanding the spatial and temporal origins of land degradation is crucial for designing targeted interventions, especially in areas lacking data on streamflow and sedimentation [24].

Cross Barrier SWC (CSB-SWC) techniques, such as Fanya Juu, soil and stone bunds, bench terraces, and plant barriers, have successfully reduced runoff by 13–71% and soil loss by 39–83% [4]. These techniques show promise in mitigating soil erosion and promoting agricultural sustainability. In Burundi, erosion is particularly severe on steep slopes, some of which exceed 100%. Nutrient-rich topsoil is lost through sheet and channel erosion, significantly depleting soil fertility. Many areas surpass the critical 12 t/ha/year erosion threshold [7], threatening sustainable agriculture [25]. Anti-erosion programs in Burundi, dating back to 1945, have included contour lines and anti-erosion hedges, with ongoing efforts to address this issue.

Pilot studies on slow-forming terraces in Burundi have been conducted for decades at research stations. Studies by Rishirumuhirwa [26] and Simonard et al. [27] demonstrated that these techniques can control erosion and improve water quality by trapping sediments. This underscores the importance of context-specific approaches for implementing erosion control techniques. While soil conservation efforts in Burundi date back several decades, there is limited research on the effectiveness of slow-forming terraces in rural fields.

Several studies have evaluated the effectiveness of slow-forming terraces in controlling runoff and soil erosion in East Africa. In Ethiopia, the implementation of Fanya Juu terraces resulted in reductions of runoff and soil loss by an average of 60% and 95%, respectively [28]. In Rwanda, a comparison between non-protected (NP) plots and plots with slow-forming terraces at Tangata demonstrated moderate effectiveness in reducing runoff (52%). However, at Murehe, PT plots generated higher runoff compared to NP plots. Soil loss was moderately reduced by up to 50% at Murehe, and by more than 80% at Tangata. Consequently, the corresponding P factor values exhibited spatial and seasonal variability [29].

In Kenya, the integration of various tillage practices with soil conservation techniques such as terracing, contouring, and contour strip-cropping has led to soil loss reductions ranging from 50% to 95% [30].

Although specific assessments of slow-forming terraces’ effectiveness in Uganda are limited, studies have been conducted on terraced plots cultivated with climbing beans in the Kabale district of southwestern Uganda. These terraced plots exhibited an annual soil loss of 121 kg ha⁻¹ yr⁻¹, compared to 548 kg ha⁻¹ yr⁻¹ on plots with traditional bush beans. This corresponds to a soil loss reduction coefficient of approximately 78% [31].

From 2019 to 2024, the World Bank’s Burundi Landscape Restoration and Resilience Project, named officially in French Projet de Restauration et de Résilience du Paysage du Burundi (PRRPB), aimed to fill this gap by implementing slow-forming terraces in rural areas of Buhinyuza (northeast) and Isare (west). Research indicates that reducing slope length can significantly decrease runoff and erosion [32], with shorter slopes and vegetation stabilizing soil [33]. Proper ditches and plant cover maintenance, particularly perennials, can further reduce erosion and runoff [34,35]. Agricultural and silvicultural practices also play a crucial role in the success of erosion control methods [36,37].

This study evaluates the effectiveness of slow-forming terraces in reducing erosion and conserving soil during the rainy season. By focusing on erosion plots in corn fields with non-vegetated anti-erosion ditches, the research aims to provide valuable insights into the practical impacts of these techniques. The PRRPB, implemented between 2019 and 2024, indeed integrates CSB-SWC techniques with innovative soil and water bioengineering (SWBE) methods in Burundi’s Bujumbura (Isare Commune) and Muyinga (Buhinyuza Commune) provinces [38,39]. Large-scale landscape restoration projects like these are crucial for achieving the Sustainable Development Goals (SDGs) and the Bonn Challenge, although they are often implemented without extensive on-the-ground monitoring [40].

2. Materials and Methods

2.1. Study Areas

The study was conducted in two communes of the PRRPB: Isare in Bujumbura Rural Province (Mumirwa region) and Buhinyuza in Muyinga Province (Eastern Depressions) (Figure 1). The project covered 22 hills in these communes, implementing anti-erosion measures such as ditches in local fields.

For erosion monitoring, six sites (listed in

Table 1. Sites retained in each BLRRP commune of the study area.

	Project Area
Province (Commune)	Bujumbura (Isare)				Muyinga (Buhinyuza)
		Slope [%]	Coordinates (lat/long)	Altitude [m a.s.l.]	Gasave and Karehe	Slope [%]	Coordinates (lat/long)	Altitude [m a.s.l.]
Study plots	Butuhurana	18.56	−3.32133° 29.43947°	1274	Karehe	43.2	−3.01784° 30.40513°	1400
	Giterama I	17	−3.32076° 29.43254°	1108	Gasave I	22.3	−2.95191° 30.42748°	1500
	Giterama II	18.1	−3.32038° 29.43239°	1089	Gasave II	24.9	−2.95191° 30.42752°	1502

) were selected with the owners’ permission: three in Isare on Benga Hill (Butuhurana and Giterama sub-hills) and three in Buhinyuza (two on Gasave Hill and one on Karehe Hill). Measurements were carried out in Isare from 16 December 2022 to 10 May 2023, and in Buhinyuza from 30 December 2022 to 17 May 2023. The monitoring was short term also due to the technical difficulties of each site.

