Understanding spatial patterns of soils for sustainable agriculture in northern Ethiopia’s tropical mountains
Jan Nyssen aff001; Sander Tielens aff002; Tesfamichael Gebreyohannes aff003; Tigist Araya aff004; Kassa Teka aff004; Johan Van de Wauw aff002; Karen Degeyndt aff002; Katrien Descheemaeker aff005; Kassa Amare aff003; Mitiku Haile aff004; Amanuel Zenebe aff004; Neil Munro aff006; Kristine Walraevens aff007; Kindeya Gebrehiwot aff004; Jean Poesen aff002; Amaury Frankl aff001; Alemtsehay Tsegay aff009; Jozef Deckers aff002
Authors place of work:
Department of Geography, Ghent University, Gent, Belgium
aff001; Department of Earth and Environmental Sciences, KU Leuven, Leuven, Belgium
aff002; Department of Earth Sciences, Mekelle University, Mekelle, Ethiopia
aff003; Department of Land Resources Management and Environmental Protection, Mekelle University, Mekelle, Ethiopia
aff004; Department of Plant Production Systems, Wageningen University, Wageningen, The Netherlands
aff005; Institute of Climate and Society, Mekelle University, Mekelle, Ethiopia
aff006; Department of Geology, Ghent University, Gent, Belgium
aff007; Research Foundation—Flanders (FWO), Brussels, Belgium
aff008; Department of Dryland Crop and Horticultural Sciences, Mekelle University, Mekelle, Ethiopia
Published in the journal:
PLoS ONE 14(10)
Knowledge of the geographical distribution of soils is indispensable for policy and decision makers to achieve the goal of increasing agricultural production and reduce poverty, particularly in the Global South. A study was conducted to better understand the soilscapes of the Giba catchment (900–3300 m a.s.l.; 5133 km2) in northern Ethiopia, so as to sustain soil use and management. To characterise the chemical and physical properties of the different benchmark soils and to classify them in line with the World Reference Base of Soil Resources, 141 soil profile pits and 1381 soil augerings at representative sites were analysed. The dominant soil units identified are Leptosol and bare rock (19% coverage), Vertic Cambisol (14%), Regosol and Cambisol (10%), Skeletic/Leptic Cambisol and Regosol (9%), Rendzic Leptosol (7%), Calcaric/Calcic Vertisol (6%), Chromic Luvisol (6%) and Chromic/Pellic Vertisol (5%). Together these eight soil units cover almost 75% of the catchment. Topography and parent material are the major influencing factors that explain the soil distribution. Besides these two factors, land cover that is strongly impacted by human activities, may not be overlooked. Our soil suitability study shows that currently, after thousands of years of agricultural land use, a new dynamic equilibrium has come into existence in the soilscape, in which ca. 40% of the catchment is very suitable, and 25% is moderately suitable for agricultural production. In view of such large suitable areas, the Giba catchment has a good agricultural potential if soil erosion rates can be controlled, soil fertility (particularly nitrogen) increased, available water optimally used, and henceforth crop yields increased.
Good land management is characterised by making optimal use of the natural resources including soils in a sustainable way. In the Giba catchment (5133 km2), north Ethiopia, poverty has been largely attributed to insufficient crop production [1, 2]. Soil degradation in this area became important when humans started deforestation almost 5000 years ago [3, 4]. The resulting reduced soil protection by vegetation cover, combined with steep slopes and erosive rainfall led to excessive soil erosion [5, 6]. Nutrients and organic matter (OM) were lost and soil depth was reduced. Expanding the agricultural areas into less suitable lands to sustain crop production would further increase soil erosion rates. Yet, the high population density allows a more intensive use of the available agricultural land. In recent decades, many soil and water conservation measures (SWCM) have been carried out to reduce soil erosion rates and to increase crop production. Ex-situ SWCM include the construction of stone bunds, infiltration trenches, check dams in gullies, micro-dams and ponds as well as a range of biological measures (e.g. exclosures), while in-situ soil management measures are being promoted (e.g. intercropping, bed and furrows, zero tillage, zero grazing) [3, 7–11]. Despite these SWCM, soil erosion still is an important problem, which results in low crop yields and biomass production. In view of all this, the Tigray region, where the Giba catchment is located, has chronically suffered of food insufficiencies. To curb such situations, soil maps have proven to be powerful tools for understanding soil processes , for the establishment of technical infrastructure, and in support of land management policies [14, 15].
Hunting Technical Services  prepared landform and land suitability maps of an area largely encompassing Giba catchment at a scale of 1:250,000, and further maps of landforms and soils at 1:50,000 for areas around Mekelle, Hawzien, Sinkata and Wuqro. Soil mapping and land evaluation have been carried out in several parts of Giba basin by student teams of IAO Firenze (led by Luca Ongaro and Valeria Alessandro) [17–21]. Other available baseline soil information for the study area comprises mainly small-scale maps based on FAO  at 1:1,000,000; derived maps include the web-based e-SOTER soil information system  and the corresponding sheets in the Soil Atlas of Africa [24, 25]. The development of a national soil model at scale of 1:500,000 has been attempted , as well as soil nutrient mapping through the EthioSIS and AfroSIS programmes , resulting among others in detailed maps of soil fertility status and recommended fertiliser blends with a resolution of 250 m for the whole Tigray region . Whereas the latter have a deliberate focus on chemical fertilizer requirements , all other mentioned maps are very generalized, allowing a regional comprehension of the soil distribution, but not at all a full understanding of the spatial patterns of the soils in a given area.
Therefore, the main objective of this study is to contribute to sustainable land management in the Giba catchment through a better understanding of the soil types and their characteristics, which is a prerequisite for analysing soil suitability for sustainable agricultural production.
A good knowledge of the geographical distribution of the soils and their chemical and physical properties is thus indispensable for policy and decision makers to improve land management and hence reduce poverty and increase the welfare of the population in north Ethiopia.
Characterisation of benchmark soils of the catchment, both in the field and in the laboratory, was combined with all available information into a comprehensive spatially explicit database of soils in Giba catchment, at a scale of 1:250,000. This allows a fundamental insight into the soil properties and the soilscapes of the Giba catchment which is needed to enhance sustainable natural resource use and management.
The Giba catchment is in Tigray region (North Ethiopia), between 13°18’N and 14°15’N and 38°38’E and 39°48’E, and comprises the region’s capital city Mekelle (Fig 1). The Giba River is a tributary of the Tekezze River, which becomes Atbara River in Sudan where it flows into the Nile. The altitude in the catchment varies from slightly over 900 m a.s.l. in the western part to more than 3300 m a.s.l. in the north. The mean elevation of the catchment is 2144 m with a standard deviation of 361 m indicating that the topography is very rugged. Because of high elevations, the climate is more temperate than would be expected at this latitude .
The geology of the catchment consists of a Precambrian basement complex, Palaeozoic (fluvio-)glacial rocks, Mesozoic sedimentary rocks, Tertiary volcanics and Quaternary deposits  (Fig 2). The landscape is characterised by a strongly incised river network. Major faults are responsible for steep cliffs. The alternation of different lithologies resulted in a stepped geomorphology due to selective erosion [32, 33].
According to the Köppen climate classification, the area is hot semi-arid (BSh) . Annual rainfall depth varies between less than 600 mm and 1000 mm, but no significant relationship with altitude exists [30, 36]. Most rains fall during the main rainy season, which typically extends from June to September. Mean annual maximum air temperature ranges from 21 to 31°C and mean annual minimum temperature from 3 to 16°C . Monthly potential evapotranspiration (PET) exceeds monthly rainfall except during the rainy season due to reduced sunshine hours and increased rainfall. However, monthly rainfall is only slightly higher than PET in the northernmost part during the rainy season while elsewhere in the catchment rainfall greatly exceeds PET. Hence, the length of the growing period (LGP), defined as the period during which the precipitation is at least half of the PET , is shortest in the northernmost part.
Recent land use maps [34, 39, 40] show that 42–50% of the Giba catchment is covered by cropland, followed by shrubland (37%). Forests are rare (2.3%), however the eastern part of the catchment holds Des’a Forest, one of the few forests in north Ethiopia, on the edge of the Rift Valley escarpment.
Based on lithology, geological structure, geomorphology, elevation and climate, the Giba catchment can be subdivided into 6 major geomorphic regions: the Atsbi horst, the Abergelle lowlands, the basalt-dominated highlands, the cuesta landscape, the severely incised Antalo Supersequence plateau with dolerite, and the Sinkata midlands (Fig 3). According to FAO  and the Soil Atlas of Africa , the catchment would be dominated by Lithic and Eutric Leptosol, Vertic and Chromic Cambisol, and Haplic Lixisol.
Part I. Soil characteristics
Materials and methods
Field work comprised ten field campaigns in the Giba catchment (1974, 1975, 2001–2011), and the analysis of 141 profile pits (Table 1). For every field campaign, district authorities issued permits, and especially, the landholders gave permission for digging profile pits. Generally they dug out the pit themselves manually and were very keen to discuss the observations with the researchers. For pits on communal lands we obtained permission from the village chairperson, who assigned nearby residents for pit excavation. All labour was paid in cash, at a rate approximately 50% above salaries paid locally for similar works. None of the data collection involved endangered or protected animal or plant species. Disturbed and undisturbed samples were taken from the different horizons for further chemical and physical analysis. Undisturbed samples were taken with Kopecki rings, 100 cm3 steel cylinders, 5 cm across, driven in the soil using a ring holder. Here we focus on 20 representative profile pits from which 46 soil horizons were sampled.
The texture of all horizons was investigated in the field using the finger method . Further, formal physical analysis was done both on the disturbed and undisturbed samples. The analysis of the undisturbed samples was done in Mekelle University (Ethiopia), the analysis of the disturbed samples was done in KU Leuven (Belgium) soil laboratories.
Texture of the soil horizons was analysed by 2 different methods: with a wet sieving followed by a decantation and tentatively with a laser diffraction particle size analyser (LDPSA). Upper texture class boundaries were conventionally set at 2 μm (clay), 50 μm (silt) and 2 mm (sand).
In preparation for the analysis by LDPSA, the samples were dried, roots and small plant remnants removed, mortared and sieved at 2 mm. With the aid of a sample splitter a very small amount (<1 g) was separated into a test tube. Distilled water was added and this mixture was boiled for at least 15 minutes to bring all particles in suspension. This mixture was analysed with the LDPSA. During the analysis, ultrasonic sound or ultrasound bath was used to break the particles apart. The laser beam reflected on the particles and this reflection is dependent on the size of the particles. Large particles (sand) provide a reflection at an angle that is smaller than small particles (clay). Each detector detects a different particle size ranging from < 0.04 μm to 1822 μm. After the analysis, the mass percentage of each particle size was known .
Standard wet sieving and decantation was also conducted on eight samples to compare the results with the LDPSA method. About 20 g of the dried and sieved (at 2 mm) soil was weighted. 50 ml of peptiser (sodium oxalate dispersing agent) was added and the mixture was diluted with distilled water to about 150 ml. Then the sample was boiled for at least 10 minutes to destroy the aggregates. After cooling, the mixture was sieved at 50 μm and the filtrate was used for the decantation. The particles that stayed behind on the sieve were collected in a pre-weighed cup and dried in an oven of 105°C for 24 hours, after which the cup was weighed again. The filtrate was put in a decantation column and diluted to 1 litre. The column was shaken well for a few seconds. Immediately after that, 50 ml was tapped and added into a pre-weighed noggin. Three more times 50 ml was tapped at 4’30”, 13’30” and 2h30’, which corresponds to the fractions smaller than 32 μm, 16 μm and 2 μm. Fifteen seconds before each tapping the tube was cleaned by tapping till the next line on the column. The noggins were also put in an oven of 105°C for 24 hours and weighed afterwards.
In the LDPSA measurement, unexpectedly, we found for almost all the samples a silt loam texture . However, the eight samples that were also analysed with the conventional decantation method were clearly finer textured, as clay percentages measured with the decantation method were much higher than those obtained through LDPSA. To further evaluate the difference in preparation method between LDPSA and the conventional method, the same eight samples that were analysed with the conventional method were also analysed again with the LDPSA but with the same preparation as in the conventional method. This includes the addition of peptiser and distilled water followed by boiling of the sample. A supplementary analysis was done with an additional treatment of standard ultrasound before the samples were analysed but without the addition of peptiser. The clay percentages obtained through all different LDPSA methods (Table 2) mostly stay well below the results obtained by decantation, particularly in the clay-rich horizons.
The large difference between the measured texture with LDPSA and the decantation measurements shows that LDPSA underestimates the finer fractions (Table 2), as also observed in earlier studies [53–55]. This can somehow be corrected by adding peptiser before LDPSA but the measured clay content still remained far below the values measured by the decantation method for six of the eight samples. Hence, texture obtained through decantation and finger methods will further be used consistently in this study.
The field capacity (FC) was measured on the undisturbed samples (in Kopecki rings) with a pressure plate apparatus. After saturation with water, the samples were weighed (Msat) and immediately placed in the pressure plate apparatus. A pressure of -1/3 bar (pF 2.53) was applied. When no more water was expelled (after about 7 days) the samples were weighed again (MFC), put in an oven at 105°C for 24 hours, and again weighed (Mdry). Field capacity was calculated as :
FC = field capacity (%),
MFC = mass of the sample at pF 2.53 (kg),
Mdry = oven-dry mass (kg),
Vkopecki = volume of the Kopecki ring (l), and
accounting for a water density of 1 kg l-1.
