Teff (Eragrostis tef) production constraints on Vertisols in Ethiopia: farmers' perceptions and evaluation of low soil zinc as yield-limiting factor

Abstract

Teff (Eragrostis tef (Zucc.) Trotter) is a major food crop in Ethiopia and Eritrea. It is well adapted to Vertisols. Yields are low (around 1000 kg ha⁻¹) despite fertilization with urea and diammonium phosphate. The objectives of this study were to understand farmers’ perception on teff production constraints and to evaluate on-farm yield response of teff to zinc (Zn) fertilization. We conducted a farm survey and a participatory fertilization experiment in three teff-based sites (peasant associations) on Vertisols in the mid highland and lowland agroecological zones in Ethiopia. Per site 10 farmers participated in the survey and on-farm experiment. Poor soil fertility in the mid highland and moisture deficit in the lowland agroecological zones were mentioned by farmers as major teff production constraints, respectively. On-farm application of Zn fertilizer at a rate of 8 kg Zn ha⁻¹ increased teff grain and straw yields by 14% and 15% on average, respectively, which could be economically profitable. Not all plots showed a positive response, however, indicating the necessity for enhanced insight in indicators for soil Zn bioavailability as a yield-limiting factor. Our study indicates the importance of Zn in teff production on Vertisols. We propose further research on management options to prepare for effective interventions based on the farm survey and on-farm experiment.

Keywords:

• Eragrostis tef (Zucc.) Trotter
• farmers’ perception
• participatory research
• Vertisol
• zinc.

Introduction

Teff (Eragrostis tef (Zucc.) Trotter) is a small-grained cereal that has been grown as food crop in East Africa for thousands of years (D’Andrea 2008). It is a staple food for the majority of the population in Ethiopia and Eritrea. Teff is adapted to a large variety of environmental conditions and widely grown from sea level up to 2800 m above sea level (a.s.l.) under various rainfall, temperature, and soil conditions (Seyfu 1997). It is cultivated in Ethiopia on about 2.59 × 10⁶ha and occupies about 28% of the total crop area allocated to cereals (Anonymous 2010a), delivering about 20% of the total cereal grain production and 17% of cereal crop residues production annually (Anonymous 2010b; FAO 1987). The average teff grain yield of 1228 kg ha⁻¹ (Anonymous 2010b) is low compared to other cereals, which is attributed to nutrient limitations, drought and water logging (Tulema et al. 2005). Farmers using improved cultivars and management practices, however, can obtain yields up till 2500 kg ha⁻¹ (Tefera and Belay 2006), while the yield potential under optimal management and when lodging is prevented, is as high as 4500 kg ha⁻¹ (Teklu and Tefera 2005).

This paper focuses on teff grown on Ethiopian Vertisols to which it is ecologically well adapted due to its tolerance to water logging (Efrem 2001). Vertisols comprise 2.5% of the world's total land area, with major areas in India, Africa and Australia. They occupy about 105 × 10⁶ha in Africa and about 12.6 × 10⁶ ha in Ethiopia (Blokhuis 1982; Debele 1985) and around 5.9 × 10⁶ha is in use by farming systems in which teff is an important crop (Bull 1988). Vertisols are dark, montmorillonite-rich clay soils with characteristic shrinking and swelling properties. They have high clay content (>30% to at least 50 cm depth from the surface) and when dry they show cracks of at least 1 cm wide and 50 cm deep (FAO 2000). They have high calcium and magnesium contents.

Nitrogen (N) and phosphorus (P) fertilizers are the major soil fertility management practices applied by Ethiopian smallholder farmers growing teff on Vertisols. These fertilizers were introduced in 1967 following four years of trials carried out by the government, with the assistance from the Food and Agriculture Organization's (FAO's) Freedom from Hunger Campaign (FAO 1995). Continuous application of N and P fertilizers without due consideration of other nutrients may have led to deficiencies of e.g. potassium (Astatke et al. 2004) or zinc (Zn). It is suspected indeed that cereals’ response to N and P fertilization is declining on central mid to high altitude Vertisols (T. Mamo, Ministry of Agriculture and Rural Development of Ethiopia, February 2008, personal communication) indicating factors other than N and P might potentially be growth-limiting.