The physical environment of the study area, covering the communes of Isare and Buhinyuza in the BLRRP project, is described as follows:

Isare Commune: located in the Mirwa natural region, one of Burundi’s eleven regions, Isare features the steepest topography in the country, with slopes often exceeding 70% and sometimes reaching 100%. The altitude ranges from about 1000 to 2000 m a.s.l. The geology consists of metamorphic rocks, predominantly gneisses, along with associated rocks, such as gneissic granites and quartzites. It also features alternating acid (granites) and basic (gabbros and dolerites) magmatic rocks. The soils are mainly recent tropical soils and kaolisols, classified as ferrisols and ferralsols, with clay derived from shales [41]. The climate, according to the Köppen classification, is primarily Aw4 near the plain, transitioning to Aw3 around 1700 m and Cw3 at higher altitudes. Annual precipitation ranges from 1000 to 1400 mm/year, increasing with altitude [41,42].

Buhinyuza Commune: situated on the western edge of the Buyogoma natural region, but classified as part of the Eastern Depressions by the BLRRP in 2020 [43]. Buhinyuza has low hills with steep slopes of around 50% near marshes. The average altitude ranges from 1300 to 1500 m a.s.l. [41]. The geology includes quartzitic summits, lateritic tabular summits, and granitic intrusive formations in the low hills. The soils are primarily oxic (ferralsols), ranging from clayey to clayey–sandy, developed on granitic substrates and quartzitic rock outcrops. These soils are generally sandy and transition to more weathered and laterized soils on intermediate flat summits [41].

Köppen’s climate data [42] suggest that Buhinyuza commune would have an Aw3 climate with an average annual precipitation of about 1100 mm. However, BLRRP data from 2020 show that the actual precipitation is approximately 925 mm, indicating regular water deficits due to a prolonged dry season [43].

In terms of vegetation, Isare Commune is now devoid of natural flora due to high population density (612 inhabitants/km²), which has led to extensive land cultivation to meet the growing demand for crops. As a result, the natural vegetation has been largely destroyed [44]. The area is now dominated by perennial crops, including bamboo in ravines, banana trees, and oil palms, along with mixed and seasonal crops [43]. In contrast, Buhinyuza, one of Burundi’s least populated communes (295 inhabitants/km²) [44], originally featured wooded savannahs, though much of it has been converted to cultivation outside Ruvubu Park. Sparse remnants of savannah trees were observed near the study plots in Gasave and Karehe.

On the hills of the two study communes, the population mainly cultivates corn and, occasionally, cassava or other perennial crops. Beans are also grown, either in association with corn or independently. Annual fallows are also common, typically belonging to elderly individuals.

Plots for this study were selected based on several criteria after a field visit to the two PRRPB communes. The criteria included (1) easy accessibility; (2) ongoing or completed earthworks; and (3) proximity of terraced and non-terraced control plots with similar slope, soil, plant cover, and exposure [45]. Following site delineation, a second visit was made to meet the landowners with the assistance of the hill chief. Of the 9 sites assessed in Isare, only 3 were secured for the study through annual rental agreements. The same process was applied in Buhinyuza. Landowners agreed to leave a 3 m strip uncultivated between the terraced and control plots and to avoid disturbing the plots with pins or nearby areas.

2.2. Erosion Pins Installation in the Plots

For each plot, the installation of erosion pins followed two phases: in the first phase, the plots were prepared for corn cultivation following the ordinary management routine, and in the second phase, the pins were established and seeds were planted.

In detail, during the first phase (shown in Figure 2), two adjacent plots were selected: one in the slow-forming (progressive) terraces plot and one in an adjacent control plot, each measuring 3 m in width. The length considered was the distance between two consecutive ditches, which varied between 6 and 8 m due to the terrain slope, which ranged from 17% to 45%. Both plots were cultivated similarly, with corn being the common crop in these communes. The plot with anti-erosion techniques featured anti-erosion ditches, while the control plot had a downstream ditch to collect runoff and sediment during erosion events.

In the second phase, eucalyptus pins were used for installation to prevent theft or damage, unlike metal pins. They were positioned as described in the following (Figure 3).

• Plot with slow-forming terraces: Two pins (1D right, 1G left) were placed immediately downstream of the upstream ditch, about 1 m apart. Two pins (2D, 2G) were set in the middle of the anti-erosion ditches, between the upstream and downstream ditches. Two pins (3D, 3G) were placed immediately upstream of the downstream ditch or near the embankment. Two pins (4D, 4G) were installed inside the erosion control ditch downstream.
• Control plot: Pins were aligned in the same positions as those in the progressive earthworks plot, with the same numbering prefixed by a “t.”

A horizontal metal bar with holes was placed above the pins to facilitate height measurement between them. After installing the erosion pins, corn was planted in both the terraced and control plots. Seeds were sown in prepared pockets in each plot.