The permanent wilting point (PWP) was measured with a pressure plate apparatus on disturbed samples. First a paste was made by adding water to about 40 g of soil. The samples were placed in the pressure plate apparatus and a pressure of -15 bar (pF 4.18) was established. Similar to FC measurement, the samples were weighed when no more water was expelled (MPWP), and weighed after drying in an oven (105°C) for 24 hours (Mdry). For each sample, the average of four analyses was taken. The PWP was calculated as :
PWP = permanent wilting point (%),
MPWP = mass of the sample at pF 4.18 (kg),
Mdry = oven-dry mass (kg), and
ρb = dry bulk density of the sample (kg / m3).
The total available water (TAW) was calculated as :
TAW = total available water (%),
FC = field capacity (%), obtained from (Eq 1), and
PWP = permanent wilting point (%), obtained from (Eq 2).
The porosity was calculated as:
Porosity = porosity of the sample (%),
Msat = mass of the saturated sample (kg),
Mdry = oven-dry mass of the sample (kg),
Vkopecki = volume of the Kopecki ring (l), and
accounting for a water density of 1 kg l-1.
The bulk density was calculated as:
ρb = bulk density of the sample (kg / m3),
Mdry = oven-dry mass of the sample (kg), and
Vkopecki = volume of the Kopecki ring (m3).
Before performing the chemical analysis, the disturbed samples were dried at 60°C, the roots were removed, the samples were crushed and sieved at 2 mm.
The pH-H20 and pH-KCl were measured after two hours of shaking in water and in a 1 M KCl 1:2.5 solution respectively, with a glass-calomel combination electrode.
The percentage CaCO3 was determined with the ‘rapid titration method’ by Piper  was performed. After adding 0.2 M HCL, the solution was titrated the next day with 0.1 M NaOH. The percentage CaCO3 was calculated based on the added NaOH. As other carbonates such as dolomite may also be dissolved by this method, the results are referred as ‘calcium carbonate equivalent’ .
The available phosphorus (Pav) was measured only on the uppermost horizons because P is not very soluble or very mobile. The total amount of available phosphorus was determined by using the Olsen-P method . In this method, the absorbance measured by a spectrophotometer at a wavelength of 720 nm is used to determine the amount P in solution (mg/l), and converted into the amount of P in the soil (mg kg-1). However, after centrifuging and filtering of the sample extracts, red-brownish colours were observed in most of the samples and no blue colouring occurred when the mixing reagent was added. This was probably due to the presence of organic matter in the sample extracts. Due to the lacking blue colours no meaningful measurements could be carried out. Three methods were used to try and remove the organic matter but all failed : (i) addition of activated coal (Norit) did not work because large amounts of phosphorus were present in the activated coal; (ii) substituting coal by the polymer polyacrylamide  failed to absorb the organic matter; and (iii) the lanthanum (La) precipitation method  led to flocculation not only of humic substances but also of phosphorus due to the high phosphate adsorption capacity of lanthanum . Because none of these methods succeeded in filtering the samples without adding or removing phosphorus we chose to measure the influence of the red-brownish colour in the extracting solution on absorbance. For this purpose, we contrasted the standard Olsen solution (standard solution; the extracting solution and the mixing reagent in a 1:1 ratio) with a water solution (the extracting solution and distilled water in a 1:1 ratio). In the water solution no colouring can occur (because no mixing reagent is added) and the spectrophotometer measured the influence of the present colour. In the standard solution the effect of the colouring is measured. By subtracting the absorbance of the water solution from that of the standard solution, the influence of the red-brownish colours was then offset. The influence of the absorbance of distilled water was also taken into account by subtracting this value from the absorbance of the water solution:
Absorbance = the final measured absorbance,
Abs* = measured absorbance of the standard solution (with mixing reagent),
Abs = measured absorbance of water solution (without mixing reagent), and
water = measured absorbance of a sample of distilled water.
The electrical conductivity was measured with a temperature-corrected conductivity meter in a 1:5 solution.
The cation exchange capacity (CEC) and exchangeable bases were measured with the ‘silver thiourea method’ . The following exchangeable bases were measured: Ca2+, Mg2+, Na+ and K+. As the measured pH differs not much from 7, it is assumed that the measured effective CEC (ECEC) is equal to CEC. Base saturation (BS) was calculated as:
BS = the base saturation (%),
exch. Ca2+ = exchangeable Ca2+ of the sample (cmolc/kg),
exch. Mg2+ = exchangeable Mg2+ (cmolc/kg),
exch. Na+ = exchangeable Na+ (cmolc/kg),
exch. K+ = exchangeable K+ (cmolc/kg), and
CEC = cation exchange capacity (cmolc/kg).
High values of exch. Ca2+ were found with this method because of partial dissolution of calcite . Due to these high values of exchangeable Ca2+, base saturations of more than 100% were found. To solve this problem the values of exchangeable Ca2+ were reduced up to reaching a base saturation of 100%.
The total organic carbon (TOC) and total nitrogen (TN) were determined by combusting pre-weighed samples in a Carlo Erba CHNS-O EA1108 elemental analyser. Before analysis, the carbonates were removed by adding HCL to the samples.
The interpretation of the measured chemical soil properties was done using Table 3.
Soil chemical properties
The pHH2O of most horizons (Table 4) is around 7 or slightly alkaline, only 5 soil profiles have a moderately to slightly acid pHH2O. In all horizons, the pHKCl is lower than the pHH2O except for horizon F2. The CaCO3 content varies between low and very high. The EC is very low for all analysed horizons.
Ca2+ is the dominant exchangeable base followed by Mg2+; both have high to very high values in most horizons (Table 4). The values for Na+ and K+ range between very low to high but most horizons have a medium value. ECEC values are high or very high for most horizons. The base saturation is very high in all horizons; most have a value close or equal to 100%.
The %C in most horizons is low to very low but some horizons have high values. The N content is low to very low for almost all the horizons. The measured available P is highly variable between the different horizons but most horizons have medium or high values.
Soil physical properties
The average field capacity is 27% (± 7%), with a range between 8 and 37%. Values per horizon are presented in the profile descriptions (see S1 File) and have been tabulated by Tielens . The average permanent wilting point is 21% (± 8%), with a range between 4 and 38%. The average total available water (TAW) is 7% (± 5%), with a range between 1 and 16%. In several horizons the calculated TAW was slighty negative, indicating that the values of PWP and FC were close to each other. In such cases, TAW was not further taken into account. The average prosity is 42% (± 7%), with a range between 30 and 54%. Like for the other soil physical parameters, these descriptive statistics concern all profiles and horizons. The average bulk density is 1.34 (± 0.24 g cm-3), with a range between 0.91 and 1.74 g cm-3.
Soil chemical properties
In neutral soils, the exchangeable base complex is dominated by Ca2+ and Mg2+, in alkaline soils Na+ and K+ are more present and in acid soils Al3+ and H+ are the most abundant . The profiles with a slightly to moderately acid pH (I, J, K, L and M) have indeed significantly lower values of exchangeable K+ and Na+ (0.32 cmolc/kg K+ and 0.16 cmolc/kg Na+) in contrast to 0.75 cmolc/kg K+ and 0.76 cmolc/kg Na+ in the other profiles. For Ca2+, also a significant difference exists between the two groups: 8.14 cmolc/kg for the acid soils compared to 22.18 cmolc/kg for the alkaline soils. For Mg2+, no significant difference was found between both groups.
When interpreting the soil chemical properties (Tables 2 and 3), the N content of all samples is low to very low (0.01–0.22%, with an outlier of 0.36 in Des’a forest), in line with limited inputs of N , high erosion rates and prolonged cultivation. N is the most limiting soil chemical property and increasing the N content is a must to increase crop yields. Even though farmers with livestock dispose of the organic form of N (manure), it was observed that this manure is stored and used in a poor way. In many cases the manure is left exposed to the weather so rain can leach all the valuable nutrients. In other cases, it is thrown away, particularly in the eastern part of the catchment, or it is dried and used as a fuel. The inorganic form (mineral fertiliser) was less popular  because of its high cost [64, 65], as well as unreliable rainfall.
Values of soil OC are also low (0.03–1.9%) except in Des’a forest where a mean value of 2.6% was found. Such low soil OC contents are the consequence of severe soil erosion, limited inputs (manure or crop residues) and overgrazing, which results in a low vegetation cover [3, 66]. In line with the OC content, the soil organic matter (SOM) plays an important role in the soil: it improves structure, water holding capacity, nutrient absorption and release to plants . The higher values in Des’a forest are the consequence of the nearly absent soil erosion, and the vegetation cover that leads to larger biomass inputs. Exclosures also trap upslope eroded sediments, which have very high organic carbon content, allowing a fast regeneration of soil productivity .
Soil physical properties
As may be expected in these landscapes that hold an extremely varied lithology, there are strong contrasts in soil texture. Sandy soils occur in profiles derived from sandstones, silty on the precambrian metamorphic rocks. Despite the fact that the inadequacy of the LDPSA laboratory analysis hampered the study, the decantation analysis showed that soils and horizons with clay contents beyond 40% are common (Table 2). For some part, they represent the pristine soil before major human interventions, either in the topsoil (profiles B, C, J and N) or as a buried horizon (profiles G and K). The mountainous nature of the topography led to frequent occurrence of colluvium, but pedogenesis on such colluvium is commonly leading to textural fining in the top horizons.
Additionally, mass movements have in many places transported materials from the basaltic uplands over the lower-lying sedimentary rocks, increasing the opportunity for clay soils to develop (see Part II).
Overall, the measured TAW (7.6% ± 4.9%) is lower than the expected 10 to 20% for silty clays and clay soils, or 15 to 25% for silt loam, loam and silty clay loam . Such low values for TAW could be the result of low values for FC or high values for the PWP. As PWP measurements, done under extremely low pressure, are subject to errors, our measurements were repeated four times and found to be consistent.
On the other hand, the relatively low FC values (average of 27%) are most probably related to the space occupied by the frequent small rock fragments in the soils (and in the undisturbed samples alike). For instance, a 40% rock fragment content in soils has been demonstrated to reduce the field capacity by 50% . Low TAW is, hence, explained by the stoniness of the soils.
Similarly, Descheemaeker et al.  analyzed TAW of top horizons of Cambisols, Calcisols and Phaeozems in exclosures and eucalyptus forest in the Giba catchment. In these soils with relatively high clay percentages (average of 37%) and relatively high SOM content (average of 3.7%) , the average TAW was 12% (± 2%) . Taking into account the higher OC content, capable of holding larger amounts of water , these values are still at the lower limit of what may be expected based on texture . Overall, the TAW for plant roots is strongly affected by the stoniness of many soils.
Part II. Soil profiles
Methodology: Soil auger and profile pit observations
In total 1381 soil auger and 141 profile pit observations were made in the catchment. The exact location of these augering sites and profile pits were determined by discussion among authors, and based on the reconnaissance studies and the interpretation of the digital data. These observations were made along soil catenas across the whole catchment.
The augerings were made with an Edelman auger. If augering depth was limited due to stoniness of the profile, multiple augerings were conducted and the deepest profile was described. For each augering the following properties were recorded: position; elevation; slope gradient; land use; depth; parent material; soil texture (assessed by finger test).
The profile pits were described in detail using the FAO guidelines for soil description . The following properties of the different soil horizons were characterised: depth; colour (Munsell Soil Colour Chart); texture; structure; stickiness; distribution of roots; reaction with HCl; surface stoniness. The profiles were classified according the World Reference Base for Soil Resources .
Local land users provided additional information–individuals who appear on profile pit photographs have given written informed consent to publish the photograph. Among the 141 soil profiles, 20 are presented in detail, well distributed over the different geomorphic regions of the Giba catchment (Fig 3).
Soils on basement and Palaeozoic (fluvio-)glacial deposits
Precambrian lithology and Palaeozoic sedimentary rocks
The basement rocks of the catchment belong to the Arabian-Nubian shield [74, 75], which was formed as a consequence of the collision of East and West Gondwana causing low-grade metamorphism of the rocks . They cover about 27% of the catchment and 6 different units exist: granitic intrusions, metalimestone, metasediments, metaconglomerate, metagreywacke and metavolcanic rocks.
The Precambrian metalimestone is blackish (Fig 4) or light grey to white and has quartz veins. It can be found in the Negash syncline where it has undergone strong folding and in the Abergelle lowlands .
The metasediments are phyllites and slates, both the result of low graded metamorphism of shale. Both are very fine grained and can be found in the same areas as the metalimestone. The slates are reddish or greyish, very cleavable and quartz veins are common. The reddish colour is due to the presence of hematite (Fe2O3). The phyllites are more metamorphosed than the slates, which explains the shiny surface, but in contrast to the slates they are not cleavable.
Metaconglomerates are only found in the northern part of the catchment, around Negash and on the Atsbi horst. The metagreywacke is mainly limited to the Atsbi horst. It is coarse grained and probably derived from pyroclastic materials ejected during back-arc volcanism [74, 77]. The sedimentary structure, which distinguishes them from the metavolcanic rocks, is explained by the transportation and reworking by running water .
As metavolcanics, both acidic and basic volcanic rocks occur in the western part, near the outlet, in the north around Negash and on the Atsbi horst and show a fine to medium grained texture .
