Structural Carbohydrates in a Plant Biomass: Correlations between

Article pubs.acs.org/JAFC

Structural Carbohydrates in a Plant Biomass: Correlations between the Detergent Fiber and Dietary Fiber Methods Bruno Godin,*,†,‡ Richard Agneessens,† Patrick Gerin,‡ and Jérôme Delcarte† †

Walloon Agricultural Research Center, CRA-W. Valorisation of Agricultural Products Department, Biomass, Bioproducts and Energy Unit, Chaussée de Namur, 146, B-5030 Gembloux, Belgium ‡ Université catholique de Louvain, Earth and Life Institute, Bioengineering group, Croix du Sud, 2 box L7.05.19, B-1348 Louvain-la-Neuve, Belgium S Supporting Information *

ABSTRACT: We compared the detergent fiber and dietary fiber methods to analyze the cellulose and hemicellulose contents of commelinid and non-commelinid magnoliophyta biomass. A good linear correlation was found between both methods. Compared to the more accurate dietary fiber method, the detergent fiber method overestimates the content of cellulose, whereas the detergent fiber method, as compared to the dietary fiber method, overestimates and underestimates the hemicellulose content in commelinid and non-commelinid magnoliophyta biomass, respectively. Because of the good linear correlations, conversion factors were determined to predict the cellulose, hemicellulose, and xylan contents to be expected from the dietary fiber method, on the basis of analyses made by the faster, cheaper, and more commonly practiced detergent fiber method. Nevertheless, the dietary fiber method offers the advantage of providing the detailed composition of the hemicelluloses (xylan, arabinan, hemicellulosic glucan, galactan, and mannan), and that is of interest for biorefining purposes. KEYWORDS: cellulose, hemicelluloses, xylan, correlation, biofuels

■

remaining structural carbohydrates of the acid detergent fiber residue are solubilized by 12.2 mol/L sulfuric acid at room temperature for 3 h.5 The residual insoluble acid detergent lignin fraction represents an indigestible portion of forages.4 The gravimetric difference between ADF and ADL insoluble residues is used to estimate the content of cellulose. The gravimetric difference between the NDF and ADF insoluble residues is used to estimate the content of hemicelluloses. The acid detergent lignin insoluble residue is used to estimate the content of lignin.4,6 The detergent fiber method shows some deficiencies in accuracy. It overestimates the contents of cellulose and hemicelluloses and underestimates the content of lignin.6−8 This can be explained by the fact that (1) the neutral detergent fiber residue contains some proteins,6,7 (2) the acid detergent fiber residue contains some pectins when the analyzed biomass (such as a non-commelinid magnoliophyta biomass) has a high content of pectins,4,6,9 (3) the acid detergent fiber residue contains some hemicelluloses,10 and (4) the acid detergent lignin residue does not contain the acid soluble lignin.6 The dietary fiber method is based on the Uppsala method.7,11 It is used to assess dietary fiber in animal and human food and in crop feedstocks that are of interest as a source of fibers, biofuels (such as bioethanol and biobutanol), and biobased chemicals.11 The dietary fiber method is used to determine more accurately the contents of cellulose and hemicelluloses, compared to the accuracy of the detergent fiber method. This method quantitates the structural polysaccharides based on their monosaccharidic

INTRODUCTION The main chemical components of a plant biomass are generally cellulose (linear structural polysaccharide homogeneously composed of D-glucose units), hemicelluloses (ramified structural polysaccharides heterogeneously composed of D-xylose, Larabinose, D-mannose, D-galactose, and D-glucose units) and lignin (phenyl propanoid polymer composed of syringyl, guaiacyl, and p-hydroxyphenyl units).1,2 A plant biomass also contains other components such as pectins (ramified heterogeneous structural polysaccharides mainly composed of Dgalacturonic acid units), starch (linear or ramified homogeneous nonstructural polysaccharide composed of D-glucose units), soluble sugars (D-glucose, D-fructose, sucrose, and fructans), proteins, and mineral compounds.1,2 The Van Soest sequential fiber solubilization method is widely used to assess the cellulose, hemicellulose, and lignin contents of plant feedstocks and to predict the nutritive value of these materials. This gravimetric method classifies the cell wall components of a plant biomass into three types of insoluble residues: neutral detergent fiber (NDF), acid detergent fiber (ADF), and acid detergent lignin (ADL) residues. The neutral detergent extraction is operated by refluxing a biomass sample in a boiling aqueous neutral detergent at pH 7 for 1 h.3 The neutral detergent fiber residue represents the incompletely digestible portion of forages that is composed of cellulose, hemicelluloses, and lignin.4 Acid detergent extraction is conducted by refluxing a biomass sample in a boiling aqueous detergent containing 0.5 mol/L sulfuric acid for 1 h.5 Cellulose and lignin are the cell wall components recovered in the insoluble acid detergent fiber residue. The acid detergent extraction has been developed as a pretreatment prior to the measurement of lignin as acid detergent lignin. After the acid detergent procedure, the © 2014 American Chemical Society

Received: Revised: Accepted: Published: 5609

February 25, 2014 May 19, 2014 May 19, 2014 May 19, 2014 dx.doi.org/10.1021/jf500924q | J. Agric. Food Chem. 2014, 62, 5609−5616

Journal of Agricultural and Food Chemistry

Article

commelinid and non-commelinid magnoliophyta biomass. The correlation between the results of the dietary fiber and detergent fiber methods was therefore investigated for the contents of cellulose, hemicelluloses, the non-individualized sum of cellulose and hemicelluloses, and the hemicellulosic components (xylan, arabinan, hemicellulosic glucan, galactan, and mannan) of a wide diversity of plant biomasses. The sum of the cellulose and hemicellulose contents (cellulose+hemicelluloses) has also been taken into account in our assessment to compare more globally the dietary fiber and detergent fiber methods.