2.3. Data Collection and Processing

Data collection focused on assessing surface conditions and measuring soil loss to erosion in the two adjacent plots.

2.3.1. Surface Conditions Data Collection and Processing

Surface condition data collection aimed to quantify coverage percentages, openness, and closure in each plot, along with estimating roughness. These factors affect soil erosion and sediment yield.

In Isare and Buhinyuza, the soil surface condition was described using the point-quadrat method [46]. This method involves defining two diagonally placed transects within the plot, extending from the boundaries, using a meter. A stake was systematically dropped every 10 cm along these two diagonals (Figure 4). For each point, we recorded whether the surface was bare or covered, as well as whether it was closed or open:

➢

The covered surfaces at ground level (CoS%) include all litter (L%), vegetation (Veg%), and stones not integrated into the mass of the soil (SNI%);

➢

The open surfaces (OS%) mainly include cracks (Crk%), galleries, and clods (Cld%) that form traps favoring infiltration;

➢

The closed surfaces (CS%) correspond to areas blocked by a film or impact crust (Cr%) or visible stones integrated into the ground (SI%), as well as areas already eroded (AE%).

It is noteworthy that CoS% is opposed to BS% and OS% is opposed to CS%, and the relationships are as follows:

CoS% = L% + Veg% + SNI%

(1)

OS% = Crk% + Cld%

(2)

CS% = Cr% + SI% + AE%

(3)

Thus, the proportion for each type of surface (TS%) was estimated using the following formula:

$$TS\%={\displaystyle \frac{\sum nTS}{Ntotal}}\times 100$$

(4)

With

-nTS: Number of constituent elements observed of each type of surface at every 10 cm on two (2) diagonals of the plot;

-Ntotal: Total number of times of measurements on two (2) diagonals considering every 10 cm for all type of surfaces.

After obtaining the proportion of each type of surface after each measurement per plot (TS%), it was necessary to calculate the monthly average of the proportions for each type of surface per study site, as there are three sites per commune.

Finally, the global average of the proportions for each type of surface across each study site was calculated after the six-month measurement period (GA%).

The roughness index (Ri%) was determined for the studied plots using the chain method [46], which involves using a metal chain laid in a straight line. In our case, we used a rope placed along the width of the plot, or laid transversely. Three repetitions were measured: the first upstream, the second in the middle, and the third downstream, close to the non-vegetated embankment for the terraced plot, and at the same level in the non-terraced plot. The index corresponds to the ratio of Ld (m), the length of the rope extended after the measurement, and l (m), the width of the plot measured when the plots were established. Equation (5).

$$Ri\%={\displaystyle \frac{\left(Ld-l\right)\times 100}{l}}$$

(5)

where Ld (m) is the length of the extended chain after measurement, and l (m) is the width of the plot measured when the plots were established.

2.3.2. Erosion Data Collection and Processing

Erosion data were collected between two pins on each plot. Monthly measurements were taken using a horizontal metal bar (1.2 m) with 10 cm spaced holes. A graduated metal rod was used to measure the distance between the bar and the ground at each hole. To detect sedimentation or erosion, we subtracted the distances measured in consecutive months. A positive difference indicated erosion, while a negative difference indicated sedimentation. Data collection followed the scheme shown in Figure 5.

The months in which there is the most erosion are the rainy season months; unfortunately, in the area under study, the collection of rainfall data is particularly arduous due to the absence of rainfall stations.

Hence, using the monthly distances between the bar and the ground at nine points, we compared measurements taken in successive months.

The difference (di) between these measurements was calculated. The average of these differences (dm) was then computed to determine erosion or sedimentation at each point on the plot. The average of the observed differences (dm) is expressed by the following equation. Equation (6):

$$dm={\displaystyle \frac{{\sum }_{i=1}^{9}di}{9}}$$

(6)

where di is the difference between the distances of two consecutive measurements (m), and i is the index for the difference obtained at each point on the horizontal bar between the two pins.

Next, the apparent density (da) was determined at each field where the plots were installed, based on surface samples at a depth of 0–10 cm. This density was obtained from Soilgrids250 2.0 in g/cm³ [47].

Subsequently, the soil losses per unit area (Pt) in g/cm² were estimated using the following. Equation (7):

$$Pt=dm \times da$$

(7)

where dm is the average of the observed differences (dm) from measurements taken in the plots between the pins, and da is the apparent density of the soil obtained from Soilgrids250 2.0 in g/cm³.

Regarding data processing, SPSS (year 2024) was used for statistical analysis to examine the data on surface conditions and sediment quantities between terraced and non-terraced plots. The Student’s t-test was used to compare means for each factor between terraced (PT) and non-terraced (n_PT) sites. Analysis of variance (ANOVA) has been used to compare the means of each parameter that characterize the surface of plots. As we have five averages to compare by parameter, we used ANOVA and results are in following tables. Also, we used it to analyze the monthly evolution of sediments and the results are reported in Figures 7–10. Additionally, the nonparametric Mann-Whitney U Test was applied to factors that did not meet the normality assumptions required for the t-test.

