Introduction

Oil palm (Elaeis guineensis) is one of the most economical oil crops. The process of palm oil production has generated empty fruit bunches, fiber and palm shell as wastes [1]. With the increasing demand for energy, biofuel from renewable raw materials has been attractive because it is easily accessible, locally abundant and rich in lignocelluloses [2]. Recently, attempts have been made to apply EFB for bioethanol production [3–7]. Generally, bioethanol production from lignocellulosic materials employs three major steps: pretreatment for breakdown of lignin and opening up the crystalline structure of cellulosic materials; hydrolysis for fermentable sugar production; and bioconversion of fermentable sugar produced to bioethanol [8]. Although chemical and physicochemical pretreatments have been widely investigated, inhibitory compounds, e.g., furfural and hydroxymethyl furfural, are released and further affect the fermentation process [9, 10]. Therefore, biological pretreatment is interesting and has the additional advantages of simple technique and low pretreatment requirements resulting in low operating cost and environmentally friendly process [11]. White-rot fungi are important microorganisms involved in lignin degradation during pretreatment [12–14]. Phanerochaete chrysosporium [15–18] and Trametes versicolor [19–22] have been reported for the pretreatment of lignocellulosic materials. Moreover, Xylariaceous fungi and Marasmius sp. have been reported as lignin-degrading microorganisms [13, 23–25]. However, different white-rot fungi differed in their capabilities of cellulose and lignin degradation from one biomass to another [26]. Some fungi not only degrade lignin effectively, but also consume cellulose simultaneously leading to low available cellulose for bioethanol production. A common measure of delignification efficiency is the selectivity value (SV) of a fungal pretreatment, defined as the ratio of lignin degradation (LD) to cellulose degradation (CD) [18, 27, 28]. A low SV means a relatively high cellulose loss during fungal pretreatment. Thus, the SV of a fungal pretreatment is used to screen white-rot fungi for biological pretreatment [28].

The limitation of biological pretreatment is a lower reaction rate and requires longer pretreatment time than chemical pretreatment [14]. Although the strategies of strain improvement may help resolve some of the drawbacks, the technical process is quite challenging. Another approach to improve the efficiency of biological pretreatment is through the optimization of nutrient and environmental cultivation to reach maximum lignin degradation and minimum cellulose degradation. Shi et al. [16] found that the moisture content and culture time affected the fungal pretreatment of cotton stalk using P. chrysosporium. Alam et al. [29] indicated that the moisture content, inoculum size and wheat flour as co-substrate affected ligninase production during the fungal pretreatment of oil palm biomass. Levin et al. [30] reported that the balance of cellulose and ligninolytic enzyme production during fungal pretreatment depended on pH, peptone and copper. Although, there have been many researches that studied the factors affecting fungal pretreatment, those involved a one-factor-at-a-time experiment or examination of only a few factors. Moreover, the screening of significant factors affecting the fungal pretreatment of EFB has not been studied. In this research, the systematic evaluation of the optimization for the fungal pretreatment was investigated. Statistically designed experiments are a powerful tool to get more information about the system being studied with a minimum number of experiments [31]. The Plackett–Burman design (PBD) has been frequently used for screening process variables that make the greatest impact on a process [32]. Response surface methodology (RSM) is a statistical and mathematical technique useful for developing, improving and optimizing the processes of an interest variable. RSM offers a large amount of information from a small number of experiments and reduces time [10, 33].

This research aims to identify the selective lignin-degrading white-rot fungus with high lignin degradation and low cellulose loss. Also, the optimum condition was studied to improve the efficiency of fungal pretreatment. PBD was used for screening the significant factors for fungal pretreatment during solid-state cultivation. Box–Behnken design was then applied to determine the optimum level of each of the significant factors for delignification with a high SV.

Methods

EFB preparation

EFBs were collected from Thai Tallow and Oil Co., Ltd., Thailand. The sample was dried and crushed into 5–10 mm fibrous length using a hammer mill, then ground to pass through a 1 mm screen (18 meshes) and kept for use in the whole experiment. The chemical composition of EFB, given on dry weight basis, was as follows: 37.6 % cellulose, 21.5 % hemicellulose and 19.0 % lignin [34].

Microorganisms

Phanerochaetechrysosporium ATCC 24725 was obtained from the Faculty of Agro-industry, Prince of Songkla University. Xylaria sp. BCC 7749, Trametes sp. BCC 8729 and Marasmius sp. BCC 9542 were received from the BIOTEC culture collection (BCC), National Center for Genetic Engineering and Biotechnology (BIOTEC), National Science and Technology Development Agency (NSTDA). Trametes versicolor TISTR 3224 was purchased from Thailand Institute of Scientific and Technological Research (TISTR).

