Optimizing application rate of nitrogen

Journal of Soil Science and Plant Nutrition, 2018, 18 (1), 13-26
RESEARCH ARTICLE

Optimizing application rate of nitrogen, phosphorus and
cattle manure in wheat production: An approach to determine
optimum scenario using response-surface methodology

Mohsen Jahan1* , Mohammad Behzad Amiri2
1

Associate Prof. of Agroecology, Ferdowsi University of Mashhad, Faculty of Agriculture, P.O. Box 91775-

1163, Mashhad, Iran. 2Assistant Prof. of Agroecology, Dep. Plant Production Engineering, Gonabad University,
Khorasan Razavi, Iran. *Corresponding author: jahan@um.ac.ir

Abstract
Optimal application rates of inorganic nitrogen (N), phosphorus (P) and cattle manure were estimated using
Response Surface Methodology (RSM). A Central Composite Design (CCD) was conducted at ield level during 2012-13 and 2013-14 growing seasons. The applied levels of N were 0, 150 and 300 kg N ha-1 in form of
urea. The levels used for P fertilizer were 0, 100 and 200 kg ha-1 (P2O5) and for cattle manure were 0, 15 and 30
t ha-1. Both seed yield (SY), biological yield (BY) were measured at harvest time. N loss (NL) and Agronomic
N Use Eficiency (ANUE) were calculated based on other measurements. Increasing N and P rates up to 200 kg

ha-1 increased SY. Optimization of N, P and manure application amount was based on economic, environmental
and eco-environmental scenarios. Under economic scenario, using 145.4 kg ha-1 N, 200 kg ha-1 P and 18.4 t
ha-1 manure resulted in 6500 kg ha-1 SY with ANUE of 10.49. For environmental scenario, by N application of
21.2 kg ha-1, no application of P and applying 16.3 t ha-1 manure, SY and ANUE of 3160 kg ha-1 and 9.08 were
obtained, respectively. Using eco-environmental scenario, by applying 144.7 N and 34.3 kg ha-1 P, plus 30 t ha-1
manure, about 4031 kg ha-1 SY and a considerable high ANUE of 16.5 were recorded. The results of this study
showed that the privilege of eco-environmental scenario compared to the other scenarios was mainly due to
higher ANUE.
Keywords: ANUE, nitrogen losses, eco-environmental scenario, seed yield, central composite design

13

14

Jahan and Amiri

1. Introduction
Higher required production of wheat has intensiied

is an important factor, which reduce nitrogen losses

consumption of agrochemicals that in turn have re-

(Gastal & Lemaire, 2002). Thus integrated use of

sulted in negative impacts on environment. Sustain-

slow release organic fertilizers such as manure could

able and suficient food production is a necessity that

reduce N losses during crops cultivation (Emilsson et

simultaneously should consider social, economic and

al., 2007). In arid and semi-arid regions, N deiciency

environmental aspects. The irst step to achieve this

occurs faster than other nutrients and usually is due

goal is optimization and improvement of resources

to low level of soil organic matter. It seems necessary

use eficiencies (Gliessman, 1998). It is reported that

to design cropping systems with high nutrient uptake

up to 50 percent of applied nitrogen would drift from

and utilization eficiencies (Fageria, 2014; Cassidy

agricultural systems as gaseous compounds and other

et al., 2013) considering global high demand for ce-

types of activated nitrogen (Jarvis et al., 2011; We-

real in coming future. It is suggested to increase the

ligama et al., 2010). At the high level application of

mean eficiency of N utilization by 0.1-0.4 % annu-

P (more than 200 kg ha-1), up to 90% of phosphorous

ally (Doberman & Cassman, 2005). Improving N use

fertilizers would be ixed in soil together with metallic

eficiency, increases economic return and reduces the

elements as insoluble forms which inally lead to fur-

potential of environmental pollution (Fageria, 2014).

ther phosphorus pollution (Adesemoye et al., 2009).

To reach such goals, research priorities might be shift-

In many crops, low absorption eficiency of fertil-

ed from maximizing yield to developing methods for

izers, is the main reason of leaching, volatilization

optimization of nutrients application.

and diffusion of soluble chemical fertilizers easily

Response Surface Methodology (RSM) was irst intro-

released to soil and air (Akiyama et al., 2000). It has

duced as optimization method for industrial use (My-

been reported that between 18-41 percent of applied

ers and Montgomery, 1995) however; this methodol-

nitrogen retain in soil after crop harvesting (Fageria,

ogy could also be used in fertilizer optimization. Central

2014). Nitrogen losses occur in various ways as am-

Composite Design (CCD), which is the most popular

monium volatilization in lime soils (10-70%), denitri-

RSM design, was implemented to design a series of tests

ication (9-22%) and leaching (14-40%) (Doberman

with least number of experiments. This approach tries to

& Cassman, 2004).

