Teff Grain Physical and Chemical Quality Responses to Soil Physicochemical Properties and the Environment

Teff is the only cultivated cereal crop from the genus Eragrostis and it is the major staple food of Ethiopians. In Ethiopia, the quality of teff and its market price are primarily determined by its grain color. The objective of this study was to evaluate the effects of soil physicochemical characteristics across multiple locations in the two main teff growing regions of Amhara and Oromia states in Ethiopia on teff grain color and nutritional quality of a single variety. Grain and soil samples were collected from 24 field sites cultivated with the popular teff variety ‘Quncho’ (DZ-Cr-387/RIL-355). The teff grain samples collected from the 24 locations were evaluated for grain color, proximate composition, amino acid composition, and grain mineral concentration and the soil samples were analyzed for their physicochemical properties. Sample location means were considered different p < 0.05. Teff grain color indices of hue (H), saturation (S), and brightness (V), grain proximate composition, amino acid composition, and mineral concentration differed among locations (p < 0.05). There were significant negative correlations between grain S color value and soil pH, SOC, Ca, Mg, S, and Na. Soils with greater pH, SOC, Ca, Mg, and S generally had lower S values and thus, whiter color teff grains. There were considerable variations in the measured parameters for soil and teff grain physicochemical properties. The results indicated an opportunity for management interventions necessary to obtain uniformity in grain color and chemical composition for the same variety of teff grown in the two major regions in Ethiopia.


Introduction
Teff [Eragrostis Tef (Zucc.) Trotter] is the only cultivated cereal crop from the genus Eragrostis [1][2][3][4]. It is believed that pre-semantic inhabitants of the Ethiopian highlands domesticated teff around 4000 BC as a food crop [1,5,6] and it has served as the major staple food for Ethiopians for thousands of years.
In Ethiopia, the main teff-producing areas have been concentrated in the Amhara and Oromia regional states (the northwestern and central highlands of Ethiopia). Teff cultivation in Amhara and Oromia regional states covered 1.45 (47.8%) and 1.14 million ha (37.6%) of land, respectively, for a combined 2.59 million ha or 87.5% of the total teff cultivation in Ethiopia during the 2017 to 2018 main growing season [7]. Teff grain production from these two main cultivated regions of Amhara and Oromia states were 2.6 (48.9%) and 2.04 million tons (38.6%), respectively, for a total of 4.64 million tons or 87.5% of the total teff production in Ethiopia during the 2017 to 2018 main growing season [7]. Teff grain production ranged from 0.84 to 1.75 tons per hectare annually in Ethiopia depending on location of cultivation and agronomic practices [7]. Almost all the teff grain produced in Ethiopia is mainly used for local consumption in the form of "injera" (a thin, malleable, with many eyes, fermented bubbly spongy type Ethiopian bread) [8].
The major determinants of quality variation of cereal grains are derived from the nutrient composition and physical attributes of the grain [9]. The grain color of teff varies from pale white to ivory white, light tan to dark brown to reddish-brown purple [10][11][12] and in Ethiopia, the quality of teff grain is determined primarily by its physical attribute of color [13,14]. There are four major categories of teff grain colors specified by the Ethiopian Standard Agency (ESA) and the grains are classified as very white (magna), white (nech), mixed (sergegna) and brown (key) teff [13]. In Ethiopia, grain color is a measure of quality that also dictates the market demand and value. The very white and white teff fetches the highest price and is consumed by the wealthiest individuals whereas, the brown teff is sold at a lower price to low-income communities [8,[15][16][17]. In Ethiopia, the production area and soil type which is closely correlated with grain color, are intricately linked to the market price of teff grain [8,18,19]. For example, teff cultivated on black and brown soils in Ethiopia has brighter grain color than those produced on red soils and are most preferred by consumers and traders [19]. However, even with the same grain color, there has been a wide range of variability in the nutrient content reported for teff [20][21][22].
