Accepted Manuscript Title: Extraction Optimization of the Polysaccharide from Adenophorae Radix by Central Composite Design Author: Xiaoqian Zhang Jiye Chen Meixin Mao Huaizhong Guo Yaru Dai PII: DOI: Reference:
S0141-8130(14)00211-6 http://dx.doi.org/doi:10.1016/j.ijbiomac.2014.03.039 BIOMAC 4251
To appear in:
International Journal of Biological Macromolecules
Received date: Revised date: Accepted date:
8-1-2014 3-3-2014 15-3-2014
Please cite this article as: X. Zhang, J. Chen, M. Mao, H. Guo, Y. Dai, Extraction Optimization of the Polysaccharide from Adenophorae Radix by Central Composite Design, International Journal of Biological Macromolecules (2014), http://dx.doi.org/10.1016/j.ijbiomac.2014.03.039 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Extraction Optimization of the Polysaccharide from
2
Adenophorae Radix by Central Composite Design
3
Xiaoqian Zhanga, Jiye Chena, Meixin Maoa, Huaizhong Guoab*, Yaru Daia a b
Key Laboratory of Pharmaceutical Quality Control of Hebei Province, Baoding 071002, Hebei, PR China
7
Abstract:The Central Composite Design (CCD) was applied to optimize the water
8
extraction of the polysaccharide from Adenophorae Radix in the paper. The three
9
variables of extraction temperature (X1), extraction time (X2) and ratio of water to raw
10
material (X3) were investigated by single factor analysis firstly. Since the presence of
11
active polysaccharides and the imperfect of its extraction, the purpose of the paper
12
was to evaluate the effects of selected variables on the yield of polysaccharide, which
13
was expected to obtain the maximum yield. By variance and regression analysis, the
14 15 16 17 18
te
d
M
an
us
6
Ac ce p
5
College of Pharmacy, Hebei University, Baoding, 071002, PR China
cr
4
ip t
1
quadratic regression equation was established as a predicted model. The R2 of 0.9825 indicated that the equation was well-fitted. The optimal conditions were 72.5
, 133
min, 1:35 (g/mL) and the predicted maximum yield of the polysaccharide was 5.78%. The predicted value was verified in triplicates under the optimum conditions, which was 5.68% and well matched with the predictive yield.
* Corresponding author: Tel.: +86 0312 5971107; fax: +86 0312 5971107. Email:
[email protected] (H.Z. Guo).
Page 1 of 21
19
Key Words: Polysaccharide; Extraction; Central Composite Design (CCD); Response
20
surface methodology (RSM); Adenophorae Radix
21
1. Introduction Adenophorae Radix (Nan shashen), a traditional Chinese medicine, is the dry
23
root of Adenophora tetraphylla (Thunb.) Fisch. or Adenophora stricta Miq. [1]. It
24
has been reported that the Adenophorae Radix has components such as cycloartenyl
25
acetate, lupenone, β-sitosterol, taraxerone, octacosanoic acid, and praeruptorin A [2].
26
A capillary zone electrophoresis method was used for the analysis of the
27
monosaccharide composition of Adenophorae Radix polysaccharides [3]. The
28
monosaccharide composition of Adenophorae Radix polysaccharides consisted of
29
xyloce, arabinose, glucose, rhamnose, mannose, galactose, glucuronic acid and
30
galacturonic acid. The related modern pharmacology research indicates that
31
Adenophorae Radix polysaccharides have the functions of anti-aging [4], reversing
32
memory impairment [5], immune-regulating and anti-radiation, which was closely
33
related to the elimination of oxygen free radicals [6]. However, little information
34
published about the quantitative and separation process optimization of Adenophorae
36 37 38
cr
us
an
M
d
te
Ac ce p
35
ip t
22
Radix polysaccharides [7,8]. So, it’s important to systematically optimize the extraction process of the polysaccharide from Adenophorae Radix. Response surface methodology (RSM) is a collection of mathematical and
statistical techniques based on the fit of a polynomial equation to the experimental
39
data, which describes the behavior of a data set with the objective of making
40
statistical previsions. It can be well applied when a response or a set of responses of
41
interest is influenced by several variables. The objective of the method is to
42
simultaneously optimize the levels of these variables to achieve the best systemic
43
performance [9]. Orthogonal experimental design and uniform design can only give
Page 2 of 21
the best combination of factors levels, and can not find the optimal value in the
45
whole area. RSM can predict the outcome (response) from the multivariate quadratic
46
equation. RSM has been applied in food [10-12] and pharmaceutical [13,14] fields.
