Identification of soil properties influencing DRIS indices of maize in the savannas of northern Nigeria

doi:10.21203/rs.3.rs-1720300/v1

Download PDF

Research Article

Identification of soil properties influencing DRIS indices of maize in the savannas of northern Nigeria

https://doi.org/10.21203/rs.3.rs-1720300/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Aims

Improper nutrient management reduces yield and affects the nutrient status of crops. This study aimed at diagnosing nutrient limitation in maize.

Methods

A three-year multi-location (348 sites) nutrient experiments were conducted in randomized blocked design to analyse nutrient limitations for maize production under conventional fertilizer recommendation system in Nigeria using DRIS and to identify soil factors that influence DRIS indices using random forest model. DRIS indices for nutrients were calculated from the results of ear leaf samples collected from the experimental plots. The DRIS indices were summed, and used to cluster the plots using k-means cluster algorithm

Results

Results show large differences in average yield between the clusters. The clusters also differed based on frequency to which nutrients are most limiting. B was the most limiting in cluster one and three, Mn in cluster two and K in cluster four. Random forest results show that soil pH, B and Mg had the largest influence on DRIS indices in cluster one. DRIS indices were most influenced by soil N and B in cluster two. To lesser extent, soil contents of Fe, K, Mg and S also influenced DRIS indices in cluster two. Soil K, B and Zn were the most significant factors that influenced DRIS indices in cluster four. Bulk Density, Fe, Na, ECEC, and organic carbon had a moderate influence on the indices in this cluster.

Conclusions

Nutrient limitation in plants can be diagnose using DRIS method. Soil properties have a definite influence on maize nutrient status.

DRIS

Soil

ear leaf

nutrient balance

maize

The area where maize is cultivated in Nigeria has remarkably increased over the last three decades (FAOSTAT 2019) due to availability of different stress-tolerant varieties (Adnan et al. 2017; Kamara et al. 2019). Consequently, maize cultivation has become more preferred among dominantly sorghum and millet farming communities. The northern Nigeria region (the savannas) is the foremost maize production zone in the country due to relatively more favorable production conditions (Shehu et al. 2018). The region is characterized by sufficient solar radiation, optimum temperature and well-defined rainy season which allows for adequate drying of maize. This region produces a combined 80% of the 10–11 MMT of maize annually produced in the country (Knoema 2021).

While several studies (Garba et al. 2020; Rurinda et al. 2020, Kamara et al, 2019, Shehu et al. 2019) have reported potential yields of > 6 t ha^-1 in experimental fields particularly within the savannas of Nigeria, national average maize yield for long has been ranging from < 1.8 to 2.5 t ha^-1; equivalent to < 30% of the potential yield of the area (FAOSTAT 2019). Low soil fertility in the area (Jibrin et al. 2012; Aliyu et al. 2020) which is compounded by improper management strategies to maintain soil nutrient stocks, and to improve yield response to nutrient application (Shehu et al. 2018; Rurinda et al. 2020) are considered the most critical factors limiting maize productivity in the region and in Nigeria as whole. Following existing blanket recommendation which ignores spatial variability in soil nutrient (or soil fertility) status and the sub-optimal application of fertilizers may further contribute to depletion of soil nutrient stock and low fertilizer use efficiency in the region (Shehu et al. 2018; Garba et al. 2020). Studies by Shehu et al (2019), Rurinda et al (2020) and Garba et al (2020) showed that fertilizer recovery efficiency was < 20% for nitrogen (N), < 15% for phosphorus (P), and < 30% for potassium (K) in some parts of the northern savanna areas. Low nutrient use efficiency reflects that a substantial portion of the nutrient(s) applied through fertilizer is not taken by the crop and either lost through leaching or being fixed in the soil.

Although the blanket recommendation acknowledges the application of secondary macronutrients and micronutrients (SMNs), however, most of emphasis are directed mainly towards three main essential macronutrients; nitrogen, phosphorous and potassium (FFD 2012). In long-term experiments in the savannas of Nigeria, Togo and Benin, Nziguheba et al (2009) reported that Ca, Mg and Zn are principally deficient as indicated by their strong negative DRIS indices. Biwe (2012) also similarly reported that the levels of Zn (0.26 mg/kg) and Cu (0.36 mg/kg) in soils of Bauchi state of North-eastern Nigeria, are far below the levels required for maize production. Weil and Mughogho (2000) reported the deficiency of S in most parts of the Guinea savanna of Nigeria, especially under the current intensifying agricultural systems. This is manifested in the suboptimal or lack of response by crops like maize to NPK fertilizers leading to a reduction in crop quality and yields.

Addressing the low maize yield in Nigeria should therefore target identification of the limiting soil nutrients and applying them in a balanced manner. Soil testing can indicate potential soil supplying capacity of nutrients, however, it is not a good in indicator of actual nutrient uptake by plants, as other factors interplay with nutrient uptake process (Fixen et al. 2005). Therefore, to effectively diagnose limiting plant nutrients, soil test needs to be complemented with plant tissue nutrient diagnosis (Roy et al. 2006; Mugo et al. 2020). Plant tissue nutrient diagnosis is based on the fact that maximum yields are associated with an optimum concentration of nutrients in the plant tissue (Maia 2012; Mangale et al. 2016). But because nutrient concentration in plant tissues is growth stage dependent (Reuter et al. 1997), and also affected by the availability of other nutrients (nutrient interaction), nutrient diagnosis based on single nutrient concentration (absolute concentration), e.g., the case of critical value approach and sufficiency range, cannot be confidently applied to diagnose plant nutrient disorder (Beaufils 1987). To overcome this defect, a bivariate nutrient diagnosis method “Diagnosis and Recommendation Integrated System” (DRIS) was proposed by Walworth and Sumner (1987). The DRIS methods diagnose nutrients in dual ratios (relative concentration), to reflect nutrient status as a function of interaction with other nutrients (Walworth and Sumner 1987). Also, because nutrient ratios in plants are fairly constant throughout crop development stages, the sensitivity of changes in nutrient concentration due to plant age is reduced using the DRIS (Singh et al. 2000; Harger et al. 2003).

Moving away from the conventional method of generating fertilizer recommendation alone based on yield response to fertilizer application, soil or foliar analysis respectively, this study aims to complement the tissue diagnosis using the Diagnosis and Recommendation Integrated System (DRIS) with soil analysis under the conventional blanket fertilizer recommendation of NPK in the savanna zones of northern Nigeria. Combining the two approaches will give the opportunity of applying nutrients in a balanced manner, and promote synergistic uptake of plant nutrients for better productivity of crops and soils. The specific objectives of this paper were: (i) to diagnose the nutritional status of maize under conventional blanket NPK recommendation, and (ii) to identify the principal soil properties influencing the maize DRIS indices in the savanna zones of northern Nigeria.