Zinc could possibly be a yield-limiting factor on Vertisols (Dang et al. 1993; Ahumada et al. 1997; Rupa and Tomar 1999; Sharma et al. 2006). It can be adsorbed to clay minerals, metal(hydr)oxides and organic matter, specifically at high pH (Catlett et al. 2002; Gao et al. 2010). In addition, Zn deficiency may have a soil physical cause, related to the limited plant rootability of Vertisols or to low transport towards the root due to drought. Contrasting information is available on the Zn status of Ethiopian Vertisols: Syers et al. (2001) reported that Zn deficiency can be common in Vertisols, Desta (1983) reported considerable variation in micronutrient contents in soils in Ethiopia, Itanna (1996) reported low availability of Zn on Vertisols and according to Haque et al. (2000) the decision on (in)sufficiency depended on the extractant used.

Zinc, being a micronutrient for plants, animals and humans, would not only be relevant as a yield-limiting factor, but also in relation to quality of food and feed. Zinc deficiency is widespread in Ethiopia's human (Umeta et al. 2000; Hotz and Brown 2001; Umeta et al. 2003) and cattle (Khalili et al. 1993) populations that depend on teff grain and straw, respectively, as major source of Zn. Teff Zn concentration has been reported to be low (Umeta et al. 2005). This could be partly due to low soil Zn supply.

Information on farmers’ soil fertility management on Vertisols and their perceptions on production constraints of teff is limited. In this study we describe teff farming systems on Vertisols in what has been classified as mid-highland and lowland agroecological zones (Anonymous 2000) and evaluate Zn as possible growth-limiting factor. To design future interventions we interviewed farmers to understand their perceptions on teff production constraints and soil fertility management, and we performed a participatory on-farm Zn fertilization experiment.

Materials and Methods

Site description

A farming system survey and an on-farm experiment were conducted in northern and central Ethiopia in 2008. The sites were selected upon advice of Debrezeit, Axum and Alamata Agricultural Research Center Managers and Extension Office Heads at district level. They suggested three peasant associations (sites) in teff growing areas where Vertisols are the dominant soil type. The three selected sites differ in management practices and are located in the sub-moist mid-highland (Ude and Hatsebo) and in the hot to warm lowland (Selam Bikalisi) agroecological zones of Ethiopia (Anonymous 2000) (Table 1; Fig. 1). Hatsebo is not located in a main Vertisol area (Fig. 1), but the farmers’ plots were on small Vertisol pockets (Gebretsadikan et al. 2009) characterized by a high clay content and visible cracks (width >1 cm, depth >50 cm). In all three sites teff is grown during the main rainy season from June to September (Fig. 2). The teff growing season of 2008 had an average rainfall in Ude and Hatsebo but rainfall was relatively low in Selam Bikalisi (Fig. 2). The rainfall and temperature data for the study sites were summarized from the nearest district data (National Meteorology Services Agency, unpublished data). Potential evapotranspiration was estimated using the LocClim 1.0 software (FAO 2002).

Figure 1. Map of Ethiopia with the three study sites. The major areas with Vertisols are in grey (FAO et al. 1998). Note that Vertisols in the Hatsebo study site are not in what is considered a main Vertisol area but have been studied and considered true Vertisols (Gebretsadikan et al. 2009).

Figure 2. Average monthly rainfall (solid lines and squares), actual monthly rainfall in 2008 (bars) and estimated potential evapotranspiration (dashed lines and triangles) for (a) Ude, (b) Hatsebo, and (c) Selam Bikalisi. Arrows indicate teff sowing (solid) and harvest (dashed) in 2008.

Table 1. Description of the study sites

	Site
	Ude	Hatsebo	Selam Bikalisi
Region	Oromia	Tigray	Tigray
District	Ada liben	Laelay Michew	Alamata
Altitude (m above sea level)	1886	2083	1500
Latitude, longitude	8°40′N, 39°01′E	14°50′N, 38°40′E	12°23′N, 39°33′E
Average annual rainfall over past 20 years (mm)	859	763	782
Mean maximum temperature over past 20 years (°C)	27	28	30
Mean minimum temperature over past 20 years (°C)	12	11	15

Farming system inventory

Ten farmers were interviewed at each site before the start of the teff growing season. Farmers were selected upon advice of development agents. The participating farmers were supposed to represent farmers with a relatively high exposure to extension services. They were selected because they are the most probable innovators and adopters of new technologies. They were 29 male and one female (from Selam Bikalisi).