47
Box-Behnken design (BBD) [9,15] and central composite design (CCD) [9,16] are
48
the two common methods in the RSM. By adjusting the asterisk arm and the number
49
of central point, CCD can acquire good properties such as orthogonality, rotation and
50
versatility.
us
cr
ip t
44
By using RSM,the present study was designed to optimize the conditions for the
52
extraction of the polysaccharide from Adenophorae Radix. A five-level, three-factor
53
Orthogonal Rotation Central Composite Design was employed to study the effects of
54
temperature, time and ratio of water to raw material on the extraction process. It can
55
provide theoretical guidance for the industrial production of polysaccharide from
56
Adenophorae Radix.
57
2. Materials and methods
58
2.1 Materials and chemicals
60 61 62
M
d te
Ac ce p
59
an
51
Adenophorae Radix, which originated from Yunnan province, purchased from a
pharmacy of Baoding (China) and identified by Guimin Tian, who is the deputy chief pharmacist of Baoding Institute for Drug Control. D-Glucose (20110114) was purchased from Tianjin Kermel Chemical Reagent Co., Ltd. (China). Other reagents
63
were all analytical grade.
64
2.2 Extraction of Polysaccharide
65
The Adenophorae Radix was ground with a high speed disintegrator, then a 60
66
mesh sieve was used to process and homogenize the particle size, and 5.0 g of the
Page 3 of 21
67
obtained powder was extracted two times using designed temperature, time and ratio
68
of water to raw material with water. The extraction solutions were combined and
69
concentrated, then precipitated by adding ethanol to a final concentration of 80% (v/v)
70
at 4
71
The precipitate was defatted by the method of sevag [17] (chloroform : butyl alcohol
72
= 4:1): the precipitate was re-dissolved in 25 mL water, and then mixed with sevag
73
reagent (v/v=5:1). After magnetic stirring for 20 min, the supernatant was obtained by
74
centrifugation at 4000 rpm for 5 min. The chloroform-protein gel was formed in the
75
middle layer after centrifugation. The process was repeated several times until the
76
protein completely removed, that is, when the chloroform-protein gel could not be
77
seen. The supernatant was concentrated and precipitated by adding ethanol to a final
78
concentration of 80% (v/v) at 4
79
constant weight to obtain the crude polysaccharide.
80
2.3 Determination of polysaccharide yield
82 83 84 85 86 87
ip t
cr
us
an
M
to a
te
d
for 8 h again. The precipitate was dried at 60
The crude polysaccharide was ground into powder. The content of the
Ac ce p
81
for 8 h. After centrifugation at 4000 rmp for 8 min, the precipitate was obtained.
polysaccharide in dried powder was measured by phenol-sulfuric method using glucose as standard at 490 nm [18]. The percentage yield (%) of polysaccharide was calculated as follows:
Yield (%) = (weight of polysaccharide in the whole dried crude polysaccharide /weight of the related Adenophorae Radix powder) × 100 2.4 Experimental design and statistical analysis
88
In this work, single factor analysis was employed to determine the central points
89
and the ranges of independent variables including X1 (extraction temperature), X2
90
(extraction time) and X3 (ratio of water to raw material). Five experimental points
Page 4 of 21
91
were setted up, which were 55, 65, 75, 85, 95
for X1, 1, 2, 3, 4, 5 h for X2 and 1:10,
92
1:20, 1:30, 1:40, 1:50 g/mL for X3. When a factor (X1, X2 or X3) was investigated, the
93
levels of other factors remained constant. Based on the single factor analysis, zero
94
levels of X1, X2 and X3 were determined to be 75
95
respectively. Table 1 displayed the coded and uncoded levels of these independent
96
variables. And then, the CCD, shown in Table 2, was applied to investigate the effects
97
of independent variables on the extraction of Adenophorae Radix polysaccharide. The
98
yield of the polysaccharide was taken as the response. Each factor was coded at five
99
levels according to the following equation [19]:
an
Xi − X z ΔX i
us
cr
ip t
, 120 min and 1:30 (g/mL),
100
xi =
101
Where xi is the dimensionless coded value of an independent variable; Xi is the real
102
value of an independent variable; Xz is the real value of an independent variable at the
103
center point; and ΔXi is the step change of the real value of the variable i.