components (D-glucose, D-xylose, L-arabinose, D-mannose, and 12−14 D-galactose). In the dietary fiber method, samples are first extracted one or several times by solvents (e.g., water, ethanol, neutral detergent, ether, and/or hexane) to remove all components that interfere with the subsequent sulfuric acid hydrolysis and chromatographic quantitation of the cellulosic and hemicellulosic monosaccharides. The total glucan, xylan, arabinan, galactan, and mannan contents are then determined by a two-stage sulfuric acid hydrolysis method (SAH). The first stage solubilizes the structural carbohydrates with 12.2 mol/L sulfuric acid at 30 °C for 1 h, and the second stage hydrolyzes these solubilized structural carbohydrates with 0.419 mol/L sulfuric acid at 121 °C for 1 or 2 h.12−14 The released monosaccharides are separated and quantitated by liquid chromatography or gas chromatography. The content of hemicellulosic glucan is determined by the same dietary fiber method except that the cellulose solubilization step (incubation with 12.2 mol/L sulfuric acid at 30 °C for 1 h) is omitted. The content of cellulose (cellulosic glucan; i.e., D-glucose of cellulose in its polymeric form) is calculated as the difference between the total glucan and hemicellulosic glucan contents.14 The monosaccharidic components (D-xylose, L-arabinose, D-glucose, D-mannose, and D-galactose) of hemicelluloses are expressed as their polymeric forms (xylan, arabinan, hemicellulosic glucan, mannan, and galactan, respectively).14 In the dietary fiber method, the hemicellulose content is considered as the sum of the xylan, arabinan, hemicellulosic glucan, galactan, and mannan contents. To correct the results for any underestimation of the monosaccharide concentrations due to acidic degradation, a standard mixture of monosaccharides is treated with sulfuric acid in parallel with the samples.12−14 The relationship between the contents of cellulose and hemicelluloses determined by the detergent fiber and dietary fiber methods has been investigated on a data set that includes miscanthus (Miscanthus sinensis Anderss.), reed canarygrass (Phalaris arundinacea L.), smooth bromegrass (Bromus inermis L.), and alfalfa leaf and stem (Medicago sativa L.).7 They showed that the detergent fiber method overestimates the content of cellulose and hemicelluloses, as compared to the dietary fiber method, except for alfalfa (a non-commelinid magnoliophyta biomass compared to other biomasses that are commelinids) where the detergent fiber method underestimates the content of hemicelluloses. The detergent fiber and dietary fiber methods have also been compared on alfalfa stems and on corn stover.4,16 In spite of the bias mentioned above, there was a very good correlation between both methods. Cellulose results of one method can be consistently and confidently converted to the expected results of the other method because of this correlation. The reliability of this conversion has not been established for the other parameters, such as hemicellulose content. The Van Soest detergent fiber method is faster, cheaper, and more commonly practiced than the dietary fiber method.4,7,16 The advantage of the chromatographic dietary fiber method, as compared to the gravimetric detergent fiber method, is that the dietary fiber method also provides the monosaccharidic composition of the hemicelluloses (xylan, arabinan, hemicellulosic glucan, galactan, and mannan) and cellulose (cellulosic glucan).14,15 Predicting the cellulose, hemicellulose, and xylan contents directly from the results of the detergent fiber method would offer an interesting alternative to the time- and resourceconsuming dietary fiber method. The purpose of this study is to check the feasibility and reliability of such a prediction on