Description of the study area

The studies were conducted in the southern Guinea, northern Guinea and the Sudan savanna zones across three States in northern Nigeria (Kano, Kaduna and Katsina States). The Sudan savanna (SS) has a length of growing season of around 120 days lasting from May to October with total of 753 ± 171 mm of rain over 57 ± 9 rainy days. Average minimum and maximum temperatures are 20.0 and 33.7 ⸰C respectively. The SS is bordered by the northern Guinea savanna (NGS) in the southern extremes. The NGS climate is characterized by a longer growing season and larger number of rainy days (63 ± 9 days) and higher total annual rainfall amount (998 ± 133 mm) than the SS. It has a growing season of 140 days. Average minimum and maximum temperatures are respectively 19.2 and 31.6 ⸰C. The southern Guinea savanna (SGS) is the wettest zone of all the savannas with an annual total rainfall of 1541 ± 270 mm and length of growing season of over 150 days covering April to October. Average minimum (21.1 ⸰C) and maximum (32.4 ⸰C) temperatures are similar to that of the SS. Description of the zones are based on 30-year averages from 1980–2009 as reported by Tofa et al (2021). These zones together present the most suitable maize growing area in Nigeria and are part of the larger maize belt of Nigeria as described by Aliyu et al (2020).

Experimental procedures

Two different field experiments were conducted across the study area from 2015–2017 rainy seasons. The first set of experiment was the nutrient omission trials (NOTs) that were conducted is the three consecutive experimental years. The sites for the NOTs were selected such that the varying maize cropping conditions across the study area are well represented. The site selection was explicitly reported in Shehu et al (2018, 2019) for 2015 and 2016. Modification of the spatial sampling frame work for 2017 is reported in Aliyu et al (2021).

Ninety-five (95), 103 and thirty (30) experiments were established on farmer fields in 2015, 2016 and 2017 growing seasons respectively. The experiments in 2015 and 2016 consisted of six nutrient treatments, which comprised of a Control (no nutrient application), PK, NK, NP, NPK and NPK + treatment which had Mg, Ca, S, Zn and B nutrients added to NPK. In 2017, the NPK + treatment was split into NPKS, NPKB, NPKZn and NPKSZnB in addition to the treatments in 2015 and 2016. During each year, the N, P and K nutrient were applied uniformly at 140 kg N ha^− 1, 50 kg P₂O₅ ha^− 1 and 50 kg K₂O ha^− 1 respectively at all trial sites. Nitrogen (N) was applied in three equal splits, i.e., at planting (basal), at 21 and 42 days after sowing (DAS), while full dosages of P and K were applied at planting. The nutrients S, Ca, Mg, Zn and B were basally applied at the rates of 10–24, 10, 10, 5–10 and 5 kg ha^− 1 respectively. The maize variety SAMMAZ 15 is the most widely adopted variety within the experimental area because of its’ tolerance to drought, Striga and maize streak virus infestation was used throughout the study. Two seeds per hole were sown at 0.25 m spacing, and later thinned to one plant per in all the studies.

In each year, each treatment plot consisted of six ridges constructed 0.75 m apart, each measuring 5 m long given a plot area of 22.5 m². Yield was estimated from a plot area of 9 m², defined by disregarding 1 row from each side of the plot and 1 m from either row side of the four middle rows. All cobs and stover in the net plot area were harvested and weighed fresh. Five cobs were then sub-sampled at random for determining moisture content, shelling percentage, and harvest index. Grain yield (in kg ha^− 1) was expressed on a dry weight basis at 15.0% moisture content adjustment using a grain moisture tester.

In the second experiment, a fertilizer response trial (FRT) was established in the 2017 rainy season. This experiment was conducted across the larger maize belt of the Nigeria savannas across eight States with site selection procedure fully described by Aliyu et al (2020). The trial was established in 935 farms from which 120 sites which belong to the focal area for this study were considered in this study. The FRT treatments included NPK, NPKSZnB, NPSZnB and a Control where no nutrient was applied. The whole plot for the NPK was made up of 20 rows of 10 m × 15 m lengths spaced at 0.75 m. The net plot was determined by leaving out the first two and last two rows of each plot and 1 m each from both ends of each row. Thus, maize yield was estimated from a plot area of 8 m × 12 m = 96 m². The same maize variety used in the NOTs was used in this experiment. Planting, fertilizer application and crop management practices were the same as the NOTs.

The NPK treatment was common between the two studies because it is the dominant/recommended fertilizer management practice in the study area. Therefore only data of the NPK treatment across the experiments were used and reported in the study.

Soil and ear leaf sampling and analysis

Prior to establishment of the fields, soil samples were collected at 20 cm depth from representative spots at each experimental site and analyzed for physical and chemical properties. Total soil organic carbon (OC_tot) was determined using modified Walkley & Black chromic acid wet chemical oxidation and spectrophotometric method (Heanes 1984). Total nitrogen content (N_tot) was measured using micro-Kjeldahl digestion method (Bremner, 1996). The pH (soil/water ratio of 1:1) was measured using a glass electrode pH meter. Available phosphorus (P_av), available sulphur (S_av), exchangeable cations (K, Ca, Mg and Na) and micronutrients (Zn, Fe, Cu, Mn and B) were extracted by Mehlich-3 procedure (Mehlich 1984) and read through inductively coupled plasma optical emission spectroscopy (ICP-OES, Optima 800, Winlab 5.5, PerkinElmer Inc.,Waltham, MA, USA). Effective cation exchange capacity (ECEC) was calculated as the sum of exchangeable cations (K, Ca, Mg and Na) and exchangeable acidity (H + Al). Soil texture was analyzed using hydrometer method (Gee and Or 2002).

For ear leaf analysis, ten maize ear leaves were randomly collected from the second and fifth rows immediately at the beginning of silking stage (female flower initiation stage) in the NOTs. For the FRT, first ear leaf was sampled from a plant selected arbitrarily at the centre of the plot. The second and third ear leaves were sampled from the fifth plants to the left and right in reference to the first sampled plant. Two rows perpendicular to the first sampled row from both sides were selected and same procedure was repeated. The tenth sample was randomly collected from within the plot.

The samples were washed with distilled water and air dried. The dried samples were then ground with agate pestle and mortar and analysed for nutrient contents. Nitrogen was analyzed by digesting the samples in hot sulphuric acid solution in the presence of Se as catalyst, followed by colorimetric N analysis using autoanalyzer (Technicon AAII, SEAL Analytical Inc.) following indophenol blue method. For the determination of Sulphur, ball-milled samples were digested with nitric acid (HNO₃) and the nutrient contents in the digest were determined in Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES Optima 3300 DV, Perkin Elmer, Norwalk, USA). Phosphorus (P), K, Ca and Mg and micronutrients (Zn, Fe, Cu, Mn and B) were analyzed by first dry-ashing the samples for four hours at 550°C and then prepared and read on ICP-OES Optima 800, Winlab 5.5 (manufactured by PerkinElmer Inc., Waltham,MA, USA).

Data analysis

DRIS analysis

Diagnosis and Recommendation Integrated System (DRIS) (Beaufils 1987) was used to assess the nutrient balance index in maize using the results of the ear leaf analysis. First step in DRIS analysis is the establishment of DRIS nutrient norms. The norms are the average nutrient pair ratios of the high yielding (reference population). The reference population was determined by sorting the data according to yield in decreasing order and a cut-off yield was determined. In this study, the cut-off point was determined at mean yield + 0.5 × standard deviation (Aliyu et al. 2021). Using this criterion, plots with yield ≥ 5645 kg ha^− 1 were considered high yielding and were used as the reference. The reference sub-population constituted 27% of the entire dataset. The Mean value for each nutrient pair, their corresponding coefficient of variation (CV), and variance (σ²) were then calculated separately for the two sub-populations. The mean value of each nutrient pair ratio in the high-yielding population were considered as the norms (Walworth and Sumner 1987). For calculating the DRIS index, we expressed all possible forms of nutrient pair expressions i.e., A/B, B/A and A×B. Accuracy of DRIS diagnosis depends on the variability of the nutrient pair ratios for the high versus the low yielding sub-populations. Thus, we again calculated the variance (σ²) for each form of nutrient pair expression separately for each sub-population. It is hypothesised that the data of the low yielding sub-population is more imbalanced and therefore should have larger variance than the high yielding one. Therefore, we divided the variance of the low yielding sub-population by that of the high yielding sub-population. Finally, the nutrient pair ratio expression that present the highest variance ratio between the low and high yielding sub-population was selected among the three nutrient pair expressions, and was used for calculating the DRIS index. The DRIS index for each nutrient was calculated as bivariate relationship between that nutrient and all other nutrients.