Inoculum preparation

Phanerochaete chrysosporium ATCC 24725 was maintained and grown on potato dextrose agar (PDA) at 37 °C for 7 days. Xylaria sp. BCC 7749, Trametes sp. BCC 8729, Marasmius sp. BCC 9542 and Trametes versicolor TISTR 3224 were maintained and grown on PDA at room temperature (28 °C) for 7 days [35].

Selection of fungi for the pretreatment of EFB

The fungal pretreatment was carried out in 250 ml Erlenmeyer flasks. The fermentation medium containing 4 g of dried EFB, 0.08 g of wheat straw and 15 ml of sodium acetate buffer (10 mM, pH 5.0) was sterilized at 121 °C for 15 min. Each flask was inoculated with four fungal pieces (1 cm2) obtained from a PDA plate [35] and incubated for 14 days at the same temperature as mentioned for the inoculum preparation.

Screening parameters for the fungal pretreatment by PBD

PBD was employed to identify the significant factors for the fungal pretreatment by T. versicolor TISTR 3224. The experiment was conducted in 250 ml Erlenmeyer flasks. From PBD, 12 experimental runs with 7 variables including the initial moisture content (%), initial pH, inoculum size (%), wheat flour concentration (%), EFB concentration (%), incubation time (week) and mineral salt solution (%) were performed with low and high levels (Table 1). Each flask was inoculated with a final spore concentration of 1 × 105 spores/ml. The initial moisture content was adjusted with the addition of 10 mM sodium acetate buffer (pH 5). The pH and moisture contents were not adjusted during the fermentation process. All experiments were incubated at room temperature (28 °C) and repeated in triplicate.

Table 1 Experimental definition for Plackett–Burman and Box–Behnken design for fungal pretreatment
Full size table

PBD is based on the first-order model (Eq. 3) and the effect of each variable was determined using Eq. 3 [32].

Y = β 0 + ∑ β i x i
(1)
E ( x i ) = 2 ∑ M i + - ∑ M i - N
(2)

where Y is the response, β0 is the model intercept, β i is the linear coefficient, x i is the level of the independent variable, Ex i is the standardized effect of the tested variable, Mi+ and Mi− are the responses from trials where the variable (x i ) is presented at high and low levels, respectively, and N is the number of experimental runs. The analysis of linear regression was carried out using SPSS version 17.

Box–Behnken design (BBD) for the optimization of T. versicolor pretreatment

The significant variables from PBD analysis were subsequently optimized using the BBD with three levels (low, medium and high, coded as −1, 0, and +1). Table 1 shows the factor codes and values used in this experiment. All experiments were carried out in triplicate and the percentages of LD, CD and SV were taken as responses. The responses were fitted to the following second-order polynomial model (Eq. 3):

Y = β 0 + β 1 X 1 + β 2 X 2 + β 3 X 3 + β 12 X 1 X 2 + β 13 X 1 X 3 + β 23 X 2 X 3 + β 11 X 1 2 + β 22 X 2 2 + β 33 X 3 2
(3)

where Y is the response, X1 the concentration of co-substrate (%), X2 the percentage of moisture content (%) and X3 the concentration of EFB (%). β0 is intercept β1β2β3β12β13β23β11β22β33.

Enzymatic hydrolysis condition

The enzymatic hydrolysis of pretreated EFB under the optimum condition obtained from BBD was carried out in triplicate to determine cellulose digestibility. The enzymatic hydrolysis was conducted in a 1.5 ml microtube with a final volume of 1 ml, containing 50 mg of pretreated EFB, 50 mM sodium citrate buffer of pH 4.8 and 50 μl of 5 % sodium azide (to inhibit microbial contamination). The samples were hydrolyzed by mixed enzymes: 10 FPU/g substrate of cellulase (Celluclast 1.5 L, Sigma) (with a cellulase activity of 61.7 FPU/ml and a xylanase activity of 1,307 U/ml), 267 U/g substrate of xylanases (Optimash BG, Japan) (with an enzyme activity of 1,374.6 U/ml) and 0.66 U/g substrate of β-glucosidase (Novozyme 188, Sigma) (with an enzyme activity of 13.2 U/ml). The reaction mixtures were incubated at 50 °C and taken after 72 h of saccharification. Then, the samples were centrifuged at 10,000 rpm for 10 min, separated and stored at −20 °C for a reducing sugar determination.

