INTRODUCTION

In guava, a crop grown successfully in a variety of
soils with pH ranging from 5.5 to 8.0, deficiency of both
major and micronutrients is reported extensively (Pathak
and Pathak, 1993). Besides, in India, the crop is scarcely
fertilized. Incipient deficiency or hidden hunger is causing
considerable damage to guava crop, resulting in economic
losses (Singh and Singh, 2007). In order to avoid yield loss,
nutrient management strategies need to be evolved for this
crop based on comprehensive nutrient diagnostic norms.
Several approaches are adopted for identification of nutrient
imbalances, a recent one being the Compositional Nutrient
Diagnosis (Parent and Dafier, 1992). CND norms are
multivariate norms that give due weightage to all the
elements, including unmeasured factors and, therefore, have
higher diagnostic sensitivity. The present investigation was
carried out to develop multivariate diagnostic norms using
CND to improve diagnostic precision and to understand
interaction among different nutrients governing yield and
quality of the guava crop.

Compositional nutrient diagnosis norms (CND) for guava  (Psidium guajava L.)

K. Anjaneyulu, H. B. Raghupathi and M. K. Chandraprakash
Division of Soil Science and Agricultural Chemistry

Indian Institute of Horticultural Research, Bangalore-560 089, India
E-mail: anjaney@iihr.ernet.in

ABSTRACT

Multivariate nutrient diagnostic norms were developed for guava using compositional nutrient diagnosis
(CND) through leaf nutrient concentration vs. yield data bank. CND norms for N (V

N
), P (V

P
) and K (V

K
) were

2.48, 0.23 and 2.13, respectively. Norms for N and K were much higher compared to P, indicating higher
requirement of these two nutrients. CND norms are multivariate norms that consider all elements, including
unmeasured factors and, therefore, has higher diagnostic sensitivity. Among micronutrients, Fe requirement
was much higher than all other nutrients. Interaction among different nutrients was explained by principal
component analysis conducted on log-transformed data which produced four significant PCs, explaining about
73.66% of the variance. The four Eigen values added up to 8.1 denoting the four significant PCs. The first PC
was positively correlated with P, Zn and R (residue, which is a reflection of dry matter accumulation in the
plant) and negatively correlated with Ca, Mg, S and Fe, indicating that P and Zn behaved in one direction and
the other elements in opposite direction. In the second PC, antagonistic effect of N, Fe with P and Cu was
evident. In PC3, P and Mg were negatively correlated with Mn and Cu.  In PC4, N and S showed their behaviour
in the same direction. Diagnostic norms developed were used for identification of yield-limiting nutrients in
low-yielding orchards. Thus, diagnostic norms and nutrient interactions help evolve nutrient management
strategies for guava to realize higher yields and better quality.

Key words: Nutrients, diagnosis, norms, CND, PCA, guava

MATERIAL AND METHODS

Sampling

During July-August, a survey was conducted in
guava orchards cultivating ‘Allahabad safeda’ in and around
Bangalore and Kolar (mostly alfisols) in Karnataka, and,
280 leaf samples were collected to develop nutrient
diagnostic norms. Samples were collected from 70 orchards
(around 15 years of age) by selecting the 3rd pair of leaves
from apex, which provides the index leaf (recently matured
leaf) in guava. From each orchard, 25 to 30 trees were
selected and 50 leaves per plant were collected randomly
at chest height from all sides of the tree to form a composite
and representative sample (Bhargava and Chadha, 1993).
The leaf samples were decontaminated by washing in a
sequentially with tap water, 0.2% detergent solution, N/10
HCl and, finally, with double distilled water. Leaf samples
were dried at 65-70oC for 48 h. The samples were then
powdered in a Cyclotec Mill and analyzed for different
nutrients by digesting 1g tissue in di-acid mixture (9:4 ratio
of nitric acid and perchloric acid) using standard analytical

J. Hortl. Sci.
Vol. 3 (2): 132-135, 2008



133

methods (Jackson, 1973).  Phosphorus was analyzed by
vanado-molybdate method, K by flame-photometer and S
by the turbidity method. Ca, Mg, Fe, Mn, Zn and Cu were
analyzed using Atomic Absorption Spectrophotometer
(Perkin-Elmer-A-Analyst-200). N was separately estimated
by micro-Kjeldhal method. Thus, nutrient ‘concentration
vs. yield’ data bank was established for developing nutrient
diagnostic norms.