As explained by Walworth and Sumner (1988), If we consider hypothetical nutrients A, B through N, then:

\(A index= \frac{\text{f}\left(\frac{\text{A}}{\text{B}}\right)+\text{f}\left(\frac{\text{A}}{\text{C}}\right)+\text{f} \left(\frac{\text{A}}{\text{D}}\right)+\dots +\text{f}\left(\frac{\text{A}}{\text{N}}\right)}{\text{N}}\) [1]

\(B index= \frac{- \text{f}\left(\frac{\text{A}}{\text{B}}\right)+\text{f}\left(\frac{\text{B}}{\text{C}}\right)+\text{f} \left(\frac{\text{B}}{\text{D}}\right)+\dots +\text{f}\left(\frac{\text{B}}{\text{N}}\right)}{\text{N}}\) [2]

\(N index= \frac{- \text{f}\left(\frac{\text{A}}{\text{N}}\right)-\text{f}\left(\frac{\text{B}}{\text{N}}\right)-\text{f} \dots -\text{f}\left(\frac{\text{M}}{\text{N}}\right)}{\text{N}}\) [3]

For, if A/B ≥ a/b;

\(\text{f}\left(\frac{\text{A}}{\text{B}}\right)=\left[\frac{\left(\frac{\text{A}}{\text{B}}\right)}{\left(\frac{\text{a}}{\text{b}}\right)}-1\right] \times \frac{1000}{\text{C}\text{V}}\) [4]

or, if A/B < a/b;

\(\text{f}\left(\frac{\text{A}}{\text{B}}\right)=\left[1- \frac{\left(\frac{\text{a}}{\text{b}}\right)}{\left(\frac{\text{A}}{\text{B}}\right)}\right] \times \frac{1000}{\text{C}\text{V}}\) [5]

Where a/b is the norm for the ratio of nutrients A and B, and CV is the coefficient of variation associated with that norm expressed as percentage. A/B denotes the ratio of average concentration of the ten ear leaves collected per plot for nutrients A and B, n is the number of nutrients considered in the diagnosis, and f (A/B) is a function of nutrients A and B ratio. The 1000 multiplier in equations 4 and 5 comprises of a factor 10 to give the resultant indices a convenient magnitude and a factor 100 to express the CV as fraction rather than as percentage.

A DRIS index value for given nutrient close to zero (“0”) indicates nutritional balance for that given nutrient relative to other nutrients in the diagnosis. A negative index value for a given nutrient, indicates lower amount relative to other nutrients and further indicates that the nutrient is yield limiting. On the other hand, positive index value of a nutrient indicates excess presence of that nutrient relative to others and could also affect yield negatively (Walworth and Sumner 1986).

Statistical analysis

After generating the DRIS nutrient index value for each nutrient, the values were summed across all the diagnosed nutrients for each plot to obtain overall nutrients DRIS index value for each plot. K-means cluster analysis was performed on the summed DRIS index of each field so that, fields with similar overall nutrient DRIS values are grouped to allow for an in-depth analysis of major soil properties influencing the DRIS index values for each cluster. Selection of optimal number of clusters used in this study was guided by highest cubic clustering criterion. This analysis was done in JMP Pro version 14 statistical package (SAS Institute Inc. 2017). Between each of the identified clusters, analysis of variance was used to compare the average levels of soil properties. The mean contents of the soil properties were separated using LSD.

Random forest (RF) regression was used to assess the major soil properties influencing maize DRIS indices in each of the identified clusters. The model considered all the analyzed soil properties as predictor variables using random forest regressor in XLSTAT statistical software. The RF model was trained with 50% of the observations, 30% was used for model validation and 20% for model testing. Feature importance was used to explain the influence of various soils properties on overall DRIS index values for each cluster. Errors in the models training were monitored using out-of-bag (OOB).

Characteristics of the soils across the experimental fields

Apart from soil pH and texture, other chemical properties of the soils across the environmental fields varied widely (Table 1). Average value of the pH (5.88) is considered moderately acidic using Esu (1991) classification of Nigerian savanna soils. About 77% of the fields however varied within the moderate to slightly acidic conditions. Few fields (19%) with pH values ranging from 4.80 to 5.50 fell within the strongly acidic condition, while the remaining 4% were either of neutral or slightly alkaline class. Average soil textural class across the sites is sandy loam according to USDA (1975) classification, this is common in about 40% of the fields. The CVs of sand (23.3%), silt (27.2%) and clay (26.4%) indicate a fairly consistent distribution pattern across the study fields. Although there was a moderate variation in total N, organic carbon, available P and ECEC contents, their average values indicate that they in low levels across the sites. Only less than 8% of the fields have moderate contents of these properties. The average contents of K, Mn, Zn and Fe are rated high, but this was highly inconsistent across the fields.

Table 1

Descriptive statistics of the chemical and physical properties of the soils at the experimental sites
Soil Property	Mean	Median	Mode	Minimum	Maximum	Standard Deviation	Kurtosis	Skewness	CV (%)
pH (Water) 1:1	5.88	5.88	5.80	4.80	7.20	0.41	0.12	0.17	6.9
Total OC (g kg^− 1)	6.81	6.31	7.84	2.02	18.0	2.71	1.20	0.96	39.7
Total N (g kg ^− 1)	0.50	0.48	0.50	0.14	1.48	0.19	4.76	1.62	37.9
Avail. P (mg kg^− 1)	2.80	2.36	2.88	0.64	9.35	1.59	2.83	1.53	56.8
K (cmol_c kg^− 1)	0.48	0.44	0.17	0.06	1.52	0.33	-0.03	0.79	69.8
Avail. S (mg kg^− 1)	9.08	8.28	15.7	4.11	38.36	3.67	16.50	2.94	40.4
Ca (cmol_c kg^− 1)	2.63	2.47	2.02	0.24	9.78	1.23	3.66	1.18	46.7
Mg (cmol_c kg^− 1)	0.94	0.81	1.37	0.08	3.53	0.52	4.32	1.72	55.7
Cu (mg kg^− 1)	1.48	1.45	1.66	0.13	5.12	0.84	2.94	1.34	56.8
Mn (mg kg^− 1)	63.2	58.1	88.6	3.71	180.7	40.4	0.58	0.97	63.9
B (mg kg^− 1)	0.05	0.03	0.10	0.00	0.25	0.05	2.83	1.74	94.2
Zn (mg kg^− 1)	4.47	1.32	3.28	0.62	69.06	6.31	31.60	4.17	141.2
Fe (mg kg^− 1)	139.7	126.4	101.5	43.4	527.2	59.7	8.90	2.24	42.8
Na (cmol_c kg^− 1)	0.13	0.10	0.07	0.04	0.55	0.09	5.07	2.17	67.3
ECEC (cmol_c kg^− 1)	5.27	4.79	7.60	0.82	18.35	2.48	2.06	1.07	47.1
Sand (%)	48.11	45.60	40.00	26.00	77.20	11.18	-0.47	0.43	23.2
Silt (%)	29.76	31.20	39.20	9.20	44.00	8.11	-0.67	-0.42	27.2
Clay (%)	22.13	20.80	20.80	11.60	41.60	5.83	0.21	0.74	26.4