■

MATERIALS AND METHODS

Biomass Material. Miscanthus giganteus (Miscanthus × giganteus J. M. Greef and Deuter ex Hodk. and Renvoize), switchgrass (Panicum virgatum L.), fiber sorghum [Sorghum bicolor (L.) Moench], spelt straw [Triticum aestivum L. ssp. spelta (L.) Thell.], “cocksfoot-alfalfa” mixture (Dactylis glomerata L.-M. sativa L.), tall fescue (Festuca arundinacea Schreb.), immature rye (Secale cereale L.), fiber corn (Zea mays L.), hemp (Cannabis sativa L.), and Jerusalem artichoke leaves and stalks (Helianthus tuberosus L.) came from randomized block designed crop trials performed in 2007, 2008, 2009, and/or 2010 at Libramont, Belgium [498 m above sea level (asl), average annual temperature of 8.6 °C, average annual precipitation of 1260 mm, 49°55′N 05°24′E], and at Gembloux, Belgium (161 m asl, average annual temperature of 9.8 °C, average annual precipitation of 856 mm, 50°33′N 04°43′E). Depending on the crop, trials were performed with different nitrogen fertilization levels, different cultivars, and/or different harvest periods. For each biomass sample, a plot between 9 and 24 m2 of the whole above ground biomass was harvested 10 cm from the ground and chopped. Reed [Phragmites australis (Cav.) Trin. ex Steud], spelt grains [T. aestivum L. ssp. spelta (L.) Thell.], onion leaves and bulbs (Allium cepa L.), yucca leaves (Yucca gloriosa L.), tulip tree wood (Liriodendron tulipifera L.), Japanese knotweed [Fallopia japonica (Houtt.) Ronse Decr.], cabbage leaves (Brassica oleracea L.), rapeseed straw (Brassica napus L.), pumpkin leaves and stalks (Cucurbita maxima Duchesne), bean leaves and stalks (Phaseolus vulgaris L.), alfalfa leaves and stalks (M. sativa L.), lupine leaves and stalks (Lupinus albus L.), oak wood (Quercus sp.), beech wood (Fagus sylvatica L.), willow wood (Salix sp.), aspen wood (Populus sp.), flax straw (Linum usitatissimum L.), nettle (Urtica dioica L.), bramble leaves and stalks (Rubus f ruticosus L.), comfrey leaves and stalks (Symphytum of f icinale L.), Jerusalem artichoke tubers (H. tuberosus L.), sunflower leaves and stalks (Helianthus annuus L.), tagetes (Tagetes patula L.), green (unforced) and white (forced) leaves of chicory (Cichorium intybus L.), unforced and forced roots of chicory (C. intybus L.), and leaves, stalks, and roots of carrot (Daucus carota L.) were harvested manually in 2010 and/or 2011 at Gembloux, Belgium, 10 cm from the ground and chopped. Immediately after the harvest, two representative subsamples of 750 g of the whole of each biomass were dried at 60 °C for 72 h in a forced air oven. After being dried, the two subsamples were first milled with a 4 mm screen BOA hammer mill (Waterleau, Herent, Belgium) followed by a second milling step with a 1 mm screen Cyclotec cyclone mill (FOSS Benelux N.V., Brussels, Belgium). The samples were stored in airtight bags at room temperature and protected from light in a dark box. Chemical Analysis. All chemicals were of analytical grade or equivalent. Duplicate aliquots were measured for each component quantitation on each of the 439 biomass samples. Structural Polysaccharides Analyzed by the Van Soest Method. The neutral detergent fiber residue (NDF, weight of the neutral detergent fiber residue) of the Van Soest method was determined as described previously,3 except that a 16.67 μkat/g of DM sample of an analytical thermostable α-amylase (Megazyme, Wicklow, Ireland) was added before the Van Soest neutral detergent step for biomass containing starch. The acid detergent fiber (ADF, weight of the acid detergent fiber residue corrected for its content of mineral compounds) residue, and acid detergent lignin (ADL, weight of the acid detergent lignin residue corrected for its content of mineral compounds) of the Van Soest method were determined as described previously,5 except 5610

dx.doi.org/10.1021/jf500924q | J. Agric. Food Chem. 2014, 62, 5609−5616


Article

Table 1. Data Set Statistics of the Structural Carbohydrate Contents in Commelinid (n = 333) and Non-Commelinid Magnoliophyta (n = 108) Biomass Determined by the Detergent Fiber and Dietary Fiber Methods compound commelinid biomass cellulose+hemicelluloses VS cellulose+hemicelluloses SAH cellulose VS cellulose SAH hemicelluloses VS hemicelluloses SAH xylan arabinan memicellulosic glucan galactan mannan non-commelinid magnoliophyta biomass cellulose+hemicelluloses VS cellulose+hemicelluloses SAH cellulose VS cellulose SAH hemicelluloses VS hemicelluloses SAH xylan arabinan hemicellulosic glucan galactan mannan

mean (g/100 g of DM)

RSD (%)

minimum (g/100 g of DM)

maximum (g/100 g of DM)

59.5 53.1 34.4 29.9 25.1 23.3 17.1 2.72 2.03 0.91 0.54

22 23 29 28 19 19 24 19 28 46 33

25.1 21.8 12.1 9.7 13.0 10.3 5.8 1.75 0.87 0.25 0.17

79.3 76.9 55.5 50.6 37.9 34.3 27.8 4.12 3.76 2.25 0.96

41.1 36.2 31.9 24.2 9.3 12.0 7.3 0.56 1.20 1.73 1.20

48 52 52 55 43 49 61 45 51 40 58

7.6 5.0 5.3 3.4 1.6 1.2 0.3 0.20 0.15 0.34 0.18

80.1 71.2 66.0 51.4 21.8 25.8 20.4 1.58 3.05 3.43 2.74

glucose, D-mannose, and D-galactose) of hemicelluloses are expressed under their polymeric form (xylan, arabinan, hemicellulosic glucan, mannan, and galactan). Hemicellulosic Components of the Acid Detergent Fiber and Weende Residues. The acid detergent fiber residue was prepared as described previously,5 except that, prior to the acid detergent extraction, an extraction with the Van Soest neutral detergent (without adding sodium sulfite) was performed as mentioned above. The Weende residue was extracted as described previously.17 Briefly, biomass samples were refluxed, in chronological order, with 0.13 mol/L boiling sulfuric acid for 30 min and with 0.13 mol/L boiling potassium hydroxide for 30 min. The xylan, arabinan, mannan, galactan, and hemicellulosic glucan contents of the acid detergent fiber and the Weende residues were determined by the same sulfuric acid hydrolysis method as described above. Statistical Analysis. The linear least-squares regression analyses of structural carbohydrate contents between the detergent fiber and dietary fiber methods and the other statistical assessments were performed using JMP 11 (SAS Institute, Tervuren, Belgium). The prediction performance made by a linear regression line is considered excellent (R2p ≥ 0.95, and RPDp ≥ 4.0), successful (R2p ≥ 0.90, and RPDp ≥ 3.0), moderately successful (R2p ≥ 0.80, and RPDp ≥ 2.3), or moderately useful (for the purpose of semiquantitative screening) (R2p ≥ 0.70, and RPDp ≥ 1.8).18 The mean comparison tests (with α = 0.05) were based on the Student’s t test. The data were paired for the comparison between the detergent fiber and dietary fiber methods. The differences (with α = 0.05) between two coefficients of correlation of prediction (Rp) were assessed using a Fisher z-transform. The differences (with α = 0.05) between two root-mean standard errors of prediction (RMSEp) were assessed using an F-test of variances. These statistical analyses were performed using JMP 11 (SAS Institute).