104
Table 1
106 107 108
te
d
M
(1)
Ac ce p
105
i=1, 2, 3....... k
Independent variables and levels (coded and uncoded) of Central Composite Design
In this study, the total runs of CCD were 23, which consisted of central point,
cubic points and axial points. The data from the design were used to establish a predicted model. The empirical second-order polynomial model was shown as
109
follows:
110
Υ = β + ∑ β x + ∑ β x2 + ∑ ∑ β x x
111
Where Y is the response; xi and xj are variables (i and j ranged from 1 to k); β0 is the
112
constant term; βj is the linear coefficient, βij are interaction coefficient, and βjj is the
k
0
j =1 j
k
j
j =1 jj
k
j
i < j = 2 ij i
j
(2)
Page 5 of 21
quadratic coefficient; k is the number of independent parameters (k=3 in this study)
114
[20-22].
115
Table 2
116
Coded and uncoded levels of independent variables and the response
cr
ip t
113
The experimental data were analyzed by multiple regression analysis through the
118
generalized least square. Pareto analysis of variance (ANOVA) was used to estimate
119
the statistical parameters. Minitab 16 software was utilized to the response surface
120
analysis.
121
3. Results and discussion
122
3.1 Model fitting and statistical analysis
M
an
us
117
The observed values of response (yield of polysaccharide) based on RSM were
124
given in Table 2. ANOVA of the experimental data was summarized in Table 3. By
125
applying multiple regression analysis on the experimental data, the quadratic
127 128 129 130
te
Ac ce p
126
d
123
polynomial model for the predicted yield of polysaccharide was shown as follows (in the form of coded values):
Y=5.6112-0.1333X1+0.3693X2+0.2971X3+0.1388X1X2-0.2137X1X3-0.1712X2X3-X12 -0.3019X22-0.2824X32
ANOVA was used to test the significance of the model. The statistical
131
significance of all the terms of the model was tested by the F-value and the P-value.
132
The corresponding variables would be more significant if the F-value became greater
133
and the P-value became smaller [23]. The model F-value of 81.11 implied that the
134
model was statistically significant, and there was only a 0.05% chance that the
Page 6 of 21
“Model F-value” was due to the noise. It could be seen from Table 3 that the liner
136
coefficients (X1, X2, X3), the quadratic term coefficients (X12, X22, X32) and
137
interaction term coefficients (X1X2, X1X3, X2X3) were statistically significant due to
138
the very small P-values (P < 0.05), which indicated that these terms were important to
139
the yield of polysaccharide.
140
Table 3
141
Estimated regression coefficient (coded) and variance analysis of the predicted model
us
cr
ip t
135
Determination coefficient (R2), adjusted determination coefficient (R2adj),
143
predicted determination coefficient (R2pre) were used to estimate the goodness of the
144
fit of the model and were listed in Table 3. The R2 value was 0.9825, which indicated
145
that 98.25% of the variations could be explained by the predicted model. The R2adj
146
value of 0.9704, indicated the high degree of correlation between the observed and
147
predicted values [19]. The R2pre value of 0.9237 was in agreement with R2 and R2adj.
148
The analysis results demonstrated that the relationship between factors and the
150 151 152 153
M
d
te
Ac ce p
149
an
142
response of the predicted model was well-correlated. The F-value and P-value of the lack of fit were 1.45 and 0.306, respectively, which implied that they were not significant. Thus, the regression equation was not lack of fit. Residual analysis is used to evaluate the adequacy of the model, since it can find
the abnormal date. Residuals are the difference between the observed value and the
154
fitted value of a model. Small residual value indicates that model prediction is
155
accurate [24]. Fig.1 was the residuals plots, which were used to evaluate whether the
156
residuals were normal or not. The normal probability plot of residuals (upper left of
157
Fig.1) was approximate along a straight line, and the residuals of fitted value plot
Page 7 of 21
(upper right of Fig.1) were randomly distributed around the zero point. It indicated
159
that the adequacy of the empirical model was satisfactory.
160
Fig.1. The residual plots of the developed model about the yield of polysaccharides.