Variation of ear leaf nutrient concentrations across the experimental fields

According to the CV values, the variation in ear leaf concentration of all the analyzed nutrients is of moderate nature according to Wilding (1985) procedure. The kurtosis and skewness values of all the nutrients were also within the acceptable limits of asymmetric data dispersion (Table 2). For N, the mean concentration was 23.89 g kg^− 1. This value was also quite similar to those of the other measures of central tendencies (median = 23.85, mode = 23.0). These values were below the critical level of N in the ear leaf. The skewness and kurtosis were respectively − 0.12 and − 0.33. Average ear leaf concentration of P was 2.43 g kg^− 1, and this also seem to be moderately consistent (CV = 28.6%) across the sites. Average K ear leaf concentration was 18.2 g kg^− 1 with slight difference across the data set. Concentration of ear leaf ranged from 2.3–9.70 g kg^− 1 for Ca, 0.3–4.96 g kg^− 1 for Mg and 0.0–3.10 g kg^− 1 for S. Among the micronutrients, average ear leaf concentration of Cu was 6.05 mg kg^− 1, 74.39 mg kg^− 1 for Mn, 13.69 mg kg^− 1 for Zn, 3.27 mg kg^− 1 for B, and 139.43 mg kg^− 1 for Fe.

Table 2

Descriptive statistics of the concentration of macro nutrients and micro nutrients in the ear leaves
Statistical Terms	Macro nutrients (g kg^− 1)						Micro nutrients (mg kg^− 1)
Statistical Terms	N	P	K	Ca	Mg	S	Cu	Mn	Zn	B	Fe
Mean	23.89	2.43	18.20	5.88	2.28	1.45	6.05	74.39	13.69	3.27	139.43
Median	23.85	2.40	17.80	5.80	2.20	1.45	5.80	67.50	12.81	3.15	132.75
Mode	23.00	2.00	11.60	4.60	1.90	1.20	4.60	80.23	9.57	2.00	195.76
Minimum	8.90	0.80	5.90	2.30	0.30	0.00	1.59	16.23	3.69	0.48	43.31
Maximum	36.70	4.50	32.57	9.70	4.69	3.10	13.44	181.14	28.35	7.61	305.04
Standard Deviation	5.18	0.70	4.99	1.47	0.73	0.43	2.32	33.96	4.92	1.27	51.61
Kurtosis	-0.33	0.12	-0.52	-0.55	0.09	0.47	-0.27	0.40	0.39	0.62	-0.11
Skewness	-0.12	0.41	0.21	0.11	0.54	0.12	0.37	0.89	0.84	0.77	0.56
CV (%)	21.7	28.6	27.4	25.1	31.9	29.6	38.4	45.7	35.9	38.7	37.0

The DRIS Analysis

For establishing the DRIS norms, nutrient dual ratios that present the highest variance among the three possible expressions (direct, indirect and product) were selected for every nutrient pair combination (Table 3). The direct and Indirect forms of expression were selected in equal cases (46.7%) for the nutrient pairs. In just 6% of the cases, the product form of the dual ratio expression (Mg*Ca, Mg*Fe and K*Zn) was selected. The results of the variances calculated for each nutrient dual ratio across the fields indicated a higher variance ratio in the low yield sub-populations than in the high yield sub-population group. This resulted in a variance ratio ≥ 1 when the variance of the low yield was divided by that of the high yield sub-population (Table 3). The variance ratio was usually higher when macro and micro nutrients were paired. The CVs were also likewise higher when a micro nutrient was involved, and again the CV of Mn/B for the low yield sub-population was the highest (76.71%) among all the (Table 3).

The results of the DRIS nutrient diagnosis indicate that the average indices for the individual nutrients widely varied across the experimental sites (Fig. 2). Although in few cases, nutrients like Mg, K, Ca, S, and Zn have the near “0” DRIS indices, they however show negative DRIS indices in many cases across the fields. The average index of N was only positive in one field across the experiment; however, N was only the most limiting nutrient in just 3% of the fields (Fig. 3). The DRIS indices of Mg and Zn were the most inconsistent across the experimental fields, with values range of 1.9 to -10.5. Copper (Cu), Mn and B had the most negative DRIS indices across the fields. Across the experiments, the DRIS analysis indicated that nutrients which were not applied (B, Mg, Mn, S and Zn) showed a consistently negative DRIS indices at individual sites (Fig. 3). In most cases (31%), Zn was found to have the most negative DRIS index among the nutrients (Fig. 3). Magnesium (Mg) and Mn with 15.4 and 14.4% respectively were the second most limiting set of nutrients, and followed by B in around 10% of the fields.

Cluster analysis

Based on attaining high cubic clustering criterion, four distinct clusters were selected using the multivariate k-means cluster analysis to best represent the highest homogeneity and heterogeneity within and between the clusters respectively. Cluster one and cluster two were the largest with 139 and 119 fields respectively. Cluster four was the third largest with 98 fields while cluster three was the smallest with 18 fields. Below are the attributes of the selected clusters in terms of soil, grain yield and DRIS indices.

Results of analysis of variance on the soil (Table 4) indicated that levels of pH, ECEC, P, Ca, S and B did not significantly vary among the clusters. The pH condition on average was moderate across the clusters. In general, ECEC, P, S and B were low, while Ca level was moderately high across the clusters. Although the contents of N and Mg in cluster four were below the critical soil levels, the cluster seem to have the higher contents of K, Cu, Mn and Na. Cluster two is the second more fertile environment with dominantly higher organic carbon level, although the organic carbon was also yet below the critical level. Cluster three had higher Zn level among the clusters. This cluster had the highest and lowest contents of sand and clay respectively. The cluster is rated the least fertile and it is characterized by lowest contents of organic carbon, N, K, Mg, S, Cu, and Na.