Further, teff has been proposed as an important dietary source of mineral nutrients to combat malnutrition for the populace of developing nations [22]. In particular, the concentrations of calcium, iron, copper, zinc, carbohydrates, and fiber are generally greater than those of other cereal crops [21,23]. Micronutrients such as iron and zinc are considered the most limiting nutrients for children in many communities in developing countries [24] and teff may be the ideal crop for biofortification as a sustainable and cost-effective strategy to curtailed micronutrient malnutrition that affects approximately 2 billion people globally [25,26]. Further, among female athletes from developing nations, only Ethiopian athletes are free of anemia and this was attributed to the consumption of teff [27]. Teff has the potential in reducing iron deficiency anemia diseases by its iron richness [21,28,29] and it is also reported to have a complete set of essential amino acids with excellent composition [21,[30][31][32][33] and richer in lysine than most other cereals [34].
The qualitative value of teff as a gluten-free grain combined with its nutritional value and health benefits have attracted global interest in its consumption [14,[35][36][37]. Compared to the other major cereal crops, the crude protein, crude fiber, fat and starch concentrations of teff grain are either similar or superior to those of maize, oat, sorghum, wheat and quinoa (false cereal) in general [38]. Globally, the demand for teff grain has increased over the last decade due to the many health benefits of this gluten-free product and as a result of the demand, the Ethiopian government has prohibited the export of teff grain raw material to ensure food security and protection of local markets in Ethiopia. However, the government of Ethiopia did not place a ban on value-added teff hence, teff bread (injera) is exported to the Middle East, Europe, and the United States of America [39].
It is possible that the reported variability in the nutrient content of teff in Ethiopia could be attributed to environmental factors (climatic and edaphic variations) rather than genetic variations. However, there are no studies that have reported on the influence of different environmental factors on variability in teff grain quality in Ethiopia. Most reported research focused on teff nutritional quality based on color [21,22,30,40] and did not address the influence of environmental factors on grain nutritional composition. Therefore, the objective of this study was to evaluate the effects of soil physicochemical characteristics across multiple locations in the main teff growing regions of Ethiopia on teff grain color and nutritional quality of a single variety.

Description of the Study Area
This study was carried out in 24 major teff producing areas of Amhara and Oromia regional states of Ethiopia (northwestern and central highlands of Ethiopia, respectively). The specific locations and known names listed in parentheses were Abrja (Denbia), Abshem (Debre-Elias), Adet 1, Adet 2, Akaki, physicochemical characteristics across multiple locations in the main teff growing regions of Ethiopia on teff grain color and nutritional quality of a single variety.

Description of the Study Area
This study was carried out in 24 major teff producing areas of Amhara and Oromia regional states of Ethiopia (northwestern and central highlands of Ethiopia, respectively). The specific locations and known names listed in parentheses were Abrja (Denbia), Abshem (Debre-Elias), Adet  The areas were selected based on their variability of altitude (1670-2570 m above sea level), soil types (Nitisols-Vertisols), annual rainfall amount (1000-1960 mm) and temperatures (Table 1). Maximum temperatures range from 39.5 °C at Debre Elias to 29 °C at Motta with a minimum temperature of 9.8 °C at Debre Elias to −1.5 °C at Dangla. Rainfall amount varies from 1964 mm (Debre Elias) to 779 mm (Bishoftu) ( Table 1). The areas were also selected from the two regions based on the information gained from Addis Ababa and other regional markets surveyed [19] and from personal communication with researchers and zonal agricultural experts. The areas were selected based on their variability of altitude (1670-2570 m above sea level), soil types (Nitisols-Vertisols), annual rainfall amount (1000-1960 mm) and temperatures (Table 1). Maximum temperatures range from 39.5 • C at Debre Elias to 29 • C at Motta with a minimum temperature of 9.8 • C at Debre Elias to −1.5 • C at Dangla. Rainfall amount varies from 1964 mm (Debre Elias) to 779 mm (Bishoftu) ( Table 1). The areas were also selected from the two regions based on the information gained from Addis Ababa and other regional markets surveyed [19] and from personal communication with researchers and zonal agricultural experts.