161
3.2 Analysis of response surface and validation of the optimized conditions
cr
ip t
158
The three-dimensional (3D) response surface plots and two-dimensional (2D)
163
contours plots were the graphical representation of quadratic regression equation
164
(Fig.2). The 3D response surface plots were established to illustrate the impact on the
165
yield of polysaccharide when the two horizontal axes represented any two in the three
166
independent variables. Simultaneously, the remaining independent variable was kept
167
at zero level. For example, Fig.2A showed the effects of temperature (X1), time (X2)
168
and their interaction on the yield of polysaccharide, and that the ratio of water to raw
169
material (X3) was kept at zero level (1:30). Fig.3 was the contour plots, in which the
170
values were equivalent in the same contour line. The shape of 2D contour plots was
171
circular or elliptical. The circular plot indicated that the interactions between the
175
Ac ce p
te
d
M
an
us
162
176
Fig.3. 2D contour plots of the developed model for the yield of polysaccharide.
172 173
174
177
corresponding variables could be negligible, while elliptical plot indicated that the interactions between the corresponding variable were significant [25].
Fig.2. 3D Response surface plots of the developed model for the yield of
polysaccharide.
With the RSM analysis, the optimized conditions that obtained from Minitab16
178
software were extraction temperature of 72.5
, extraction time of 133 min and ratio
179
of water to raw material of 1:35 (g/mL). Under the optimized conditions, the
Page 8 of 21
predicted yield of polysaccharide was 5.78%. The replication experiment was
181
performed under the optimized conditions and the mean value (n=3) was 5.68%,
182
which was well in agreement with the predicted value. Experimental and predicted
183
values of the relative error and relative standard deviation (RSD) are 1.76% and
184
1.24%, respectively. This result showed that the model could be used to optimize the
185
extraction of polysaccharide from Adenophorae Radix.
186
4.Conclusion
us
cr
ip t
180
In this study, CCD was successfully used to optimize the extraction conditions of
188
polysaccharide from Adenophorae Radix. The independent variables (extraction
189
temperature, extraction time, and ratio of water to raw material) had great impact on
190
the yield of polysaccharide. ANOVA and multiple regression analysis were applied to
191
achieve the predicted model. The predicted optimum yield was 5.78% with the
192
extraction temperature of 72.5 , extraction time of 133 min and ratio of water to raw
193
material of 1:35. The result of replication experiment (5.68%, n=3) matched the
194
predicted value under the optimized conditions. The increase of polysaccharide yield
196 197
M
d
te
Ac ce p
195
an
187
is cost savings and improves the economic benefit. The research provides the basis of further research and industrial development of polysaccharide from Adenophorae Radix.
References
[1] Chinese Pharmacopoeia Commission, The Chinese Pharmacopoeia, Part 1, China Medical Science Press, China, 2010, 228. [2] S.S. Roh, S.H. Kim, Y.C. Lee, Y.B. Seo, Effects of Radix Adenophorae and Cyclosporine A on an OVA-Induced Murine Model of Asthma by Suppressing to T Cells Activity, Eosinophilia, and Bronchial Hyperresponsiveness, Mediators of
Page 9 of 21
Inflammation 2008 (2008) 1-11. [3] H.Z. Guo, F.L. Liu, G.Y. Jia, W.Q. Zhang, F. Wu, Extraction optimization and analysis of monosaccharide composition of fucoidan from Saccharina japonica by
ip t
capillary zone electrophoresis, Journal of Applied Phycology 25 (2013) 1903-1908.
cr
[4] C.H. Li, Y. Li, X.F. Li, Y. Li, Anti-aging action of Radix Adrenophorae Potaninii Korsh polysaccharides in mice and fruit-flies, Chinese Pharmacological Bulletin
us
18 (2002) 452-455.
an
[5] C.M. Zhang, X.F. Li, Improving effect of Radix Adenophorae Potanini Korsh Polysaccharides (RAPS) on scopolamine-induced impairment of learning and
M
memory in rats, Pharmacology and Clinics of Chinese Materia Medica 17 (2001) 19-21.
d
[6] Y. Li, C.H. Li, F.T. Tang, X.F. Li, Pharmacological Action of Adenophora
te
Polysaccharides, Chinese Journal of Integrative Medicine 10 (2004) 78-80. [7] Z.A. Feng, H.M. Gong, T.J. Shao, M.Q. Liao, Establishment and Comparison of
Ac ce p
Content Determination Methods for Polysaccharides in Radix Adenophorae, China Pharmacist 15 (2012) 290-293.