Table 3. Selected forms of nutrient dual ratios, DRIS norms, variance (σ²) for low and high yield sub-populations and variance ratios
Form of expression	High yield sub-population			Low yield sub-population						σ²_low/ σ²_high		Form of expression		High yield sub-population						Low yield sub-population						σ²_low/ σ²_high
Form of expression	Norm	CV (%)	σ²_high	Norm		CV (%)		σ²_low		σ²_low/ σ²_high		Form of expression		Norm		CV (%)		σ²_high		Norm		CV (%)		σ²_low		σ²_low/ σ²_high
Mg/N	0.09	35.86	0.00		0.10		37.42		0.00		1.27		S/Ca		0.27		31.56		0.01		0.25		39.12		0.01		1.38
K/N	0.80	29.42	0.06		0.79		34.80		0.08		1.36		Ca/Cu		0.01		45.29		0.00		0.12		54.23		0.00		1.75
Ca/N	0.24	29.40	0.00		0.26		31.16		0.01		1.39		Mn/Ca		1.37		50.19		46.96		1.49		62.58		71.23		1.52
N/P	10.24	23.44	5.76		10.47		32.60		11.64		2.02		Ca/Fe		0.00		45.58		0.00		0.00		55.43		0.00		1.40
S/N	0.06	28.31	0.00		0.63		31.00		0.00		1.31		Zn/Ca		2.57		40.66		1.09		2.46		48.57		1.43		1.31
N/Cu	0.05	47.02	0.00		0.05		54.28		0.00		1.34		Ca/B		0.02		36.14		0.00		0.22		64.50		0.00		4.40
Mn/N	0.35	40.08	1.49		0.29		50.03		2.71		1.82		S/P		6.14		27.48		0.03		0.64		41.05		0.07		2.43
Fe/N	0.52	40.78	4.57		0.39		46.18		8.71		1.91		Cu/P		0.25		35.27		0.80		2.69		46.76		1.58		1.98
N/Zn	0.02	27.97	0.00		0.19		34.78		0.00		1.52		Mn/P		3.12		44.80		195.79		34.65		66.53		531.5		2.71
N/B	1.62	39.00	0.00		0.87		65.41		0.00		3.14		P/Fe		0.00		39.80		0.00		0.00		54.21		0.00		1.57
Mg/K	0.12	35.44	0.00		0.41		48.18		0.00		2.43		Zn/P		0.58		34.88		4.03		6.15		48.85		9.03		2.24
Mg*Ca	0.14	47.72	0.00		0.14		49.21		0.00		1.09		P/B		0.01		43.35		0.00		0.09		63.97		0.00		2.48
Mg/P	0.93	32.18	0.09		1.04		43.54		0.20		2.26		S/Cu		0.00		44.36		0.00		0.00		53.70		0.00		1.50
Mg/S	1.63	45.89	0.39		1.68		37.24		0.56		1.44		Mn/S		53.12		50.43		717.57		5.66		58.03		1043.3		1.45
Mg/Cu	0.00	37.01	0.00		0.00		49.67		0.00		2.15		Fe/S		4.56		56.31		2835.3		1.73		52.46		3168.0		1.12
Mg/Mn	0.00	54.21	0.00		0.00		58.79		0.00		1.26		S/Zn		0.00		37.62		0.00		0.00		45.50		0.00		1.48
Mg*Fe	31.3	51.80	262.9		33.0		52.17		295.9		1.13		S/B		0.05		41.20		0.00		0.01		51.52		0.00		1.75
Mg/Zn	0.00	37.60	0.00		0.00		48.91		0.00		1.93		Mn/Cu		1.37		57.93		0.63		1.48		68.66		1.04		1.64
Mg/B	0.01	42.16	0.00		0.01		63.79		0.00		2.97		Fe/Cu		2.33		49.03		1.30		2.76		64.64		3.17		2.43
K/Ca	3.54	36.21	1.64		3.24		43.81		2.02		1.23		Cu/Zn		0.05		44.31		0.00		0.05		51.00		0.00		1.39
K/P	7.96	29.86	5.65		8.04		40.43		10.57		1.87		Cu/B		2.02		46.14		0.01		0.21		68.50		0.02		2.47
K/S	13.99	43.75	32.48		3.53		42.11		37.49		1.15		Fe/Mn		2.07		57.11		1.40		2.29		61.34		1.97		1.41
K/Cu	0.04	51.34	0.00		0.36		64.43		0.00		1.55		Mn/Zn		5.88		50.89		8.96		6.29		67.46		17.98		2.01
K*Mn	147.6	50.87	5475.4		130.41		56.74		5638.1		1.03		Mn/B		25.18		53.88		184.05		28.29		76.71		470.8		2.56
K/Fe	0.00	39.97	0.00		0.00		46.03		0.00		1.02		Fe/Zn		10.33		44.52		21.15		11.84		52.59		38.79		1.83
K*Zn	28.28	45.56	166.00		24.65		54.70		181.88		1.10		Fe/B		42.69		55.52		561.81		52.57		61.61		1049.2		1.87
K/B	0.06	42.78	0.00		0.64		49.62		0.00		1.35		Zn/B		4.76		56.67		7.28		5.04		65.33		10.85		1.49
Ca/P	2.39	29.07	0.48		2.66		38.59		1.05		2.18

Grain yield generally varied moderately within each cluster except cluster one which varied widely (Fig. 4). Highest yield was obtained from cluster two, with a mean yield of approximately 7,000 kg ha^− 1. Yield of cluster one and four did not significantly vary. Also statistically, distribution of yield was identical between the two clusters. The cluster three had yield between 2,000–8,000 kg ha^− 1, and mean yield of 4,700 kg ha^− 1, which make it the second highest yield cluster after cluster two (Fig. 4).

When the DRIS indices for the nutrients were clustered, the result show that Mn and B were the most important limiting nutrients with lowest DRIS indices (most negative indices) occurring in 27.34% and 25.18% instances respectively in cluster one (Table 5). Mg (18.71%), P (13.67%) and K (10.07%) were also found to limit maize productivity in this cluster. For cluster two, Mn (26.89%) and Mg (21.01%) were the dominant limiting nutrients according to the DRIS analysis. Boron (B) and Fe were the second important yield limiting nutrients in 13.45 and 10.08% of the sites in the cluster two. The third group of limiting nutrients in this cluster include P, K, Ca, S and Cu. Cluster three had the lowest number of limiting nutrients. In this cluster, B was the most important limiting nutrient in 43.75% of the fields. Each of P and Mn was found as most limiting in 25% of the fields in cluster three. Yield is indicated to be limiting by almost all nutrients in the same magnitude in cluster four. The indices of K were the most negative in 18.37% percent of the fields in the cluster. That of P, Zn and Fe were second after K with each recurring in 12.24% of the case.

Table 4

Analysis of variance results comparing soil properties of different clusters across the study
Cluster	pH (water) 1:1	ECEC (cmol/kg)	OC (%)	N (%)	MehP (mg/kg)	K (cmol/kg)	Ca (cmol/kg)	Mg (cmol/kg)	S (mg/kg)
1	5.91	5.13	0.62b	0.47b	2.83	0.44b	2.44	0.88b	8.85
2	5.88	5.05	0.75a	0.51ab	2.94	0.44b	2.73	0.91b	8.97
3	5.90	5.16	0.53b	0.44ab	2.75	0.34b	2.76	0.79b	7.48
4	5.85	5.78	0.70a	0.53a	2.62	0.59a	2.74	1.07a	9.78
LSD	0.281	1.285	0.136	0.097	0.827	0.170	0.636	0.268	1.893
Cu (mg/kg)	Mn (mg/kg)	Zn (mg/kg)	B (mg/kg)	Na (cmol/kg)	Fe (mg/kg)	BD (g/cm3)	Sand (%)	Silt (%)	Clay (%)
0.58b	58.39bc	1.00b	0.05	0.13b	140.49	0.90	49.76b	28.29b	21.95a
0.53b	59.96bc	1.02b	0.05	0.12b	135.83	0.96	44.87c	32.75a	22.39a
0.40b	66.56b	1.40a	0.05	0.11ab	153.41	0.96	60.45a	21.46c	18.1b
0.74a	73.55a	1.03b	0.06	0.16a	141.12	0.75	47.69bc	29.24b	23.07a
0.221	12.089	0.270	0.026	0.046	31.078	0.305	5.603	4.000	3.010