that, prior to the acid detergent extraction, an extraction with the Van Soest neutral detergent without adding sodium sulfite was conducted as described for the Van Soest neutral detergent extraction. Cellulose +hemicelluloses of the Van Soest method was calculated as NDF − ADL. The cellulose VS content was calculated as ADF − ADL, and the hemicellulose VS content was calculated as NDF − ADF.3,5 The term “detergent fiber method” will be used to refer to data generated by the Van Soest fiber data (NDF, ADF, ADL, and their combinations). Structural Polysaccharides Analyzed by the Sulfuric Acid Hydrolysis Method. The samples were fractionated by the Van Soest neutral detergent extractions, with extraction 1 being 0.1 mmol/L phosphate buffer at pH 7 for 15 min at 90 °C where a 16.67 μkat/g of DM sample of an analytical thermostable α-amylase (Megazyme) was added when the sample contained starch and extraction 2 being Van Soest neutral detergent at 100 °C for 1 h.14 The xylan, arabinan, galactan, mannan, and total glucan contents of the insoluble residue left after these extractions were determined by a two-stage sulfuric acid hydrolysis method (SAH), with stage 1 being solubilization by 12.2 mol/L sulfuric acid at 30 °C for 1 h and stage 2 being hydrolysis by 0.419 mol/L sulfuric acid at 121 °C for 2 h.14 The released monosaccharides were separated and quantitated by a liquid chromatography system with an analytical liquid chromatography column [300 mm × 7.8 mm (inside diameter), 7 μm, Carbo Sep CHO-682 Pb (Interchrom, Montluçon, France)] at 80 °C using deionized water at a rate of 0.4 mL/min as the mobile phase, with a charged aerosol detector, as described previously.14 The hemicellulosic glucan content was determined by the same sulfuric acid hydrolysis method except that the cellulose solubilization step (incubation with 12.2 mol/L sulfuric acid at 30 °C for 1 h) was omitted. The cellulose SAH (cellulosic glucan, i.e., D-glucose of cellulose in its polymeric form) content is calculated as the difference between the total glucan and hemicellulosic glucan contents. The hemicellulose SAH content is calculated as the sum of the xylan, arabinan, galactan, mannan, and hemicellulosic glucan contents. The content of cellulose+hemicelluloses SAH is calculated as the sum of the cellulose and hemicellulose contents. The term “dietary fiber method” will be used to refer to data generated by the dietary fiber data (total glucan, xylan, arabinan, galactan, mannan, hemicellulosic glucan, and their combinations). The monosaccharidic components (D-xylose, L-arabinose, D-

■

RESULTS AND DISCUSSION Main Structural Carbohydrates. The summary statistics of the mean structural carbohydrate contents among the 333 commelinid and the 108 non-commelinid magnoliophyta 5611



Article

Figure 1. Relationship between the detergent fiber and dietary fiber methods for the contents of cellulose+hemicelluloses (left), cellulose (center), and hemicelluloses (right) in commelinid (n = 333) (A) and non-commelinid magnoliophyta (n = 108) (B) biomass. The black dashed line in each plot is the line of equality (y = x). The black plain line in each plot is the regression line. The black lines with smaller dashes in each plot are the 95% confidence lines of the regression.

Table 2. Linear Regression Lines and Their Prediction Performances for Structural Carbohydrate Contents in Commelinid (n = 333) and Non-Commelinid Magnoliophyta (n = 108) Biomass between the Detergent Fiber and Dietary Fiber Methodsa

a

Figure

independent variable

1A, left 1A, center 1A, right

cellulose+hemicelluloses VS cellulose VS hemicelluloses VS NDF NDF NDF ADF

1B, left 1B, center 1B, right

cellulose+hemicelluloses VS cellulose VS hemicelluloses VS NDF NDF NDF ADF

dependent variable

slope

Commelinid Biomass cellulose+hemicelluloses SAH 0.906 ± 0.010 cellulose SAH 0.844 ± 0.008 hemicelluloses SAH 0.856 ± 0.020 cellulose+hemicelluloses SAH 0.758 ± 0.065 cellulose SAH 0.508 ± 0.010 hemicelluloses SAH 0.250 ± 0.008 cellulose SAH 0.655 ± 0.006 Non-Commelinid Magnoliophyta Biomass cellulose+hemicelluloses SAH 0.945 ± 0.011 cellulose SAH 0.803 ± 0.012 hemicelluloses SAH 1.381 ± 0.049 cellulose+hemicelluloses SAH 0.818 ± 0.013 cellulose SAH 0.565 ± 0.014 hemicelluloses SAH 0.252 ± 0.005 cellulose SAH 0.673 ± 0.015

intercept

R2p

RMSEp

RPDp

−0.78 ± 0.58b 0.81 ± 0.29 1.82 ± 0.51 3.96 ± 0.44 −3.07 ± 0.66 7.03 ± 0.52 3.81 ± 0.27

0.96 0.97 0.85 0.98 0.89 0.76 0.97

2.23 1.41 1.61 1.84 2.64 1.94 1.45

5.8 5.9 2.7 6.6 3.2 2.3 5.8

−2.64 ± 0.49 −1.37 ± 0.44 −0.79 ± 0.49b −3.10 ± 0.70 −2.98 ± 0.76 −0.12 ± 0.29b −1.90 ± 0.65

