New Biotechnology  Volume 28, Number 6  October 2011 RESEARCH PAPER

Statistical optimization of xylitol

production from corncob hemicellulose

Research Paper
hydrolysate by Candida tropicalis HDY-02
LingHongzhi 1, ChengKeke 2, GeJingping
§ 1
and PingWenxiang 1

1
Key Laboratory of Microbiology, College of Life Sciences, Heilongjiang University, Harbin 150080, PR China
2
Institute of Nuclear and New Energy Technology, Tsinghua University, Beijing 100084, PR China

The statistical experimental designs were adopted to optimize the culture medium in xylitol production
by Candida tropicalis HDY-02 with corncob hemicellulose hydrolysate as substrate. In the first step,
Plackett–Burman design was used for screening the important variables. KH2PO4, yeast extract,
(NH4)2SO4 and MgSO47H2O were found to significantly affect xylitol yield. In the second step, central
composite design (CCD) was used to determine the optimum level of each of the significant variables. A
second-order polynomial was determined by the multiple regression analysis of the experimental data.
The interactive effects of yeast extract and MgSO47H2O on xylitol yield of C. tropicalis HDY-02 were
determined to be significant. The validation experimental was consistent with the prediction model. The
optimum combinations for xylitol yield were 5 g l1 (NH4)2SO4, 1.3 g l1 KH2PO4, 4.6 g l1 yeast extract
and 0.6 g l1 MgSO47H2O. Under these optimal conditions, the continuous fed-batch experiments could
produce xylitol of 58 g l1 with a yield of 0.73 g g1 xylose.

Introduction Northeast China, corncob is an ideal raw material to produce

Xylitol, a five-carbon sugar polyalcohol, can be applied to the
food, pharmaceutical and odontological industries [1–3]. Large
scale of xylitol production is carried out by a chemical process
consisting of catalytic xylose hydrogenation from xylan-contain-
ing plant materials in the presence of nickel catalysts [4]. This
process cost a lot, because of the requirement of the purification of
xylitol from the mixture with other sugars and polyols after the
chemical reaction [5]. In recent years, more and more people are
attracted by bioconversion of xylose to xylitol and other chemical
products, using hydrolysate from various lignocelluloses, such as
Eucalyptus grandis chips [6], sugarcane bagasse [7] and corncob [8].
Corncob contains approximately 35% hemicellulose fraction,
which can be easily hydrolyzed to constituent carbohydrates.
These carbohydrates mainly consist of the xylose and other minor
pentose [9–11]. As the most abundant agricultural crop residues in
xylitol by bioconversion.
Biological xylitol production is influenced by the factors of the
concentration of various ingredients in culture medium; therefore,
their optimization study is very important. Response surface
methodology (RSM) is a mathematical and statistical analysis,
which is useful for the modeling and analysis of problems that
the response of interest is influenced by several variables [12]. RSM
was utilized extensively for optimizing different biotechnological
processes [13,14]. In the present study, the screening and the
optimization of variables for xylitol production by Candida tropi-
calis HDY-02 using Plackett–Burman and RSM are reported. The
Plackett–Burman screening design is applied for knowing the most
significant variables enhancing xylitol production. Then, the cen-
tral composite design (CCD) was applied to determine the opti-
mum level of each of the significant variables.

§
Supported by ‘863’ Hi-Tech Research and Development Program of China
Materials and methods
(2007AA100702-6) and Science Fund of Young Scholars, Heilongjiang Uni- Corncob hydrolysate preparation
versity, China. Corncob from Heilongjiang province in Northeast China was used
Corresponding author: Ping, W. (wenxiangp@yahoo.com.cn) as raw material. Particles in the size range from 0.45 mm to 0.9 mm

1871-6784/$ - see front matter ß 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.nbt.2010.05.004 www.elsevier.com/locate/nbt 673
 RESEARCH PAPER New Biotechnology  Volume 28, Number 6  October 2011