3. Results

3.1. Impacts of Slow-Forming Terraces on the Surface Condition in Plots with Corn Cultivation

3.1.1. Isare Commune

According to our results (

Table 2. Monthly evolution of the surface condition of plots covered and not covered with corn in Isare.

Monitoring	Surface Conditions in PT at Isare					Surface Conditions in n_PT at Isare
	OS%	CS%	CoS%	BS%	Ri%	OS%	CS%	CoS%	BS%	Ri%
16 to 22 December 2022	97.17	2.83	3.43	96.57	1.76	97.60	2.40	2.40	96.43	1.30
26 to 28 January 2023	78.33	21.67	47.33	52.67	5.37	72.33	27.67	27.67	55.00	10.37
2 to 4 March 2023	79.00	21.00	66.67	33.33	4.45	68.33	31.67	31.67	35.00	11.12
1 to 9 April 2023	77.33	22.67	90.67	9.33	3.61	61.67	38.33	38.33	17.67	11.86
6 to 10 May 2023	80.33	19.67	96.83	3.17	2.41	61.33	38.67	38.67	4.00	10.93

), the OS% in terraced sites with Crk% and Cld% remained consistently high, around 80%. In contrast, the open surfaces in non-terraced sites significantly reduced to about 60% by the fourth and fifth periods, after being around 70% during the second and third periods. This reduction suggests progressive changes in the landscape.

Regarding surface features, in non-terraced plots, closed surfaces (with SI% and impact crusts) doubled by the last measurement period, reaching 38.67%, compared to 19.67% in terraced plots. This increase might indicate a shift caused by soil erosion processes in non-terraced sites.

Both terraced and non-terraced sites showed a notable increase in vegetation (Veg%) and litter (L%) cover over time, rising from 3% to 97% in terraced sites and from 3% to 96% in non-terraced sites. This uniform progression suggests that soil cover through vegetation growth was not significantly influenced by the presence of terraces.

Regarding surface roughness, in non-terraced sites, the roughness index (Ir) progressively increased from 1% to 12% over the first four periods before stabilizing at 11% by the fifth period. This increase in roughness is likely due to the erosion of clods, which exposed channels and integrated pebbles. In contrast, in terraced sites, the roughness index decreased from 5% in the second period to 2% by the fifth period, indicating a more stable surface with less erosion.

In terms of erosion dynamics, in non-terraced sites, the erosion of clods and galleries (Cld%), despite the presence of vegetation (Veg%), exposed underlying pebbles (SNI% and SI%) and facilitated the formation of impact crusts (Cr%). These crusts, combined with slope length, led to increased surface roughness. Additionally, the formation of claw-like channels in non-terraced sites as a result of clod erosion further contributed to this roughness. On the other hand, the terraced sites maintained greater stability with less roughness over time, due to controlled erosion processes.

At Isare, terraced sites had over 80% of open surfaces (OS%), compared to 70% in non-terraced sites. Vegetation cover (Veg%), litter (L%), and surface not integrated into the soil (SNI%) were present in 60% of terraced areas and 50% of non-terraced ones. Slow-forming terraces helped reduce runoff and erosion while increasing infiltration, which preserved more clods, galleries, and cracks. Additionally, the lower proportion of open surfaces in non-terraced sites is likely due to steeper slopes (

Table 3. Global average (GA%) of the surface conditions of terraced (PT) and non-terraced plots (n_PT) with corn cultivation in Isare.

	Global Average (GA%) of Surfaces Conditions
	OS%%	CS%	CoS%	BS%	Ri%
Surface conditions in PT at Isare	82.43	17.57	60.99	39.01	3.52
Surface conditions in nPT at Isare	72.25	27.75	58.18	41.62	9.12

).

3.1.2. Buhinyuza Commune

At Buhinyuza, monthly observations showed that some surface parameters improved, while others declined at both sites. Results are reported in

Table 4. Monthly evolution of the surface condition of plots covered and not covered with corn in Buhinyuza.

Monitoring	Surface Conditions in PT					Surface Conditions in n_PT
	OS%	CS%	CoS%	BS%	Ri%	OS%	CS%	CoS%	BS%	Ri%
30 to 31 December 2022	84.00	16.00	18.17	81.84	1.03	83.90	16.10	18.70	81.30	1.30
5 February 2023	67.33	32.67	62.33	37.67	15.09	73.33	26.67	59.67	40.33	14.91
6 March 2023	69.00	31.00	90.33	9.67	15.65	75.00	25.00	92.67	7.33	14.63
11 April 2023	71.33	28.67	98.00	2.00	15.84	76.33	23.67	97.00	3.00	15.84
17 May 2023	73.33	26.67	99.17	0.83	15.56	80.00	20.00	98.87	1.13	15.35

and

Table 5. Global average (GA%) of the surface conditions of terraced (PT) and non-terraced plots (n_PT) with corn cultivation in Buhinyuza.

Global Average in Surface Conditions at Buhinyuza
	OS%	CS%	CoS%	BS%	Ri%
Surface conditions in PT	73.00	27.00	73.60	26.40	12.63
Surface conditions in n_PT	77.71	22.29	73.38	26.62	12.41

and illustrated below. CoS% increased from 18.17% to 99.17% in terraced sites and from 18.70% to 98.17% in non-terraced sites, likely due to the growth of corn foliage, herbaceous plants, and litter.