0.99 0.98 0.88 0.97 0.94 0.95 0.95

2.20 2.11 2.04 3.05 3.26 1.24 2.91

8.6 6.3 2.9 6.2 4.1 4.7 4.6

Uncertainties are expressed as ± the standard deviation. bThe intercept of regression is not significantly different from zero (p ≥ 0.05).

samples are listed in Table 1. These two types of biomass were separated because of the differences between their cell wall components. The statistical difference between these two types of biomass for each of the contents of the analyzed structural carbohydrate components is significant (p < 0.001, except cellulose VS that has a p value between 0.01 and 0.05) for each type of method. Commelinid magnoliophyta have higher cell wall contents of arabinoxylans, β-glucans, and lignin.1,2 Noncommelinid magnoliophyta have higher cell wall contents of xyloglucans, mannans, and pectins.1,2 The structural carbohydrates contained in these samples were determined by the detergent fiber and dietary fiber methods. Table 1 shows that

cellulose and hemicelluloses are the main chemical components of the dry matter. The average content of cellulose+hemicelluloses represents more than the half of the dry matter of the analyzed commelinid biomass, while it is less than the half of the dry matter for the non-commelinid magnoliophyta biomass. The higher content of cellulose+hemicelluloses in commelinid magnoliophyta biomass can be explained by the higher content of hemicelluloses and lower content of pectins in their cell walls.1,2 Figure 1 illustrates the relationship among the contents of cellulose+hemicelluloses, cellulose, and hemicelluloses determined by the detergent fiber and dietary fiber methods. Table 2 5612



Article

Figure 2. Hemicellulosic components detected in the acid detergent fiber residue (ADF) (A) and in the Weende residues (B).

hemicelluloses by the detergent fiber method.1,6,9 This information confirms that the relationship among the contents of cellulose+hemicelluloses, cellulose, and hemicelluloses determined by the detergent fiber and dietary fiber methods for commelinid and non-commelinid magnoliophyta biomass must be determined separately. The slope of the regressions (Table 2) is an assessment of the bias between the detergent fiber and dietary fiber methods. The bias between these two methods affects, by order of increasing importance, the contents of cellulose+hemicelluloses, cellulose, and hemicelluloses. This could be explained by the fact that the acid detergent fiber residue of the detergent fiber method contains some residual hemicelluloses, as also suggested previously.10 Indeed, an average of 16.3 g of hemicelluloses per 100 g was detected in the acid detergent fiber residue of the analyzed biomass (Figure 2A). An average of 16.8 g of hemicelluloses per 100 g was also detected in the Weende residue (Figure 2B). The Weende method aims to gravimetrically determine the cellulose content by means of an acid and alkaline extraction of all the other components, including hemicelluloses. Hemicelluloses are made of four types of structural polysaccharides: arabinoxylan, xyloglucan (which also contains galactan), β-glucan, and mannan (which also contains galactan and hemicellulosic glucan).19,20 In the acid detergent fiber and the Weende residues, the relative hemicellulosic components showed increased relative contents of hemicellulosic glucan and

presents the corresponding regression information. The contents of cellulose+hemicelluloses, cellulose, and hemicelluloses determined by the detergent fiber method are always higher than those determined by the dietary fiber method, except in the case of the content of hemicelluloses in non-commelinid magnoliophyta biomass (Tables 1 and 2 and Figure 1). The statistical difference between these two methods for the contents of cellulose+hemicelluloses, cellulose, and hemicelluloses is significant (p < 0.001) for each type of biomass. The overestimation of cellulose and hemicelluloses determined by the detergent fiber method can be explained by the fact that it is based on gravimetric differences between residues that are contaminated by undesired compounds, while the dietary fiber method measures specifically the monosaccharidic contents of cellulose and hemicelluloses.6 The bias of the estimations of the contents of cellulose and hemicelluloses by the detergent fiber method depends mainly on the neutral detergent fiber residue, which also contains some proteins, and the acid detergent lignin residue, which does not contain the acid soluble lignin.6 The acid soluble lignin fraction is higher in commelinid than in noncommelinid magnoliophyta biomass.7 In commelinid magnoliophyta, lignin and hemicelluloses are bound through ferulate esters, which are indeed more sensitive to acid hydrolysis.1,6 Non-commelinid magnoliophyta have cell walls with a content of pectins higher than that of cell walls of commelinids.1 The presence of some pectins in their acid detergent fiber residue also explains an overestimation of cellulose and an underestimation of 5613



Article

Table 3. Data Set Statistics of the Structural Carbohydrates Contents in Commelinid Biomass Determined by the Detergent Fiber and Dietary Fiber Methods, Expressed as the Relative Content of the Neutral Detergent Fiber (NDF) or Acid Detergent Fiber (ADF) Residues (n = 333) compound relative content in the NDF residue cellulose+hemicelluloses VS cellulose+hemicelluloses SAH cellulose VS cellulose SAH hemicelluloses VS hemicelluloses SAH xylan arabinan hemicellulosic glucan galactan mannan relative content in the ADF residue cellulose VS cellulose SAH


RSD (%)