Compositional Nutrient Diagnosis

CND norms were developed by adopting the
procedure outlined by Parent and Dafir (1992). Full
composition array for nutrient proportions (D) in plant
tissues was described by the following simplex (SD)
contained to 100%:

S D=[(N, P, K,… R): N>O,P>O,K>O,…,R>O;
N+P+K+…+R =100%]          — 1

where, 100% is dry matter content (i.e., the invariable sum
of all the components or full relative composition of the
diagnostic tissues); N, P, K are nutrient concentrations and
R is filling value between 100% and sum of the nutrient
concentrations. The value of R is, thus, composed of
undetermined components as well as experimental error,
and was required to linearize compositional data.

Bounded sum constraint to 100% of compositional
data was alleviated by correcting nutrient concentrations by
geometric mean  (G) of all the D components, including R.

G = [ N x P x K x ….. x R]1/D       —— 2

Row centered log-ratios were generated for V
N
 to V

Zn
 as

follows:

V
N 

= ln (N/G),…..,V
Zn 

= ln (Zn/G)      —— 3

Expressions such as N/G,… Zn/G are multi-nutrient
ratios, since, each nutrient is divided by the geometric mean
of all components (the nutrients determined and the filling
value). Row-centered log-ratios are linearized (undistorted)
estimates of the original components fully compatible with
PCA.

V*
N
 to V*

Zn 
and SD*

N 
to SD*

Zn
 are CND norms

(indicated by asterisks), i.e., mean and standard deviation
of each row-centered log ratios in the high-yielding
population. The standardized variables (V

N 
- V

N
*) / SD*

N

to (V
Zn 

- V*
Zn

) / SD*
Zn

 are CND nutrient indices.

I
N
 = (V

N
 - V*

N
) / SD*

N
,… I

Zn
 = (V

Zn 
- V

Zn
*) / SD*

Zn  
— 4

 Independent values for V
N
 to V

Zn
 were introduced

in the equation for diagnostic purpose.

Principal component analysis (PCA)

PCA application could lead to greater
understanding of nutrient interactions in the plant. PCA
reduces the number of interdependent variables to a smaller
number of independent PCs that are linear combinations of
original variates. Therefore, PCA was performed on log-
transformed data of the original nutrient concentrations,
prior to statistical computation that followed normal
distribution.

Selection criteria

To be declared significant, PCs must have Eigen
values >100/P, where P is the total number of varieties under
diagnosis. Alternatively, PCs showing Eigen values <1 were
considered not significant. PC loadings in Eigen vectors
having values greater than selection criterion (SC) only are
considered significant. Selection criterion was computed
as follows:

SC = 0.50 / (PC eigen values) 0.5

RESULTS AND DISCUSSION

Nutrient concentration range

The mean N concentration was 1.91% and ranged
from 1.33 to 2.48% (Table 1). Maximum yield in guava
was reported when N concentration in the leaf ranged
between 1.40 and 2.0% (Singh and Rajput, 1981). The mean
P concentration was 0.20% and varied from 0.14 to 0.26%,
which was comparable to the values (0.18 to 0.24%)
published earlier by Khanduja and Garg (1980). The mean
K concentration was 1.35% and varied widely between 0.90
and 1.85%. Higher range of Ca concentration (1.50 to
2.60%) was observed, whereas, a majority of the samples
were in the optimum range with regard to Mg (0.30 to
0.75%). The mean S concentration was 0.37% and was
comparable to the earlier report. Concentrations of Fe and
Mn ranged from 104 to 197 and 25 to 98 ppm, respectively.
The mean Zn concentration was 56 ppm.