The roughness index (Ri%) rose until the fourth period but slightly declined in the fifth period at both sites. In terraced sites, it increased from 1.03% to 15.84% and remained constant at 15.56%, while in non-terraced sites, it rose from 1.30% to 15.84% and stabilized around 15.34%.

Regarding OS% (clods and galleries: Cld), it decreased until the third period (March 2023), then increased again in the fourth (April 2023) and fifth (May 2023) periods at both sites. During the third period, we observed high Veg% and L% along with the presence of insects, such as ants and termites. These insects contributed to the formation of galleries after vegetation and litter helped reduce erosion.

In all plots installed in Buhinyuza, the average proportion of Ri% was higher in the terraced plots (12.63%) compared to the non-terraced plots (12.41%). This was similar to CoS%, which remained higher in the terraced plots (27%) compared to 22.29% in the non-terraced plots. The high proportions of SI% and Cr%, as well as the roughness in the terraced plots, can be explained by water leaks observed at the level of the upstream slow-forming terraces. These ditches, with water leaks, were located on sites with slopes greater than 40%, which likely favored erosion and caused damage.

This situation affected OS%, particularly the clods and galleries (Cld%), in the terraced sites (73%) compared to 77.71% in the non-terraced sites. However, despite this initial erosion, CoS%, along with its components (Veg%, L%, and SNI%), gradually increased at both sites during the monitoring period. The proportion of CoS% in the terraced sites (73.60%) slightly exceeded that in the non-terraced sites (73.38%). Thus, terracing likely facilitated the establishment of plant cover, including herbaceous plants in erosion channels and the foliage of corn plants. In contrast, the non-terraced sites maintained lower plant cover, likely due to the slope length, which promotes progressive erosion [33].

3.2. Effect of Slow-Forming Terraces on the Quantity of Sediments Inside Plots and in Ditches with Corn Plants

3.2.1. Isare Commune

Regarding the monthly evolution of the quantity of eroded sediments, field data (Figure 6) at Isare showed consistent differences between PT and n_PT. Erosion remained higher in n_PT (7.44 g/cm²) compared to terraced ones (3.42 g/cm²) throughout the monitoring period. Monthly erosion also declined, with sediment levels dropping from 4.30 g/cm² to 0.19 g/cm² in n_PT and from 1.97 g/cm² to 0.02 g/cm² in PT.

The monthly evolution of sediment retention in slow-forming terraces downstream of corn plots at Isare is detailed in Figure 7. Sediment retention was higher in the ditches of n_PT (37.11 g/cm²) compared to the anti-erosion ditches in terraced sites (8.79 g/cm²). Despite the Fanya Juu embankment, sediment deposits were still found in the anti-erosion ditch, likely due to water leaks and rain splash erosion, exacerbated by the lack of vegetation and litter on the embankments.

In n_PT, sediment deposits decreased monthly in both ditches, from 22 g/cm² to 3 g/cm² in non-terraced sites and from 5 g/cm² to 0.5 g/cm² in terraced sites.

3.2.2. Buhinyuza Commune

In Buhinyuza, sediment erosion in both PT and n_PT corn plots showed a steady decline throughout the monitoring period. The results regarding the effect of non-vegetated terraces on sediment quantity inside the plots and in the ditches with corn plants are shown in Figure 8. Sediment levels decreased from 5.91 g/cm² to 0 g/cm² in non-terraced sites and from 4.29 g/cm² to 0.35 g/cm² in terraced sites. The increase in plant cover and litter likely facilitated sedimentation. However, an unusual trend emerged: PT plots lost a higher quantity of sediment than n_PT plots starting in January. But that quantity was not significantly different between PT and n_PT. This was likely due to water overflowing the anti-erosion ditches, which eroded the plots and embankments. Poor maintenance of the ditches likely contributed to this issue. Despite these challenges, total sediment loss in terraced sites (7.37 g/cm²) was similar to that in non-terraced sites (8.11 g/cm²).

Regarding the comparison of the monthly evolution of the quantity of sediments retained in the ditches downstream of the plots with corn in Buhinyuza, results are shown in Figure 9. Sediment retention in ditches showed a continuous monthly decrease. In non-terraced sites, sediment levels dropped from 19.72 to 0.12 g/cm², and from 11.28 to 0.15 g/cm² in terraced sites. Similar to Isare, this reduction would be due to the increase in plant cover and litter.

However, non-terraced sites retained more sediment (26.54 g/cm²) compared to terraced sites (12.32 g/cm²).

3.3. Effect of Slow-Forming Terraces on Erosion Reduction

Regarding the effect of slow-forming terraces in the reduction of eroded sediments inside the plots, results are reported here. Data from Isare and Buhinyuza revealed that n_PT experienced higher sediment erosion than PT. In Isare, sediment levels in n_PT reached 7.44 g/cm², compared to 3.42 g/cm² in PT. Similarly, Buhinyuza’s n_PT had 8.11 g/cm² of sediment, while PT recorded 7.37 g/cm².