92.2 82.3 52.6 45.8 39.6 36.5 26.5 4.43 3.33 1.48 0.87

3 4 7 8 15 11 9 28 35 48 28

85.0 74.4 44.2 35.4 23.7 26.9 19.1 2.12 0.97 0.28 0.19

96.6 94.1 61.8 56.9 49.4 47.2 32.0 7.93 6.87 3.64 1.39

87.3 76.0

4 6

80.1 65.4

93.9 89.5

Table 4. Data Set Statistics of the Structural Carbohydrate Contents in Non-Commelinid Magnoliophyta Biomass Determined by the Detergent Fiber and Dietary Fiber Methods, Expressed as the Relative Content of the Neutral Detergent Fiber (NDF) or Acid Detergent Fiber (ADF) Residues (n = 108) compound relative content in the NDF residue cellulose+hemicelluloses VS cellulose+hemicelluloses SAH cellulose VS cellulose SAH hemicelluloses VS hemicelluloses SAH xylan arabinan hemicellulosic glucan galactan mannan relative content in the ADF residue cellulose VS cellulose SAH


RSD (%)



85.5 73.5 65.3 48.9 20.3 24.6 13.9 1.60 2.56 3.99 2.50

5 10 9 14 23 13 31 86 35 32 29

63.3 44.7 43.3 29.7 9.2 12.7 3.2 0.28 1.11 1.42 1.05

96.9 87.0 76.4 62.0 38.8 32.8 23.7 8.94 7.45 8.91 3.95

81.8 61.3

7 12

54.2 38.2

95.0 75.5

RPDp ≥ 2.3) (Table 2), the regression lines seem to be reliable enough to use the relatively simple detergent fiber method to predict quantitatively the values to be expected by the more expensive and time-consuming dietary fiber method.4,7,16 This is especially the case for the prediction made by cellulose +hemicelluloses VS and cellulose VS. The suitability of using directly the neutral detergent fiber content to predict the contents of cellulose+hemicelluloses SAH, cellulose SAH, and hemicelluloses SAH to be expected from the dietary fiber method was also tested (Table 2). The neutral detergent fiber residue-based predictions are generally less reliable for the commelinid magnoliophyta biomass, especially for the prediction of hemicelluloses SAH, which is not reliable. This last regression line for commelinid magnoliophyta biomass has prediction performances that are too low (R2p < 0.80, and/or RPDp < 2.3) (Table 2). The generally lower prediction performances made by the neutral detergent fiber residue can be explained by the fact that it also contains cellulose and lignin. This lower prediction performance is only significant (p < 0.05) in terms of the coefficient of correlation of prediction (Rp) and the root-mean standard error of prediction (RMSEp) for the

mannan and decreased relative contents of xylan, arabinan, and galactan compared to the contents of the relative hemicellulosic components of native hemicelluloses of the analyzed samples, as described previously.2 This suggests that interaction of cellulose is stronger with xyloglucan and mannan than with arabinoxylan. The high content of mannan in larch wood (Figure 2) is explained by its high content in the native hemicelluloses of the larch wood.2 Larch wood is a part of pinophyta biomass, which is known to have a high content of mannan in their hemicelluloses.1,2 The higher relative bias for cellulose+hemicelluloses of commelinid biomass, as compared to non-commelinid magnoliophyta biomass, is mainly due to the acid soluble lignin fraction that is higher in commelinid biomass as explained above.7 The higher relative bias for cellulose and hemicelluloses of noncommelinid magnoliophyta biomass, as compared to commelinid biomass, is mainly due to the presence of some pectins in the acid detergent fiber residue of non-commelinid magnoliophyta as explained above.6,9 Because of the large range of the contents of the various components and good prediction performances (R2p ≥ 0.80, and 5614



Article

Table 5. Best Linear Regression Lines and Their Prediction Performances for the Monosaccharidic Content of Hemicelluloses in Commelinid (n = 333) and Non-Commelinid Magnoliophyta (n = 108) Biomass between the Detergent Fiber and Dietary Fiber Methodsa independent variable

a

dependent variable

NDF

xylan

hemicelluloses VS NDF

xylan xylan

slope

intercept

Commelinid Magnoliophyta Biomass 0.245 ± 0.005 1.21 ± 0.37 Non-Commelinid Magnoliophyta Biomass 1.043 ± 0.038 −2.35 ± 0.38 0.185 ± 0.006 −1.57 ± 0.33

R2p

RMSEp

RPDp

0.86

1.45

2.8

0.88 0.89

1.48 1.39

3.0 3.2

Uncertainties are expressed as ± the standard deviation.

The relationship between the detergent fiber and dietary fiber methods is driven by the acid detergent solubilized fraction (components such as hemicelluloses, nonstructural carbohydrates, pectins, proteins, organic acids, alcohols, pigments, lipids, and mineral compounds) for the cellulose content. When it is expressed without taking into account the solubilized fraction by the acid detergent (i.e., as the relative content of the acid detergent fiber residue) (Tables 3 and 4), as compared to the cellulose content expressed on the total dry matter, the same observations are made as mentioned above for the neutral detergent fiber residue. Monosaccharidic Components of Hemicelluloses. The mean composition of the hemicelluloses and the corresponding summary statistics are listed in Table 1 for the data set of the 333 commelinid and 108 non-commelinid magnoliophyta samples. The composition of the hemicelluloses can be determined by only the dietary fiber method. Xylan is the major hemicellulosic component of the analyzed samples, with contents more than 2 times higher in commelinid than in non-commelinid magnoliophyta biomass. The content of hemicelluloses is indeed known to be higher in commelinid cell walls, where hemicelluloses substitute for the pectins that are more present in noncommelinid magnoliophyta cell walls.1,21−23 In addition, commelinid cell walls are known to have hemicelluloses with higher contents of arabinoxylan and β-glucan, whereas noncommelinid magnoliophyta cell walls have higher contents of xyloglucan and mannan.1 The average content of hemicellulosic components in commelinid biomass is, by order of decreasing importance, xylan, arabinan, hemicellulosic glucan, galactan, and mannan, while it is xylan, hemicellulosic glucan, galactan, mannan, and arabinan for non-commelinid magnoliophyta biomass (Table 1). This is consistent with the known hemicellulose compositions of these two types of biomass as mentioned above.1,21−23 Table 5 indicates that xylan is the only individual hemicellulosic component that is sufficiently correlated with the hemicelluloses VS and neutral detergent fiber residue to perform quantitative predictions (R2p ≥ 0.80, and RPDp ≥ 2.3), except for the hemicelluloses VS of commelinid magnoliophyta biomass. Further investigation showed that no useful linear regression lines with quantitative prediction performances (R2p < 0.80, and/or RPDp < 2.3) can be detected between hemicelluloses VS or neutral detergent fiber residue and other individual SAH hemicellulosic components (data not shown). The fact that only xylan is highly correlated can be explained by its dominating contribution in the hemicellulosic component (73 mass percent in commelinids and 61 mass percent in noncommelinids magnoliophyta biomass) (Table 1) and its large range compared to those of the other hemicellulosic components.