Table 1. Mean and range of nutrient concentration for guava

Nutrient Unit Mean Minimum Maximum
N % 1.91 1.33 2.48
P % 0.20 0.14 0.26
K % 1.35 0.90 1.85
Ca % 1.14 0.53 2.41
Mg % 0.43 0.33 0.57
S % 0.38 0.21 0.53
Fe ppm 148.90 104.00 197.00
Mn ppm 56.86 25.00 98.00
Zn ppm 31.66 21.00 47.00
Cu ppm 8.37 4.30 13.60

CND norms for guava

J. Hortl. Sci.
Vol. 3 (2): 132-135, 2008



134

CND norms

CND norms for N (V
N
), P (V

P
) and K (V

K
) for

guava were 2.48, 0.24 and 2.13, respectively. CND norms
are multivariate norms with due weightage to all the other
elements, including unmeasured factors. Sum of the tissue
components is 100% and, therefore, the sum of row-centered
log ratios (including filling value) is zero (Table 2). CND
norm values developed were difficult to comprehend
compared to nutrient concentrations, expressed as % or ppm.
However, CND norms have higher diagnostic precision
compared to the bivariate values, as in the case of diagnosis
and recommendation integrated system (Walworth and
Sumner, 1987). The Ca norm (1.91) was twice as high as
that of Mg (0.99). Among micronutrients, Fe had higher
norm value of –2.377 and, therefore, its requirement was
much higher, compared to Mn, Zn or Cu.

Principal component analysis

PCA conducted on log-transformed data produced
four significant PCs and four Eigen values added up to 8.10
explaining about 73.66% of variance (Table 3). Since PCs
are the linear contrasts among nutrients, interpretation of
PCs considers the sign of the variate. The first PC was
positively correlated with K, Fe, Zn and R (residue, which
is a reflection of dry matter accumulation in the plant) and
was negatively correlated with Ca, Mg, and S indicating
that K, Fe, and Zn behaved in one direction and the rest of
the elements in an opposite direction. In the second PC,
antagonistic effect of N and Fe with P and Cu was evident.
In PC3, P and Mg were negatively correlated to Mn. In

Table 2. Compositional nutrient diagnosis norms for guava

CND variate CND norm SD
V

N
2.480 0.120

V
P

0.236 0.140
V

K
2.131 0.191

V
Ca

1.906 0.319
V

Mg
0.987 0.117

Vs 0.844 0.164
V

Fe
-2.377 0.172

V
Mn

-3.383 0.311
V

Zn
-3.932 0.201

V
Cu

-5.291 0.283
V

R
6.394 0.083

Table 3. Principal component analysis (PCA) loading performed
on log- transformed data

Nutrient PC1 PC2 PC3 PC4
N 0.196 0.572* 0.322 0.547*
P 0.242 -0.414* 0.591* 0.026
K 0.800* -0.127 -0.034 0.354
Ca -0.837* 0.145 0.117 -0.139
Mg -0.598* 0.099 0.631* -0.274
S -0.714* 0.045 0.058 0.556*
Fe 0.431* 0.635* 0.091 -0.358
M n 0.002 0.349 -0.795* -0.139
Zn 0.677* -0.280 -0.028 0.077
Cu -0.094 -0.813* -0.123 -0.142
Residue 0.712* 0.198 0.432* -0.268
Eigen values 3.46 1.852 1.713 1.075
% Variance 31.49 49.32 63.89 73.66
  Selection criteria 0.268 0.367 0.382 0.482
* Significant over selection criteria

Anjaneyulu et al

J. Hortl. Sci.
Vol. 3 (2): 132-135, 2008

Table 4. CND indices for selected, low-yielding orchards of guava

N P K Ca Mg S Fe Mn Zn Cu R
-0.80 1.23 1.44 0.53 -0.28 0.95 -0.61 -0.45 -0.07 -0.77 -1.53
2.24 -0.53 -0.52 0.92 2.32 -0.08 0.00  1.70 -0.61 -0.77 0.16

-0.32 -0.35 0.76 -0.38 -0.88 -0.21 0.10 -0.25 1.27 0.16 -0.01
-1.09 -1.07 -0.85 -0.05 -1.49 -0.87 -0.33 0.75 1.22 1.40 -0.06
0.90 0.70 0.29 -0.61 -0.57 -0.86 -0.81  1.58 1.03 2.15 -0.11
2.83 -1.22 1.25 -0.58 -1.12 -0.54 0.05 0.40 -0.96 -1.35 1.08

-0.02 0.17 1.86 -0.73 -1.21 -1.28 0.24 -0.19 1.27  0.42 1.49
-0.08 0.21 2.18 -0.79 -1.55 -0.48 0.64 -1.07 1.66 -0.27 0.97
-0.21 -0.55 1.57  1.94 0.21 -0.29 -0.82  0.47 -0.64 2.01 0.85
0.53 1.21 1.38  2.20 -0.58 -0.78 0.56  1.56 2.52 0.67 1.57