In Isare, terracing reduced erosion by over 54%. However, in Buhinyuza, erosion reduction was lower (10%), a circumstance that could be due to the absence of anti-erosion techniques.

For the effect of the non-vegetated embankments in the increase in retained sediments in the anti-erosion ditch downstream of the plots, results are reported below. In Isare, the sediment eroded within the terraced plot (3.42 g/cm²) was lower than that retained in the downstream anti-erosion ditch (8.79 g/cm²), leaving an excess of 5.37 g/cm², or over 61%. Similarly, Buhinyuza showed more than 67% of excess. This excess likely came from the freshly created, unprotected Fanya Juu embankments, which were eroded by rain splash and claw effects.

In Isare’s non-terraced plot, the downstream retention pit held significantly more sediment (37.11 g/cm²) than inside the plot (7.44 g/cm²), an excess of 30 g/cm², about 80%. In Buhinyuza’s non-terraced plot, the excess sediment in the pit was 18.43 g/cm² (8.11 versus 26.54 g/cm²), about 69%. The length of unterraced plots explains that excess of sediments observed.

Under typical conditions, sediment eroded within a plot should equal the amount collected in the downstream retention ditch for n_PT. However, for PT plots, the eroded sediment within the plot is expected to exceed the sediment measured in the downstream anti-erosion ditch, indicating the effectiveness of these ditches and their embankments.

In contrast, current results show that both PT and n_PT plots retain more sediment in the downstream ditches than what is eroded within the plot area.

This implies an excess of sediment accumulation in the ditches, which requires explanation. In PT plots at Isare and Buhinyuza, the excess is due to erosion of unprotected Fanya Juu embankments caused by rain splash and animal/claw activity.

For n_PT plots, the excess sediment is attributed to the extended plot length, which increases erosion potential.

This discrepancy suggests that external sources contribute to sediment yield beyond intra-plot erosion. The retention ditches are capturing sediment not only from within the plot but also from adjacent or upslope areas. These findings indicate that erosion control measures must also target embankments and nearby surfaces.

4. Discussion

4.1. Impacts of Slow-Forming Terraces on Surface Conditions

At the Isare commune, in the plots, OS% was over 72%, which aligns with findings by Nsabiyumva et al. [36] in a drier case study from Morocco, where water retention techniques, such as holes dug around fruit plants, were used. However, OS% values in n_PT were lower than those observed in Morocco. In Tunisia, on lands with terraces, OS% varied between 20% and 85%. This variation was attributed to the different presence of cracks (Crk%) and clods (Cld%).

Concerning the proportions of CoS%, they increased in both PT and n_PT plots at Isare and Buhinyuza sites, reaching more than 95%. However, this value was higher than the 80% found in forest, matorral, and plowed plots in Morocco [48]. The high value of CoS% was due to the growing corn foliage, herbaceous plants, and litter. The same circumstance occurred, e.g., in Italy, where plant cover increased over time [34]. By the final month, anti-erosion ditches appeared to help maintain plant cover, which aligns with Sabir et al. [33], who found that shorter slopes reduce erosion and support vegetation growth.

Despite the high values of OS% and CoS% in the studied plots at Isare, the SI% and impact crusts were higher in n_PT than in PT. The high SF% value in n_PT can be explained by soil erosion processes. These processes are confirmed by the increase in Ri% from 1% to 12%. Moreover, studies have shown that Ri% increases due to the transport of fine particles in runoff, leading to higher roughness [49]. The decrease in Ri% observed in PT suggests a reduction in porosity due to surface sealing, resulting from the breaking and disintegration of clods after wetting and the deposition of sediments in depressions [48].

These results suggest that slow-forming terracing effectively controls erosion, maintaining higher open surfaces and reducing roughness. In contrast, non-terraced sites, despite similar vegetation growth, experience more erosion, leading to increased closed surfaces, roughness, and the formation of impact crusts. These findings align with previous studies that identify slope length as a critical factor in erosion dynamics [33]. Additionally, it is known that anti-erosion ditches reduce runoff and erosion, preserving more clods, galleries, and cracks, which is consistent with Roose’s findings [32]. The lower open surface proportions in non-terraced sites are likely due to steeper slopes, which, as Sabir et al. noted, accelerate erosion and prevent vegetation growth [33].

At Buhinyuza, regarding the surface condition parameters studied, including BS% and Ri%, the opposite situation was observed, as they remained higher in PT compared to n_PT. This indicates that the PT plots experienced erosion, likely caused by water leaks in the upstream anti-erosion ditches. These ditches, with water leaks, were located in sites with slopes higher than 40%, which would have favored erosion and caused damage. Similar cases were described by Bougère et al. [50], who showed that the effectiveness of anti-erosion ditches depends on their monitoring and maintenance. Moreover, PRRPB [43] specified that this progressive terracing technique is more recommended for lands with slopes ranging from 6% to 25%. The high Ri% in PT suggests that erosion, possibly caused by overflowing silted ditches, led to increased roughness, as noted by Bougère et al. [50].