Coarse-grained granitic intrusions are the youngest Precambrian formation in the Giba catchment . Its spatial coverage is limited, the most extensive exposures are found west of Negash (almost 50 km2), around Abiy Addi and at the outlet of the catchment. These intrusions are a possible source of the granite boulders in the Edaga Arbi tillites.
Palaeozoic (fluvio-) glacial deposits are the oldest sedimentary rocks in the catchment. They are unconformably overlying the first planation surface formed on the basement rocks . Two different units can be found: the Edaga Arbi tillites and the Enticho sandstone. Generally, the tillites overlay the Enticho sandstone, but the two are often interfingering .
The Edaga Arbi tillites consist of poorly sorted, unstratified and poorly consolidated fine-grained sediments (silt- to claystones) with colours varying from red, purple to dark grey and black [31, 79]. At some locations varved proglacial deposits can be found. Another evidence of a glacial environment is given by the presence of dropstones of various sizes [79–81]. Glacial landforms like roches moutonnées also occur, with the presence of striations, grooves and chattermarks in the underlying rocks, indicating direct ice contact [79, 81, 82]. The area covered by tillites is small (<1%), and is located near Abiy Addi, Wuqro and Idaga Hamus.
The Enticho sandstone is a white, medium- to coarse-grained sandstone and is characterised by cross beddings . Deposited as glacial outwash , it unconformably overlies the basement rocks. Precipitation of iron at the contact between layers made it very resistant to erosion and is the reason why plateaus of Enticho sandstone stand out in the landscape. The Enticho sandstone covers a large area (8%), particularly in the northern part of the Giba catchment.
Soils in the Abergelle lowlands
The Abergelle lowlands are in the western part of the catchment (Fig 3). This area is dominated by Precambrian rocks including metalimestone, metasediments (phyllite and slate), metavolcanic rocks and granitic intrusions . In the east, the geomorphic region is confined by a steep Adigrat Sandstone cliff (Fig 5, left). Typical for the Precambrian rocks are the occurrence of many small rounded hills that are mainly aligned in a NE-SW direction. The vegetation cover is limited due to lower precipitation and higher mean annual temperatures, and provides little protection against erosion by water which resulted in shallow soils. The dry climatic conditions and shallow soils make this area not suitable for cultivation. Most of the land is bare land or rangeland. Deeper, mostly cultivated, soils are observed in the valley bottoms, often corresponding to areas with metalimestone as parent material.
Towards the outlet, the landscape is strongly incised by the Giba River which is reflected in a very rugged terrain with steep slopes, shallow soils and very limited cultivation. At the foot and toeslopes of the steep Adigrat Sandstone cliff, Palaeozoic tillites outcrop and a 0.01 to 10 m thick layer of colluvium has been deposited.
Profile A: Calcaric Rendzic Leptosol. Profile A is situated on the shoulder of a small metalimestone hill (Fig 4) in Taget, Abergelle. Despite the rather steep slope, the shallow soil and the very high topsoil stoniness, this area is used for (marginal) cultivation. The parent material is strongly weathered metalimestone. Chemically this soil is rather rich and characterised by very high CaCO3 values (19%). The contents of Ca2+ (17.15 cmolc/kg) is very high and Mg2+ (3.21 cmolc/kg) high (Table 3). The available P (15.19 ppm) is high and the C (1.9%) and N (0.22%) contents are medium. The texture is silt loam, but many small rock fragments are present, which reduce the water holding capacity. This soil profile, like all others, is described in detail in the S1 File, which contains also all analytical data.
Profile B: Epileptic Proto-vertic Cambisol. This soil profile is at the toeslope of the same metalimestone hill as profile A. The soil depth is limited to 45 cm. The rock fragment content is less and almost no rocks occur in the B horizon. The parent material is the same strongly weathered metalimestone. All the soils in the area are used for cultivation. The CaCO3 content is medium (2.9%) and the pH (7.2) is slightly alkaline. The higher clay percentages result in very high effective cation exchange capacity (ECEC) values (35.5 cmolc/kg) which makes this soil rather fertile. The available P (8.57 ppm) and total N (0.07%) contents are however low and very low. The clay has swell-shrink properties as evidenced by small but not completely developed slickensides. Soil that is located near one of the numerous termite mounds gives better crop yields according to the farmers and the mounds are therefore not destroyed .
Profile C: Pellic Vertisol. Profile C is situated about 0.5 km east of profile B in the valley bottom where thick black clays cover metalimestone. Small limestone fragments can be found throughout the profile. The whole area is intensively used for cultivation (sorghum). The very high ECEC value (ca. 38 cmolc/kg) makes this soil chemically very rich. The organic carbon content (1.4%) is medium but the available P (9.58 ppm) and N (0.09%) contents are low and very low respectively. The clay percentage in the B horizon (61%, in contrast to 46% in the A horizon) is very high and clearly developed slickensides are visible. The swell-shrink properties of the clays result also in an angular blocky structure. At the surface, black metalimestone fragments (up to 1 m across) can be found. Five-centimetre wide and up to one-metre deep cracks develop in the topsoil during the dry season. The surface horizon meets the requirements of the qualifier ‘Grumic’ (strong fine granular structure) but it was chosen not to use this qualifier because the structure might have been caused by recent ploughing.
Profile D: Colluvic Calcic Luvisol. This profile is situated in a gently sloping area about 2.5 km south of the Adigrat Sandstone cliff (indicated by an arrow on Fig 5). The parent material is Adigrat Sandstone colluvium. Besides sandstones, also basalt fragments are found at the surface which confirms the colluvial origin. Maize is cultivated on this location. The colluvial origin can explain the loamy nature of the A horizon. A clear clay jump occurs in the B horizon, which meets the requirements of an ‘argic’ horizon. In this horizon many small CaCO3 concretions were observed. Chemically this soil, and particularly the B horizon, is dominated by a high CaCO3 content (12.1%) and Ca2+ (20.3 cmolc/kg) values; it was classified as calcic. The ECEC (24 cmolc/kg) is also high but the exchange complex is mainly dominated by Ca2+. The organic C (0.71%) and total N (0.07%) contents are low to very low, as reported earlier on for the wider region . The available P (14.61 ppm) is medium.
Profile E: Chromic Vertisol. This profile is situated less than 1 km downhill (south) of profile D (Fig 5). In contrast to profile D, no sandstones but only basalt fragments were found. The area is used for irrigated agriculture; cotton and red pepper are intercropped. Chemically the soil has very high ECEC values (31.24 cmolc/kg). The very high values for Ca2+ (23.4 cmolc/kg) and Mg2+ (6.4 cmolc/kg) are typical for soils derived from basalt. The clay percentages are also high (33% to 41%) and clear slickensides can be observed in the B horizon which indicates the swell-shrink properties of the clay. Like the other soils, the levels of organic C (0.90%) and N (0.09%) are low to very low.
Profile E was classified as a Vertisol. Other Vertisols were also found in the area (profile C) but the location of this Vertisol is remarkable: it is found a few km downslope from the Adigrat Sandstone cliff whereas the whole surrounding area is covered by reddish, sandy colluvial material of this cliff. In stead, the Vertisol developed on materials derived from basalt; as a mafic rock, it is rich in Ca and Mg and therefore an ideal parent material for smectites . The nearest potential source of basalt are the Hagere Selam highlands: 5 to 6 km north of profile E and are almost 1 km higher (on top of the escarpment depicted in Fig 5). Most probably, a debris flow transported these sediments to the lowlands, the path of which can be identified on aerial photographs . Debris flows in the Hagere Selam highlands  are capable of transporting debris over much larger distances than local landslides. The fact that a Vertisol developed on the deposited basalt-derived sediments indicates that the debris flow occurred at least several thousand years ago [87, 88]. Humid periods favourable to pedogenesis existed in the study area roughly between 10 000 and 5000 yr BP and between 2500 and 1500 yr BP . It is therefore likely that the Vertisol formed during one of those two periods.
Profile F: Lithic Leptosol. Profile F is situated about 1 km uphill (north) of profile D, closer to the Adigrat Sandstone cliff (Fig 5). The parent material is strongly weathered Adigrat Sandstone. Because of the very limited soil depth this area is used as rangeland but the vegetation cover is sparse. The sandy characteristics result in a chemically very poor soil with ECEC values around 10 cmolc/kg. The amounts of organic C (0.12%) and N (0.01%) are very low.
Profile G: Haplic Planosol. This profile is in a large gully, close to the Adigrat Sandstone cliff south of Abiy Addi (Fig 6). The area is, like profile F, used as rangeland and dominated by small trees and shrubs. The A horizon is sandy, but at a depth of 40 cm a very abrupt textural change occurs, with 42% clay in the B horizon. Chemically this soil is very poor. The lower, clayey layer (below a depth of 40 cm) has a higher ECEC (15.9 cmolc/kg) but organic C (0.17%) and N (0.02%) content is very low. We classified this soil as a Planosol, given the abrupt textural change in the soil profile , between the coarser uppermost layer and the underlying one.
No mineralogical analysis was conducted in this research, but the high pH (around 7) and high base saturation (BS > 71%) indicate that the actual conditions are not suitable for ferrolysis, in line with suggestions elsewhere in Ethiopia . Given the location of these Planosols, the geogenetic process  is the most likely cause: they are situated very close (within 1 km) of the Adigrat Sandstone cliff. The coarse surface layer is a colluvial deposit from the cliff, similar to the surface layer of profile D. The finer textured layer underneath is a clayey layer of the Edaga Arbi tillites that outcrop under the sandstone cliff. Hunting T.S.  presents a description of a profile (PE/9) that is very similar to the above-described Planosol, in a similar geomorphic setting.
Soil types observed in Aqushala in the Abergelle lowlands  comprise (1) in the metamorphosed black limestone, Endoleptic Calcisol at the upper slope; Endoleptic Cambisol and Vertic Leptosol at the middle slope, Hypercalcic Calcisol at the footslope and Grumic Vertisol in the valley bottom; (2) in the metasediments, Leptosol at the upper and footslope, Regosol at the mid slope position and Fluvisol in the valley bottom; and (3) in metamorphosed banded marl, Leptic Calcisol at the upper slope, Haplic Calcisol at the foot slope, and Fluvisol in the valley bottom. The soil-landscape model was successfully tested in the Taget area (where profiles A, B and C are located).
Soils on the Precambrian and Palaeozoic rocks of the Atsbi horst
The Atsbi horst is in the north-eastern part of the catchment (Fig 3). At the west, it is demarcated by the Negash geosynclinal fold and a major normal fault . Both the fault and fold lines are running north-south . In the south, the horst is bordered by the younger Wuqro fault belt. In the eastern, northern and locally in the central part Enticho sandstone outcrops occur. Due to differential erosion, the Enticho sandstone now stands out in the landscape and forms mesas or smaller buttes [16, 49]. However, most of the area is covered by Precambrian rocks: metaconglomerate, metagreywacke, and dominant metavolcanic rocks. Almost all the land except for the steep slopes is under cropland (Fig 7).
Profile H: Leptic Luvisol. This profile is situated on the Atsbi Horst on the level top of a ridge. The parent material is strongly weathered metasediment. Despite of the level topography, soil depth is limited to 60 cm. Chemically, this soil has high ECEC values and in the B horizon even very high values (38 cmolc/kg) were measured. The values of Ca2+ (22.0 cmolc/kg) and Mg2+ (10.7 cmolc/kg) are high to very high and the pH (7.3) is slightly alkaline. In the B horizon, a clay jump occurs and this horizon is classified as ‘argic’. The organic C (0.6%) and total N (0.06%) contents are low to very low. The top horizon has the highest measured value of available P (112.7 ppm) of all surface horizons. Despite the loamy nature of the C horizon, this horizon has a very high ECEC value (37.4 cmolc/kg).
Profile I: Leptic Cambisol. Profile I is situated at the footslope of the same metasediment hill as profile H (Fig 7). Despite its 10% slope gradient, this area is used for cultivation. The surface stoniness is very high and besides metasediments and metavolcanic rocks, quartz fragments are abundant. Chemically, this soil differs greatly from profile H. The amounts of exchangeable cations and the ECEC values (10.4 cmolc/kg) are much smaller. Over the entire profile, this soil has one of the lowest measured base saturation (52.7%). The CaCO3 content (1.0%) is low and the pH is slightly acidic (6.8). Both the organic C (0.35%) and total N (0.03%) are very low, and decrease with depth.
Profile J: Haplic Cambisol. Profile J is 1.5 km west of profiles H and I, almost in the valley bottom, covering metavolcanic rock. The slope is very gentle and the soil is much deeper than the previous two profiles. Similar to profile I, quartz fragments occur at the surface but are less abundant. At the surface 2–3 cm of overwash was observed. Small nutty structures were observed in the B horizon, but the clay percentages are too low to consider it as a nitic horizon. Consequently, this soil is chemically not as rich as profile H in ECEC and Ca2+. The pH is slightly more acid (6.5) than profile I. Like most soils in Giba catchment, this soil has very low to low values of C (0.5%) and N (0.05%). The available P (24.9 ppm) content, however, is very high.
Lithic Leptosol, Leptic Cambisol and Leptic Regosol were also observed on and near rock outcrops in the Ruba Feleg and Kuret sub-catchments whereas in areas where thin colluvium of Enticho Sandstone covers the Precambrian, associations of Haplic Cambisol, Haplic Regosol and Skeletic Regosol were observed .