Data Collection, Grain and Soil Nutrient Analysis
At each of the 24 field sites, the popular teff variety named 'Quncho' (DZ-Cr-387/RIL-355) was earmarked for grain collection due to its white grain color, adaptability to the different agro-ecological zones, and its widespread cultivation across Ethiopia [41]. The recommended fertilizer rate for Vertisols, in Ethiopia, is 60 kg N and 80 kg P 2 O 5 ha −1 by Debre Zeit Agricultural Research Center and for Nitisols (red soils), is 40 kg N and 60 kg P 2 O 5 ha −1 by Adet Agricultural Research Center (AARC) in Ethiopia. Diammonium phosphate (DAP) and urea fertilizer materials are the typical sources of the nutrients. While DAP is usually applied once at the time of sowing, split application of urea is the general N management practice, one half at sowing and the other at early tillering. However, in the study areas, the extent to which farmers follow these fertilizer recommendations is not verifiable.
Grain sampling was carried out at each of the selected field site (approximately 1 ha in size) and a total of five 1-m 2 quadrat samples were collected randomly along a diagonal transect across the field at fixed steps intervals at the time of harvest or physiological maturity (November to December, 2017). The five quadrats samples from each field were then stored separately in a moisture-free area in preparation for seed cleaning. The samples were hand threshed inside cloth bags in order to reduce soil contamination. The threshed grain samples were then manually cleaned by sifting and winnowing to remove dust, chaff, and other materials.
Soil samples were randomly collected down to a depth of 20 cm from each field site at the time of grain sampling using a 1.9-cm diameter probe and composited. Further, independent undisturbed core soil samples were taken for bulk density determination to help in soil characterization. The collected samples for soil nutrient analysis were air-dried while those for bulk density determination were oven-dried at 110 • C for 24 h at the Adet Agricultural Research Center Soils Laboratory. The soil samples were then ground and sieved through a 2-mm screen before chemical analysis. The soil pH, soil organic carbon (SOC) and cation exchanging capacity (CEC) were determined at the same Laboratory. Total nitrogen (TN) was determined using the Kjeldahl method [42] at Horticoop Soil and Water lab, Bishoftu, Ethiopia. The soil color at each site was determined using the Munsell color chart in dry soil bases. Each sample was analyzed in triplicates for macro and micronutrients (phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), sulfur (S), iron (Fe), manganese (Mn), zinc (Zn), copper (Cu), molybdenum (Mo), and sodium (Na)) using the Mehlich 3 extraction method (0.2M CH 3 COOH, 0.25M NH 4 NO 3 , 0.015M NH 4 F, 0.013M HNO 3 , and 0.001M EDTA adjusted to pH 2.5) [43]. The soil samples were analyzed using inductively coupled plasma atomic emission spectroscopy (Spectro CIROS ICP-AES, Spectro Analytical Instruments, Kleve, Germany).
Grain color images were captured using a Tecno-Camon mobile 24 mm pixel camera (Techno Mobile, Hong Kong). The images were first analyzed using RGB (red, green, blue) color detector free software. The RGB color was then converted to HSV (hue, saturation, and value) by RGB to HSV color converter software. The preference for HSV color space was out of the ease of understanding color with the human eyes than other color spaces [44,45]. Ibraheem et al. [45] and Deswal and Sharma [44] described HSV color space as H (hue) measures the purity of a particular color, S (saturation) measures the degree of white color embedded in a specific color and V (value or brightness) detects the intensity of colors. They also noted that V can be used as a luminance which detects what the color brightness is (brightness/lightness or darkness). The value of H is in degree and the S and V values are in percent.