4.2. Impacts of Slow-Forming Terraces’ Erosion and Sedimentation

In the plots at Isare and Buhinyuza, there was a steady decrease in sediment erosion over the course of the monitoring period. However, in terraced plots (PT), soil loss was 3.42 g/cm² at Isare and 7.37 g/cm² at Buhinyuza, compared to 7.44 g/cm² in non-terraced plots (n_PT) at Isare and 8.11 g/cm² at Buhinyuza. These values were still higher than the sediment quantities found by Samake et al. [51] in Mali, where developed fields experienced soil losses of 0.0134 g/cm² and 0.0179 g/cm², and non-developed fields had losses of 0.0356 g/cm² and 0.0424 g/cm². Similarly, in Rwanda, the average seasonal soil losses in sites with slow-forming terraces ranged from 0.0009 g/cm² to 0.043 g/cm² at Murehe and from 0.0104 g/cm² to 0.0210 g/cm² at Tangata [29], which are lower than the six-month results obtained in Isare and Buhinyuza. Moreover, Rutebuka et al. [29] found soil losses in non-terraced sites in Rwanda ranged from 0.0022 g/cm² to 0.08 g/cm² at Murehe and from 0.179 g/cm² to 0.288 g/cm², which were still lower compared to the values found in the n_PT plots of our study. This discrepancy can be attributed to the different erosion factors between regions.

In Spain, soil losses greater than 0.4 g/cm²/year are considered severe enough to warrant intervention [52]. However, in Burundi, the values observed in Isare are consistent with those found by Rishirumuhirwa [26] at Mumirwa. Using the Wischmeier equation, he found sediment losses under crops ranged from 3.7 to 977.6 t/ha, which corresponds to 0.037 g/cm² to 9.776 g/cm². The sediment losses in the plots at Buhinyuza were much higher than the values reported by Rishirumuhirwa [26] for the dry eastern plains of Burundi, where soil losses ranged from 2.3 t/ha to 134.8 t/ha, corresponding to 0.023 g/cm² to 1.348 g/cm². In our study, however, sediment losses were greater than 7 g/cm².

The observed monthly decline in soil losses in Isare and Buhinyuza can be explained by Roose [32], who noted that anti-erosion ditches help reduce runoff and erosion. The gradual decrease in erosion at both sites can be attributed to the growing plant cover from corn, herbaceous plants, and litter, which enhance water infiltration and slow runoff [53]. Roose [32] demonstrated that vegetation can reduce erosion by up to 100%. Similarly, e.g., Apollonio et al. [34] found that perennial vegetation reduced erosion by up to 300 times compared to bare soil, and plant cover helps filter runoff and trap sediments [54].

However, an unusual trend emerged at Buhinyuza: ealthough the difference was not significant, PT plots lost more sediment than n_PT plots from January onward. This was likely due to water overflowing from the anti-erosion ditches, eroding the plots and embankments. Poor maintenance of the ditches contributed to this issue, as emphasized by Bougère et al. [50], who highlighted the importance of regular maintenance to prevent erosion. A similar situation was observed by Rutebuka et al. [29] in Rwanda, where sites with slow-forming terraces generated more sediment than non-terraced plots due to inadequate maintenance after two years.

Regarding the effect of non-vegetated terraces on sediment retention in ditches at Isare and Buhinyuza, sediment quantities continued to decrease monthly, similar to the trend inside the plots. However, the sediment quantity in the ditches of n_PT was higher than in those of PT. This monthly reduction in sediment is due to the increase in plant cover and litter, which helps reduce runoff and promote sedimentation inside the plots [54,55,56]. This is consistent with findings that increased plant cover, foliage, and litter help reduce erosion [32].

In non-terraced sites, the slope length likely contributed to the increased erosion in the ditches, which is why the quantity of sediment retained in the ditches of n_PT was higher, as noted by [32,33,57]. Erosion caused by slope length can promote channel and gully erosion [32].

In terraced plots, despite the Fanya Juu embankment, sediment deposits were still found in the anti-erosion ditches, likely due to water leaks and rain splash erosion, exacerbated by the lack of vegetation and litter on the embankments. Unprotected embankments can be eroded and contribute to increased soil loss in ditches. Vegetation helps reduce the energy of erosive agents and splash effects, protecting the soil [56]. Vegetated embankments also promote sedimentation [58].

4.3. Effect of the Non-Vegetated Embankments of PT in the Reduction and Increase in Sediments

4.3.1. Effect of the Non-Vegetated Embankments of PT in the Reduction of Eroded Sediments Inside the Plots

Concerning our results on the effect of non-vegetated embankments in reducing eroded sediments in the plots: at Isare, terracing reduced erosion by over 54%, which aligns with findings that shorter slopes reduce erosion [32,33]. The effectiveness of anti-erosion techniques in reducing erosion has been studied in other countries as well. For instance, Wickama et al. [17] demonstrated that annual cropping on steeper landscapes reduced soil erosion by 16.4% at Sunga and 13.3% at Soni, compared to unprotected sites, which had higher erosion rates than our study sites. Wickama et al. [17] also compared these sites with those employing soil and water conservation (SWC) technologies, where effectiveness ranged from 84.6% to 87.1% at Soni and Sunga sites in the Usambara Highlands from 2009 to 2011.