Soil profiles in the Sinkata midlands
The Sinkata midlands are in the north-western part of the catchment. They start north of the Wuqro fault and extend northwards to the basalt-dominated highlands of Mugulat. In the East, the midlands are bounded by the Atsbi horst. The midlands can be subdivided into three units with their own geological and geomorphic characteristics.
The first unit, in the northern part, is covered by Enticho sandstone, with occurrences of Edaga Arbi glacials in the lower positions. The area is a very gently undulating plain and most of the area is used as agricultural land (Fig 8, left). A relatively high base saturation was observed in these soils despite a high quartz content which has been attributed to the calcitic cement in the sandstone . Leaching of the bases on the higher sites and accumulation in the depressions together with selective downslope transport of finer particles resulted in coarse-textured and more acidic soils on the higher sites and finer textured and base saturated soils in the depressions . At some locations the soil is very shallow where the Enticho sandstone outcrops with its protective ferruginous sandstone cap. Frequent occurrence of rounded pebbles does mostly not indicate fluvic properties, as such pebbles occur as lenses in the Enticho sandstone, and throughout the tillites.
The second unit can be found south of the previous unit and is covered by Precambrian rocks and some smaller Enticho sandstone outcrops. The metavolcanic rocks cover the largest area, followed by metaconglomerate. In the centre, a large granite batholith is present  which forms a series of concentrically oriented ridges with narrow intervening valleys . These ridges are composed of very rocky tors and are thus not suitable for cultivation; in the wider valleys the soil depth allows cropping.
The last unit is the Negash synclinorium  and consists of metasediments and metalimestone. The landscape is very steep with locally alluvial terraces with deeper soils which are used for cultivation.
Profile L: Arenic Lixisol. Profile L is situated near the town of Sinkata on a small plateau. Weathered Enticho sandstone is the parent material. Just like most of the northern part of the Sinkata midlands this area is used for cultivation. Chemically this soil is the poorest of all described profiles. The ECEC of the B horizon is low (5.47 cmolc/kg). Due to the presence of a clay jump, the B horizon was qualified as an ‘argic’ horizon. Organic C (0.4%) and total N (0.04%) are again low to very low due to the agriculture practices, but the available P content (38.25 ppm) is very high. In view of the presence of an argic horizon, and base saturation of more than 50%, this soil is classified as a Lixisol. In the entire profile small, reddish iron nodules/coatings were observed.
Profile M: Haplic Fluvisol. This profile is situated almost in the valley bottom of the Sinkata midlands, on grazed fallow land, but the adjacent lands were cultivated. Fluvic material was found beneath the shallow A horizon. Two different horizons, separated by a gravel layer (sandstone and metavolcanics) at 40 cm depth, can be distinguished in the fluvic material. Chemically the A horizon differs from the underlying horizons. The ECEC values in the fluvic horizons (21 cmolc/kg) are almost double of the uppermost (12 cmolc/kg) horizon. The percentages of organic C (0.4%) and total N (0.04%) are very low in the entire profile. Like profile L, the entire profile is slightly acid (pH of 6.5). At the bottom of the profile strongly weathered sandy material with purple colours was present.
Profile N: Mazic Sodic Vertisol. This profile is situated a few metres from profile M but in the thalweg (Fig 8). In contrast to profile M this area is used as permanent grassland. A few metres further, ground water was observed in a small depression but no groundwater nor gleyic colour patterns were observed in the profile. Due to the hardness of the soil, the profile was only excavated to a depth of 80 cm. Chemically and physically this soil differs strongly from profile M. The texture is more clayey and becomes very hard upon drying. In the B horizon, not fully developed slickensides could be observed, indicating vertic properties. At the surface, a gilgai microrelief was observed although it was not very distinct.
ECEC values (30.1 cmolc/kg) are high to very high as can be expected with swell-shrink clays. However, the most prominent characteristic are the very high values of exchangeable Na+. In the B horizon, this value is even extremely high (more than 14 cmolc/kg). 42% of the exchange complex is occupied by Na.
In the Tsinkaniet catchment of the western Midlands, Tesfu Woldegerima (48) found Endoeutric Cambisol and Haplic Regosol on 10–15% slopes, Haplic Leptosol on 5% slopes, Cutanic Luvisol and Mazic Vertisol on 1% slopes, and Arenic Fluvisol along the rivers. The soil-landscape model was applied in another similar area (Sendeda Guims), with a rate of accuracy of 30%. Thin-section analysis showed the presence of soils with an argic horizon in the plains and on plateaux which were mainly classified as Lamelli-Arenic Luvisol . Hunting TS  presents profile descriptions for Chromic, Eutric, and Vertic Cambisols as well as for a Chromic Vertisol and a Cambic Arenosol in the wider Hawzien area (just outside and northwest of the Giba catchment, but part of the Sinkata Midlands).
Soils on Mesozoic sedimentary rock
The Mesozoic rocks cover more than half the catchment (53%) and are the result of a transgression-regression cycle. During the transgression and regression two sandstone formations were deposited: the Lower (Adigrat) and Upper (Amba Aradam) Sandstone formations. During the transgression period, the wider area was below sea level and the Antalo Formation was deposited, which consists of limestone, shale, marl and minor intercalations of gypsum layers .
The Adigrat Sandstone has a maximum thickness of nearly 700 m around Abiy Addi and it is further exposed around Wuqro, Hayki Meshal, north of Idaga Hamus and in some deeply incised gorges of the Giba R. and its tributaries [31, 93].
The Antalo Formation can be found south of the Wuqro fault belt, all the way to the southern part, and in the western part it stops around Hagere Selam. The thickest depositions in the east are estimated at 1100 m; the formation pinches away towards the west .
The upper sandstone formation (Amba Aradam) is a near-shore deposition with cross beddings. The sandstone is reddish coloured which indicates the presence of oxidised iron (hematite and magnetite). It is fine grained and overlain by flood basalts which created a ‘baked’ contact in the upper part of the formation [94, 95], leading to induration and low permeability. The maximum thickness of the Amba Aradam Sandstone is around 50 m; it covers <1% of the catchment, particularly in the Hagere Selam highlands and on the Amba Aradam mountain in the southern part of the catchment .
The cuesta landscape west of Wuqro stretches till Hawzien (outside the Giba catchment). The cuestas, with their back slopes dipping towards the south-southeast, occur both in the Adigrat and the Antalo Formations. The Suluh River cuts consequently through the cuesta fronts. Overall, the slopes are rather steep in this landscape and agriculture is limited to the flatter areas where alluvio-colluvial deposits occur , on which Hunting TS  described a Cambic Arenosol. The steep slopes, especially on the cuestas, are transformed to exclosure or used as grazing land (Fig 9). Due to these steep slopes, the soils are very shallow and stony. The soils in the colluvial deposits are deeper but still rather shallow with generally a depth around 30 to 40 cm.
Soils of the incised Antalo Supersequence plateau
This unit covers the largest area in the catchment; to the north it is confined by the Wuqro fault, in the west it is limited by the Hagere Selam highlands and the steep Amba Aradam Sandstone cliff. The whole area is severely incised due its rapid uplift which started around 25 million years ago and amounted roughly to 2000 metres . The dominant lithologies are limestone, shale and marl (part of the Antalo Supersequence) with many dolerite sills and dykes. When exposed, dolerite forms steep cliffs or flat mountain tops. The dykes are mainly found in the major fault areas, like the Mekelle and Chelekot fault. Because of the high resistance to erosion, these dykes mostly form cliffs when exposed to the surface, Dykes are often associated with tufa dams, when transversal to a river channel. Locally, Adigrat Sandstone outcrops in deeply incised gorges of the larger rivers.
Three main half-grabens are confined by the 3 major faults, from north to south: the Wuqro, the Mekelle and the Chelekot basins (Fig 2). At many places along these fault belts, dolerite has been injected. A more dense and lush vegetation covers the fault escarpments that are lined by dolerite as compared to the limestone cliffs.
In the Wuqro basin, limestone and shales each cover half of the area, the terrain morphology is characterised by entrenched river valleys bordered by steep cliffs and separated by undulating to rolling interfluves . On steeper slopes, soil depth is limited and limestone is often outcropping, for instance in Des’a forest (Fig 10). These areas are less suitable for agriculture and mainly used for grazing or transformed to exclosures. On gentle slopes, the soils are rather deep and vertic properties are common especially in the valley bottoms which are often deeply incised. These soils are fertile and almost all of them are under cropland.
The dominant lithology in the Mekelle basin is Agula shale, and dolerite dykes and sills are a common feature. A clear distinction can be made between the western and the eastern part. In the west, the terrain is more hilly and incised. Here, the soils are similar to those of the Wuqro basin. Soil depth is limited on the steeper slopes and not favourable for agriculture; on the gentler slopes soil depth increases and vertic properties are common. The alluvial plains of the larger rivers hold well-developed Fluvisols. The Mekelle basin in the east is rather flat to gently undulating. Vertisol is very common in this semi-graben and provides excellent agricultural lands. The hills that stand out in the landscape consist of dolerite, which is more resistant to erosion compared to the shales.
The Chelekot basin is more incised and more rugged, especially in the area south of the Hagere Selam highlands. South of the Giba River the presence of dolerite resulted in the formation of steep cliffs which demarcate a gentle undulating plateau.
Profile P: Vertic Calcaric Phaeozem. This profile is in Des’a forest (dominated by Juniperus procera and Olea europaea ssp africana ) in an open spot in the forest. The parent material is Antalo Limestone. Profile depth is rather limited with bedrock at 70 cm. The A and B horizons have a very good structure and roots are abundant. The A horizon has an organic C content of almost 4% and was classified as a ‘mollic’ horizon. ECEC values (35.7 cmolc/kg) are very high in the entire profile and Ca2+ (33.4 cmolc/kg) is dominant on the exchange complex, in line with the high percentages of CaCO3 (10.1%). This soil has some vertic properties (seasonal cracks).
Profiles Q and R: Mollic Calcaric Cambisols. These two profiles are also located in Des’a forest but in more densely vegetated areas (Fig 10). Profile depth is limited to 75 cm with Antalo Limestone as parent material. ECEC values (18.5 cmolc/kg) are lower than in profile P but still high. Organic C content (2.4–3.7%) is medium to high in the A horizons which were also classified as mollic horizons. In the topsoil, rock fragments up to 5 cm are common. Very high CaCO3 values (18.7%) are found in the entire profiles.
Profile S: Rendzic Leptosol. This profile in Des’a forest is situated in a densely vegetated area. Profile depth is limited (25 cm) due to the outcrop of Antalo Limestone. Like all soils in Des’a forest, this one has also high C values but the highest values are found in the lower layer, below 5 cm (4.1%) which is also darker coloured. The CaCO3 content is very high in the topsoil (19.3%) and decreases strongly in the lower layer (4.5%). Both observations tend to indicate that the upper 5 cm may be considered as overwash. ECEC values (34.8 cmolc/kg) are again very high and Ca2+ (30.4 cmolc/kg) is dominant on the exchange complex.
Profile T: Mollic Calcaric Cambisol. The above described forest soils can be considered as the baseline from which most currently occurring soils in the Antalo Supersequence plateau have developed under longstanding human activity, through either truncation of the topsoil or burial by colluvium. Such is for instance the case of profile T along the road between Agula’e and Birki, very close to Birki. The profile is situated on an old river terrace of the Agula’e River at the convex border to the lower lying current terrace. The area is used for cultivation. Profile depth is very limited; at 40 cm depth Antalo Limestone is found.
Chemically this soil is rich with a high ECEC (29.5 cmolc/kg), mainly dominated by Ca2+, but the high to very high values of exchangeable K+ are remarkable (3.69 cmolc/kg in the top horizon). As expected by the parent material, the percentages of CaCO3 (19.1%) are very high and the pH (7.7) is slightly alkaline. Organic C (1.7%) values are medium, total N (0.18%) values are low and available P (42.8 ppm) is very high. The 25-centimetre thick A horizon has enough organic C (1.9%) to be classified as a mollic horizon.
Similar degraded soils have been described near Mekelle : a Lithic Cambisol and a “Vertic Lithosol”, as well as a Pellic Vertisol. Similarly, two Calcaric Regosol profiles were described in the lower part of the May Zegzeg catchment, which is at the western margin of the Antalo Supersequence plateau, as well as a Phaeozem under forest . In the same area, besides the forest Phaeozems, Calcisol, Calcaric Regosol and Calcaric Cambisol profiles were described on degraded steep slopes .
Soils on Cenozoic volcanics
Basalt and dolerite
During the Cenozoic, northern Ethiopia was exposed to very intense magmatic and tectonic activity. It is also in this period that the Ethiopian rift valley formed which caused the uplift of the northern Ethiopian highlands . Two kinds of volcanic depositions can be found in the catchment: flood basalts and dolerite dykes and sills.
The flood basalts were extruded during the Oligocene, and different series of eruptions led to a trap (stepped) landscape . During periods of lesser activity sediment was deposited in lakes that formed in the basalt landscape. Such geological layers are white coloured and consist mainly of very fine grained lacustrine sedimentary rocks which have been silicified [100, 101].
Dolerite is a mafic intrusive rock which has comparable chemical properties as basalt. However, as it did not reach the surface it had more time to crystallise and the crystals are thus bigger compared to basalt. Typical for dolerite is the rounded weathering. Because it is an intrusive rock it is present in the form of sills and dykes. The sills are mainly found in the Antalo Supersequence and may reach a thickness of 80 to 130 m . The dykes are mainly found in the major fault areas, like the Mekelle and Chelekot faults.