Analytical methods

After the pretreatment, the residual biomass was filtered and dried at 105 °C to a constant weight. Weight losses in total solids were calculated from the initial and final dry weights. Then the samples were analyzed for acid detergent fiber, lignin and cellulose [34]. LD and CD were defined as the percentage of total lignin and total cellulose reduced during pretreatment [14, 18]. SV of degradation was calculated as the ratio of LD to CD [27].

Glucose concentration was analyzed using a high-performance liquid chromatography system (Ultimate-3000 RS) equipped with a refractive index detector (RI-Shodex) and an Aminex HPX-87H column (BioRad). The column temperature was set at 65 °C. Samples were eluted at a flow rate of 0.5 ml/min with 5 mM H2SO4. All analyses were performed in triplicate. Cellulose digestibility was calculated using the following equation [7]:

Cellulose digestibility ( % ) = amount of glucose produced × 0.9 × 100 amount of cellulose in pretreated biomass

Statistical analysis

The software SPSS version 17 was used for statistical and linear regression analysis. All component degradation data were subjected to analysis of variance (ANOVA). Multiple comparison tests were performed with Turkey’s test.

Results and discussion

Effect of fungal stains on the pretreatment of EFB

The weight and component losses of EFB pretreated with white-rot fungi are shown in Table 2. To select the selective lignin-degrading white-rot fungus, a high SV greater than 1.0 was considered [27]. Although P.chrysosporium ATCC 24725 had the highest cellulose and lignin degrading ability (31.9 and 18.8 %, respectively), it gave an SV of 0.59. It indicated that P. chrysosporium could simultaneously degraded both lignin and cellulose. It was presumed that cellulose might be hydrolyzed and consumed by P. chrysosporium during pretreatment of lignocellulosic biomass [14].

Table 2 Weight loss and degradation efficiency after 2 weeks of pretreatment of empty fruit bunch by different white-rot fungi
Full size table

In this study, only T. versicolor TISTR 3224 was the selective lignin-degrading fungus with the highest SV of 1.57. A high SV means better prospects for preferential lignin degradation and a low SV means a relatively high cellulose degradation during biological pretreatment leading to low available cellulose [36]. However, a high amount of remaining cellulosic biomass after fungal pretreatment is desired for further hydrolysis and ethanol production. Consequently, T. versicolor TISTR 3224 was selected for further studying the optimization of environmental conditions for fungal pretreatment.

Screening parameters for fungal pretreatment by T. versicolor TISTR 3224 using PBD

The experimental matrix and the values of responses are presented in Table 3. The statistical analysis of the responses for the fungal pretreatment is shown in Table 4. All the examined factors did not significantly affect SV and had a low confidence value (<85 %). It indicated that the total variations were not satisfactorily explained by this model and these variables were considered insignificant. The factors that had a significant effect on the LD were moisture content and wheat flour as co-substrate, with a high confidence value (>95 %). In the case of CD, the co-substrate was the only significant factor with a high confidence value (>95 %). It was observed that the co-substrate affected both LD and CD with a standardized effect of 8.23 and 7.10, respectively. These positive effects mean that if the concentrations of wheat flour are increased, the cellulose loss can be increased as well. The efficiency of pretreatment is not considered but only lignin degradation; however, the availability of cellulose is also an important criterion for further hydrolysis and ethanol production. Consequently, the optimum value of wheat flour concentration was further studied.

Table 3 PBD variables in code levels with lignin degradation (LD), cellulose degradation (CD) and selectivity value (SV) as responses
Full size table
Table 4 Level of the variables and statistical analysis of PBD on lignin degradation (LD), cellulose degradation (CD) and selectivity value (SV) as responses
Full size table

Also, the moisture content was considered because it had a significantly positive effect on LD, with a standardized effect of 3.57. The moisture content was a key factor for fungal growth in solid-state cultivation. Too low moisture contents could limit fungal delignification without providing sufficient water to fungal growth [14]. Higher moisture contents cause clogging interparticle spaces, limited oxygen circulation, often inhibited aerobic solid-state cultivation and increased susceptibility to bacterial contamination [16, 37]. Another effect of high moisture content is reduced solid loading for the fungal pretreatment [14]. Therefore, the optimum amount of moisture content for the high LD was further investigated.

Although EFB concentration gave a low confidence value for both LD and CD (5.8 and 90 %, respectively), it gave the highest negative effect (Ex i value of −3.24) for CD (Table 4). The negative effect means that if the EFB concentration was added on a decreasing trend, the CD could be improved. Accessibility of cellulose for enzymatic hydrolysis is one of the major factors influencing the hydrolysis process. The optimum EFB concentration was also examined.