Each farmer was asked to list and rank the major production constraints of teff on his/her fields. For each production constraint mentioned we counted the number of farmers who mentioned it among the top three constraints.

On-farm zinc fertilization experiment

At the beginning of the growing season all selected farmers agreed to provide a 100 m² (10 m × 10 m)-experimental plot in one of their teff fields on Vertisol, based on their preference. All plots were on level fields.

Before planting 10 surface soil samples (0–20 cm depth) were collected from each plot in a diagonal pattern using an auger. These samples were bulked into one composite sample. The composite soil samples were air-dried, cleaned from stones and plant residues, ground—using pestle and mortal—and passed through a 2-mm sieve. Sub-samples were taken after thorough mixing with a stainless steel spoon. The soil sub-samples were analyzed for pH (Jackson 1967), organic carbon [modified Walkley and Black method (Jackson 1967)], texture [hydrometer method procedure of Bouyoucos (Day 1965)], available phosphorous (Olsen et al. 1954) and Zn (DTPA-Zn; Lindsay and Norvell 1978). The Zn concentration in the extract was analyzed using an atomic absorption spectrophotometer (Varian Spectra AA 240 FS).

Teff was grown in each farmer-managed plot. Half of each plot (5 m × 10 m) received Zn fertilizer at a rate of 8 kg Zn ha⁻¹ in the form of ZnSO₄ · 5H₂O. The rate of 8 kg ha⁻¹ Zn was based on a preliminary field experiment in 2007 on Vertisols in Ethiopia (unpublished data). In the presence of the researcher the Zn fertilizer was broadcasted and incorporated in the top soil using a rake at the day of sowing. Nitrogen and P were applied to all experimental plots in Ude and Hatsebo in the form of urea and diammonium phosphate (DAP, 46% P₂O₅) according to farmers’ common practice. These fertilizers were not applied in the case of Selam Bikalisi because the participating farmers were neither used to nor willing to apply. The teff varieties used for the experiment were Dz-Cr-387 and Dz-Cr-196 in Ude and Hatsebo, respectively. Farmers’ varieties were sown in Selam Bikalisi. All other cultural practices were according to the participating farmers’ practices. Harvesting was conducted when grains were hard enough. A composite sample was made by combining plants from two random sample quadrants of 1 m × 1 m per sub-plot. Aboveground biomass per net plot area was weighed after 15 days sun-drying. The shoots were threshed and cleaned and the grain yield was weighed. The straw yield was calculated by subtracting grain yield from biomass yield. Yields are reported on air dry basis as these are the best indicators of marketable yields.

Farming system survey after experiment

After the harvest, individual interviews and a group discussion with the 10 farmers of each site were held on their farming system and crop management, using a semi-structured questionnaire and a semi-structured guide or check list, respectively. During individual interviews, the farmers described the major crops and livestock, crop residue utilization, crop rotation, possible crop rotation patterns, teff production constraints on their individual farm, characteristics of teff fields on Vertisols in relation to fertility status and production objectives and soil fertility management, and gave their opinion on the effects of the application of Zn fertilizer on their plots. The farmers assessed the yield of the Zn-fertilized and control plots while the crop was standing at maturity. Estimations for area, urea and DAP rates were based on the assumption that an oxen plough day is equivalent to a quarter of a hectare land. For those who applied manure or compost, application rates were estimated from the farmers’ indication on the weight and number of bags applied per unit area. The group discussions mainly focused on production constraints and were moderated by facilitators from the development agents and the research centers. The production constraints of teff were listed and ranked. While comparing the teff production constraints (two at a time), there was sometimes a debate before consensus was reached.

Data analysis

Both the quantitative and qualitative data from the individual interviews and soil were analyzed using SPSS 15.0 for Windows. Grain and straw yield data from the on-farm Zn fertilization experiment were analyzed considering site and Zn application rate as fixed effects, and farmers and their interaction with the fixed effects as random effects. The variance components and their maximum standard errors of difference were generated with GenStat 12.1 using the linear mixed models (REML) procedure.