The basalt-dominated highlands comprise the southern edge of the catchment, the northern Mugulat Mountains (Fig 2) and the central-western part near Hagere Selam. There, the selective erosion of basalt flows led to a trap landscape covering the underlying sedimentary rocks. In the north, the basalt covers the Adigrat Sandstone while around Hagere Selam the Amba Aradam Sandstone is underlying the flood basalts. In both areas, the baked contact of the basalt and the sandstone has (i) increased the resistance against erosion which resulted in a steep cliff, and (ii) induced the local occurrence of perched water tables .
Due to the presence of many nutrients in basalt, the soils in these areas are chemically very rich and suitable for cultivation. However, the weathering of basalt may lead to high clay contents, which makes it physically hard to cultivate these soils but if managed properly they are excellent agricultural land. In the Hagere Selam highlands almost all the land is used for cultivation. Especially in the basalt areas, even the steep slopes (up to 30%) are used for cultivation .
Hunting TS  present a profile description of a Pellic Vertisol some kilometres west of Hagere Selam, most probably in the upper part of the May Zegzeg subcatchment where a detailed soil study was carried out later on . These Vertisols are part of a “red-black soil catena”: from Leptosol over Skeletic Regosol, Cumuli(skeletic) Regosol, Vertic Cambisol to Vertisol [43, 47]. Remnant forest patches also here typically have conserved Phaeozems .
Landslide and debris flow deposits are a common feature in this area due to the presence of swelling clays (smectites) derived from basalt, the presence of the lacustrine marl-clayish deposits, the presence of steep slopes and the less permeable baked contact [86, 104]. Basaltic material has been displaced downhill over the sandstone cliff and locally covers sandstone, limestone and marls. A consequence of the occurrence of such ancient landslide and debris flow deposits is the transfer of fertile material to the poorer soils on the sandstone and limestone which provides better conditions for growing crops. In the May Leiba sub-catchment, Van de Wauw  noticed that areas covered with basaltic debris were more cultivated than the adjacent fields on limestone and marl of the Antalo Supersequence, and they describe soil profiles in the landslide material that originated from the basalt highlands and covering adjacent limestone: a Vertic Cambisol, a Skeletic Cambisol and a Haplic Vertisol. Therefore, landslides have a significant impact on geomorphology and the spatial pattern of soils in this the landscape and a correct mapping of them is important when making a detailed soil map .
Soils in valley bottoms with Quaternary deposits
Two types of recent Quaternary deposits can be found in the catchment: alluvial sediments and carbonate precipitates. The area covered by these deposits comprises only 1% of the catchment, mainly in the floodplains of the Giba River and its tributaries. These deposits may range from well-sorted to poorly-sorted mixtures of clay, silt, sand and pebbles  (Fig 11).
Carbonate precipitates include tufa which is rather rare and small in extent, however tufa dams may have a significant impact on landscape evolution. Waterfalls at knickpoints in the longitudinal river channel profile created favourable conditions for CaCO3 precipitation due to degassing . Tufa deposition needs more humid conditions than the current climate and is therefore an indication of a wetter period in the past; the dams are generally dated early Holocene [106–108]. At some locations, these tufa deposits could grow up to 10 m high and more, resulting in wide dams, which blocked the course of rivers leading to the formation of small lakes. These tufa dams and their backfill deposits are mainly found on the Mesozoic sedimentary rocks in the central part of the Giba catchment.
Profile K: Haplic Fluvisol. In this profile in a river bank on the Atsbi horst, two layers were distinguished in the fluvic material. The upper layer is of lighter colour and in the lower layer four distinct gravel deposits occur. The area is used as grazing land for the nearby village. The upper layer (C1 horizon) has a lower pH, a lower ECEC value, a lower CaCO3 content and a lower organic carbon percentage compared to the underlying stony horizon C2. The darker colour of the lower layer reflects a period of more stability during which organic material could accumulate. The different gravel layers indicate periods of larger stream power that allowed these gravels to be transported.
Profile O: Mollic Calcaric Fluvisol. This profile is in the Agula’e river bank between Agula’e and Birki (Fig 11). The farmland of this profile was not under cultivation, but further away from the river bank almost the entire area is used for cropping. At the bottom of these at least 4 m thick alluvial deposits, large boulders are very abundant, but the deposits become finer towards the top. The uppermost 130 cm was described and three different horizons were distinguished. Chemically and physically the three horizons are quite alike, with high ECEC values (29.9 cmolc/kg); CaCO3 values (15%) are high to very high and the pH (7.9) indicates slight to moderate alkalinity. The percentages of organic C (1.3%) and BS (100%) are high enough to classify the upper (C1) horizon as a mollic horizon.
Part III. Soil geography and soil use
Materials and methods
Using the description of 141 profile pits, and 1381 soil augerings (Table 1), the soil geography was analysed and mapped. A field-based approach was used in which representative sub-catchments were first mapped in detail (using expert-based delineation of soil polygons ), and the obtained recurring land systems and soil groups extrapolated to the larger corresponding geomorphic region. Land systems, conceptualised by CSIRO , are areas with specific and unique geomorphic and geological characteristics, and which can be characterised by a particular soil distribution as specified by the soil catena. In line also with the “pédopaysages” approach [111, 112], all available soil information was combined into a comprehensive map at 1:250,000. Given the complex geology and topography of the catchment, this method was preferred over digital or predictive soil mapping [113, 114]). The following digital data were used: the Aster DEM of the catchment, the geological map , Landsat images (February 2003), SPOT images (January 2005) and aerial photographs (January 1994). Earlier baseline soil information for the study area was consulted, mainly small-scale maps based on FAO  at 1:1,000,000; derived maps include the e-SOTER map  and the corresponding sheets in the Soil Atlas of Africa [24, 25] (Fig 12). All these documents follow a different concept from ours, i.e. soil types are overtly generalized and mapped as exclusive polygons.
Soil suitability assessment
A soil suitability assessment for agricultural field crops (wheat, barley, teff, lentil, field peas, horse beans and sorghum) was then carried out, in which soil limitations were derived from the soil units following the Soil Fertility Capability System of Sanchez, Palm , and the suitability of each soil type was interpreted by the limitation approach [116, 117]. After qualitative assessment, the soil types were grouped into three soil suitability classes (very suitable, moderately suitable and not suitable soils for rainfed cultivation of annual crops), following the principles of the scale-independent FAO Framework for Land Evaluation .
S1: Very suitable soils are soils which do not restrict the expected yield much. The yield is thus not negatively affected by their soil properties (both chemically and physically), nor by a steep relief. These are soils that are deep enough, have a good natural fertility and can store sufficient amounts of water.
S2: Moderately suitable soils are soils that restrict the expected yield considerably due to their chemical and/or physical properties or by being located in steep terrain. These can be shallow soils, soils with a limited natural fertility or soils which are not capable of holding and releasing adequate amounts of soil moisture.
N: Not (or marginally) suitable soils are soils that greatly restrict and reduce the expected yields. In our catchment, these are very shallow and stony soils with limited soil depth (generally on steep slopes) or soils with impeded drainage.
Land systems and soil map
Based on soil profile descriptions, augerings, and available soil studies, typical soil units were defined (Table 5), with their main characteristics and classification. The studied region shows a large variability, as can be expected in a mountainous, lithologically contrasted region that has been subject to millennia of land degradation. Within the major geomorphic regions (Fig 3), land systems were defined considering the regional soil-landscape relationships . Soil unit 1 (Leptosol and bare rock) is by far the most dominant soil unit in this very rugged and strongly incised catchment (18.7% coverage). In total, the shallow soils (soil units 1 to 9) cover 39.7% of the area. Soil unit 10 (Vertic Cambisol) is the second most dominant soil unit (13.9% coverage). Together with soil units 11 and 12 (both Vertisol), they cover 24.9% of the catchment. Another dominant soil unit is unit 21 (Eutric Regosol and Cambisol) with a coverage of 9.8%. These are young soils which are mainly found at footslopes.
Soil suitability for agricultural field crops
The different soil units were classified according to their suitability for agricultural field crops [117, 118] (Table 6). Very suitable soils, i.e. soils where the crop yield is not limited by their chemical (fertility) and physical (depth, water holding capacity) soil properties, include, in the study area, soils with vertic properties (Vertisol, Vertic Cambisol), Phaeozem, Luvisol and well-drained Fluvisol. These cover 40.2% of the whole catchment. Moderately suitable soils, i.e. soils that hamper the expected crop yield considerably such as shallow soils, soils with limited natural fertility or other soils which are not capable of holding/releasing large amounts of soil moisture, in the Giba catchment include Leptic Phaeozem and Leptic Luvisol, Rendzic Leptosol, Regosol, Cambisol and shallow soils with vertic properties. They cover 25.1% of the catchment. The soils that are not suitable are very shallow and stony soils (e.g. Leptosol, Skeletic Cambisol), soils with a hard layer which prevents root penetration (Petric Plinthosol) or soils with impeded drainage (e.g. strongly expressed gleyic properties). These cover 34.7% of the entire catchment.
Soil distribution and its controlling factors
The soil catenas of each land system  indicate that topography (relief) and geology (parent material) are the most importing controlling factors that determine the spatial distribution of the different soil units (Fig 13, Table 5, S2 File). Besides these two major factors, vegetation (or land cover) may not be overlooked.
Leptosol and bare rock are found on the steepest slopes (>40%) (soil units 1, 2 and 3). On slopes between 20 and 40% soil depth is still limited but besides Leptosol also shallow and stony Cambisol, Regosol and Phaeozem (if a denser vegetation cover is present) might be found (soil units 4, 5, 6).
On the lower slopes/foot slopes (10–20%), with overal more vegetation cover, more developed but still young soils like Cambisol or Regosol (soil units 4, 5, 21, 22) can be found. The parent material will determine the specific soil unit: a limestone parent material will result in a Calcic or Calcaric qualifier.
On the gentler slopes (2–10%) in the lower situated areas, deeper and the most developed soils can be found. Parent material strongly determines the soil type. Soils with vertic properties (soil units 10, 11, 12, 14, 15, 17, 18) are found on limestones, shales or mafic material. The closer to the valley bottom, the better the vertic properties are developed and the deeper the soil becomes. More reddish soils, Luvisols (soil units 20, 26), can also be found in these areas on more convex areas which results in the typical ‘red-black’ soil catena  (Fig 14). If sandstone or Precambrian rocks are the parent material, mainly Cambisol, Regosol and Luvisol/Lixisol are found (soil units 19, 20, 21, 24, 25, 26, 28).
On plateaus, soil depth is also often limited and parent material determines strongly the soil types. On limestone, Rendzic Leptosol (soil unit 3) is very common while on Enticho sandstone Petric Plinthosol (soil unit 9) or shallow soils like Leptosol and shallow, stony Cambisol/Regosol (soil units 1, 4) are found.
In the valley bottoms, fine-textured soils occur with alluvial, stagnic or vertic properties like Fluvisol, Gleysol and Vertisol (soil units 12, 13, 29, 30, 31, 32, 33).
Areas with sufficient vegetation cover have deeper soils than those without vegetation. If the cover is barely touched by humans, the original soils are at the surface. These are moderately deep to deep Phaeozem on the plateaus and slopes and Vertisol in the lower areas. Such areas have become very scarce, but they can still be found in Des’a forest and the wetter depressions that it holds (“dambos”, sensu ; Fig 15), as well as in old church forests, where the soils and the vegetation have been protected since a long time .
Soil suitability for cropping
Despite its strong relief, 40.2% of the Giba catchment is classified as very suitable for agriculture and another 25.1% as moderately suitable. On the other hand, long-standing cropping and soil erosion in this mountainous catchment [3, 4] have led to the presence of large unsuitable areas (26% of the area is covered by bare rock and Leptosols). Furthermore, relatively less erodible clay and sand dominate the soil texture; the high rock fragment contents of topsoils after prolonged tillage and erosion also provide a partial protection against soil erosion [122–124]. Such positive feedback effects have led to a new dynamic equilibrium of the soilscape, not only in relation to tectonic uplift but also in relation to longstanding human impact [125, 126]. In total, 65% would be suitable for crop production. This value is larger than the 42–50% classified as agricultural land in earlier studies [34, 39, 40]. One may however not conclude that the 65% of suitable land indicates that some space is left to be used for cultivation, as some part of these suitable lands are occupied by villages, forest and regenerating semi-natural vegetation. Furthermore, a soil unit might be suitable for agriculture at this moment but it may not be sustainable in the future. For instance, soil units 3, 6 and 7 (7.5% coverage) are moderately suitable for agriculture but these are shallow soils. If used as agricultural land, water and tillage erosion will reduce soil depths even more which may convert them into unsuitable soils in the mid-term. However, the high percentage of suitable soils indicates that the Giba catchment, despite the long-standing soil degradation, still has good agricultural potential.
Optimal land use for increased crop productivity
‘Optimal’ land use should minimise soil erosion rates to values less than a tolerable soil loss, i.e. “the maximum level annual amount of soil, which can be removed before the long-term natural soil productivity is adversely affected” . Because soil depth is rather limited, except in the flatter areas towards and in the valley bottoms, this tolerable soil loss should not be larger than the soil formation rates, which, for the semi-arid midlands in northern Ethiopia, were modelled at 6 Mg ha-1 year-1 , but less at higher (due to cold) and lower elevations (due to drought). Measured soil loss rates, at plot, but also at catchment level are well beyond that value [10, 129, 130]. The most efficient way to drastically reduce these erosion rates would be to convert all the cropland that exceeds a critical slope gradient to exclosures [44, 131]. A critical slope gradient of 10% would mean that more than 60% of the whole catchment should be converted into exclosures. This might be a sustainable solution from a long-term view but it is not possible under the current agricultural productivity conditions. At this moment crop yields are even too small for the local population to be self-sufficient–for instance, 66.2% of the population in Ethiopia depended on agriculture for its livelihood in 2018 . Reducing the cropland area cannot be done without increasing specific crop yields.