[8] S.P. Wang, X.H. Xing, Study on separation technology optimization to ladybell root polysaccharide, Journal of Changchun Institute of Technology (Nature Science Edition) 11 (2010) 114-116.
[9] M.A. Bezerra, R.E. Santelli, E.P. Oliveira, L.S. Villar, L.A. Escaleira, Response surface methodology (RSM) as a tool for optimization in analytical chemistry, Talanta 76 (2008) 965-977. [10] Y.Y. Chen, H.Y. Luo, A.P. Gao, M. Zhu, Extraction of Polysaccharides from Mango (Mangifera indica Linn.) Seed by Response Surface Methodology and
Page 10 of 21
Identification of their Structural Characteristics, Food Analytical Methods 5 (2012) 800-806. [11] X.L. Liu, T.H. Mu, H.N. Sun, M. Zhang, Optimisation of aqueous two-phase
ip t
extraction of anthocyanins from purple sweet potatoes by response surface methodology, Food Chemistry 141 (2013) 3034-3041.
cr
[12] S.H. Wu, G.L. Gong, Y.Y. Wang, F. Li, et al, Response surface optimization of enzyme-assisted extraction polysaccharides from Dictyophora indusiata,
us
International Journal of Biological Macromolecules 61 (2013) 63-68.
an
[13] M.B. Lan, J. Guo, H.L. Zhao, H.H. Yuan, Optimization of the Extraction of the Magnolia officinalis Polysaccharides Using Response Surface Methodology,
M
Asian Journal of Chemistry 24 (2012) 2290-2294.
[14] Y.K. Hong, W.J. Liu, T. Li, S.Y. She, Optimization of extraction of Eucommia
d
ulmoides polysaccharides by response surface methodology, Carbohydrate
te
Polymers 92 (2013) 1761-1766.
[15] J. Prakash Maran, S. Manikandan, K. Thirugnanasambandham, C. Vigna Nivetha, Dinesh,
Box–Behnken
Ac ce p
R.
design
based
statistical
modeling
for
ultrasound-assisted extraction of corn silk polysaccharide, Carbohydrate Polymers 92 (2013) 604-611.
[16] T. Zhu, H.J. Heo, K.H. Row, Central Composite Design for Optimized Extraction of Polysaccharides from Undaria pinnatifida, Chemical Research in Chinese Universities 28 (2012) 620-623.
[17] M.G. Sevag, D.B. Lackman, J. Smolens, The Isolation of The Components of Streptococcal Nucleoproteins In Serologically Active Form, The Journal of Biological Chemistry 124 (1938) 425-436. [18] DuBois Michel, K.A. Gilles, J.K. Hamilton, P.A. Rebers, Colorimetric Method
Page 11 of 21
for Determination of Sugars and Related Substances, Analytical Chemistry 28 (1956) 350-356. [19] J. Prakash Maran, V. Mekala, S. Manikandan., Modeling And Optimization of
ip t
Ultrasound-Assisted Extraction Of Polysaccharide From Cucurbita moschata, Carbohydrate Polymers 92 (2013) 2018-2026.
cr
[20] G.E.P. Box, J.S. Hunter, Multi-factor Experimental Designs for Exploring Response Surface, The Annals of Mathematical Statistics 28(1957) 195-241.
us
[21] V. Samavati, Central composite rotatable design for investigation of
an
microwave-assisted extraction of okra pod hydrocolloid, International Journal of Biological Macromolecules 61 (2013) 142-149.
M
[22] J. Prakash Maran, V. Sivakumar, R. Sridhar, V. Prince Immanuel, Development of model for mechanical properties of tapioca starch based films, Industrial Crops
d
and Products 42 (2013) 159-168.
te
[23] A.C. Atkinson, A.N. Donev, Optimum experimental designs, Clarendon Press, Britain, 1992, 132-189.
Ac ce p
[24] V. Samavati, Polysaccharide extraction from Abelmoschus esculentus: Optimization by response surface methodology, Carbohydrate Polymers 95 (2013) 588-597.
[25] Q.S. Shao, Y.M. Deng, Optimization of polysaccharides exteaction from Tetrastigma hemsleyanum Dielset using response surface methodology, International Journal of Biological Macromolecules 49 (2011) 958-962.