In preparation for nutrient analysis, 100 g of teff grain from each field site sample was ground separately using a RIRIHONG high-speed multifunction rotary grain grinder (Shanghai Yuanwo Industrial and Trade Co. LTD, Shanghai, China). One half (50 g) of the ground sample was used for nutrient analysis [43] at the Horticoop Soil and Water Laboratory, Bishoftu, Ethiopia and the other for composition analysis (crude fiber, fat, crude protein (CP), and starch analysis at the Ethiopian Agricultural Research Institute (Addis Ababa, Ethiopia). Grain proximate and amino acids composition were analyzed using 3 g of homogenized teff flour in duplicate. The flour samples were scanned using a Near Infrared Reflectance Spectrophotometer (NIRS) spinning system (FOSS, Model: NIRS system 5000, Hilleroed, Denmark). Samples were placed in ring cups and their spectra were recorded in reflectance mode in the range from 400 to 2500 nm, at 2 nm intervals as described by Agza et al. [46]. The prediction for the collected spectra was carried out using plant-based and aqua feed calibrations developed by the International Livestock Research Institute in collaboration with Ethiopian Institute of Agricultural Research. As described by Agza et al. [46] the coefficient of determination (R 2 ) for the calibration and validation ranged between 0.93-0.99 and 0.93-0.98 with corresponding standard error values ranging between 0.03-0.25 and 0.04-0.37. The grain mineral elements determination was carried out using Mehlich 3 extraction method (0.2M CH 3 COOH, 0.25M NH 4 NO 3 , 0.015M NH 4 F, 0.013M HNO 3 , 0.001M EDTA and adjusted to pH 2.5) [43] and gas chromatography-mass spectrometry.

Statistical Analysis
Soil physicochemical parameters were not subjected to ANOVA because samples collected were composited for each location thus excluding replications. Therefore, location means for each soil physicochemical parameters were presented based on the triplicate analysis in the laboratory. Analysis of variance was performed to detect differences in grain parameters among the different locations and block (northwestern vs. central highlands) using PROC GLM in SAS software [47]. Location and block mean for grain nutrient and color data were separated using the Fisher's protected least significant difference at alpha = 0.05. Pearson's correlation coefficients among soil physicochemical and teff grain parameters were generated using the PROC CORR procedure in the SAS software [47].

Soil Physicochemical Properties
Based on the Munsell color chart, soil color varied across the different sample collection sites in the two main teff growing regions of Ethiopia however, the majority were black colored soils (Table 2). Soil pH based on the numerical value did not vary widely, however, six out of the 24 locations were considered slightly acidic ( Table 2). The CEC value varied widely among locations and for Abrja, Arerti 2, Bichena, and Simada the values were above 60 but for the Kudmi location, CEC was relatively low (Table 2). Both SOC and TN also varied across locations and the SOC values were considered high at the Birzakani, Gumdri, and Abshem locations relative to Wondata location (Table 2). Total N based on numerical value was highest at Abshem followed by Akaki location and some of the lowest values were obtained at Abrja, Atsedemariam and Kosheshila locations ( Table 2). In this study, the numerical values for mineral concentration varied across locations (Table 3). For example, the concentration of soil P ranked highest at the Bishoftu location (43 mg/kg) followed by Arerti 2, Alemtena (almost 24 mg/kg), and Arerti 1 (about 15 mg/kg) respectively, while the numerical values for P concentration at most other locations were relatively close ( Table 3). The K concentration values showed wider variations at the different locations relative to P, and K concentration numerical values were greatest at Alemtena followed by Arerti 1, Arerti 2, and Bishoftu locations respectively and the lowest was from Kudmi location ( Table 3). The Ca concentration numerical value was greatest at Kosheshila and lowest at Gumdri and Zenzelima locations while for Mg concentration, the numerical value was highest at Abrja and lowest at Abshem and Gumdri locations (Table 3). For S concentration, wide variations also occurred across locations based on numerical value and soil S concentration at Abrja, Adet 2, Akaki, Areti 1, Areti 2, Atsedema, Birzakani, Fetlokus, Gejageda, Kosheshila, Simada, Wondata, and Yubodegabati were among the lowest (Table 3). In relation to the concentration of soil Fe concentration, the greatest numerical value was obtained from soils collected from Bichena and lowest at Arerti 2 location (Table 3). There were also, differences in numerical value for all other soil micronutrient concentrations across the locations used in this study and there were wider variations in Mn, Zn, and Cu compared to the variation among locations for soil Mo concentration ( Table 3). The numerical values for soil Na concentration varied across locations and were ranked highest at Abrja field site and lowest at Gumdri and Kudmi locations (Table 3).