prediction of cellulose SAH of both types of biomass. In the case of the non-commelinid magnoliophyta biomass, the prediction performance of hemicelluloses SAH is more reliable on the basis of the neutral detergent fiber residue, as compared to that based on the hemicelluloses VS. This can be explained by the overestimation of hemicelluloses VS due to the presence of some pectins in the neutral detergent fiber residue.6,9 This higher prediction performance is significant (p < 0.05) in terms of the coefficient of Rp and RMSEp. The acid detergent fiber residue content can also be used to predict the cellulose SAH of the dietary fiber method because of the excellent prediction performances (R2p ≥ 0.95, and RPDp ≥ 4) (Table 2). This prediction is generally slightly less reliable than the cellulose VS-based prediction probably because of the presence of lignin in the acid detergent fiber residue. This difference of prediction is significant (p < 0.05) in terms of the coefficient of Rp for the non-commelinid magnoliophyta biomass and is not significant (p ≥ 0.05) in terms of Rp for commelinid magnoliophyta biomass or in terms of RMSEp for both types of biomass. The acid detergent fiber residue always provides a slightly better prediction of the cellulose SAH compared to the same prediction given by the neutral detergent fiber residue, probably because of the higher cellulose content in the acid detergent fiber residue than in the neutral detergent fiber residue. This difference in prediction is significant (p < 0.05) in terms of the coefficient of Rp for the commelinid magnoliophyta biomass and not significant (p ≥ 0.05) in terms of Rp for the noncommelinid magnoliophyta biomass or in terms of RMSEp for both types of biomass. The relationship between the detergent fiber and dietary fiber methods is driven by the neutral detergent solubilized fraction (components such as nonstructural carbohydrates, pectins, proteins, organic acids, alcohols, pigments, lipids, and mineral compounds) for the contents of cellulose+hemicelluloses, cellulose, hemicelluloses, and xylan. Indeed, the neutral detergent fiber residue showed quite constant contents of these components for both commelinid and non-commelinid magnoliophyta biomass (Tables 3 and 4). When these contents are expressed without taking into account the solubilized fraction by the neutral detergent (i.e., as relative contents of the neutral detergent fiber residue, which is also called “extractives-free basis”), as compared to these contents expressed on the total dry matter, the relative standard deviation (RSD) and the range (maximum to minimum) of these contents decreased significantly (Tables 3 and 4). In addition, as compared to these contents expressed on the total dry matter, the neutral detergent fiber residue-based predictions of these contents expressed as relative contents of the neutral detergent fiber residue are no longer reliable for commelinid or non-commelinid magnoliophyta biomass (data not shown). 5615