investigate the effect of parameters involved (i.e. contact

Wheat crop shows a strong positive correlation be-

time, dosage, pH, initial concentration) on responses in a

tween productivity and NPK fertilizers (Hawkesford

cost- and time-effective way. The CCD makes it feasible

& Barraclough, 2011; Osborne, 2007). On the other

to observe the possible interaction of the parameters and

hand, applying organic fertilizers in integration with

their inluences used RSM methodology to optimize N,

chemical ones could be an eficient management

water and plant density in canola cultivation. reported

practice to reduce application rate of chemical fertil-

that the application of 93.48 kg N ha-1 based on eco-envi-

izers and subsequently reduce their negative impacts

ronmental scenario was an optimized N use, was able to

on environment. Moreover, improved nutrient bal-

reduce environmental hazards and produced acceptable

ance in soil and plant, enhance crop productivity and

onion yield. It has been claimed that oxalic acid and lactic

yield stability in intensiied cultivation. Synchrony

acid are the major acids responsible for enhancing the P

between application time and crop nutrient demand,

solubilization .

Journal of Soil Science and Plant Nutrition, 2018, 18 (1), 13-26

Optimizing application rate of nitrogen

15

This study was aimed to optimize the chemical and

Soil samples were taken from 0-15 and 15-30 cm

organic fertilizer use in winter wheat production and

depths and analyzed for some physiochemical prop-

determine the best applicable scenario in Kashaf-rood

erties before beginning the experiment (Table 1). To

watershed in northeast of Iran. It has also been stud-

determine soil nitrogen content, soil sampling was

ied the application trend of different N, P and cattle

repeated at the end of growing season.

manure levels and their effects on wheat production.
Furthermore, the effectiveness of manure compared

2.2. Experimental design

to chemical fertilizer was studied based on NUE and
CCD (sometimes called Box-Behnken design) with

wheat yield improvement.

two replicates was used for itting response surface
2. Material and Methods

to experimental data (Tawik et al., 2017). Two years

2.1. Site description

suring uniformity of the error mean squares. The re-

data where subjected to combined analysis after ensults of combined analysis indicated that year × treatField studies conducted during 2012-13 and 2013-14

ment interaction was not signiicant (P >0.05), there-

growing seasons at the Research Station of Agricul-

fore the two years data were joined before exposing to

ture Faculty, Ferdowsi University of Mashhad, Iran

response surface analysis.

(latitude: 36° 15¢ N; longitude: 59° 28¢ E; elevation:

The experimental factors were the combination of

985 m above sea level). Experiment station was lo-

different amounts of nitrogen, phosphorus and cattle

cated in Kashaf-rood watershed in northeast of the

manure. The total number of experimental runs for a

country in a semi-arid region with mean annual pre-

3-factor CCD is 15 including 12 factorial points and 3

cipitation of 252 mm and temperature of 15 °C.

replications for center points.

Table 1. Soil properties of the experimental ield (mean of two years).
Soil depth (cm)
Soil properties
Total N (%)
Available P (ppm)
Available K (ppm)
C/N
pH (saturation extract)
EC (dS m-1)
Water storage capacity (%)
Bulk density (g cm-3)
Texture grade

0-15

15-30

0.078
19
380
12.8
7.7
1.3
22.3
1.34
Loamy-silt

0.065
15
372
12.3
7.5
1.3
20
1.41
Loamy-silt

Journal of Soil Science and Plant Nutrition, 2018, 18 (1), 13-26

16

Jahan and Amiri

By conducting CCD, it is possible to obtain all in-

kg ha-1) and manure (0, 30 t ha-1) (Table 2). The N, P

formation from the least operational practices due to

and K content of manure were determined as 1.18%,

distribution of experimental points through treatments

0.29% and 1.04%, respectively. The high and low

conined. The design points were deined based on the

levels of manure were determined based on nutrient

low and high levels of N (0, 300 kg ha-1), P (0, 200

content and local recommendations.

Table 2. Actual and coded values of expeimental factors for CCD.
Treatment values*
Runs
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