Teff Grain Physical and Chemical Quality
The grain of the teff variety Quncho sampled across the main regions of its cultivation in Ethiopia varied in color ( Figure 2). There was a significant effect of location (p < 0.05) on teff grain color indices of hue (H), saturation (S), and brightness (V) in this study (Table 4). Based on the color space classification scheme, grains which showed high V and low S were generally very bright white and those with low V and high S values were generally darker in color based on Deswal and Sharma, [44] color classification. The pureness of the teff grain color represented by the H value was generally higher in the locations Adet 1, Akaki, and Alementa than most other locations (Table 4). In relation to the S values, teff grown in Abshem and Kudmi had the highest values among the different locations and were of a pale white color while the lowest S value was for teff grain collected from the Bishoftu location and the grains were very white (Table 4). Other locations that had teff grain classified as very white based on the S value were Abrja, Arerti 1, Arerti 2, Atsede Mariam, and Kosheshila (Table 4). Teff grain V value was greatest for teff grain collected from the Bishoftu location (brighter color) and lowest for grain sampled from Zenzelima location (darker color) while for all other locations, the values were intermediate (Table 4). Teff grown on red colored soils (Nitisols) had lower V and higher S values while teff grown on the black colored soils (Vertisols) had higher V and lower S values (Table 4). Pertaining to grain color of teff grown in the two main regions (northwestern highlands vs. central highlands) both the S (p = 0.004; SD = 5.3) and V (p = 0.001; SD = 3.7) color value were different. However, the H color value was not different (p = 0.237; data not shown) between the two main teff growing regions in Ethiopia. The greater grain S color value was obtained from teff grown in the northwestern highlands (39%) compared to teff grown in the central highlands (32%) and this indicates that the lower saturated grain color of teff was linked to locations within the central highlands while the higher saturated grain color was from locations within the northwestern highlands. The greater grain V value of 81% was obtained from teff grown in the central highlands compared to the northwestern highlands (75%) and is indicative of whiter grains from the central highlands of Ethiopia. In our study, there were significant negative correlations between grain S value and soil pH, SOC, Ca, Mg, S, and Na (Table 5). Therefore, the soils in our study with greater pH, SOC, Ca, Mg, and S generally exhibited lower S values and thus very whiter color teff grains were observed. For the V value, there were significant positive correlations with black soil color, pH, Ca, Mg, Na, and a negative association with SOC and S (Table 5).   In addition, there was some significant negative relationships between H color value and grain starch (r = −0.41; p < 0.05), S value and CP (r = −0.44; p < 0.05), V value and fiber concentration (r = −0.48; p < 0.05) but a significant positive association between S value and grain fat (r = 0.44; p < 0.05). The higher V value indicates brighter color and lower V value, darker color teff grain. Only the S value was significantly associated with the amino acids alanine (r = −0.50; p < 0.05), arginine (r = −0.45; p < 0.05), aspartic acid (r = −0.56; P < 0.01), methionine (r = −0.51; p < 0.05), serine (r = −0.46; p < 0.05), threonine (r = −0.48; p < 0.05), tyrosine (r = −0.44; p < 0.05) and valine (r = −0.46; p < 0.05). However, there was no significant correlation (p > 0.05; data not shown) between the grain HSV color space values and grain mineral concentrations in this study. For teff grain, the other qualitative traits evaluated were the proximate composition, amino acid composition, and concentration of the different mineral nutrients (P, K, Ca, Mg, S, Fe, Mn, Zn, Cu, B, Mo, and Na) across the 24 locations in the two main teff growing regions of Ethiopia. For grain proximate composition parameters of fiber, fat, crude protein, and starch, there was a main effect of location (p < 0.05) in this study (Table 4). Grain fiber concentration ranked highest at Zenzelima and lowest at Arerti 1, Bichena, Bitiejersalafo, Kudmi, and Yubodegabati locations (Table 4). While for grain fat