Table 5: Percentage of fields with negative DRIS indices of nutrients for each cluster

	Cluster 1	Cluster 2	Cluster 3	Cluster 4
P	13.67	7.56	25.00	12.24
K	10.07	6.72	0.00	18.37
Ca	0.72	5.04	0.00	5.10
Mg	18.71	21.01	6.25	10.20
S	0.72	4.20	0.00	2.04
Cu	0.00	4.20	0.00	6.12
Mn	27.34	26.89	25.00	11.22
Zn	1.44	0.84	0.00	12.24
B	25.18	13.45	43.75	10.20
Fe	0.72	10.08	0.00	12.24

Major soil factors influencing the DRIS Indices

Results of the random forest regression analysis show that the algorithm stopped growing trees for clusters 1, 2 and 4 when the trees reach 200. For cluster four, the prediction error reduces and stabilized after more than 800 trees were grown in the forest (results not shown). The variable importance features for each cluster are presented in Fig. 5a − 5d. The results indicated that nutrient DRIS indices are influenced by unique soil factors for respective clusters. For cluster one, pH appears to have the largest influence on nutrient DRIS indices (Fig. 5a). Boron and Mg were the second most important soil variables which affected the DRIS indices. Other important factors were ECEC, sand content, P and organic carbon. Nutrient DRIS indices were most influenced by N and then by B contents in cluster two (Fig. 5b). Soil contents of Fe, K, Mg and S also influenced the nutrients DRIS indices at similar magnitude in this cluster. For cluster three, pH was the major variable, and Na as the second (Fig. 5c). Bulk density (BD) and Cu were also critical influencers of the DRIS indices. More than in any cluster, multiple variables influenced the DRIS indices of the nutrients in cluster four (Fig. 5d). Potassium (K), B and Zn seem to be the most significant factors in this cluster. Iron (Fe) and Na, ECEC, BD and organic carbon had a moderate influence on the nutrient’s indices in descending order respectively.

Studies on plant, soil, and nutrients interaction are few in Nigeria, and the focus is mainly on outcome produced by individual factor, but how these factors interact to produce a given outcome have not received a significant attention. Previous studies on soils (Aliyu et al. 2020) indicated a wide variability of soil nutrients within the major maize production domain in Nigeria. Relative response to nutrient application has also been reported with a high variation across a study area by Shehu et al (2018), Garba et al (2020) and Aliyu et al (2022). Nziguheba et al (2009) and Aliyu et al (2021) both used the DRIS to diagnose nutrients that limit maize yield in Nigeria. In the studies some of the diagnosed nutrients which were found in lower levels (negative DRIS indices) in the maize plant tissues were actually in sufficient levels in the soils. Aliyu et al (2021) concluded that tissue nutrient diagnosis should be complemented with soil analysis for targeting a proper nutrient recommendation. Sampling population in previous studies on nutrient diagnosis is either formed from different nutrient treatments (Nziguheba et al. 2009; Aliyu et al. 2021), or from a farm survey across different cropping systems (Serra et al. 2014; Carneiro et al. 2015). Using those type of population for nutrient diagnosis may likely complicate the diagnosis and compromise its’ quality and reliability. In this study we used population which had received similar nutrient treatment so that the difference in performance of the crop may be attributed more to variation in soil properties. Our findings indicated a wide variation in inherent soil properties across the experimental fields. Variation in pH is the lowest among the soil properties. Over 77% of the fields are within the suitable pH limit for maize production (FFD 2012; Shehu et al. 2019). The fields with lower pH values (strongly acidic condition) are few across the study fields and could have resulted from improper application of mineral nitrogen fertilizer over time which usually leads to soils becoming more acidic (Schumann 1999). These specific fields have high sand content, and it is possible the situation would worsen with continuous application of high nitrogen since sandy soils have low buffering capacity (Nelson and Su 2010). Our study environment is generally low in nitrogen, organic carbon and P contents. This finding is consistent with that of Ekeleme et al (2014) and Shehu et al (2015). Aliyu et al (2020) reported a high correlation between organic carbon and N, and highlighted that their availability is dependent on pH condition of soil. The same study also attributed the variability of the N, P and K to inherent soil forming minerals. Therefore, our sites present a suitable environment for conducting nutrient experimentation.

Deriving our experimental data from the uniform treatment set had helped in reducing variability in ear leaf nutrients across the fields. However, the small variability was also sufficient to allow for data categorization in the diagnosis process. The ear leaf concentrations of N, P, B, Zn and S were below the critical levels established by Reuter et al (1997). The concentrations of K, Mg, Ca, Fe and Cu are notably within the sufficiency levels established by Reuter et al (1997). While some of the values for the ear leaf nutrients; N, P, Cu and B were similar to those reported by Aliyu et al (2021), those of K, Ca and Fe were significantly higher in this study.

The overall variability of the DRIS indices across the experimental fields could be attributed to the variability of the soil and ear leaf nutrient concentration. Nutrient DRIS index is direct reflection of tissue concentration of the nutrient (McCray et al. 2013). Magnesium had the highest variation in of the DRIS index across the experiment and this can be related to higher range of distribution of the nutrient, both in the soil and ear leaf across the fields. On the contrary, lower variation in N DRIS index across the fields is attributed to the low variation of soil nitrogen across the fields. The highest occurrence of Zn as the most negative nutrient in the diagnosis is contrary to the findings of Aliyu et al (2021) under the NPK treatment. However, this is possible because Zn had the most consistent negative DRIS indices distribution across the sites. In addition, the soils across the fields were uniformly low in Zn content. A similar finding of low Zn in these soils was also reported by Huising and Mesele (2021). Zn was predominantly the most negative nutrient irrespective of treatments in the long-term experiment reported by Nziguheba et al (2009). Aliyu et al (2021) also reported that application of Zn did not significantly increase Zn ear leaf content, thus resulting in negative DRIS index of Zn in many cases. The case of negative B and Mg indices corresponds to low soil contents of the nutrient as also reported in other studies (Garba et al. 2020). The high percentage of negative Mn DRIS indices despite high amount of Mn in the soil could be attributed to high antagonistic effect of Fe (content of Fe in our experiment fields was high) on Mn during uptake as discovered by Kobraee and Shamsi (2011).

Cluster analysis

The DRIS indices for the nutrients in cluster one was mainly influenced by pH and B, and to larger extent Mg, ECEC and Sand. Highest number of negative DRIS index of B could be linked to the low soil B content of the cluster. Cluster one is the least fertile among the four clusters and it is therefore not surprising to also found that almost all nutrients have negative DRIS indices in this cluster. The ECEC and soil pH are highly correlating soil properties (McKenzie et al. 2004). Low ECEC in this cluster is the result of high sand content which easily leach away the cations, this is the reason that this cluster is associated with low levels of the cations. The yield limitation and high negative DRIS indices of Mg could have been created by the low Mg levels in the soil. In addition to that, the cluster has sufficient levels of K and Ca and both have stronger binding affinity which gives them edge over Mg at the exchange site during uptake. This situation was more aggravated because of the higher level of pH of the cluster (René et al. 2017).