(20–40 mesh) were used in the experiments. The biomass at a solid TABLE 1
loading of 10% (w/w) was mixed with dilute sulfuric acid 1% (w/w) Variables for screening using Plackett–Burman design
and pretreated in an autoclave at 1208C for one hour. The liquid Variables Parameter High level (+) Low level ()
fraction was separated by filtration and the unhydrolyzed residue 1
X1 (NH4)2SO4 (g l ) 4 2
was washed with warm water (608C). The filtrate and washings
X2 KH2PO4 (g l1) 3 1
were pooled together. The average compositions of the hemicel-
1
X3 Yeast extract (g l ) 6 4
lulose acid hydrolysate were 17.1 g xylose l1, 7.2 g glucose l1,
2 g arabinose l1, 0.5 g cellobiose l1, 0.9 g galactose l1, 0.3 g X4 NH4NO3 (g l1) 1 0
mannose l1, 4 g acetic acid l1, 1.4 g furfural l1. X5 Peptone 2 0
X6 Urea (g l1) 0.6 0
Detoxification X7 MgSO47H2O (g l1) 0.6 0.2
Research Paper

Hemicellulose acid hydrolysate was heated to 1008C, and then X8 Dummy – –

maintained for 15 min to remove or reduce the volatile compo-
X9 Dummy – –
nents. The hydrolysate was then over-limed with solid Ca(OH)2 up
X10 Dummy – –
to pH 10, in combination with 0.1% sodium sulfite, filtered to
remove the insoluble material. The filtrate was concentrated under X11 Dummy – –
vacuum at 508C to give 46.3 g l1 xylose for storage and later usage.

Microorganism
C. tropicalis HDY-02 was isolated from the soil samples. The culture
was maintained on yeast extract peptone dextrose (YEPD) agar
slant at 48C and subcultured every four weeks.
Growth medium and culture conditions
Loopfuls of cells from the plates were transferred to 250 ml Erlen-
meyer flasks containing 100 ml of the growth medium containing
5 g l1 KH2PO4, 2 g l1 (NH4)2SO4, 1 g l1 peptone, 3 g l1 yeast
extract, 0.2 g l1 MgSO4, 20 g l1 D-xylose. The flasks were main-
tained at 308C under agitation at 200 rpm for 16 h and subse-
quently inoculated into the fermentation medium at 5% (v/v).
The pretreated corncob hemicellulose hydrolysate, adjusted to
pH 7 with 2 M H2SO4 or 3 M NaOH and supplemented with
different KH2PO4, (NH4)2SO4, yeast extract, MgSO4 concentra-
tion for tests according to the selected factorial designs, was used
as the fermentation medium. The 250 ml flasks containing
100 ml medium were fermented at 358C and 200 rpm for 48
hours.
The fed-batch cultivations were conducted in a 5 l stirred-vessel
bioreactor (Biostat, B. Braun, Germany) containing 4 l fermenta-
tion medium under 0.6 vvm air flow. The pH was controlled at six
by automatic addition of 3 M NaOH and all fermentation experi-
ments were carried out at 358C and 200 rpm.
Analytical methods
The liquid samples were analyzed by HPLC, equipped with UV and
RI detectors. The concentrations of glucose, xylose, galactose,
mannose and arabinose were determined using refractive index
detector and Aminex HPX-87P column at 858C with H2O as mobile
phase at 0.8 ml min1. Cellobiose, and acetic acid were analyzed
using refractive index detector and Aminex HPX-87H column at
658C with 5 mM H2SO4 as mobile phase at 0.8 ml min1. Furfural
was detected by UV absorbency at 250 nm. Cell growth was
monitored at 600 nm and converted to cell dry weight (CDW)
by an appropriate calibration curve.
Experimental design
Plackett–Burman experimental design
The design in 12 runs was used for screening the important
parameters affecting xylitol yield. In the design, 11 variables
comprising 7 independent variables and 4 dummy variables were