concentration, teff grain from Abshem, Bitiejersalafo, and Gumdri locations were ranked among the highest and the lowest grain crude fat concentration was obtained from Birzakani location ( Table 4). The grain CP concentration was highest at the Birzakani location and lowest at Simada location (Table 4). Teff grain sampled from the locations Adet 2, Alemtena, and Zenzelima were ranked highest in starch concentration and those collected from Abrja, Adet 1, and Demdengay locations were among the lowest in starch concentration (Table 4). Overall, there was no consistent trend where anyone location ranked highest in all the proximate composition parameters in this study which indicated how variable these proximate composition parameters were in this study. Among the proximate composition parameters, only crude fat was different (p = 0.009; SD = 4.4) between the two main teff growing regions and was greater for teff grown in the northwestern highlands (29.0 g/kg) relative to those grown in the central highlands (23.0 g/kg). The grain CP concentration was positively correlated with SOC, and TN but negatively associated with soil Mg, Fe, and Na (Table 6). Teff grain crude fiber concentration was positively correlated with TN and S, but negatively correlated with soil pH, K, Ca, Mg, and Na while grain fat concentration was only negatively correlated with SOC and positively associated with Cu (Table 6). However, grain starch had no significant associations with any of the soil physicochemical properties ( Table 6). There was a significant effect (p < 0.05) of location on all 18 amino acids analyzed in the teff grain of this study (Table 7). No one location ranked highest for all the 18 amino acids analyzed in the teff grain collected, however, teff grain collected from the Bishoftu location ranked highest for 12 out of the 18 amino acids evaluated (Table 7). Also, between the two main teff growing regions (northwestern vs. central highlands), none of the amino acids were different (p > 0.05). There was a location effect (p < 0.05) on teff grain minerals concentration in this study (Table 8). Among the macronutrients, grain P concentration was ranked highest for teff grown at Fetlokuskuam and Yubodegabati and lowest at Zenzelima location ( Table 8). The grain K concentration followed a similar trend like P and grain Ca concentration was highest at the Yubodegabati location and lowest at Gejagedanba and Zenzelima locations ( Table 8). The grain Mg concentration was highest at Fetlokuskuam and Yubodegabati locations and lowest at Zenzelima location while for grain S concentration, highest at Atsedemariam and lowest at Akaki and Zenzelima locations (Table 8). Additionally, the grain Na concentration ranked highest at Atsede Mariam, Bishoftu, and Simada but lowest at Akaki, Bichena, Gejagedanba, Wondata, and Zenzelima while all other locations were intermediate and also varied ( Table 8). In relation to the teff grain micronutrients, Fe concentration was greatest at Abrja location and ranked lowest at Abshem, Alemtena, Arerti 1, Bichena, Birzakani, Bitiejersalafo, Demdengay, Gumdri, Kosheshila, Wondata, Yubodegabati, and Zenzelima locations (Table 8). There were differences in all other grain micronutrients concentration with wider variations in micronutrients concentration for B and Cu and the least variation occurred for Mo across locations (Table 8). Between the two main teff growing regions, grain Mg concentration differed (p = 0.011; SD = 12.1) and it was greater for teff grown in the northwestern highlands (160.2 mg/100 g) compared to those grown in the central highlands region (144.5 mg/100 g). There were trends for both grain P (p = 0.07; SD = 25.9) and grain S concentration (p = 0.06; SD = 2.9) difference between the two main teff growing regions. Both grain P (320.7 mg/100 g) and S (26.5 mg/100 g) for teff grown in the northwestern highlands region tended to be greater than those grown in the central highlands regions P (291.1 mg/100 g) and S (23.9 mg/100 g) respectively. For teff grain minerals concentration, the significant correlations that occurred with soil physicochemical parameters were all below the 50% threshold and indicated that other external variables would have played a more significant role in the variation of grain mineral concentration among locations (Table 6)    Alemtena  460mn  520j  660jk  175hi  1630m 330def 190n  230lm  