-0.74 0.12 1.10 -1.52 -0.15 -1.24 0.36 -1.38 2.02 1.73 0.89
-0.41 1.51 -1.19 1.76 2.04 0.58 -1.29 -0.70 -0.63 -1.64 -0.19
0.64 1.48 1.62  1.05 -1.01 2.10 0.07  1.55 -0.10 0.32 -1.07

-1.66 0.09 1.01 -1.09 -2.38 -2.69 -0.63 2.15 1.47 1.17 -1.09
-2.46 0.30 1.16 -0.61 -1.61 -1.63 -0.43 1.45 1.38 0.43 -0.74
2.29 -0.71 -0.14  1.13 -1.32 0.72 0.69 1.81 1.23  2.07 -0.66

-0.27 -2.06 -1.05 1.49 0.36 1.44 -0.41 0.23  -0.32 -0.36 -1.91
1.37 -3.19 -0.64 0.82 -0.86 2.29 -0.53 0.27 -0.04 0.41 -2.38

R = Residue

PC4, N and S were isolated (Raghupathi et al, 2002). These
nutrient interactions need to be considered for correction of
nutrient deficiencies and for evolving nutrient management
strategies for guava for realizing higher yield and better
quality.



135

CND indices

Independent values were introduced from low-
yielding orchards for the purpose of diagnosis of a nutrient
that limits the yield. Among the eighteen low-yielding
orchards studied, N was found to limit yield in as many as
ten orchards, whereas, P was low in nine and K in six
orchards. The micronutrients were also found to be either
low or deficient, as reflected by the indices (Table 4).
However, no single nutrient was found solely responsible
for low yield. Correlation among indices indicated that N
indices correlated with none of the nutrient indices, whereas,
there was an overwhelming negative correlation between
K and Ca indices, indicating their antagonism. Indices for
Zn were negatively correlated with S and positively
correlated with Fe. Multi-nutrient diagnosis developed
through CND and nutrient interactions elucidated through
PCA were found to have higher precision in diagnosing
nutrient imbalance in guava and, are thus helpful in evolving
nutrient management strategies.

REFERENCES

Bhargava, B.S. and Chadha, K.L. 1993. Leaf nutrient guides
for fruit crops. In: Advances in Horticulture-Fruit
crops, Vol. 2. pp 973-1030. Chadha, K. L. and Pareek,
O. P. (eds), Malhotra Publ. House, New Delhi

Jackson, M.L. 1973. Soil Chemical Analysis, Prentice Hall
of India, New Delhi, p 408

Kanduja, S.D. and Garg, V.K. 1980. Nutritional status of
guava (Psidium guajava L.) trees in North India. J.
Hortl. Sci., 55:433–435

Parent, L.E. and. Dafir, M. 1992. A theoretical concept of
compositional nutrient diagnosis.  J. Amer. Soc. Hortl.
Sci., 117:239-242

Pathak, R.A. and Pathak, R.K. 1993. Guava nutrition. In:
Advances in Horticulture-Fruit crops, Vol. 2. pp 867-
878. Chadha, K.L. and Pareek, O.P. (eds), Malhotra
Publ. House, New Delhi

 Raghupathi, H.B., Reddy, B.M.C. and Srinivas, K. 2002.
Multivariate diagnosis of nutrient imbalance in
banana. Commun. Soil. Sci. Pl. Analysis, 33:2131-
2144

Singh, H.P. and Gorakh Singh. 2007. Nutrient and water
management in guava. Procs. First Int’l. Guava
Symp. G. Singh, R. Krishun, E. Chandra (eds). Acta
Hort.,735:389 - 398

Singh, N.P. and Rajput, C.B.S. 1981. Leaf analysis as a
guide to nitrogen nutrition of guava (Psidium guajava
L.). Prog. Hort., 13:27- 39

Walworth, J.L. and Sumner, M.F. 1987. The diagnosis and
recommendation integrated system. Adv. Soil Sci.,
6:149-188

(MS Received 17 June 2008, Revised 1 November, 2008)

J. Hortl. Sci.
Vol. 3 (2): 132-135, 2008

CND norms for guava