TABLE 2
The Plackett–Burman design for 11 variables
X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 Xylitol yield (g g1)
1            0.14
2 + +  + + +    +  0.43
3  +  + +  + + +   0.40
4 + +    +  + +  + 0.33
5 + + +    +  + +  0.61
6 +  + +  + + +    0.42
7 +  + + +    +  + 0.55
8  + + +    +  + + 0.49
9 +    +  + +  + + 0.37
10  + +  + + +    + 0.59
11    +  + +  + + + 0.30
12   +  + +  + + +  0.33

674 www.elsevier.com/locate/nbt
New Biotechnology  Volume 28, Number 6  October 2011 RESEARCH PAPER

TABLE 3 The results of the experimental design were analyzed using

Levels of variables used in the experiment design Design Expert Version 7.0.2 (Stat-Ease Inc., Minneapolis, MN,
Variables (g l1) Range and levels USA) statistical software. The estimates of the coefficients with
confidence levels above 95% (P < 0.05) were accepted in the final
2 1 0 +1 +2
models. The F-test was used to evaluate the significance of the
X1, (NH4)2SO4 1 2 3 4 5
models.
X2, KH2PO4 1 2 3 4 5
X3, yeast extract 3 4 5 6 7 Results
X4, MgSO47H2O 0.2 0.3 0.4 0.5 0.6 Screening experiment
Effect of parameter, standard error (SE), t-value and probability (P)
for the design are given in Table 4. It can be seen that effect Eðxi Þ of

used (Tables 1 and 2). All experiments were performed in tripli-
cates and the average values of the observations were used. In the
design, a high (+) and a low () level were set at the point which
was close to its concentration in reference [15–17]. The effect of
each variable was determined with the equation in reference [18].
Central composite design
CCD was chosen to show the statistical significance effects of
KH2PO4, (NH4)2SO4, yeast extract and MgSO47H2O concentrations
on xylitol yield by C. tropicalis HDY-02. CCD allows estimating the
second degree polynomial of the relationships between the factors
and the dependent variable and gives information about interaction
between variables (factors) in their relation to the dependent vari-
able. The ranges of the variables, KH2PO4, (NH4)2SO4, yeast extract
and MgSO47H2O concentrations, were chosen from experiments
carried out previously according to the PB design. The lowest and
the highest levels of variables are given in Table 3.
A 24 factorial CCD with eight star points, and six replicates at
the center points leading to 30 runs were employed for the
optimization of the culture media. The variables were coded
according to the following equation:
ðXi  X0 Þ
xi ¼ i ¼ 1; 2; . . . ; K
DX
where xi is the dimensionless value of a variable, Xi is the real value
of a variable, X0 is the value of Xi at the center point and DX is the
step change.
The second-order polynomial, Eqn (2), which includes all inter-
action terms that were used to calculate the predicted response
X
4 X
4 X
4 where Ŷi is the predicted xylitol yield, X1, X2, X3 and X4 are the
Ŷi ¼ b0 þ bi x i þ bii x2i þ bi j x i x j
i¼1 i¼1 i; j¼1
Ŷi is the predicted response, xi and xj are the input variables, b0 is
the intercept term, bi is the linear effect, bii is the squared effect and
bij is the interaction term.
Research Paper
nitrogen sources (NH4NO3, peptone and urea) from low level to
high level was insignificant and confidence level was less than 95%.
It means that at these levels of NH4NO3, peptone and urea did not
improve xylitol yield, low levels of them were used for further
optimization studies. The effects of (NH4)2SO4, KH2PO4, yeast
extract and MgSO47H2O were significant (P < 0.05). The effects
Eðxi Þ of (NH4)2SO4, KH2PO4, yeast extract and MgSO47H2O were
calculated to be 0.41, 0.038, 0.085 and 0.035. This means that from
low level to high level of (NH4)2SO4, KH2PO4, yeast extract and
MgSO47H2O, xylitol yield increased by 0.41 units, 0.038 units,
0.085 units and 0.035 units. The variables selected for optimization
studies were (NH4)2SO4, KH2PO4, yeast extract and MgSO47H2O.
Central composite design
Table 5 lists the design matrix of experiment results by tests
planned according to the 24 full factorial designs. The xylitol yield
was selected as the response due to the different cycles of the runs.
Thirty experiments were performed in triplicate and the central
point was repeated six times (runs 25–30).
Results from analysis of variance (ANOVA) are given in Table 6.
Regression equation showed that the xylitol yield was an empirical
function of test variables in coded unit as shown in the following
(1) equation:
Ŷi ¼ 0:51 þ 2:5E  003X1  0:029X2  0:043X3  0:012X4
 0:015X1 X2  1E  002X1 X3 þ X1 X4 þ 3:75E
 003X2 X3 þ 1:25E  003X2 X4 þ 0:016X3 X4
þ 0:016X1 X1 þ 0:022X2 X2 þ 0:017X3 X3 þ 0:011X4 X4 (3)
(2)
concentration of (NH4)2SO4, KH2PO4, yeast extract and
MgSO47H2O, respectively.
Table 6 lists the results from ANOVA for this experiment. The
Fisher’s F-test (Fð14;15Þ ¼ S2m =S2s ¼ 7:64 > Ft ð14;15Þ ¼ 3:56) with a