The use of a wide set of SWCM in the Giba catchment  has been proven to drastically reduce the erosion rates [7–9, 134, 135], improve soil quality as well as environment [136, 137] and increase crop yields if they are implemented correctly [103, 138–140]. The participatory approach that is generally implemented in the Tigray Region of Ethiopia combines scientific knowledge and the local knowledge of the farmers, which is likely the most successful approach and is strongly recommended [6, 141].
Deficit or supplementary irrigation, tailor-cut to soil type, is another way to increase crop yields . With the use of crop growth models, like Aquacrop [142, 143], it is possible to develop guidelines for deficit and supplementary irrigation to increase the crop yields. Besides specific information about the crops (e.g. canopy development and transpiration), the climate and the used management practices, Aquacrop requires specific information about the soils in order to calculate the soil water balance .
Finally, Integrated Soil Fertility Management (ISFM), i.e. “the application of soil fertility management practices, and the knowledge to adapt these to local conditions, which maximise fertiliser and organic resource use efficiency and crop productivity”  would allow to increase the agronomic efficiency, the ratio between the increase in crop yield and the applied nutrients. In the Giba catchment, most of the soils are lacking nitrogen. Incorporating legumes in the rotation system is a simple way to increase the N content , which is commonly done by the farmers in the study area . Furthermore, legume-cereal intercropping is especially beneficial in areas with low-input/high-risk environments, such as moisture and nutrient stress in the study area . Despite slow take-off , mineral fertilizer has become more popular in recent years. A bottleneck that still needs to be tackled is the insufficient and inappropriate use of manure [29, 147].
The in-depth study of the Giba catchment soils involved soil profile and augering descriptions, soil type characterisation and comprehension of soil-landscape relations. On the steepest slopes, shallow soils (e.g. Leptosol) and bare rock are found; on the footslopes, more developed but younger soils occur (e.g. Cambisol and Regosol); on the more gentle slopes, the most developed and deeper soils occur but the parent material strongly determines the soil type (e.g. Vertisol, Luvisol, Cambisol); in the valley bottoms, more fine-textured soils with alluvial, stagnic or vertic properties are present; on the plateau, soil depth is often more limited (e.g. Leptosol, Plinthosol).
The geographical distribution of the soil types was determined using land systems, i.e. areas with specific and unique geomorphic and geological characteristics. In this study 41 different and unique land systems were demarcated (Fig 13), each characterised by a particular soil distribution in line with the soil catena. Thirty-four different soil units were distinguished and characterised (Table 1). The most dominant soil units are: unit 1 (Leptosol and bare rock, 19% coverage), unit 10 (Vertic Cambisol, 14% coverage), soil unit 21 (Regosol and Cambisol, 10% coverage), unit 4 (Skeletic/Leptic Cambisol and Regosol, 9% coverage), unit 3 (Rendzic Leptosol, 7% coverage), unit 11 (Calcaric/Calcic Vertisol, 6% coverage), unit 20 (Chromic Luvisol, 6% coverage) and soil unit 12 (Chromic/Pellic Vertisol, 5% coverage). Together these eight soil units cover almost 75% of the catchment.
Topography and parent material are the most important driving factors explaining the soil distribution while vegetation (or land cover) has a less important role, as most parts of the Giba catchment are deforested since many centuries. Without human-induced erosion, the soil distribution would be much more homogenous and topography and parent material would be of lesser importance in controlling the soil distribution. In the new dynamic equilibrium of the soilscape, after major human impacts, younger soils dominate.
Determining the optimal land use for the catchment based on the soil map, strongly depends on the degree to which sustainability is taken into account. Our results show that approximately 65% of the catchment is suitable for agricultural purposes at this moment but not all of these soils can sustain agriculture in the long term. Erosion rates should be reduced to ensure that soil depth will not decrease any further. Yet, the high percentage of suitable soils clearly shows that the Giba catchment has certainly high agricultural potential if correct land management decisions are made.
For most soils, besides water, nitrogen is the most limiting factor for crop growth. Increasing the nitrogen content by e.g. integrated soil fertility management in combination with the ongoing ex-situ and recommended in-situ soil and water conservation measures and irrigated farming will most likely yield the best results.
S2 File [zip]
Digital map of soil units (.KML file, to be opened in Google Earth).
1. van der Veen A, Tagel Gebrehiwot. Effect of Policy Interventions on Food Security in Tigray, Northern Ethiopia. Ecology and Society. 2011;16(1):18.
2. Pender J, Berhanu Gebremedhin. Determinants of agricultural and land management practices and impacts on crop production and household income in the highlands of Tigray, Ethiopia. Journal of African Economies. 2007;17(3):395–450.
3. Nyssen J, Poesen J, Moeyersons J, Deckers J, Mitiku Haile, Lang A. Human impact on the environment in the Ethiopian and Eritrean highlands—a state of the art. Earth-Science Reviews. 2004;64(3–4):273–320. doi: 10.1016/S0012-8252(03)00078-3
4. Blond N, Jacob-Rousseau N, Callot Y. Terrasses alluviales et terrasses agricoles. Première approche des comblements sédimentaires et de leurs aménagements agricoles depuis 5000 av. n. è. à Wakarida (Éthiopie). Géomorphologie: Relief, Processus, Environnement. 2018;24(3):277–300.
5. Teketay Demel. Deforestation, wood famine, and environmental degradation in Ethiopia's highland ecosystems: urgent need for action. Northeast African Studies. 2001;8:53–76.
6. Nyssen J, Frankl A, Amanuel Zenebe, Deckers J, Poesen J. Land management in the northern Ethiopian highlands: local and global perspectives; past, present and future. Land Degradation & Development. 2015;26(7):759–64.
7. Nyssen J, Veyret-Picot M, Poesen J, Moeyersons J, Mitiku Haile, Deckers J, et al. The effectiveness of loose rock check dams for gully control in Tigray, Northern Ethiopia. Soil Use and Management. 2004;20:55–64. doi: 10.1111/j.1475-2743.2004.tb00337.x
8. Nyssen J, Poesen J, Desta Gebremichael, Vancampenhout K, D'Aes M, Gebremedhin Yihdego, et al. Interdisciplinary on-site evaluation of stone bunds to control soil erosion on cropland in Northern Ethiopia. Soil and Tillage Research. 2007;94(1):151–63.
9. Tewodros Gebreegziabher, Nyssen J, Govaerts B, Fekadu Getnet, Mintesinot Behailu, Mitiku Haile, et al. Contour furrows for in situ soil and water conservation, Tigray, Northern Ethiopia. Soil and Tillage Research. 2009;103(2):257–64. doi: 10.1016/j.still.2008.05.021
10. Tesfay Araya, Cornelis WM, Nyssen J, Govaerts B, Tewodros Gebregziabher, Tigist Oicha, et al. Effects of conservation agriculture on runoff, soil loss and crop yield under rain fed conditions in Tigray, Northern Ethiopia. Soil & Tillage Research. 2011;27:404–14.
11. Alemtsehay Tsegay, Vanuytrecht E, Berhanu Abrha, Deckers J, Kindeya Gebrehiwot, Raes D. Sowing and irrigation strategies for improving rainfed tef (Eragrostis tef (Zucc.) Trotter) production in the water scarce Tigray region, Ethiopia. Agricultural Water Management. 2015;150:81–91.
12. Ngatunga EL, Cools N, Dondeyne S, Deckers JA, Merckx R. Buffering capacity of cashew soils in South Eastern Tanzania. Soil Use and Management. 2001;17(3):155–62. doi: 10.1111/j.1475-2743.2001.tb00022.x
13. Jarvis MG, Hedges MR. Use of Soil Maps to Predict the Incidence of Corrosion and the Need for Iron Mains Renewal. Water and Environment Journal. 1994;8(1):68–75. doi: 10.1111/j.1747-6593.1994.tb01094.x
14. Vitharana UW, Van Meirvenne M, Simpson D, Cockx L, De Baerdemaeker J. Key soil and topographic properties to delineate potential management classes for precision agriculture in the European loess area. Geoderma. 2008;143(1–2):206–15.
15. Nzeyimana I, Hartemink AE, Geissen V. GIS-based multi-criteria analysis for Arabica coffee expansion in Rwanda. PloS one. 2014;9(10):e107449. doi: 10.1371/journal.pone.0107449 25299459
16. Hunting Technical Services. Tigrai Rural Development Study, Annex 1, Land and vegetation resources. Hemel Hempstead (G.B.): Hunting Technical Services Ltd; 1976. 419 p, 15 maps. p.
17. Rabia AH, Figueredo H, Huong T, Lopez B, Hishe S, Alessandro V. Land suitability analysis for policy making assistance: a GIS based land suitability comparison between surface and drip irrigation systems. International Journal of Environmental Science and Development. 2013;4(1):1–6.
18. IAO. Land evaluation in Kilte Awulaelo—Tigray Region, Ethiopia. Firenze, Italy: Ministry of Foreign Affairs, Istituto Agronomico per l'Oltremare; 2009. 231 p.
19. IAO. Land evaluation in Enderta District—Tigray Region, Ethiopia. Firenze, Italy: Ministry of Foreign Affairs, Istituto Agronomico per l'Oltremare; 2008. 229 p.
20. IAO. Land evaluation in the May Gabat watershed Enderta-Hintalo Wejirat Districts (Northern Ethiopia). Firenze, Italy: Ministry of Foreign Affairs, Istituto Agronomico per l'Oltremare; 2014. 229 p.
21. Yeshi Hadgu. GIS-based land suitability evaluation for irrigation in the Semha watershed (Enderta Woreda, Tigray, Ethiopia). MSc dissertation. Firenze: Universià degli studi Firenze, Scuola di Agraria; 2016. 138 p.
22. FAO. Ethiopia—geomorphology & soils map and legend—assistance to land-use planning: FAO Ethiopia; 1983.
23. Batjes NH. ISRIC-WISE derived soil properties on a 5 by 5 arc-minutes global grid (ver. 1.2). Report 2012/01. Wageningen, The Netherlands: ISRIC—World Soil Information; 2012.
24. ESDAC. Soil Atlas of Africa and its associated Soil Map (data): European Soil Data Centre (ESDAC), European Commission, Joint Research Centre; 2014.
25. Dewitte O, Jones A, Spaargaren O, Breuning-Madsen H, Brossard M, Dampha A, et al. Harmonisation of the soil map of Africa at the continental scale. Geoderma. 2013;211–212:138–53. doi: 10.1016/j.geoderma.2013.07.007
26. Brunner M. A national soil model of Ethiopia. A geostatistical approach to create a national soil map of Ethiopia on the basis of an SRTM 90 DEM and SOTWIS soil data. (MSc thesis). Bern, Switzerland: University of Bern; 2012. 126 p.
27. Hengl T, Leenaars JG, Shepherd KD, Walsh MG, Heuvelink GB, Tekalign Mamo, et al. Soil nutrient maps of Sub-Saharan Africa: assessment of soil nutrient content at 250 m spatial resolution using machine learning. Nutr Cycl Agroecosyst. 2017;109(1):77–102.
28. MoA ATA. Soil Fertility Status and Fertilizer Recommendation Atlas for Tigray Regional State, Ethiopia. Addis Ababa: Ministry of Agriculture, Agricultural Transformation Agency; 2014. 91 p.
29. Nyssen J, Birhanu Biruk, Zbelo Tesfamariam, Frankl A, Biadgilgn Demissie, Tesfaalem Gebreyohannes, et al. Geographical determinants of inorganic fertiliser sales and of resale prices in north Ethiopia. Agriculture, Ecosystems & Environment. 2017;249:256–68.
30. Virgo KJ, Munro RN. Soil and erosion features of the Central Plateau region of Tigrai, Ethiopia. Geoderma. 1978;20:131–57.
31. Tesfamichael Gebreyohannes, De Smedt F, Miruts Hagos, Solomon Gebresilassie, Kassa Amare, Kurkura Kabeto, et al. Large-scale geological mapping of the Geba basin, northern Ethiopia. Mekelle, Ethiopia: VLIR-Mekelle University IUC Programme; 2010.
32. Nyssen J, Poesen J, Moeyersons J, Deckers J, Mitiku Haile. Processes and rates of rock fragment displacement on cliffs and scree slopes in an amba landscape, Ethiopia. Geomorphology. 2006;81(3–4):265–75.
33. Coltorti M, Dramis F, Ollier C. Planation surfaces in northern Ethiopia. Geomorphology. 2007;89(3–4):287–96.
34. Tesfamichael Gebreyohannes, De Smedt F, Walraevens K, Solomon Gebresilassie, Abdelwasie Hussien, Miruts Hagos, et al. Application of a spatially distributed water balance model for assessing surface water and groundwater resources in the Geba basin, Tigray, Ethiopia. Journal of Hydrology. 2013;499:110–23.
35. Peel MC, Finlayson BL, McMahon TA. Updated world map of the Köppen-Geiger climate classification. Hydrology and Earth System Sciences. 2007;11(5):1633–44. PubMed PMID: ISI:000251516100009.