Results

Soil properties of the farmers’ fields on Vertisols

Seven, ten and nine out of ten farmers’ fields on Vertisols were low in soil Zn (DTPA-Zn < 1.0 mg kg⁻¹) in Ude, Hatsebo and Selam Bikalisi, respectively. Mean DTPA‐Zn was low in all three sites (Table 2). Mean soil pH varied from neutral to moderately alkaline. The mean soil P status was high (P-Olsen > 10 mg kg⁻¹) in Ude and Hatsebo and low (P-Olsen < 10 mg kg⁻¹) in Selam Bikalisi. The mean soil organic carbon content was low (< 2%) in all sites.

Table 2. Soil properties of teff fields on Vertisols from the participating farmers

Soil parameters	Site
	Ude	Hatsebo	Selam Bikalisi
pHwater (1:2.5)	6.8 ± 0.1	7.8 ± 0.1	8.0 ± 0.1
Organic carbon (%)	1.3 ± 0.1	0.8 ± 0.0	1.2 ± 0.1
P-Olsen (mg kg^−1)	12 ± 2	11 ± 2	6 ± 1
DTPA-Zn (mg kg^−1)	0.92 ± 0.07	0.59 ± 0.02	0.54 ± 0.08
Clay (%)	57 ± 1	58 ± 1	41 ± 1
Data represent means ± standard error (n = 10).
P-Olsen, available phosphorous using procedure by Olsen et al. (1954).
DTPA-Zn, Zinc (Zn) determined by diethylene triamine pentaacetate extraction (Lindsay and Novell, 1978).

Farming systems of the study sites

The major farming system was a mixed cereal/legumes-livestock system in Ude and Hatsebo and a mixed cereal-livestock system in Selam Bikalisi (Table 3).

Table 3. Summary description of farming systems as provided by the participating farmers at the three sites (n = 10 per site)

Descriptor	Site
	Ude	Hatsebo	Selam Bikalisi
Farming system	Cereal/legume-livestock	Cereal/legume-livestock	Sorghum/teff-livestock
Major crops	Teff, wheat, chickpea, faba bean, grass pea and lentil	Teff, wheat, barley, chickpea, faba bean, grass pea, fenugreek, field pea and lentil	Sorghum and teff
Major livestock	Cattle, sheep, goat and donkey	Cattle, sheep, goat and donkey	Cattle, goat, sheep and donkey
Crop rotation	Yes	Yes	No
Rotation pattern on Vertisols	Year 1: Chickpea or grass pea Year 2: Teff or wheat	Year 1: Chickpea or fenugreek or grass pea Year 2: Teff or wheat	No
Rotation pattern on others soils	Year 1: Faba bean or lentil Year 2: Teff or wheat	Year 1: Faba bean or lentil or field pea Year 2: Teff or wheat or barley	No
Crop residue removal	Yes	Yes	Yes
Average own land holding (ha)	2.3 ± 0.2	1.1 ± 0.2	0.8 ± 0.1
Average operated land 2008† (ha)	3.7 ± 0.4	1.7 ± 0.2	1.6 ± 0.2
Average plot size of teff fields on Vertisols 2008 (ha)	0.74 ± 0.10	0.38 ± 0.06	0.40 ± 0.05
Average livestock holding, TLU‡	8.5 ± 1.0	3.4 ± 0.6	3.6 ± 0.8
†Own plus rented or shared land.
‡Tropical livestock unit.

Farmers in Ude and Hatsebo were well aware of the importance of crop rotation to replenish soil fertility and skillfully used this option. They choose the crop to be grown in rotation depending on rainfall (total amount and distribution), soil type, market demand for the legumes, expenditures such as loan and ceremonial as well as the need to meet their food and livestock feed requirements. Farmers prefer to grow chick pea and grass pea as leguminous rotation crops and wheat as cereal rotation crop in both sites and in addition fenugreek in Hatsebo in rotation crop with teff on Vertisols. The leguminous rotation crops are usually planted towards the end of August and partly grow on residual moisture. Farmers in Ude and Hatsebo reduce their N fertilizer rate for teff after leguminous crops, mostly for one year. Farmers in Selam Bikalisi are not practicing crop rotation. The main reason is traditional land allocation to teff and sorghum, based on seasonal soil moisture conditions and soil fertility: relatively fertile lands receiving seasonal floods from upstream are allocated to sorghum.