690n  200ij  170kl  410m  610h  330kl  230k  70kl  280no  330o  Arerti 1  490jk  610efg 710ghi 202cd 1650m 320efgh 210ij  240jk  760ij  190ji  170lm  430l  650g  350j  270ih  70kl  290lm  360lmn  Arerti 2  590d  740b  990b  194de 2680bc 380ab  270d  430b  990c  340b  220b  610c  710d  500b  370c  120c  400b  510c  Atsede Mariam  550e  620def  820r  177gh 2420d 350bcd 250e  360d  890e  300c  210de  550e  700de  450e  320e  90f  360f  444ef  Bichena  530g  610fgh  740f  202cd 1820jkl 380av  230fg  280fg  790hg  270d  180gh  460ij  680ef  390gh  290g  80h  310ij  390hi  Birzakani  480kl  540j  660k  174hij 1470n  270i  200k  220m 700mn 190ij  160m  390n  560k  330l  240jk  60n  270ef  330o  Bishoftu  680a  820a  1140a  190ef  2710b  360bc  280c  490a  1100ab 340b  240a  670a  640g  550a  420a  140a  420a  580a  Bitiejersalafo  580d  630dce 880d  216a  2460d 370ab  300b  380c  990c  370a  200e  580d  780ab  480c  360c  100e  380d  480d  Demdengay  630b  760b  930c  211ab  2820a  270i  320k  430b  1110a  310c  210c  640b  740c  510b  410ab  130b  390de  560b  Fetlokuskuam  540ef  640cd  820e  206bc 2470d 380ab  280c  350d  930d  320bc  190f  560e  790a  470cd  340d  90g  370ef  450ef  Gejagedanba  510hi  580hi  730fg  185fg  1900h 340cde 230fg  280gh  780hi  270d  180i  4700hi  650g  390g  280gh  70ij  310ijk  380jk  Gumdri  480l  580i  680ijk 166jkl  1770l  280i  210j  260ji  740kl  210ih  170k  430l  590hij  350j  260i  70l  290mn  360n  Simada  480kl  570i  690hij  167i-l  1440n  280i  180mn 230klm 690n  180j  170jk  390n  510l  320l  240jk  60lm  260q  340o  Kosheshila  510h  620def  820e  162klm 2300e  370ab  210j  330e  810g  260de 200de  530f  640g  410f  290fg  90fg  350g  410g  Kudmi  550ef  660c  820e  213ab  2180f  390a  270d  320e  870ef  310c  190g  530f  760bc  440e  330e  90f  350g  440f  Motta  500i  610efg  740f  165klm 1870hij 300ghi 230gh 280gh  750jk  250def 180hi  450k  570k  360ij  280gh  80h  300jkl  390ij  Wondata  460m  570i  730fg  140n  1620m 320efgh 160o  240kl  680n  150k  170k  420l  510l  350jk  230k  60mn  280op  330o  Yubodegabati  550ef  620def  820e  205bc 1840ijk 330defg 240f  290f  860f  250def  180ij  490g  660fg  400f  300f  80h  310i  400gh  Zenzelima  540fg 630def 880d  159lm 2630c  330cd 230gh  390c  940d  270d  210c  580d  680ef  460de  320e  110d  380de  450e  LSD (0.05)  10  30  30  10  60  30  10  10  10  30  10  10  30  20  20  40 10 14

Discussion
The characterization of soil properties helps provide an understanding of the variation in crop production and quality in different agro-ecological zones [48]. While Na is not an essential nutrient element required for plant growth its presence in high concentration in the soil can hinder nutrient uptake, growth, and development of crop plants [49]. The common trend of spatial variation in soil physicochemical properties are generally the result of soil formation factors and variation in agricultural soil management in cultivated areas [50] and this may have been the principal factors responsible for the variation in soil physicochemical properties among the locations evaluated. Our results for teff grain color evaluated across multiple locations in the main teff growing regions of Ethiopia is in confirmation with the results of Wu et al. [51] who reported significant color variation in wheat grain from a single variety cultivated over diverse production conditions. Further, Baye [14] reported that the very white and white teff are grown on the vertisols of the central highlands of Ethiopia which corresponds to our results. The results suggested that there could be a direct relationship between grain nutrient and grain color of teff. Tańska et al. [52] reported that variation in grain color of wheat was reflected by changes in its chemical composition but not the environment which is in partial agreement with the results of our study. In addition, the accumulation of carotenoids and anthocyanins pigments in wheat grain accounted for the large variation in its grain color [52][53][54][55] and may have played a role in the grain color differences for the single variety of teff sampled across multiple locations in our study. Overall, the average of 63.5% of the variation in grain color can be attributed to the variation in the soil physicochemical properties that influenced several parameters that were correlated with grain color in this study (Table 5).