TABLE 4
Effect, standard error, t-value and probability (P) for the Plackett–Burman design
S. no. Variables (g l1) Effect Eðx i Þ Standard error (SE) t-Value P-value
1 (NH4)2SO4 0.41 7.546E003 54.33 0.0366
2 KH2PO4 0.038 7.546E003 5.036 0.0146
3 Yeast extract 0.085 7.546E003 11.26 0.0078
4 NH4NO3 0.018 7.546E003 2.39 0.1358
5 Peptone 0.032 7.546E003 4.24 0.0524
6 Urea 0.013 7.546E003 1.72 0.2193
7 MgSO47H2O 0.035 7.546E003 4.64 0.0435

www.elsevier.com/locate/nbt 675
 RESEARCH PAPER New Biotechnology  Volume 28, Number 6  October 2011

TABLE 5 and 0.0463 (P > 0.1), respectively, which indicated that ‘lack of fit’
Experiment design with real values and predicted values of xylitol was significant. The determination coefficient (R2 = 0.877) and
production adjusted determination coefficient (Adj R2 = 0.7622) were also
Runs X1 X2 X3 X4 Xylitol yield (g g1) satisfactory to confirm the significance of the model. Value of
Experimental Predicted coefficient of variation (CV = 5.4%) is shown in Table 6, which
indicates that the conducted experiments were precise and
1 1 1 1 1 0.68 0.65
reliable.
2 1 1 1 1 0.68 0.71 The significance of regression coefficients was determined by
3 1 1 1 1 0.64 0.62 P-value which are presented in Table 7. It could be concluded
4 1 1 1 1 0.65 0.61 that X2 and X3 had the significant linear effect on the response.
5 1 1 1 1 0.55 0.55 In comparison, X1 and X4 played less significant effect on xylitol
Research Paper