Nitrogen and B appears as the major influencers of the DRIS indices of cluster two. This cluster had a sum of -22.94 (not shown) DRIS balance index. It is clear that this cluster is responsive to nutrients going by the highest mean yield recorded. The high nutrient response in this cluster seems to be related with low amount of nutrients and the ability of the soil to hold nutrients due to higher clay content. Boron being another important influencer of the DRIS indices could also be explained by its very low content in the soil. The pH is the most important soil property in cluster three. The nutrients with highest number of negative DRIS indices in this cluster were P, Mn and B. These nutrients were in low supply in the soil and they therefore contributed to the cluster having the highest overall negative DRIS balance index of -155.

Cluster four had the second lowest grain yield though it has the highest contents of most of the soil fertility properties. Soil K, B and Zn were the dominant factors that influenced the DRIS indices of the nutrients. Low yield in this cluster could be related to luxury consumption of nutrients by the maize. This is reflected in the overall nutrient balance index of this cluster going up to 27.4. Luxury nutrient consumption had been reported to decrease nutrient use efficiency in maize (Janssen et al. 1990; Shehu et al. 2019). The roles of K and B in plants are similar and they interact synergistically to influence the uptake of each other. In this cluster, K was in very high amount in the soil and possibly upset the nutrient balance of the other nutrients in the diagnosis.

Nutrient diagnosis using DRIS method proves to be useful in identifying nutrient limitation in maize production in the maize belt region of Nigeria. The diagnosis was also found to be consistent with results of soil analysis of the diagnosed nutrients. Both the DRIS and soil analysis indicated consistent limitation of B. Both high (more positive) and low (more negative) DRIS indices resulted to low maize yield respectively due over and under supply of nutrients respectively. Apart from the nutrients, soil chemical and physical properties also play important roles in DRIS nutrient balance. The Soil pH and ECEC were the chemical properties which consistently seem very significant. Percentage of sand and bulk density was also to an extent important in determining nutrient availability to maize in this study.

Acknowledgments

We greatly thank the technical staff of International Institute of Tropical Agriculture (IITA) Kano, Centre for Dryland Agriculture (CDA), Bayero University Kano, Nigeria, agricultural development projects (ADPs) of Kano, Kaduna, Katsina, Niger, Bauchi, Nassarawa, Plateau and Taraba states for coordinating the site selection, soil sampling, trials establishment, field management and data collection.

Funding

This work was supported by the Bill and Melinda Gates Foundation (BMGF) for funding this research through “Taking Maize Agronomy to Scale in Africa (TAMASA)” (Contract ID: OPP1113374) Project managed by the International Institute of Tropical Agriculture (IITA).

Competing interests

The authors have no relevant financial or non-financial interests to disclose.