Article

(6) Jung, H.-J. Analysis of forage fiber and cell walls in ruminant nutrition. Nutr. J. 1997, 127, 810S−813S. (7) Theander, O.; Westerlund, E. Quantitative analysis of cell wall components. In Forage cell wall structure and digestibility; Jung, H., Buxton, D., Hatfield, R., Ralph, J., Eds.; American Society of Agronomy: Madison, WI, 1993; pp 83−104. (8) Godin, B.; Agneessens, R.; Gofflot, S.; Lamaudière, S.; Sinnaeve, G.; Gerin, P.; Delcarte, J. Revue sur les méthodes de caractérisation des polysaccharides structuraux des biomasses lignocellulosiques. Biotechnol., Agron., Soc. Environ. 2011, 15, 165−182. (9) Ampuero, S. Détermination de la teneur en fibres dans les aliments pour animaux; ALP Posieux: Posieux, France, 2008; pp 1−7. (10) Morrison, I. Hemicellulosic contamination of acid detergent residues and their replacement by cellulose residues in cell wall analysis. J. Sci. Food Agric. 1980, 31, 639−645. (11) Sluiter, J.; Ruiz, R.; Scarlata, C.; Sluiter, A.; Templeton, D. Compositional analysis of lignocellulosic feedstocks. 1. review and description of methods. J. Agric. Food Chem. 2010, 58, 9043−9053. (12) Theander, O.; Aman, P.; Westerlund, E.; Andersson, R.; Pettersson, D. Total dietary fiber determined as neutral sugar residues uronic acid residues, and Klason lignin (the Uppsala method): Collaborative study. J. AOAC Int. 1995, 78, 1030−1044. (13) Sluiter, A.; Hames, B.; Ruiz, R.; Scarlata, C.; Sluiter, J.; Templeton, D.; Crocker, D. Determination of structural carbohydrates and lignin in biomass; National Renewable Energy Laboratory: Golden, CO, 2008; pp 1−16. (14) Godin, B.; Agneessens, R.; Gerin, P.; Delcarte, J. Composition of structural carbohydrates in biomass: Precision of a liquid chromatography method using a neutral detergent extraction and a charged aerosol detector. Talanta 2011, 85, 2014−2026. (15) Englyst, H.; Quigley, M.; Hudson, G. Determination of dietary fibre as Non-starch polysaccharides with gas-liquid chromatographic, high-performance liquid chromatographic or spectrophotometric measurement of constituent sugars. Analyst 1994, 119, 1497−1509. (16) Wolfrum, E.; Lorenz, A.; de Leon, N. Correlating detergent fiber analysis and dietary fiber analysis data for corn stover collected by NIRS. Cellulose 2009, 16, 577−585. (17) European Union. Commission Regulation No 152/2009. Off. J. Eur. Communities: Legis. 2009, L54, 1−130. (18) Malley, D.; McClure, C.; Martin, P.; Buckley, K.; McCaughey, W. Compositional analysis of cattle manure during composting using a field-portable near-infrared spectrometer. Commun. Soil Sci. Plant Anal. 2005, 36, 455−475. (19) Ishii, T. Structure and functions of feruloylated polysaccahrides. Plant Sci. 1997, 127, 111−127. (20) Vogel, J. Unique aspects of the grass cell wall. Cur. Opin. Plant Biol. 2008, 11, 301−307. (21) Godin, B.; Lamaudière, S.; Agneessens, R.; Schmit, T.; Goffart, J.P.; Stilmant, D.; Gerin, P.; Delcarte, J. Chemical characteristics and biofuels potentials of various plant biomasses: Influence of the harvesting date. J. Sci. Food Agric. 2013, 93, 3216−3224. (22) Godin, B.; Lamaudière, S.; Agneessens, R.; Schmit, T.; Goffart, J.P.; Stilmant, D.; Gerin, P.; Delcarte, J. Chemical characteristics and biofuel potential of several vegetal biomasses grown under a wide range of environmental conditions. Ind. Crops Prod. 2013, 46, 1−12. (23) Godin, B.; Ghysel, F.; Agneessens, R.; Schmit, T.; Gofflot, S.; Lamaudière, S.; Sinnaeve, G.; Goffart, J.-P.; Gerin, P.; Stilmant, D.; Delcarte, J. Détermination de la cellulose, des hémicelluloses, de la lignine et des cendres dans diverses cultures lignocellulosiques dédiées à la production de bioéthanol de deuxième génération. Biotechnol., Agron., Soc. Environ. 2010, 14, 549−560.

Relevance of the Detergent Fiber and Dietary Fiber Methods in the Biomass Valorization Sector. Because of the reliable conversion factors determined in this study, the relatively simple detergent fiber method was used to predict the values to be expected by the more time- and resource-consuming dietary fiber method.4,7,16 Nevertheless, the advantage of the chromatographic dietary fiber method is that it also provides the composition of the hemicelluloses (xylan, arabinan, hemicellulosic glucan, galactan, and mannan).14,15 The detergent fiber method can then be very useful in the sector of biofuels (such as bioethanol and biobutanol) and biobased chemical production to rank plant feedstock samples with a good accuracy according to their contents of cellulose+hemicelluloses, cellulose, hemicelluloses, and xylan, while the dietary fiber method may be required for an optimal valorization of hemicellulosic components.

■

ASSOCIATED CONTENT

S Supporting Information *

Agronomic details about the mechanically and manually harvested biomass (Tables S1 and S2, respectively). This material is available free of charge via the Internet at http:// pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Telephone: 00 32 81 62 71 49. Fax: 00 32 81 61 58 47. E-mail: b. [email protected]. Funding

This research was funded by the Walloon Agricultural Research Center (CRA-W) with the support of the Belgian Science Policy. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS We are grateful to Bridoux A for her technical support. ABBREVIATIONS USED ADF, acid detergent fiber; ADL, acid detergent lignin; asl, above sea level; DM, dry matter; NDF, neutral detergent fiber; RSD, relative standard deviation; RMSEp, root-mean-square error of prediction; Rp, coefficient of correlation of prediction; R2p, coefficient of determination of prediction; RPDp, ratio of the standard deviation of the dependent variable to the RMSE of prediction; SAH, sulfuric acid hydrolysis; VS, Van Soest

■

REFERENCES

(1) Carpita, N.; McCann, M. The cell wall. In Biochemistry and molecular biology of plants; Buchanan, B., Gruissem, W., Jones, R., Eds.; American Society of Plant Physiologists: Rockville, MD, 2000; pp 52− 108. (2) Godin, B.; Lamaudière, S.; Agneessens, R.; Schmit, T.; Goffart, J.P.; Stilmant, D.; Gerin, P.; Delcarte, J. Chemical composition and biofuel potentials of a wide diversity of plant biomasses. Energy Fuels 2013, 27, 2588−2598. (3) Van Soest, P.; Wine, R. Use of detergents in the analysis of fibrous feeds. IV. Determination of plant cell wall constituents. J. AOAC Int. 1967, 50, 50−55. (4) Jung, H.-J.; Lamb, J. Prediction of cell wall polysaccharide and lignin concentrations of alfalfa stems from detergent fiber analysis. Biomass Bioenergy 2004, 27, 365−373. (5) Van Soest, P. Collaborative study of acid-detergent fiber and lignin. J. AOAC Int. 1973, 56, 781−784. 5616


Structural Carbohydrates in a Plant Biomass: Correlations between

Recommend Documents