According to statistical analysis, the initial pH, inoculum size, incubation time and mineral solution were the insignificant factors with low confidence levels (<90 %) for all responses, except mineral solution in the LD model (Table 4). However, this factor gave a negative effect on LD, which means that if a lower mineral solution was added LD could be improved. This might be due to the sufficient inorganic salts in natural lignocellulosic materials required for fungal growth [16]. For the inoculum size between 5 and 15 %, this study indicated that only a very small quantity of mycelium was enough to inoculate the substrate for fungal pretreatment. An increase of inoculum quantity could shorten the time required for substrate colonization and also aid the inoculated fungus to displace any other microbes that may be present [37]. However, a high inoculum size might lead to an exhaustion of nutrients in the fermentation medium [29].

Optimization of key parameters using Box and Behnken design

BBD was adopted to establish the optimum point of each variable affecting LD, CD and SV. The results of the observed and predicted responses are presented in Table 5. A second-order polynomial model was fitted to the experimental data and constructed using the Design Expert software. The models were developed in the following equations (Eqs. 3, 5, 3) with only significant coefficients (P < 0.05):

Table 5 Box–Behnken design matrix along with the experimental (exp.) and predicted (pred.) values of lignin degradation (LD), cellulose degradation (CD) and selectivity value (SV) after the pretreatment of EFB by T. versicolor TISTR 3224
Full size table
Y 1 = 12.21 + 3.61 X 1 + 1.09 X 2 + 0.11 X 3 - 3.44 X 1 2 - 0.08 X 2 2 - 0.99 X 3 2 - 0.62 X 1 X 2 + 1.08 X 1 X 3 - 2.29 X 2 X 3
(4)
Y 2 = 13.29 + 3.98 X 1 - 0.43 X 2 - 0.74 X 3
(5)
Y 3 = 1.01 + 0.08 X 1 + 0.12 X 2 + 0.04 X 3 - 0.27 X 1 2 - 0.05 X 2 2 - 0.19 X 3 2 - 0.20 X 1 X 2 + 0.03 X 1 X 3 - 0.11 X 2 X 3
(6)

where Y1 = lignin degradation (%), Y2 = cellulose degradation (%), Y3 = selectivity value (%), X1 = wheat flour concentration, X2 = moisture content and X3 = EFB concentration.

Table 6 shows the ANOVA for the response surface model. The significance of the models was also confirmed by high F values of the regression with low probability values (P value <0.05). However, a lack of fit F values for LD, CD and SV models was 0.43, 1.34 and 0.074, respectively, with high P value (>0.05). It implied that it was not significant relative to the pure error. There were 75.0, 44.7 and 97.0 % chances that a lack of fit F value could occur due to experimental errors from the LD, CD and SV models, respectively. Non-significant lack of fit would be appropriate for this experiment. Also, a low probability value (P value <0.05) obtained from the regression ANOVA demonstrated that the quadratic model of LD (P value = 0.0244) and linear model of CD (P value = 0.0057) were significant. In contrast, a probability value of the quadratic model of SV (P value = 0.0916) was not significant. For this reason, the SV model was not considered.

Table 6 The ANOVA of Box–Behnken design for the lignin degradation (LD), cellulose degradation (CD) and selectivity value (SV) after the pretreatment of EFB by T. versicolor TISTR 3224
Full size table

The goodness of fit of the model was checked using the determination coefficient (R2). The models of LD and CD were given R2 of 0.893 and 0.636, respectively. This explained the high level of correlation between the experimental and predicted values for the LD model. The adequate precision of 7.24 was high compared to the desirable value (greater than 4) [38, 39]. Therefore, this model can be used to navigate the design space. In contrast, the model of CD gave low R2, though a high adequate precision (7.509) was obtained. Therefore, the CD model cannot be used for the prediction of fungal pretreatment.