36. Jacob M, Frankl A, Mitiku Haile, Zwertvaegher A, Nyssen J. Assessing spatio-temporal rainfall variability in a tropical mountain area (Ethiopia) using NOAAs Rainfall Estimates. International Journal of Remote Sensing. 2013;34(23):8305–21.
37. Araya Alemie, Keesstra S, Stroosnijder L. A new agro-climatic classification for crop suitability zoning in northern semi-arid Ethiopia. Agricultural and Forest Meteorology. 2010;150(7–8):1057–64.
38. FAO. Report on the agro-ecological zones project. Vol 1: Results for Africa. Roma, Italy: FAO; 1978. 158 p.
39. Amanuel Zenebe. Assessment of spatial and temporal variability of river discharge, sediment yield and sediment-fixed nutrient export in Geba river catchment, northern Ethiopia. Leuven: Department of Earth and Environmental Sciences, K.U.Leuven; 2009. 346 p.
40. Etefa Guyassa, Frankl A, Lanckriet S, Biadgilgn Demissie, Gebreyohannes Zenebe, Amanuel Zenebe, et al. Changes in land use/cover mapped over 80 years in the Highlands of Northern Ethiopia. Journal of Geographical Sciences. 2018;28(10):1538–63.
41. Tigist Araya. Soil landscape relationship modeling of the Atsbi Horst, Tigray, Ethiopia. Unpub. MSc thesis. Mekelle, Ethiopia: Department of Land Resources Management and Environmental Protection, Mekelle University; 2006.
42. De Geyndt K. Bodemkartering ten behoeve van bodemconserveringsonderzoek in Hagere Selam, Ethiopië. Unpub. M.Sc. thesis. Leuven, Belgium: Department of Land Management, University of Leuven; 2001.
43. Nyssen J, Naudts J, De Geyndt K, Mitiku Haile, Poesen J, Moeyersons J, et al. Soils and land use in the Tigray highlands (Northern Ethiopia). Land Degradation & Development. 2008;19(3):257–74. doi: 10.1002/ldr.840
44. Descheemaeker K, Nyssen J, Rossi J, Poesen J, Mitiku Haile, Moeyersons J, et al. Sediment deposition and pedogenesis in exclosures in the Tigray Highlands, Ethiopia. Geoderma. 2006;132:291–314. doi: 10.1016/j.geoderma.2005.04.027
45. Kassa Teka, Nyssen J, Nurhusen Teha, Mitiku Haile, Deckers J. Soil, land use and landform relationship in the Precambrian lowlands of northern Ethiopia. Catena. 2015;131:84–91.
46. Van De Wauw J. Soil-landscape relationships in the basalt-dominated highlands of Tigray, Ethiopia. Unpub. M.Sc. thesis. Leuven, Belgium: Department of Land Management, K.U. Leuven; 2005.
47. Van de Wauw J, Baert G, Moeyersons J, Nyssen J, De Geyndt K, Nurhussen Taha, et al. Soil-landscape relationships in the basalt-dominated highlands of Tigay, Ethiopia. Catena. 2008;75:117–27.
48. Tesfu Woldegerima. Soil landscape relationship modelling: A tool for identifying and mapping soils of the watershed (A case study of the midland sandstone dominated soils of Tsenkaniet watershed). Unpublished MSc Thesis. Mekelle, Ethiopia: Mekelle University; 2006.
49. Hunting Technical Services. Central Tigre Development Study–Tigre Province Ethiopia, Working Paper I: Soils and land classification. Hemel Hempstead (U.K.): Hunting Technical Services Ltd; 1975. 64 p.
50. Tielens S. Towards a soil map of the Geba catchment using benchmark soils. MSc thesis. Leuven and Brussels, Belgium: KULeuven and VUB; 2012.
51. Blume H-P, Stahr K, Leinweber P. Bodenkundliches Praktikum: Eine Einführung in pedologisches Arbeiten für Ökologen, Land-und Forstwirte, Geo-und Umweltwissenschaftler: Springer-Verlag; 2011.
52. Zobeck TM. Rapid soil particle size analyses using laser diffraction. Applied Engineering in Agriculture. 2004;20(5):633.
53. Buurman P, Pape T, Reijneveld J, De Jong F, Van Gelder E. Laser-diffraction and pipette-method grain sizing of Dutch sediments: correlations for fine fractions of marine, fluvial, and loess samples. Netherlands Journal of Geosciences. 2001;80(2):49–57.
54. Fisher P, Aumann C, Chia K, O'Halloran N, Chandra S. Adequacy of laser diffraction for soil particle size analysis. PloS one. 2017;12(5):e0176510. doi: 10.1371/journal.pone.0176510 28472043
55. Pieri L, Bittelli M, Pisa PR. Laser diffraction, transmission electron microscopy and image analysis to evaluate a bimodal Gaussian model for particle size distribution in soils. Geoderma. 2006;135:118–32.
56. Kirkham MB. Chapter 10—Field Capacity, Wilting Point, Available Water, and the Nonlimiting Water Range. In: Kirkham MB, editor. Principles of Soil and Plant Water Relations (Second Edition). Boston: Academic Press; 2014. p. 153–70.
57. van Reeuwijk LP. Procedures for soil analysis. Wageningen, The Netherlands: International Soil Reference and Information Centre (ISRIC); 2002.
58. Banderis A, Barter D, Henderson K. The use of polyacrylamide to replace carbon in the determination of ‘Olsen’s’ extractable phosphate in soil. Journal of Soil Science. 1976;27(1):71–4.
59. Geraedts K, Maes A. The lanthanum precipitation method. Part 1: A new method for technetium (IV) speciation in humic rich natural groundwater. Chemosphere. 2008;73(4):484–90. doi: 10.1016/j.chemosphere.2008.06.041 18682308
60. Zhang L, Wan L, Chang N, Liu J, Duan C, Zhou Q, et al. Removal of phosphate from water by activated carbon fiber loaded with lanthanum oxide. Journal of Hazardous Materials. 2011;190(1–3):848–55. doi: 10.1016/j.jhazmat.2011.04.021 21530079
61. Dohrmann R. Cation exchange capacity methodology III: correct exchangeable calcium determination of calcareous clays using a new silver–thiourea method. Applied Clay Science. 2006;34(1–4):47–57.
62. Hazelton P, Murphy B. Interpreting soil test results: what do all the numbers mean? Australia: CSIRO Publishing; 2007. 160 p.
63. Juo AS, Franzluebbers K, Backman CR. Tropical soils: properties and management for sustainable agriculture. Oxford (UK): Oxford University Press; 2003. 281 p.
64. Getachew Alemu, Wondimu Bayu. Effects of farmyard manure and combined N and P fertilizer on sorghum and soil characteristics in northeastern Ethiopia. Journal of sustainable agriculture. 2005;26(2):23–41.
65. Amare Haileslassie. Soil nutrient stocks and fluxes under smallholders’ mixed farming system in the central highlands of Ethiopia: research experiences from the Galessa and Gare areas. In: Kindu Mekonnen, Glatzel G, Habermann B, editors. Indigenous tree and shrub species for environmental protection and agricultural productivity; Holetta Agricultural Research Centre, Ethiopia: Commission for Development Studies, Austrian Academy of Sciences; 2006. p. 62–75.
66. Girmay Gebresamuel, Singh BR, Mitiku Haile, Borresen T, Lal R. Carbon stocks in Ethiopian soils in relation to land use and soil management. Land Degradation & Development. 2008;19(4):351–67.
67. Schumacher BA. Methods for the determination of total organic carbon (TOC) in soils and sediments. Las Vegas, NV, USA 2002. 25 p.
68. NRCS. Soil quality resource concerns: available water capacity. USA: USDA, Natural Resources Conservation Service; 1998. 2 p.
69. Paruelo J, Aguiar M, Golluscio R. Soil water availability in the Patagonian arid steppe: gravel content effect. Arid Land Research and Management. 1988;2(1):67–74.
70. Descheemaeker K, Nyssen J, Poesen J, Raes D, Haile M, Muys B, et al. Runoff on slopes with restoring vegetation: A case study from the Tigray highlands, Ethiopia. Journal of Hydrology. 2006;331(1–2):219–41.
71. Olness A, Archer D. Effect of organic carbon on available water in soil. Soil Science. 2005;170(2):90–101.
72. Jahn R, Blume H, Asio V, Spaargaren O, Schad P. Guidelines for soil description (4th ed.). Roma, Italy: FAO; 2006.
73. IUSS Working Group WRB. World Reference Base for Soil Resources 2014, first update 2015. Roma, Italy: FAO; 2015.
74. Mulugeta Alene, Ruffini R, Sacchi R. Geochemistry and geotectonic setting of Neoproterozoic rocks from northern Ethiopia (Arabian-Nubian Shield). Gondwana Research. 2000;3(3):333–47.
75. Swanson-Hysell NL, Maloof AC, Condon DJ, Jenkin GR, Mulugeta Alene, Tremblay MM, et al. Stratigraphy and geochronology of the Tambien Group, Ethiopia: evidence for globally synchronous carbon isotope change in the Neoproterozoic. Geology. 2015;43(4):323–6.
76. Chen T. The Mozambique Belt: the link between East and West Gondwana in the Neoproterozoic and possible connection with the Trans-Antarctic Mountains. Gondwana Research. 2001;4(4):594–5.
77. Tarekegn Tadesse, Hoshino M, Sawada Y. Geochemistry of low-grade metavolcanic rocks from the Pan-African of the Axum area, northern Ethiopia. Precambrian Research. 1999;96(1–2):101–24.
78. Miller NR, Mulugeta Alene, Sacchi R, Stern RJ, Conti A, Kröner A, et al. Significance of the Tambien Group (Tigrai, N. Ethiopia) for snowball Earth events in the Arabian–Nubian shield. Precambrian Research. 2003;121(3–4):263–83.
79. Bussert R, Schrank E. Palynological evidence for a latest Carboniferous-Early Permian glaciation in Northern Ethiopia. Journal of African Earth Sciences. 2007;49(4–5):201–10.
80. Sacchi R, Mulugeta Alene, Barbieri M, Conti A. On the Palaeozoic Tillite of the Adigrat Group (Tigrai, Ethiopia). Periodico di Mineralogia. 2007;76(2–3):241–51.
81. Kumpulainen R. The Ordovician glaciation in Eritrea and Ethiopia, NE Africa. In: Hambrey MJ, Christoffersen P, Glasser NF, Hubbard B, editors. Glacial Sedimentary Processes and Products: John Wiley & Sons; 2009. p. 321–42.
82. Bussert R. Exhumed erosional landforms of the Late Palaeozoic glaciation in northern Ethiopia: Indicators of ice-flow direction, palaeolandscape and regional ice dynamics. Gondwana Research. 2010;18(2–3):356–69.
83. Alemu Tilahun, Fassil Kebede, Yamoah C, Erens H, Mujinya B, Verdoodt A, et al. Quantifying the masses of Macrotermes subhyalinus mounds and evaluating their use as a soil amendment. Agriculture, Ecosystems & Environment. 2012;157:54–9.
84. Mitiku Haile, Berhanu Gebremedhin, Amare Belay. The status of soil fertility in Tigray. In: Berhanu Gebremedhin, Pender J, Ehui S, Mitiku Haile, editors. Policies for sustainable land management in the highlands of Tigray, northern Ethiopia. EPTD Workshop Summary. Washington, DC, USA; Nairobi, Kenya; Mekelle, Ethiopia: International Food Policy Research Institute (IFPRI); International Livestock Research Institute (ILRI); Mekelle University; 2002.
85. Driessen P, Deckers J, Spaargaren O, Nachtergaele F. Lecture notes on the major soils of the world: Food and Agriculture Organization (FAO); 2000.
86. Moeyersons J, Van Den Eeckhaut M, Nyssen J, Gebreyohannes T, Van de Wauw J, Hofmeister J, et al. Mass movement mapping for geomorphological understanding and sustainable development: Tigray, Ethiopia. Catena. 2008;75(1):45–54. doi: 10.1016/j.catena.2008.04.004
87. Gustavson TC. Buried Vertisols in lacustrine facies of the Pliocene Fort Hancock Formation, Hueco Bolson, West Texas and Chihuahua, Mexico. Geological Society of America Bulletin. 1991;103(4):448–60.
88. Coulombe CE, Dixon J, Wilding L. Mineralogy and chemistry of Vertisols. Developments in Soil Science. 1996;24:115–200.
89. Van Ranst E, Alemayehu Regassa, Dumon M, Cornelis J-T, Deckers J, editors. On the origin of Planosols–the process of ferrolysis revisited. 19th World congress of Soil Science: Soil solutions for a changing world; 2010; Brisbane, Australia: International Union of Soil Sciences (IUSS).
90. Tesfaye Kidane, Bachtadse V, Mulugeta Alene. Quaternary remagnetization of the Neoproterozoic limestone of Negash Synclinorium (Arabian–Nubian Shield, northern Ethiopia): With implications of no paleomagnetic testing for the proposed Snowball Earth events. Physics of the Earth and Planetary Interiors. 2014;235:1–12.
91. Mulugeta Degie. Pedogenetic significance of Enticho sandstone as parent material for soil formation in the landscapes of Rubafeleg and Tsenkaniet catchments, Eastern Tigray. MSc Dissertation. Mekelle, Ethiopia: Mekelle University; 2007.