Farmers in all three sites carry crop residues away from their field. The aftermath is grazed by their animals. The major uses of crop residues are livestock feed, sale and construction purposes. Farmers in Hatsebo and Selam Bikalisi take part of their straw to the nearby markets of Axum and Alamata, respectively, where it is sold to feed urban livestock. Farmers in Ude sell part of their teff straw to wholesalers who in turn sell the crop residue for urban cattle fattening and dairy farms in the nearby towns of Debrezeit and Nazreit. Some of the crop residue is sold in towns bordering Djibouti and Somalia. Crop residue, especially of teff, is also used locally for plastering after mixing with clay, and as manure fuel cakes in all three sites. Rare farmers in Hatsebo use crop residues to make compost particularly from the remains of livestock feed. Farmers in Selam Bikalisi use crop residue of sorghum for roofing and fuel. Here, sorghum stumps are also an important source of fuel for both the rural and nearby urban dwellers.

The average farm size of teff fields on Vertisols and livestock holding were higher in Ude than Hatsebo and Selam Bikalisi sites (Table 3). In all sites, the average operated land of the farmers involved in the study in 2008 was greater than the average owned landholding.

Farmers’ perception on teff production constraints

At the first individual interview, before any hint to Zn as a potential yield-limiting factor was made, poor soil fertility (Ude and Hatsebo) and moisture deficit (Selam Bikalisi) were mentioned among the three major constraints of teff production by all participating farmers (Fig. 3). In Ude and Hatsebo, the high cost of fertilizers ranked second.

Figure 3. Initial individual farmers’ identification of teff production constraints at the three investigated sites.

In the group discussions organized per site after the Zn fertilizer experiment, poor soil fertility as constraint was again ranked first both in Ude and Hatsebo. Farmers in both sites noticed that the yield response to urea and DAP application was less than they remembered from the past. Farmers tried to put this into words by expressing the feeling that their land “is getting old”. High cost of fertilizer was ranked as the second most important constraint in both sites. Weeds in Ude and pests in Hatsebo were ranked as third most important constraints. The results were almost identical to the initial individual interviews as reported in Fig. 3.

Farmers in Selam Bikalisi ranked moisture deficit (rainfall amount and pattern) as the leading teff production constraint followed by pests as second and weeds as third constraint. This was again identical to the results of the initial individual interviews (Fig. 3).

Farmers’ classification and use of teff fields on Vertisols

Farmers were asked to classify their teff fields on Vertisols according to fertility and production objective. Farmers in all the three sites classified the more productive fields as fertile. Farmers in Ude and Hatsebo tended to apply more chemical fertilizer to what they consider their more productive fields in order to get more yield. Farmers classified the majority of their teff fields on Vertisol as fertile in Ude, against medium to fertile in Hatsebo and Selam Bikalisi (Table 4).

Table 4. Classification of the different teff fields on Vertisols during the 2008 cropping season as provided by the 10 participating farmers per site

Plot characteristics	Class	Site†
		Ude (28)	Hatsebo (18)	Selam Bikalisi (21)
Soil fertility	Unfertile	0	7	1
	Medium	3	3	9
	Fertile	25	8	11
Production objective	Market	23	6	1
	Home consumption	2	8	9
	Both purposes	3	4	11
†Figures in parentheses indicate the number of teff fields evaluated.

Teff production was mainly market-oriented in Ude, mainly home consumption-oriented in Selam Bikalisi and while farmers roughly considered market and home consumption equally important in Hatsebo (Table 4).

Farmers’ soil fertility management on teff fields on Vertisols

Farmers were used to apply chemical fertilizers to all teff fields in Ude and Hatsebo. The average rates in 2008 were 73 and 86 kg ha⁻¹ urea, and 101 and 82 kg ha⁻¹ DAP respectively (Table 5). No chemical fertilizer was applied to any teff field in Selam Bikalisi. The main reason not to apply any chemical fertilizer was the perception that moisture deficit was so strongly limiting yield that fertilizer application was not worth the investment.