Figure 1 shows the interactive effect of each variable on the percentage of LD in relation to the moisture content, wheat flour and EFB concentration. The highest LD of 14.7 % was predicted at the optimum condition of an initial moisture content of 80 %, wheat flour of 2.29 % and EFB concentration of 23.3 %. To verify the optimized results, a set of experiments was performed. LD of 15.6 % and CD of 12.5 % were obtained with SV of 1.25 and less than 5.0 % solid loss. The observed LD was slightly higher than the predicted LD with an error of 0.06 %. Physicochemical pretreatment of EFB such as ionic liquid treatment [7], NH3 treatment [5], bisulfate treatment [40] and sequential acid/alkaline treatment [41] achieved an LD of 33.5–70.7 % at 60–121 °C. Although the delignification of EFB (15.6 %) obtained from this study was lower than from physicochemical pretreatments, this fungal pretreatment was conducted under low temperature (at 28 °C) leading to lower energy input and lower operating cost. Delignification of EFB using T. versicolor TISTR 3224 in this study seems rather average compared to other lignocellulosic biomasses, eucalyptus wood chip [21], oil palm trunk [22], wheat straw [35], rubber wood [36], corn straw [42], hornbeam wood chip [43] and bamboo [44], which achieved a delignification of 9.35–24.4 % in 10–32 days. Compared with other fungi for the pretreatment of EFB, the ability of T. versicolor TISTR 3224 for delignification was higher than that of Pleurotus floridanus (a lignin removal of 0.48 %) as reported by Isroi et al. [45] and Phanerochaete chrysosporium ATCC32629 (a lignin removal of 5.89 %) as reported by Hamisan et al. [46]. However, Harmini et al. [47] found that Pleurotus floridanus LIPIMC996 removed lignin from EFB in approximately 25–27 %. Different strains might have different abilities of delignification. Consequently, it indicated that T. versicolor TISTR 3224 has the potential to be used for biological pretreatment of EFB.

Fig. 1
figure 1

Three-dimensional surface plot showing interactive effects between (a) EFB concentration and wheat flour at an initial moisture content of 80 %, (b) moisture content and wheat flour at EFB concentration of 23.3 % and (c) moisture content and EFB concentration at wheat flour 2.29 % on lignin degradation (LD)

Full size image

In this study, SV cannot be predicted by the model and the optimized value cannot be statistically analyzed. However, the highest SV of 1.29 was observed with the 28 % EFB concentration and 2 % wheat flour concentration at 70 % moisture content (Trial 15) (Table 2). It indicated that there was no difference between the SV obtained under this condition and under optimum condition (23.3 % EFB concentration and 2.29 % wheat flour concentration at 80 % moisture content). For an economic approach, low wheat flour concentration and moisture content might be preferred.

Structure analysis and enzymatic digestibility of fungal-treated EFB (fEFB)

The surface morphology of the untreated EFB and fEFB was investigated by scanning electron microscopy (SEM). SEM images show the smooth and packed fibers of the untreated EFB (Fig. 2a). After fungal pretreatment, the mycelium covered and penetrated into the surface of EFB, which led to more porosity and roughness than the untreated EFB (Fig. 2b). After an enzymatic hydrolysis, the cellulose digestibility of the fungal pretreated EFB was higher than that of the untreated EFB (Fig. 3). After 72 h of saccharification, the untreated EFB and fEFB achieved 9.04 and 12.6 % of cellulose digestibility, respectively. It indicated that the fungal pretreatment increased cellulose digestibility by about 1.34-fold from the untreated EFB. According to Isroi et al. [45], fungal pretreatment of EFB affected the structural changes in the lignin and loss of aromatic units. Consequently, fungal pretreatment caused a further increase in the surface area and susceptibility of EFB to enzymatic saccharification.

Fig. 2
figure 2

Scanning electron micrographs (×1,000) of the untreated EFB (a) and EFB pretreated with T. versicolor TISTR 3224 (b)

Full size image
Fig. 3
figure 3

Cellulose digestibilities of untreated EFB and fEFB after saccharification (50 °C, pH 4.8) of 10 FPU/g substrate with cellulase

Full size image

However, a long fungal pretreatment (3 weeks) was required to achieve high lignin degradation. Physical or chemical pretreatments may take a shorter time than a biological method [14]. However, the solid loss from those methods was high with generation of pollutants. Also, the high energy input or production cost must be taken into account [48]. The combination of fungal pretreatment with physical and/or chemical methods has been reported to synergistically improve enzymatic digestibility with an environmental friendly and energy-efficient process [27]. Therefore, a future study for the suitable combination of fungal pretreatment with other methods might be required.

Conclusions

Trametes versicolor TISTR3224 showed a great selective lignin-degrading ability on biological pretreatment of EFB under solid-state cultivation. With statistical analysis using PBD and BBD, maximum LD was achieved with 23 % EFB concentration and 2.29 % wheat flour concentration at 80 % moisture content. With the consideration of the high SV and an economic benefit, 28 % EFB concentration and 2 % wheat flour concentration at 70 % moisture content might be preferred. However, a relatively low efficiency and long residence time were still the major disadvantages of e fungal pretreatment. New strategies should be used to overcome these weak aspects. A two-step pretreatment consisting of fungal pretreatment and other methods may be a good approach to enhance the efficiency of biological pretreatment and lower the requirements.