92. Bosellini A, Russo A, Fantozzi PL, Getaneh A, Solomon T. The Mesozoic succession of the Mekele outlier (Tigre Province, Ethiopia). Memorie di Scienze Geologiche. 1997;49:95–116.
93. Bussert R, Nyssen J. Rock-Hewn Sandstone Churches and Man-Made Caves in and Around Dogu’a Tembien. In: Nyssen J, Jacob M, Frankl A, editors. Geo-trekking in Ethiopia’s Tropical Mountains. GeoGuide. Cham (CH): Springer; 2019. p. 121–37.
94. Arkin Y, Beyth M, Dow D, Levitte D, Temesgen Haile, Tsegaye Hailu. Geological map of Mekele sheet area ND 37–11, Tigre province, 1:250.000. Addis Ababa: Imperial Ethiopian Governement, Ministry of Mines, Geological survey; 1971.
95. Merla G, Abbate E, Azzaroli A, Bruni P, Canuti P, Fazzuoli M, et al. A geological map of Ethiopia and Somalia (1973) 1:2.000.000 and comment. Firenze, Italy: University of Florence; 1979.
96. Walraevens K, Tesfamichael Gebreyohannes, Kassa Amare, Baert R, Ronsse S, Van Hulle L, et al. Water balance components for sustainability assessment of groundwater-dependent agriculture: example of the Mendae plain (Tigray, Ethiopia). Land Degrad Develop. 2015;26(7):725–36.
97. Williams M, Williams F. Evolution of the Nile basin. In: Williams M, Faure H, editors. The Sahara and the Nile Quaternary Environments and Prehistoric Occupation in Northern Africa. Rotterdam: Balkema; 1980. p. 207–24.
98. Hadgu Hishe, Kidane Giday, Mulugeta Neka, Teshome Soromessa, Van Orshoven J, Muys B. Detection of Olea europaea subsp. cuspidata and Juniperus procera in the dry Afromontane forest of northern Ethiopia using subpixel analysis of Landsat imagery. Journal of Applied Remote Sensing. 2015;9(1):095975.
99. Kieffer B, Arndt N, Lapierre H, Bastien F, Bosch D, Pecher A, et al. Flood and shield basalts from Ethiopia: magmas from the African superswell. Journal of Petrology. 2004;45:793–834.
100. Miruts Hagos, Kassa Amare, Koeberl C, Nyssen J. The Volcanic Rock Cover of the Dogu’a Tembien Massif. In: Nyssen J, Jacob M, Frankl A, editors. Geo-trekking in Ethiopia’s Tropical Mountains. GeoGuide. Cham (CH): Springer Nature; 2019. p. 139–51.
101. Merla G, Minucci E. Missione geologica nel Tigrai. Roma: Reale Accademia d’Italia; 1938.
102. Vandecasteele I, Nyssen J, Clymans W, Moeyersons J, Martens K, Van Camp M, et al. Hydrogeology and groundwater flow in a basalt-capped Mesozoic sedimentary series of the Ethiopian highlands. Hydrogeology journal. 2011;19(3):641–50.
103. Nyssen J, Poesen J, Descheemaeker K, Nigussie Haregeweyn, Mitiku Haile, Moeyersons J, et al. Effects of region-wide soil and water conservation in semi-arid areas: the case of northern Ethiopia. Zeitschrift für Geomorphologie. 2008;52:291–315. doi: 10.1016/j.catena.2008.04.009
104. Van Den Eeckhaut M, Moeyersons J, Nyssen J, Amanuel Zenebe, Poesen J, Mitiku Haile, et al. Spatial patterns of old, deep-seated landslides: A case-study in the northern Ethiopian highlands. Geomorphology. 2009;105(3–4):239–52.
105. Ford T, Pedley H. A review of tufa and travertine deposits of the world. Earth-Science Reviews. 1996;41(3–4):117–75.
106. Dramis F, Umer M, Calderoni G, Mitiku Haile. Holocene climate phases from buried soils in Tigray (northern Ethiopia): comparison with lake level fluctuations in the Main Ethiopian Rift. Quat Res. 2003;60(3):274–83. doi: 10.1016/s0033-5894(03)00107-8
107. Berakhi O, Brancaccio L, Calderoni G, Coltorti M, Dramis F, Umer MM. The Mai Maikden sedimentary sequence: a reference point for the environmental evolution of the Highlands of Northern Ethiopia. Geomorphology. 1998;23(2–4):127–38. doi: 10.1016/s0169-555x(97)00108-6
108. Moeyersons J, Nyssen J, Poesen J, Deckers J, Mitiku Haile. Age and backfill/overfill stratigraphy of two tufa dams, Tigray Highlands, Ethiopia: Evidence for Late Pleistocene and Holocene wet conditions. Palaeogeography, Palaeoclimatology, Palaeoecology. 2006;230(1–2):165–81. doi: 10.1016/j.palaeo.2005.07.013
109. Wielemaker W, De Bruin S, Epema G, Veldkamp A. Significance and application of the multi-hierarchical landsystem in soil mapping. Catena. 2001;43(1):15–34.
110. Bui EN. Soil survey as a knowledge system. Geoderma. 2004;120(1–2):17–26.
111. Bornand M, Legros J-P. Principes de la cartographie des pédopaysages dans les Alpes. Écologie. 1998;29(1/2):49.
112. Laroche B, Arrouays D, Olivier D, Lecerf N. Essai de cartographie numérique des pédopaysages à 1/250 000 dans les départements de l'Aube et de la Marne: Enseignements, limites et perspectives. 2011.
113. Scull P, Franklin J, Chadwick O, McArthur D. Predictive soil mapping: a review. Progress in Physical Geography. 2003;27(2):171–97.
114. Hengl T, de Jesus JM, MacMillan RA, Batjes NH, Heuvelink GB, Ribeiro E, et al. SoilGrids1km—global soil information based on automated mapping. PloS one. 2014;9(8):e105992. doi: 10.1371/journal.pone.0105992 25171179
115. Sanchez PA, Palm CA, Buol SW. Fertility capability soil classification: a tool to help assess soil quality in the tropics. Geoderma. 2003;114(3–4):157–85.
116. Sys C, Van Ranst E, Debaveye J. Land Evaluation. Part I: principles in land evaluation and crop production calculations. Brussels, Belgium: General Administration for Development Cooperation; 1991.
117. Sys C, Van Ranst E, Debaveye J. Land Evaluation. Part II: methods in land evaluation. Brussels, Belgium: General Administration for Development Cooperation; 1991.
118. FAO. A Framework for Land Evaluation. Roma, Italy: FAO; 1976.
119. Deckers J, Tielens S, De Geyndt K, Van de Wauw J, Mitiku Haile, Poesen J, et al. Understanding Soil Spatial Patterns for Sustainable Development. In: Nyssen J, Jacob M, Frankl A, editors. Geo-trekking in Ethiopia’s Tropical Mountains. GeoGuide. Cham (CH): Springer Nature; 2019. p. 361–72.
120. Koohafkan P, Nachtergaele F, Antoine J. Use of agro-ecological zones and resource management domains for sustainable management of African wetlands. Wetland Characterization and Classification for Sustainable Agricultural Development. Harare, Zimbabwe: FAO/SAFR; 1998.
121. Moeyersons J, Nyssen J, Deckers J, Mitiku Haile, Poesen J. Geomorphic Processes in Late-Pleistocene and Holocene Environments. In: Nyssen J, Jacob M, Frankl A, editors. Geo-trekking in Ethiopia’s Tropical Mountains. GeoGuide. Cham (CH): Springer Nature; 2019. p. 179–93.
122. Poesen JW, Torri D, Bunte K. Effects of rock fragments on soil erosion by water at different spatial scales—a review. Catena. 1994;23(1–2):141–66. PubMed PMID: ISI:A1994PJ57200010.
123. Nyssen J, Poesen J, Moeyersons J, Lavrysen E, Mitiku Haile, Deckers J. Spatial distribution of rock fragments in cultivated soils in northern Ethiopia as affected by lateral and vertical displacement processes. Geomorphology. 2002;43(1–2):1–16.
124. Nyssen J, Mitiku Haile, Poesen J, Deckers J, Moeyersons J. Removal of rock fragments and its effect on soil loss and crop yield, Tigray, Ethiopia. Soil Use Manag 2001;17:179–87.
125. Kirkby M. A conceptual model for physical and chemical soil profile evolution. Geoderma. 2018;331:121–30.
126. Willgoose G. Principles of soilscape and landscape evolution: Cambridge University Press; 2018.
127. Wall G, Coote D, Pringle E, Shelton I. RUSLEFAC—Revised universal soil loss equation for application in Canada: A handbook for estimating soil loss from water erosion in Canada. Ottawa: Agriculture and Agri-Food Canada; 2002. 117 p.
128. Hurni H. Soil formation rates in Ethiopia (with scale 1: 1,000,000). Roma, Italy: FAO; 1983.
129. Nigussie Haregeweyn, Poesen J, Nyssen J, Govers G, Verstraeten G, de Vente J, et al. Sediment yield variability in Northern Ethiopia: a quantitative analysis of its controlling factors. Catena. 2008;75(1):65–76.
130. Nyssen J, Clymans W, Poesen J, Vandecasteele I, De Baets S, Nigussie Haregeweyn, et al. How soil conservation affects the catchment sediment budget—a comprehensive study in the north Ethiopian highlands. Earth Surface Processes and Landforms. 2009;34:1216–33. doi: 10.1002/esp.1805
131. Aerts R, Nyssen J, Mitiku Haile. On the difference between "exclosures" and "enclosures" in ecology and the environment. Journal of Arid Environments. 2009;73:762–3.
132. World Bank. TheGlobalEconomy.com; 2019.
133. Munro RN, Teweldeberhan Woldegerima, Berhane Hailu, Amanuel Zenebe, Gebremedhin Z, Abrha Hailemichael, et al. A History of Soil and Water Conservation in Tigray. In: Nyssen J, Jacob M, Frankl A, editors. Geo-trekking in Ethiopia’s Tropical Mountains. Cham (CH): Springer Nature; 2019. p. 477–93.
134. Tigist Oicha, Cornelis W, Verplancke H, Nyssen J, Deckers J, Mintesinot Behailu, et al. Short-term effects of conservation agriculture on Vertisols under tef (Eragrostis tef (Zucc.) Trotter) in the northern Ethiopian highlands. Soil & Tillage Research. 2010;106:294–302. doi: 10.1016/j.still.2009.12.004
135. Berhane Grum, Kifle Woldearegay, Hessel R, Baartman JE, Abdulkadir M, Yazew E, et al. Assessing the effect of water harvesting techniques on event-based hydrological responses and sediment yield at a catchment scale in northern Ethiopia using the Limburg Soil Erosion Model (LISEM). Catena. 2017;159:20–34.
136. Gebremeskel Kassa, Teka Kassa, Birhane Emiru, Negash Emnet. The role of integrated watershed management on soil-health in northern Ethiopia. Acta Agriculturae Scandinavica, Section B—Soil & Plant Science. 2019:1–7.
137. Gebremeskel G, Gebremicael T, Girmay A. Economic and environmental rehabilitation through soil and water conservation, the case of Tigray in northern Ethiopia. Journal of Arid Environments. 2018;151:113–24.
138. Mulubrhan Balehegn, Mitiku Haile, Fu C, Liang W. Ecosystem-Based Adaptation in Tigray, Northern Ethiopia: A Systematic Review of Interventions, Impacts, and Challenges. In: Leal Filho W, editor. Handbook of Climate Change Resilience. Cham (CH): Springer Nature; 2019. p. 1–45.
139. Wuletawu Abera, Lulseged Tamene, Degefie Tibebe, Zenebe Adimassu, Habtemariam Kassa, Habtamu Hailu, et al. Characterizing and evaluating the impacts of national land restoration initiatives on ecosystem services in Ethiopia. Land Degradation & Development. 2019;in press.
140. Tesfay Araya, Cornelis WM, Nyssen J, Govaerts B, Fekadu Getnet, Bauer H, et al. Medium-term effects of conservation agriculture based cropping systems for sustainable soil and water management and crop productivity in the Ethiopian highlands. Field Crops Research. 2012;132:53–62. doi: 10.1016/j.fcr.2011.12.009
141. Dondeyne S, Emmanuel L, Deckers JA. Mr Napite's botanical knowledge: bridging farmers' and scientists' insights during participatory research. Indilinga African Journal of Indigenous Knowledge Systems. 2003;2(2):45–57.
143. Van Gaelen H, Alemtsehay Tsegay, Delbecque N, Shrestha N, Garcia M, Fajardo H, et al. A semi-quantitative approach for modelling crop response to soil fertility: evaluation of the AquaCrop procedure. The Journal of Agricultural Science. 2015;153(7):1218–33.
144. Sanginga N, Woomer PL. Integrated soil fertility management in Africa: principles, practices, and developmental process. Nairobi: CIAT; 2009. 263 p.
146. Willey R, Natarajan M, Reddy M, Rao M, Nambiar P, Kannaiyan J, et al. Intercropping studies with annual crops. In: Ciba Foundation, editor. Better crop for food. London, U.K.: Pitman; 1983. p. 88–100.
147. Tittonell P, Giller KE. When yield gaps are poverty traps: The paradigm of ecological intensification in African smallholder agriculture. Field Crops Research. 2013;143:76–90. doi: 10.1016/j.fcr.2012.10.007