Nitrogen
(kg ha-1)
300
150
300
300
0
150
150
0
150
150
150
150
0
300
150

Phosphorus
(kg ha-1)
0
0
100
100
100
100
0
100
0
200
200
100
200
200
100

Coefficients
Manure
(t ha-1)
15
15
30
0
30
15
30
0
0
0
30
15
15
15
15

Nitrogen
(X1)
+1
0
+1
+1
-1
0
0
-1
0
0
0
0
-1
+1
0

Phosphorus
(X2)
-1
-1
0
0
0
0
-1
0
-1
+1
+1
0
+1
+1
0

Manure
(X3)
0
0
+1
-1
+1
0
+1
-1
-1
-1
+1
0
0
0
0

* +1, -1, and 0 indicates up, down and medium level of each factor

A normality test was already performed and transfor-

were arranged to sow wheat seeds on double side

mation was also performed for numerical data where

of rows. Manure was well mixed with soil one

needed. To ensure uniformity of treatment variances,

month before planting. The first N installment and

the Bartlett's test was used. Since there was no statis-

the whole P amount were applied at sowing. The

tical difference between both years of data, the mean

sowing dates (October 10, 2013-14) were the same

value of each trait was reported. Data analysis and

for both years of experiment. Plots were immedi-

graph plotting were performed using Minitab® Statis-

ately irrigated after sowing and after that irrigation

tical Software Ver. 16.1.1, and Microsoft Excel Ver. 14

continued at 10-day intervals. An eco-climate appropriate cultivar (Gascogne) was employed in this

2.3. Crop management

study. The plant density was 350 plants per square
meter. Weeds were controlled two times during

Plots of 3 × 4 m with a distance of 1 meter be-

the growth period by hand. No agrochemical

tween, to avoid nutrients mixing due to irrigation,

was used during soil cultivation, planting and

were prepared. Each plot consists of 6 rows that

growing season.

Journal of Soil Science and Plant Nutrition, 2018, 18 (1), 13-26

17

Optimizing application rate of nitrogen

2.4. Measurements

ANUE =

Y Grain
N initial + N fertilizer

Eq.2

Each plot was partitioned into two sections, one for
seed yield and its components and the second sec-

YGrain: seed yield (kg m-2)

tion used for destructive time series sampling during

To determine relative water content (RWC) of leaves,

the crop growth season. Before the inal harvest, 5

the samples were prepared between 9:00 to 10:00 AM

plants from each plot were randomly selected and

at lowering stage. The samples were submerged in

the number of fertile tillers was recorded. Seed yield

distilled water for 6 hours and then turgor weight was

was determined from 4 m2 of each plot which was

measured and their dry weight were also recorded in

kept untouched, considering marginal effect. The air

after drying samples in 75 oC in oven. Finally RWC

dried plants were weighed and biological yield (dry

was calculated using Equation 3 (Kramer, 1988):

matter yield), seed yield and harvest index were also
measured.
The N contents in plant tissue were measured using

RW C =

AOAC Oficial Method by a Kajehldal Semi Auto-

(FW − DW )
(TW − DW )

Eq.3

mated Distillation Unit (Horwitz & Latimer, 2005).
The total N was determined for each soil plot (AOAC

2.5. Statistical analysis

oficial method 968.06 (4.2.04)). Nitrogen loss was
Response of measured variables (y) to experimental

calculated using Equation 1 (Jarvis et al., 2011):

factors (X) was estimated using second order polynomials including the interaction (Equation 4):
Eq.1

Nloss= Ninitial+ Nfertilizer – (Nplant+ Nsoil)

Where, Nloss: nitrogen loss (kg m-2), Ninitial: soil avail-

y = β 0 + ∑ βi X i + ∑ β i j X i X j + ∑ β ii X i 2
m

m

m

i =1

i< j

i =1

Eq.4

able nitrogen content at the early season (kg m-2)
which was calculated by: NTotal End ×0.03
0.03= availability coeficient of total N in soil (as

Where b0 is constant and bi, bij and bii are coeficients

mineral for crop use) (Fageria, 2014).

for linear, interaction and quadratic terms, respec-

Nfertilizer: applied nitrogen (kg m ), Nplant: plant nitrogen

tively.

content at the end of season, and Nsoil: the available ni-

The result was a second order polynomial which de-

-2

trogen content in soil after inal harvest (kg m ) which

scribes the estimated of response (yield) as a func-

was calculated as:

tion of inputs variables. Finally, after optimizing

NTotal End ×0.03.

the resulted function and eliminating the low effect

Agronomical nitrogen use eficiency (ANUE, kg

terms using statistical tests and criteria such as, F

Grain/kg Nfertilizer) was calculated using Equation 2

test, lack of it test, coeficient of determination (R2),

(Rathke et al., 2006):

a inal function was calculated to predict yield and

-2

other expected variables as below (Equation 5):

Journal of Soil Science and Plant Nutrition, 2018, 18 (1), 13-26

18

Jahan and Amiri

Y = a0 + a1X 1 + a2 X 2 + a3 X 3 + a4 X 12 + a5 X 2 2 + a6 X 3 2 + a7 X 1X 2 + a8 X 1X 3 + a9 X 2 X 3

Eq.5

In this function, Y is dependent variable, X1 is N fer-

The RMSE percentage states the different be-

tilizer, X2 is P fertilizer, X3 is manure, and a0 to a9 are

tween predicted versus observed values. When

coeficients. The equation is only functional in the de-

RMSE