Table 5. Farmers’ soil fertility management to teff fields on Vertisols during the 2008 cropping season

Type	Site†
	Ude (28)	Hatsebo (18)	Selam Bikalisi (21)
Urea	73 ± 4 (28)	86 ± 10 (18)	0
DAP‡	101 ± 8 (28)	82 ± 11 (18)	0
Compost	930 (1)	2000 (3)	1000 (1)
Manure	0	2000 (1)	1160 ± 275 (6)
Data (in kg ha^−1) represent means ± standard error.
†Figures in parentheses indicate the number of teff fields evaluated.
‡Diammonium phosphate (46% P2O5).

On-farm yield response of teff to zinc fertilization

There was no interaction between Zn fertilizer application and site for either grain or straw yield of teff. Average grain and straw yields of teff were significantly (P < 0.001) increased by application of 8 kg ha⁻¹ Zn with 14% and 15% over the control, respectively (Table 6). Regardless of Zn fertilizer application, grain and straw yields of teff were lowest in Selam Bikalisi. A comparison of the appreciation of the Zn fertilizer effects by farmers on the basis of visual inspection prior to harvesting and the measured effects showed correspondence on 12 and 13 plots for grain and straw yield respectively (Table 7). On the other plots farmers appreciated yield effects differently from actual, measured effects. On two or three of the plots appreciation for grain and straw yield, respectively, were opposite to measured effects.

Table 6. Main effects of zinc fertilizer and site on grain and straw yields of teff studied on 10 farmers’ Vertisol fields per site

Independent variable	Level		Yield (kg ha^−1)
			Grain	Straw
Zinc	Control		1675	3921
	8 kg ha^−1		1914	4515
		P-Value	<0.001	<0.001
		SED†	57	151
Site	Ude		2002	5256
	Hatsebo		1853	4643
	Selam Bikalisi		1528	2756
		P-Value	0.08	<0.001
		SED†	203	383
Zinc × site		P-Value	0.86	0.55
†SED, standard error of the differences.

Table 7. Matrix showing yield effects for both grain and straw of the application of 8 kg zinc (Zn) ha⁻¹ to teff plots in a participatory field experiment on 10 farmers fields each in three sites comparing measured effects and effects as indicated by farmers on the basis of visual inspection prior to actual harvesting

	Measured	Indicated by farmers
		Positive	No effect	Negative
Grain	Positive†	9	3	2
	No effect	10	3	3
	Negative†	0	0	0
Straw	Positive†	9	2	2
	No effect	10	4	2
	Negative†	1	0	0
†More than 198 kg ha^−1 and 523 kg ha^−1 difference for grain and straw, respectively, between the Zn-fertilized and control plot (i.e. 2 × standard error of the differences of the treatment by location interaction effect at the same level of location in the Zn fertilization experiment).

Discussion

To our knowledge, this is the first paper identifying low soil Zn as a teff yield-limiting factor on Vertisols under farmers’ conditions. The on-farm experiments showed a significant average yield increase upon Zn fertilization at all three sites investigated (Table 6) and no interaction between the Zn application and location. At two of the three sites farmers indicated poor soil fertility as a major production constraint, both prior to and after the experiments. This offers excellent perspectives for interventions that may not only lead to increased crop production, but also to improved quality of a major staple food and feed for populations at risk of Zn deficiency.

Farmers’ perception in Ude and Hatsebo on poor soil fertility as the major teff production constraint (Fig. 3) corresponded well with the results of our soil analyses and the results of the on-farm experiment. The soil organic carbon content was low (Table 2), probably because of the continuous and total removal of residues from all crops in the rotation. The mean soil P status was high after long-term application of considerable amounts of DAP (Table 5). The mean soil Zn status, however, was low (Table 2) and teff grain and straw yields increased significantly upon Zn fertilization (Table 6). Farmers’ perceptions on poor soil fertility and their statements that their land “is getting old,” despite application of urea and DAP, may therefore be related to depletion of their Vertisols from bioavailable Zn. We cannot exclude, however, that other nutrients may be depleted and have become yield-limiting as well. Additional trials would be needed to find out.