6 1 1 1 1 0.59 0.56 yield in the ranges of this study. The quadric effect of X22 was
significant at 1% level while the quadric effect of X21 , X23 and the
7 1 1 1 1 0.50 0.53
interaction effect of X3X4 were significant at 5% level; others
8 1 1 1 1 0.49 0.48
were insignificant. It could be concluded that the linear effects of
9 1 1 1 1 0.58 0.60
X2, X3 and X22 were the primary determining factors of the
10 1 1 1 1 0.70 0.65 response on Ŷi , the quadric effect of X21 , X23 and the interaction
11 1 1 1 1 0.56 0.56 effect of X3X4 were the secondary determining factors, other
12 1 1 1 1 0.55 0.56 terms of the model had insignificant effect on Ŷi as shown in
13 1 1 1 1 0.54 0.56 Table 7 and Eqn (3). Among them, the quadric term X21 , X22 , X23
14 1 1 1 1 0.54 0.57
and the interaction term X3X4 were positive coefficient implied
an enhancement on Ŷi .
15 1 1 1 1 0.56 0.54
The contour plot in Fig. 1 presents the effect of yeast extract and
16 1 1 1 1 0.49 0.49
MgSO47H2O on the xylitol yield, while the other two variables
17 2 0 0 0 0.57 0.57 were held at a constant level. An elliptical profile of the contour
18 2 0 0 0 0.56 0.58 plots indicated remarkable interaction between the independent
19 0 2 0 0 0.66 0.66 variables. The design expert presented the maximal numerical
20 0 2 0 0 0.52 0.54 solution with the predicted xylitol yield up to 0.70 g g1.
21 0 0 2 0 0.63 0.66
Verification
22 0 0 2 0 0.51 0.49
Verification of the calculated optimum conditions for xylitol yield
23 0 0 0 2 0.55 0.58
was done by performing the experiment at optimized conditions.
24 0 0 0 2 0.54 0.53 Under these conditions C. tropicalis produced xylitol yield
25 0 0 0 0 0.48 0.51
26 0 0 0 0 0.52 0.51
TABLE 7
27 0 0 0 0 0.51 0.51
Model coefficient estimated by multiplies linear regression
28 0 0 0 0 0.52 0.51
Factor Coefficient estimated P-value
29 0 0 0 0 0.52 0.51
Intercept 0.51 0.0002**
30 0 0 0 0 0.52 0.51
b1 2.5E003 0.6932
b2 0.029 0.0003**
very low probability value [(Pmodel > F) = 0.0002] indicated the
b3 0.043 <0.0001**
model was highly significant. The P-value of the model was
0.0002, which was used as a tool to check the significance of each b4 0.012 0.0801
coefficient, and indicated that the model was suitable for use in b21 0.016 0.0169*
this experiment. The F-value and P-value of ‘lack of fit’ were 4.92 b22 0.022 0.0019**

b23 0.017 0.0109*

TABLE 6 b24 0.011 0.0876
ANOVA for full quadratic model b1b2 0.012 0.0675
Source SS DF MS F-value Probe > F b1b3 1E002 0.2087
Model 0.099 14 7.085E003 7.64 0.0002 b1b4 0 1.0000
Residual 0.014 15 9.272E004 b2b3 3.75E003 0.6294
Lack of fit 0.013 10 1.263E003 4.92 0.0463 b2b4 1.25E003 0.8718
Pure error 1.283E003 5 2.567E004 b3b4 0.016 0.497
Total 0.11 29 *
Significant at 5% level.
**
CV = 5.4%; R2 = 0.877; Adj. R2 = 0.7622. Significant at 1% level.