Author Contributions

All authors contributed to the study conception and design. Field trial establishment, ear leaf and grain samples collections and ear analysis were performed by Kamaluddin T. Aliyu, Bello M. Shehu and Adam M. Adam. The first draft of the manuscript was written by Kamaluddin T. Aliyu and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Data Availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Adnan AA, Jibrin MJ, Kamara AY et al (2017) CERES-maize model for determining the optimum planting dates of early maturing maize varieties in northern Nigeria. Front Plant Sci 8:1118. https://10.3389/fpls01118
Aliyu KT, Huising J, Kamara AY, Jibrin MJ, Mohammed IB, Nziguheba G, Adam AM, Vanlauwe B (2021) Understanding nutrient imbalances in maize (Zea mays L.) using the diagnosis and recommendation Integrated system (DRIS) approach in the maize belt of Nigeria. Sci Rep 11:16018. https://doi.org/10.1038/s41598-021-95172-7
Aliyu KT, Kamara AY, Huising J, Jibrin MJ, Shehu BM, Rurinda J, Adam AM, Mohammed IB, Vanlauwe B (2022) Maize nutrient yield response and requirement in the maize belt of Nigeria. Environ Res Lett 17 (2022) 064025. https://doi.org/10.1088/1748-9326/ac5bb1
Aliyu KT, Kamara AY, Jibrin MJ, Huising J, Shehu BM, Adewopo JB, Mohammed IB, Solomon R, Adam AM, Samndi AM (2020) Delineation of soil fertility management zones for site-specific nutrient management in the maize belt region of Nigeria. Sustain 12:9010. doi:10.3390/su12219010
Alley MM, Vanlauwe B (2009) The role of fertilizers in Integrated Plant Nutrient Management, 1 edn. Edn IFA Paris, France
Beaufils ER (1987) Research for rational exploitation of Hevea brasiliensis using a physiological diagnosis based on the mineral analysis of various parts of the plant. Fertilizer 3:27
Biwe ER (2012) Status and distribution of available micronutrients along a topo sequence at Gubi Bauchi North Eastern Nigeria. Int Res J Agric Sci and Soil Sci 2(10):436–439
Bremner JM (1996) Nitrogen Total. In: Sparks DL (ed) Methods of soil analysis: chemical methods. Am Soc Agron Soil Sci, Madison, WI, USA
Carneiro AC, Óscar P, Mário C, Jorge Q (2015) The diagnosis and recommendation integrated system (DRIS) - first approach for the establishment of norms for vineyards in Portugal. Sci Tech vitiv 30(2):53–59
Ekeleme F, Jibrin MJ, Kamara AY, Oluoch M, Samndi AM, Fagge AA (2014) Assessment of the relationship between soil properties, Striga hermonthica infestation and the on-farm yields of maize in the dry savannas of Nigeria. Crop Prot 66:90–97
Esu IE (1991) Detailed Soil Survey of NIHORT Farm at Bunkure Kano State, Nigeria. Ahmadu Bello University Zaria, Kaduna, Nigeria
Federal Fertilizer Department (FFD) (2012) Fertilizer Use and Management Practices for Nigeria, Federal Fertilizer Department, Federal Ministry of Agriculture and Rural Development, 4th edn. Abuja, Nigeria
Fixen PE, Jin J, Tiwari KN, Stauffer MD (2005) Capitalizing on multi-element interactions through balanced nutrition - a pathway to improve nitrogen use efficiency in China, India and North America. Sci China Ser C Life Sci 48:1–11
FAO (2009) FAOSTAT. FAO Statistics Division. https://faostat.fao.org/. Accessed 25 June 2018
FAOSTAT (2019) Production statistics of the food and agriculture organization of the United Nations. https://faostat3.fao.org/download/Q/QC/E. Accessed 1 February 2019
Garba II, Jibrin MJ, Kamara AY, Adnan AA, Abdulrahman BA (2020) Response of maize to secondary nutrients and micronutrients in the Guinea savanna of Nigeria. J Agron 19:120–130
Gee GW, Or D(2002) Particle-size analysis. In: Dane JH, Topp GC (ed) Methods of soil analysis. Book Series 5 Am Soc Agron Soil Sci, Madison, WI, USA, pp 255–293
Harger NR, Fioretto R, Ralisch R (2003) Nutritional assessment of soybean crop by DRIS methods and sufficiency levels. Sowing: Agric Sci 24:219–224
Heanes DL (1984) Determination of total organic-C in soils by an improved chromic acid digestion and spectrophotometric procedure. Commun Soil Sci Plant Anal 15:1191–1213. https://doi.org/10.1080/00103628409367551
Huising J, Mesele MS (2021) Soil Quality Assessment and Management Plans for IITA Research Farm, Nigeria. IITA Ibadan, Nigeria
Janssen BH, Guiking FCT, van der Eijk D, Smaling EMA, Wolf J, van Reuler H (1990) A system for quantitative evaluation of the fertility of tropical soils (QUEFTS). Geoderma 46:299–318
Jibrin MJ, Kamara AY, Friday E (2012) Simulating planting date and cultivar effect on dryland maize production using CERES maize model. Afr J Agric Res 7:5530–5536. 10.5897/AJAR12.1303
Kamara AY, Ewansiha SU, Tofa AI (2019) Yield, N uptake N utilization of early maturing, drought and striga-tolerant maize varieties under low N conditions. Commun Soil Sci Plant Anal. 10.1080/00103624
Kobraee S, Shamsi K (2011) Effects of Zn, Fe and Mn on soybean production. Ecol Environ Conser 17(2):191–196
Knoema(2021) Nigeria maize production quantity. Knoema statistics. http//knoema.com/atlas/Nigeria/topics/Agriculture/Crops-Production-Quantity-tonnes/Maize-production. Accessed 15 November 2021
Maia CE (2012) Fertilization response likelihood for the interpretation of leaf analyses. R Bras Ci Solo 36:437–445
Mangale N, Muriuki A, Kathuku-Gitonga AN, Kibunja CN, Mutegi JK, Esilaba AO
Ayuke FO, Nguluu SN, Gikonyo EW (2016) Field and laboratory research manual for Integrated soil fertility management in Kenya. Kenya Soil Health Consortium KALRO, Kabate, Nairobi Kenya
McKenzie NJ, Jacquier DJ, Isbell RF, Brown KL (2004) Australian Soils and Landscapes: An Illustrated Compendium. CSIRO Publishing, Collingwood, Victoria
McCray J, Rice R, Ezenwa I, Lang T, Baucum L (2013) Sugarcane Plant Nutrient. Diagnosis University of Florida – IFAS Extension, Florida
Mehlich A (1984) Mehlich 3 soil test extractant: A modification of Mehlich 2 extractant. Commun Soil Sci Plant Anal 15:1409–1416
Mugo JN, Karanja NN, Gachene CK, Klaus D, Shadrack NO, Schulte-Geldermann E (2020) Assessment of soil fertility and potato crop nutrient status in central and eastern highlands of Kenya. Sci Rep 10:7779. https://doi.org/10.1038/s41598-020-64036-x
Nelson PN, Su N (2010) Soil pH buffering capacity: a descriptive function and its application to some acidic tropical soils. Aust J Soil Res 48:210–207
Nziguheba G, Tossah BK, Diels J, Franke AC, Aihou K, Iwuafor ENO, Nwoke C, Merckx R (2009) Assessment of nutrient deficiencies in maize in nutrient omission trials and long-term field experiments in the West African savanna. Plant Soil 314:143–157
René PJJ, Rietra MH, Chistian OD, Prem SB (2017) Effects of nutrient antagonism and synergism on yield and fertilizer use efficiency. Commun Soil Sci Plant Anal 48(16):1895–1920. DOI:10.1080/00103624.2017.1407429
Reuter DJ, Edwards DG, Wilhelm NS (1997) Temperate and tropical crops. In: Reuter DJ, Robinson JB (eds) Plant analysis: an interpretation manual. CSIRO, Australia, pp 83–284
Roy RN, Finck A, Blair GJ, Tandon HLS (2006) Plant nutrition for food security, FAO fertilizer and plant nutrition bulletin 6. Food and Agriculture Organization (FAO) of the United Nations, Rome, Italy
Rurinda J, Shamie Z, Jibrin MJ, Tesfaye B, Kenneth M, Jens AA, Pampolino MF, Mohammed IB, James M, Kamara AY, Vanlauwe B, Craufurd PQ (2020) Science-based decision support for formulating crop fertilizer recommendations in sub-Saharan Africa. Agric Sys 180:102790. https://doi.org/10.1016/j.agsy.2020.102790
Institute Inc SAS (2017) JMP® 13 Documentation Library. SAS Institute Inc., Cary, NC
Schumann B (1999) The causes of soil acidity. New South Wales Acid Soil Action Program
Queanbeyan, leaflet no. 5
Serra AP, Marchetti ME, Ensinas SC, de Morais HS, do Amaral VC, Guimarães FCN, Barbosa GPO(2014) Diagnosis and Recommendation Integrated System (DRIS) to assess the nutritional state of cotton crop in Brazil. Am J Plant Sci 5:508–516. https://dx.doi.org/10.4236/ajps.2014.54065
Shehu BM, Jibrin JM, Samndi AM (2015) Fertility status of selected soils in the Sudan Savanna biome of Northern Nigeria. Int J Soil Sci 10:74–83. https://doi.org/ 10.3923/ijss.2015.74.83
Shehu BM, Merckx R, Jibrin MJ, Kamara AY, Rurinda J (2018) Quantifying variability in maize yield response to nutrient applications in the northern Nigerian savanna. Agron 818:1–23
Shehu BM, Bassam AL, Jibrin MJ, Kamara AY, Mohammed IB, Rurinda J, Zingore S, Craufurd PQ, Vanlauwe B, Adam MA, Merckx R (2019) Balanced nutrient requirements for maize in the northern Nigerian savanna: Parameterization and validation of QUEFTS model. Field Crop Res. https://doi.org/10.1016/j.fcr.2019.107585
Singh NP, Awasthi RP, Sud A (2000) Diagnosis and recommendation integrated system (DRIS) norms for apple (Malus x Domestica Borkh. L. CV. Starking Delicius) in Himachal Pradesh. Indian J Hotic 53(3):96–204
Tilman D, Christian B, Jason H, Belinda LB (2011) Global food demand and the sustainable intensification of agriculture. Proceedia Nat Academic Sci 108(50):20260–20264
Tofa AI, Kamara AY, Babaji BA, Akinseye FM, Bebeley JF (2021) Assessing the use of a drought–tolerant variety as adaptation strategy for maize production under climate change in the savannas of Nigeria. Sci Rep 11:8983. https://doi.org/10.1038/s41598-021-88277-6
USDA (1975) Textural Classification System. Soil Conservation Service of the United States Department of Agriculture USDA
Walworth JL, Sumner ME (1986) Foliar diagnosis - A review. In: Tinker BP (ed) Advances in plant nutrition. Elsevier, New York
Walworth JL, Sumner ME Foliar diagnosis: A review In: Tinker B, Lauchli AV (eds) Advances in plant nutrition.Praeger Publishers New York, pp193–241
Walworth JL, Sumner ME (1987) The Diagnosis and Recommendation Integrated System (DRIS). Adv Soil Sci 6:148–185
Wilding LP(1985) Spatial variability: Its documentation, accommodation and implication to soil surveys. In: Nielsen DR, Bouma J (ed) Soil Spatial Variability, Wageningen, Netherlands
Weil R, Mughogho S (2000) Sulphur nutrition of maize in four regions of Malawi. Agron J 92:649–665

Download PDF

Version 1

posted

You are reading this latest preprint version

Identification of soil properties influencing DRIS indices of maize in the savannas of northern Nigeria

Status:

Version 1

Abstract

Figures

Introduction

Materials And Methods

Data analysis

Statistical analysis

Results

Discussion

Conclusion

Declarations

References

Status:

Version 1