In Rwanda, Rutebuka et al. [29] studied the effectiveness of terracing over two years (2015–2016 and 2016–2017). It reached 93% reduction in soil loss control at Tangata, confirming its significant potential as an erosion control measure, even in mountainous areas. Similarly, in Portugal, Valle Junior et al. [59] found that terraces effectively reduced nearly 30% of the soil erosion risk. However, terraces located on steep slopes or in areas with soil losses exceeding 50 tons per hectare per year were five times more likely to slump.

Durán et al. [60] reported a 58–98% reduction in soil loss in areas covered with different types of vegetation compared to areas without vegetation during one year (1998 to 1999), while Chirino et al. [61] found a reduction of up to 70–95% over 4 years (1996 to 1999). In Paraná, IAPAR [62] reported a 50% reduction in soil losses through terracing, regardless of the cultivation system used during ten years (1972–1982).

Chow et al. [63] observed dramatic decreases in soil loss, from an average of 20 t/ha to less than one by combining terracing with grassed waterways and contour planting of potatoes between 1978 and 1991. Therefore, terraces play a crucial role in reducing soil losses, and in many countries, this reduction is greater than the values observed in our study sites.

However, at Buhinyuza, erosion reduction was lower (10%), due to the absence of permanent monitoring for ditch maintenance in PT, which caused leaks and hindered effectiveness. Bougère et al. [50] have emphasized that one of the most important activities for reducing erosion is the maintenance of terrace walls. Abandoning terraces can create a major risk of massive soil loss, as shown by Vogel [64] and Harden [65].

Some authors argue that terraces may retain too much water, leading to saturation and, consequently, storm runoff [66]. Poor management of terrace drainage, combined with steep-slope gradients and high amounts of runoff, are key factors in the lack of efficacy of terracing in combating erosion [67].

4.3.2. Effect of the Non-Vegetated Embankments of PT in the Increase in Retained Sediments in the Anti-Erosion Ditch Downstream of Plots

Regarding the effect of non-vegetated embankments on the increase in sediment retention in the anti-erosion ditch downstream of the plots at Isare and Buhinyuza: the excess sediment retention was more than 60% at Isare and 65% at Buhinyuza. Lasanta et al. [67], between 1956 and 1991 for 35 years, observed that erosion at the foot of the terrace slope could lead to the deterioration of the terrace as a whole, as well as gully formation, which eventually leads to increased erosion. According to these authors, this erosion is due to the steepness and the sparse vegetation cover, or its complete absence. In our study, unprotected Fanya Juu embankments were eroded by rain splash and clod effects. These embankments should be protected, as Cosandey et al. [57] highlighted between 1988 and 1992, showing how vegetation and litter prevent soil erosion by reducing rain impact, a finding supported in 2013 by Mostephaoui et al. [55] in their study by GIS. Dan et al. [35] also demonstrated that plant cover can reduce sediment loss by over 80%.

In contrast, in the ditch of non-terraced plots, the excess of sediments—80% at Isare and 70% at Buhinyuza—was higher by 20% compared to the terraced plots. This difference is likely due to the longer slope length in these areas, which exponentially increases erosion [32].

5. Conclusions

This study was conducted in two ecologically contrasting regions of Burundi—Isare, located in the Mumirwa highlands, and Buhinyuza, situated in the Eastern Depressions—to assess the influence of non-vegetated terraces on surface characteristics and sediment dynamics. A total of 12 experimental plots were established, with 132 erosion measurement pins installed (72 in non-terraced plots and 60 in terraced plots) to enable systematic monitoring of changes in surface cover, surface roughness, and sediment deposition throughout the rainy season. The study confirms the reliability of the pin-based method for quantifying the impact of non-vegetated terraces on erosion processes. Results indicate that these structures effectively reduced erosion across both study sites. However, the findings also highlight the critical importance of regular maintenance to ensure the long-term functionality of terraces, particularly on steep slopes. Although non-vegetated terraces significantly mitigated soil erosion, their protective efficacy over time depends strongly on consistent upkeep. While the study confirmed the beneficial role of non-vegetated anti-erosion ditches in modifying soil surface dynamics, it also emphasized maintenance as a key factor in preserving terrace integrity. In the absence of routine maintenance, terrace structures are prone to functional decline due to sediment accumulation within ditches and embankment degradation. In Isare, for instance, terraces reduced erosion by 54%, suggesting their potential applicability in other areas with comparable slope gradients and land management conditions. However, the relatively lower effectiveness observed in Eastern Burundi underscores the need for targeted maintenance interventions, especially on slopes exceeding 40%.

Future research should focus on evaluating the performance of vegetated embankments under similar environmental conditions using the same pin-based methodology. Such studies would help clarify the added benefits of vegetation in enhancing terrace stability and erosion control. Additionally, integrating cartographic and remote-sensing techniques could complement traditional pin-based measurements, enabling broader comparative analyses and improving the scalability and applicability of erosion-monitoring approaches.