676 www.elsevier.com/locate/nbt
New Biotechnology  Volume 28, Number 6  October 2011
FIGURE 1
Isoresponse contour plot for yeast extract and MgSO47H2O.
0.71 g g1 which is in agreement with the predicted value
0.70 g g1 suggesting that Eqn (3) is valid for xylitol yield.
As the high cost of xylitol recovery from aqueous solution, an
economical production of xylitol from corncob hydrolysate
requires the improvement of both product concentration and
productivity. Fed-batch cultivations were studied deeper to
increase the final xylitol concentration in both. The substrate
xylose level was regulated to 20–30 g l1 with concentrated
hydrolysate containing 113.5 g xylose l1, 19.7 g glucose l1,
14.6 g arabinose l1 during the fed-batch fermentation period.
As shown in Fig. 2, the fed-batch experiment in the 5 l fermenter
led to improved xylitol production. In the 78 hours with the
optimum culture medium conditions, the xylitol concentration,
FIGURE 2
Time course of xylitol fed-batch fermentation by Candida tropicalis with
treated corncob hydrolysate as substrates: xylose (&), glucose (~), arabinose
(!), xylitol (*), acetate (~), CDW (*).
RESEARCH PAPER
productivity and yield were 58 g l1, 0.74 g l1 hour1 and
0.73 g g1 xylose, respectively.
Discussion
KH2PO4 and MgSO47H2O as the essential nutrients influencing
the growth and chemical composition of polyols in yeasts have
been reported by Mahler et al. [19]. Vandeska et al. observed that
the xylitol yield of C. boidinii increased when the fermentation
medium was supplemented with ammonium sulfate [20].
Vongsuvanlert and Tani obtained the highest productivity
when the medium was supplemented with the yeast extract
Research Paper
as a nitrogen source [21]. In this study, we investigated
the important ingredients in culture medium which can affect
the xylitol yield using statistically based experimental design.
The significant effects of KH2PO4, MgSO47H2O, yeast extract
and (NH4)2SO4 on the xylitol yield were found. The positive
effects of yeast extract and MgSO47H2O on the xylitol yield
obtained from this study support the above findings. Further,
the analysis of the experimental responses revealed that yeast
extract and MgSO47H2O had significant interactive effects on
xylitol yield.
Carla et al. studied the effect of xylose concentration and
inoculum level in rice straw hydrolysate by using Candida guillier-
mondii FTI 20037. They reported that the maximum values of
xylitol batch productivity (0.54 g l1 hour1) and xylitol yield
(0.65 g g1) after 96 hours fermentation [22]. Mussatto and
Roberto investigated the kinetic behavior of C. guilliermondii dur-
ing the xylitol production from the rice straw hemicellulosic
hydrolysate which was treated with activated charcoal.
0.32 g l1 hour1 xylitol productivity and 0.53 g g1 yield were
obtained after 116 hours fermentation [23]. In this study, the
higher xylitol productivity of 0.74 g l1 hour1 and yield of
0.73 g g1 were obtained. Rivas et al. reported that higher xylitol
productivity of 1.49 g l1 hour1 in the high cell density cultures
with membrane bioreactors using Debaryomyces hansenii grown in
synthetic media containing corncob hemicellulose hydrolysate
[8]. If a similar membrane bioreactor can be applied in xylitol
fermentation by C. tropicalis HDY-02, there would be a promising
possibility to improve xylitol productivity.
According to our previous study, C. tropicalis HDY-02 can pro-
duce 180 g l1 xylitol in fed-batch cultures using xylose as the sole
carbon source. However, the fermentation parameters for these
treated corncob hemicellulose acid hydrolysate are much lower
than those obtained with a synthetic medium. This result showed
that there were some leftover toxic components in the treated
hemicellulose acid hydrolysate that negatively affected the fer-
mentation performance of C. tropicalis. To further improve the
fermentation efficiency, efforts should be made to establish more
efficient hydrolysate detoxification strategies.
Conclusions
Plackett–Burman and CCD were proved to be a powerful tool for
the optimization of xylitol yield by C. tropicalis HDY-02 using the
corncob hemicellulose hydrolysate as a carbon source. The impor-
tant parameters selected by Plackett–Burman screening experi-
ments were KH2PO4, yeast extract, (NH4)2SO4 and MgSO47H2O.
By CCD, the interactions among the significant factors screened,
which was not found in previous reports, was determined. The
www.elsevier.com/locate/nbt 677
 RESEARCH PAPER
interactive effects of yeast extract and MgSO47H2O on xylitol