For the proximate composition parameters, the grain crude fiber, crude fat, CP, and starch concentrations were aligned with the range reported in several studies for different varieties of teff across different geographic locations [20,22,32,55]. This result is indicative of how these nutritional parameters are highly variable in space. In this study, the seven out of the 18 amino acids had significant correlations with soil P, while only four out the 18 correlated with Mg, and 10 out of the 18 correlated with Na (Table 9). However, with the exceptions of aspartic acid and soil P, and cysteine and soil Na, all other significant correlations were below the 50% threshold ( Table 9) and indicates that other environmental variables may have played a role in the differences in teff grain amino acids concentration. The amino acids concentration of teff grain in this study were in line with those reported for the same variety of teff in a study by Daba [22]. For teff grain minerals concentration, the significant correlations that occurred with soil physicochemical parameters were all below the 50% threshold and indicated that other external variables would have played a more significant role in the variation of grain mineral concentration among locations (Table 6). In our study, high soil mineral concentration did not necessarily correspond to high grain mineral concentration particularly, the grain micronutrients of Fe and Zn showed no significant correlations with soil physicochemical properties and may be indicative of a dilution effect because of higher grain yield in those nutrient-rich locations similar to the trends reported by Fan et al. [56]. The grain mineral concentration in this study was similar to those reported for both white and red teff [57] but often times lower grain mineral concentration than that of the major cereal crop wheat [58,59]. These variabilities in the grain proximate composition parameters, amino acids, and mineral concentration are generally modified by the unique soil physicochemical characteristics, climatic conditions, and the type of agronomic management practices instituted at the location of crop plants cultivation [58][59][60].

Conclusions
Grain color of teff dictates its quality and demand in Ethiopia and was influenced by location for the same variety. In this study, teff grown on the red colored soils of the five locations in Amhara (Abshem, Adet 1, Kudmi, Motta, and Zenzelima), had higher S (saturation) and lower V (brightness) color space values, which indicates the production of pale white grains compared to the whiter grains produced on the black colored soils of several locations in both Amhara and Oromia regional states. The soil properties of pH, soil organic carbon, Ca, Mg, and S were strongly associated with grain S value, however, among those soil variables based on the correlation coefficient, soil S concentration had the strongest relationship with teff grain color S value and a similar trend occurred for grain V (brightness) value. The locations of Abshem, Adet 1, Gumdri, Kudmi, Motta, and Zenzelima with high soil S concentration were all in Amhara state which indicates that these locations produced pale white teff grains and for the lower soil S recorded in the other locations in both Amhara and Oromia regional states produced whiter teff grains based on the significant positive correlation between grain S color space value and soil S concentration. There was no strong association between grain proximate composition parameters (grain fiber, fat, CP, starch) and grain color in this study. The grain amino acid concentration of teff was only associated with grain color S value particularly, for aspartic acid and alanine. Grain crude fiber was strongly associated with soil pH and Ca but not crude fat, crude protein, and starch. Also, in this study, soil physicochemical properties had minimal influence on grain mineral concentration and since those relationships were below the 50% threshold level, it is indicative that other factors had played a more significant role in altering grain mineral nutrient concentration. Soil P, Mg, and Na were the only soil properties associated with grain amino acids concentration but none of these relationships were considered strong and therefore, suggested that other major factors were involved in determining the level of grain amino acid concentration present in teff grain. Based on the results of this study, grain color and nutrient concentration were influenced by location within the two regional states of Amhara and Oromia and this may help dictate management interventions necessary to obtain uniformity in grain color, chemical composition, and market equity for the same variety of teff in these two major teff cultivation regions in Ethiopia.
Author Contributions: A.A. Graduate student work on the development of the proposal, implementation of the project, data collection, data analysis, and manuscript preparation.; E.A., the conceptualization of project from proposal development to implementation on fieldwork, and editing.; B.Y., involved in proposal development and fieldwork. G.A., involved in proposal development, statistical advising, and editing. K.A., involved in proposal development and fieldwork. J.K.Q.S.; data analysis, reviewing, and writing of the manuscript, W.P.; review and editing, supervision, and funding acquisition.
Funding: This research was funded by the College of Agriculture, Biotechnology and Natural Resources, University of Nevada, Reno.