Page 12 of 21
Table 1. Independent variables and levels (coded and uncoded) of Central Composite
199
Design
200
Table 2. Coded and uncoded levels of independent variables and the response
201
Table 3. Estimated regression coefficient (coded) and variance analysis of the
202
predicted model
203
Fig.1. The residual plots of the developed model about the yield of polysaccharides.
204
Fig.2. 3D Response surface plots of the developed model for the yield of
205
polysaccharide.
206
Fig.3. 2D contour plots of the developed model for the yield of polysaccharide.
an
us
cr
ip t
198
Ac ce p
te
d
M
207
Page 13 of 21
207
Table. 1 levels of variables Symbol -1.682
-1
0
﹢1
﹢1.682
92
X1
58
65
75
85
Time (min)
X2
70
90
120
150
X3
1:13
1:20
1:30
raw materia l(g/mL)
1:40
1:47
us
ratio of water to
170
cr
Temperature ( )
ip t
Independent Variables
Ac ce p
te
d
M
an
208 209
Page 14 of 21
209
Table. 2 Coded and uncoded variables Run
ratio of water to raw material(g/mL) X3 -1(1:20)
1
Temperature(℃) X1 1(85)
Time(min) X2 1(150)
2
0(75)
0(120)
1.682(1:47)
3
0(75)
0(120)
0(1:30)
4
0(75)
0(120)
0(1:30)
5
0(75)
0(120)
-1.682(1:13)
4.20
6
0(75)
0(120)
0(1:30)
5.59
7
0(75)
0(120)
8
1(85)
-1(90)
9
-1.682(58)
0(120)
10
-1(65)
-1(90)
11
0(75)
12
Yield(%)
ip t
5.10
5.37
5.73
us
5.74
-1(1:20)
3.85
0(1:30)
4.72
-1(1:20)
3.97
-1.682(70)
0(1:30)
4.06
1(85)
-1(90)
1(1:40)
4.22
13
-1(65)
1(150)
1(1:40)
5.34
14
0(75)
0(150)
0(1:30)
5.54
15
1(85)
1(150)
1(1:40)
4.92
16
0(75)
0(120)
0(1:30)
5.53
17
0(75)
0(120)
0(1:30)
5.52
18
0(75)
0(120)
0(1:30)
5.57
19
-1(65)
1(150)
-1(1:20)
4.80
20
0(75)
0(150)
0(1:30)
5.76
21
0(75)
1.682(170)
0(1:30)
5.40
22
1.682(92)
0(120)
0(1:30)
4.44
23
-1(65)
-1(90)
1(1:40)
5.33
Ac ce p
te
M
an
0(1:30)
d
cr
5.53
210 211
Page 15 of 21
Table. 3 SSa
MSb
DFc
Estimated coefficient
F-value
P-value
Model
8.71904
0.96878
9
5.6112
81.11
0.000
X1
0.24278
0.24278
1
-0.1333
20.33
0.000
X2
1.86266
1.86266
1
0.3693
155.94
X3
1.20564
1.20564
1
0.2971
100.94
X 1 X2
0.15401
0.15401
1
0.1388
12.89
0.003
X 1 X3
0.36551
0.36551
1
-0.2137
30.60
0.000
X 2 X3
0.23461
0.23461
1
-0.1712
19.64
0.001
X 12
2.00191
2.00191
1
-0.3549
167.60
0.000
X 22
1.44842
1.44842
1
-0.3019
121.26
0.000
X 32
1.26786
1.26786
1
-0.2824
106.15
0.000
Residual error
0.15528
0.01194
Lack of fit
0.07372
0.01474
1.45
0.306
Pure error
0.08156
0.01019
212 213 214 215 216
cr
us
an
M
0.000
5
R2
0.9825
0.677103
R2adj
0.9704
0.109291
R2pre
0.9237
8
Ac ce p
S
0.001
13
d
PRESS
ip t
Source
te
211
a Sums of squares. b Mean square.
c Degree of freedom.
Page 16 of 21
ip t cr us an
fig. 1
M
216 217
Ac ce p
te
d
218
Page 17 of 21
fig. 2
us
cr
ip t
218
220
Ac ce p
te
d
M
an
219
Page 18 of 21
ip t cr us
221
Ac ce p
te
d
M
an
222
Page 19 of 21
fig. 3
us
cr
ip t
222
224
Ac ce p
te
d
M
an
223
Page 20 of 21
ip t cr us
225
Ac ce p
te
d
M
an
226
Page 21 of 21