yield of C. tropicalis HDY-02 were determined to be significant. A
second-order model was obtained to describe the relationship
between xylitol yield and the parameters (the concentration of
(NH4)2SO4, KH2PO4, yeast extract and MgSO47H2O in culture
medium). Validation experiments verified the availability and
References
1 Ahmet, S. and Gurbuz, G. (2006) Influence of some bulk sweeteners on rheological
properties of chocolate. LWT-Food. Sci. Technol. 39, 1053–1058
Research Paper
2 Takahashi, Y. et al. (2007) Formulation and evaluation of lactoferrin bioadhesive
tablets. Int. J. Pharmacol. 343, 220–227
3 Massoth, D. et al. (2006) The effect of xylitol on Streptococcus mutans in children.
Calif. Dent. Assoc. J. 34, 231–234
4 Mikkola, J.P. and Salmi, T. (2001) Three-phase catalytic hydrogenation of xylose to
xylitol-prolonging the catalyst activity by means of on-line ultrasonic treatment.
Catal. Today 64, 271–277
5 Nigam, P. and Singh, D. (1995) Processes for fermentative production of xylitol – a
sugar substitute. Process. Biochem. 30, 117–124
6 Silvio, S.S. et al. (1998) Acid hydrolysis of Eucalyptus grandis chips for microbial
production of xylitol. Process. Biochem. 33, 63–67
7 Santos, J.C. et al. (2005) Influence of aeration rate and carrier concentration on
xylitol production from sugarcane bagasse hydrolyzate in immobilized-cell
fluidized bed reactor. Process. Biochem. 40, 113–118
8 Rivas, B. et al. (2002) Bioconversion of posthydrolysed autohydrolysis liquors an
alternative for xylitol production from corncob. Enzyme. Microb. Technol. 31,
431–438
9 Lisbeth, O. and Bärbe1, H.H. (1996) Fermentation of lignocellulosic hydrolysates
for ethanol production. Enzyme. Microb. Technol. 18, 312–331
10 Balan, V. et al. (2009) Ligocellulosic biomass pretreatment using AFEX. Methods
Mol. Biol. 581, 61–77
11 Liaw, W.C. et al. (2008) Xylitol production from rice straw hemicellulose
hydrolyzate by polyacrylic hydrogel thin films with immobilized Candida
subtropicalis WF79. J. Biosci. Bioeng. 105, 97–105
12 Montgomery, D.C. (2001) Design and Analysis of Experiments. John Wiley and Sons
678 www.elsevier.com/locate/nbt
New Biotechnology  Volume 28, Number 6  October 2011
the accuracy of the model. Under optimal condition (the concen-
tration of (NH4)2SO4, KH2PO4, yeast extract and MgSO47H2O of
5 g l1, 1.3 g l1, 4.6 g l1 and 0.6 g l1, respectively), a fed-batch
experiment produced xylitol of 58 g l1 at a fermentation of 78 h.
The study also provided useful information and reference for the
optimization of other biotechnological processes.
13 Li, W. et al. (2007) Optimization of whole cell-catalyzed methanolysis of soybean
oil for biodiesel production using response surface methodology. J. Mol. Catal. B-
Enzym. 45, 122–127
14 Naveena, B.J. et al. (2005) Direct fermentation of starch to l(+) lactic acid in SSF by
Lactobacillus amylophilus GV6 using wheat bran as support and substrate: medium
optimization using RSM. Process. Biochem. 40, 681–690
15 Pessoa, A. et al. (1996) Cultivation of Candida tropicalis in sugar cane hemicellulosic
hydrolyzate for microbial protein production. J. Biotechnol. 51, 83–88
16 Kwon, S.G. et al. (2006) Increase of xylitol productivity by cell-recycle
fermentation of Candida tropicalis using submerged membrane bioreactor. J. Biosci.
Bioeng. 101, 13–18
17 Walther, T. et al. (2001) The influence of aeration and hemicellulosic sugars on
xylitol production by Candida tropicalis. Bioresour. Technol. 76, 213–220
18 Jiang, Y. et al. (2008) Statistical study on the effects of environmental factors on the
growth and microcystins production of bloom-forming cyanobacterium –
Microcystis aeruginosa. Harmful Algae. 7, 127–216
19 Mahler, G.F. et al. (1994) Influence of Magnesium ion on growth, ethanol and
xylitol production by Pichia stipitis NRRL 7124. Biotechnol. Lett. 16, 407–412
20 Vandeska, E. et al. (1995) Effects of environmental condition on production of
xylitol by Candida boidinii. World J. Microbiol. Biotechnol. 11, 213–218
21 Vongsuvanlert, V. and Tani, Y. (1989) Xylitol production by a methanol yeast,
Candida boidinii (Kloeckera sp.) No. 2201. J. Ferment. Bioeng. 67, 35–39
22 Carla, J.S.M. et al. (2001) Optimization of xylitol production by Candida guilliermondii
FTI 20037 using response surface methodology. Process. Biochem. 36, 1119–1124
23 Mussatto, S.I. and Roberto, I.C. (1995) Kinetic behavior of Candida guilliermondii
yeast during xylitol production from highly concentrated hydrolysate. Process.
Biochem. 39, 1433–1439