Regardless of the Zn fertilizer application, grain and straw yields of teff were lowest in Selam Bikalisi. This is probably related to the lower availability of P (Table 2), water (Fig. 2) and N due to the absence of legumes in the rotation (Table 3). Yet, there was an average increase in yield upon Zn fertilization and a positive response on half of the plots (Table 7) despite the absence of any history in urea and DAP fertilization that could have resulted in depletion of the soil from Zn, and even though 2008 was considered as a relatively dry year (Fig. 2). This could mean that Zn bioavailability is inherently low in these Vertisols which is in line with previous case studies from other Vertisol areas in the world (Dang et al. 1993; Ahumada et al. 1997; Rupa and Tomar 1999; Sharma et al. 2006). Consequently, also other crops grown on these soils are at risk of Zn deficiency, even at low N and P fertilizer rates.

Average yield increases of 239 kg ha⁻¹ grain and 594 kg ha⁻¹ straw were recorded as a response to 8 kg ha⁻¹ Zn fertilization (Table 6). Taking into account the local market prices of teff grain [700–1000 US dollars (USD) ton⁻¹; Berhe 2009] and teff straw [49–150 USD ton⁻¹, as converted from local currency data provided by Gebremedhin et al. (2009)], the additional production would mean an additional income of 167 to 239 USD ha⁻¹ from grain and of 29 to 89 USD ha⁻¹ from straw. The current price of Zn sulphate (heptahydrate, 21% fertilizer grade) is 520 USD ton⁻¹ (cost, insurance and freight) at the ports in Djibouti or Sudan [W. Chen, Liuzhou Glory Zinc Minmetchem Co., Ltd. (Guangxi, P.R. China), April 2010, personal Communication], which for other inputs is roughly half of the farm gate price in Ethiopia (Jayne et al. 2003). This would indicate an average benefit/cost ratio for Zn fertilization between 5:1 and 9:1, costs for application not included. This is a ratio generally considered attractive for market-oriented farmers, who are the majority among the participants (Table 4). Agronomical optimization of the ratios of N/P/Zn fertilizer could lead to a reduction of N and P fertilizer rates in favor of Zn, which would further improve the cost/benefit ratio of Zn fertilizer application.

There was a fair correspondence between observed positive yield effects on teff plots and the farmers’ assessment (Table 7). There were, however, also ample false positive assessments of the Zn fertilizer effect on grain and straw yield. On two or three of the plots appreciation for grain and straw yield, respectively, were opposite to measured effects. This implies that simple demonstration plots may not be enough to inform farmers about effects especially as false positive assessment would imply an additional financial risk to farmers. It should be noted that the participating farmers are probably not representing the majority of subsistence teff growers in the studied areas. They were selected because they are relatively well exposed to extension services and in conjunction their yields are relatively high. Further research and intervention strategies should not only focus on this category of farmers but consider how to scale out by involving other teff growers to learn from these potential early adopters.

Not all plots showed a positive yield response to Zn fertilization (Table 7). One plot showed a negative straw yield response, while on roughly half the plots no yield effect was observed. The possible reasons could be variation in actual Zn limitation related to variation in bioavailability, interference with other yield-limiting factor(s) or/and variation in handling and management of the experimental plots by the farmers. This implies that blanket recommendations are bound to lead to disappointments and these are thus inappropriate. Additional insight in predictors of success of Zn fertilization that are plot specific should be sought to support possible introduction of Zn fertilization practices, in addition to verification studies over years and more locations.

We also suggest further research on optimum N, P and Zn utilization including options such as seed coating/priming and use of Zn-enriched fertilizers in sites similar to Ude and Hatsebo. In Selam Bikalisi-like sites, with low soil fertility and current levels of manure and fertilizer application to teff fields (Table 5), integrated organic and inorganic soil fertility management could be an entry point for further research.

Acknowledgments

The authors are grateful to the Debrezeit, Axum and Alamata Agricultural Research Centers and Extension Offices, participant farmers and development agents in Ude, Hatsebo and Selam Bikalisi, the Mekelle Soil Laboratory Staff and Yirgalem Weldegebriel from Ezana Mining PLC for their support. The authors thank Professor Dr Oene Oenema (WU) and Professor Dr Tekalign Mamo (Ministry of Agricultural and Rural Development, Addis Ababa) for their constructive ideas. The authors would like to thank Dr Alfred Hartemink (ISRIC-World Soil Information) for supplying digital soil data and Dr Arnaud Temme (WU) for making the map with the study sites. This study was funded by the Wageningen University sandwich PhD and INREF “From Natural Resources to Healthy People” programs and the Tigray Agricultural Research Institute.