Toward a global calibration for acid detergent fiber in rapeseed by

visible and near-infrared spectroscopy

R. Font1, A. G. Badani2, B. Wittkop2, M. del Río-Celestino3, W. Lühs2, W.

Friedt2, and A. de Haro-Bailón1,4

1 Department of Agronomy and Plant Breeding, Institute of Sustainable

Agriculture, CSIC, 14080 Córdoba, Spain; 2 Institute of Crop Science and

Plant Breeding I, Justus-Liebig-University, Heinrich-Buff-Ring 26-32, D-

35392 Giessen, Germany; 3 C.I.F.A., Junta de Andalucía, Alameda del

Obispo s/n, 14080, Córdoba, Spain; 4 Corresponding author, E-mail:

cs9habaa@uco.es

Abstract

Visible―near-infrared spectroscopy (NIRS) calibrations for acid detergent

fiber (ADF) in rapeseed (Brassica napus L.) were performed over two

different seed volumes (10 ml, 500 seeds approx.; and 1 ml, 50 seeds

approximately). The inclusion of brown and yellow-seeded cultivars in this

work has allowed covering the whole range currently described in literature

for this character. Chemometric techniques have been used for developing

calibration equations for both sample presentations. On the basis of the R2cv

values obtained for the 10 ml and 1 ml models (0.80 and 0.73), and

SECV/SEL ratios (2.30 and 2.57), respectively, both equations showed an

accuracy sufficient for screening purposes.

/tt/file_convert/577cce5a1a28ab9e788dd659/document.doc

1

Key words: brown and yellow-seeded rapeseed, acid detergent fiber, near

infrared spectroscopy, intact and ground seed, seed volume

Introduction

Near-infrared spectroscopy (NIRS) has been formally accepted by

International Standards Committees for the analysis of many compounds,

including acid detergent fiber (ADF) (Barton and Windham 1988), a

fraction of the total fiber which is related to plant quality (Bjergegaard et al.

1991). Different authors have reported successful predictions of ADF by

NIRS mainly in forages (Marten et al. 1984; García-Ciudad et al. 1993), but

only few NIR studies (Michalski et al. 1992; Font et al. 2003) have

approached this issue over the commercially important rapeseed (Brassica

napus L). These studies were performed over rapeseed cultivars showing

medium or high ADF contents, without including yellow-seeded plants, a

trait related to low fiber and also high oil and protein. In a previous report

(Font et al. 2003), it was stated the need of extending the NIR study on ADF

by adding new variability for this character coming from cultivars not

considered previously, to increase robustness of the equation and avoidance

of biases in predicting composition of cultivars showing extreme low ADF

contents. In the present work we approach this objective from the

perspective of a single species, considering in the study not only brown-

2

seeded rapeseed cultivars, which usually show medium or high contents, but

also yellow-seeded cultivars exhibiting low ADF contents. On the other

hand, due to the fact that rapeseed plants grown under extreme

environmental conditions could produce an insufficient number of seeds per

plant to fully fill the standard cup for NIRS analysis, we developed an

additional equation which was performed over a low seed volume,

providing in this work a comparative study of the results obtained for both

presentations.

Materials and Methods

Segregating doubled haploids (DH) and F2 populations of two different

crosses between yellow and brown-seeded rapeseed cultivars (n= 98) were

used to conduct this study. Seed samples were analysed in duplicate for

ADF following the method proposed by Goering and Van Soest (1970). The

same accessions were then placed in the NIRS sample holder (standard ring

cup of 10 ml volume, 500 seeds approx., and then in the standard ring cup

equipped with an adaptor device of 1 ml volume, 50 seeds, approx.), being

scanned in an NIR spectrometer (NIRSystems model 6500, Foss-

NIRSystems, Inc., Silver Spring, MD, USA) in reflectance mode. Spectra

were acquired at 2 nm wavelength resolution over a wavelength range from

400 to 2500 nm. Using the GLOBAL v. 1.50 software (WINISI II, Infrasoft

International, LLC, Port Matilda, PA, USA), the reference values for ADF

were regressed against NIR spectra for the two different seed volumes by

3

using modified partial least square regression. The equations obtained,

namely, is/10 eqa and is/1 eqa for 10 and 1 ml, respectively, were computed

using the raw optical data (log 1/R, where R is reflectance), or first or

second derivatives of the log 1/R data. In addition, standard normal variate

and de-trending algorithms (SNV-DT) (Barnes et al. 1989) were applied to

raw spectra to correct baseline offset. Cross-validation was used to validate

the equations.

The standard deviation (SD) to standard error of cross-validation (SECV)

ratio, coefficient of determination in the cross-validation R2cv, and standard

error of cross-validation (SECV) to standard error of laboratory (SEL) ratio

for duplicate ADF analyses, were used to assess performances of the

calibration equations.

Results and Discussion

ADF contents in the rapeseed samples showed a higher range (6.80 to 13.46

% dry wt) than those previously reported in NIRS studies, the lower values

being close to those reported for B. juncea and B. carinata low ADF

cultivars. The mean and SD were, respectively, 9.42 and 1.28 % dry wt.

Calibration resulted in R2c of 0.85 and 0.77 for is/10.eqa and is/1.eqa,

respectively. These equations were validated by cross-validation giving an

R2cv of 0.80 (Figure 1) and 0.73, respectively, which are considered as

indicative of good quantitative information equations (Shenk and

Westerhaus 1996). Both sample presentations produced equations exhibiting

4

similar accuracy, as it can be deduced from the standardized SECV values

obtained (SD/SECV of 2.13 and 1.91, for is/10.eqa and is/1.eqa,

respectively). These ratios were for both equations similar to those

previously reported by Font et al. (2003). The ratio standard error of cross-

validation to standard error of laboratory (SECV/SEL) exhibited by

is/10.eqa and is/1.eqa (2.30 and 2.57) (Table 1) supported the high

prediction capability of both equations.

References

Barnes, R. J., M. S. Dhanoa, and S. J. Lister, 1989: Standard normal variate

transformation and de-trending of near-infrared diffuse reflectance spectra.

Appl. Spectrosc. 43, 772―777.

Barton, F. E. II., and W. R. Windham, 1988: Determination of acid

detergent fiber and crude protein in forages by near infrared spectroscopy:

Collaborative study. J. Assoc. Off. Anal. Chem. 71, 1162―1167.

Bjergegaard, C., B. O. Eggum, S. K. Jensen, and H. Sorensen, 1991: Dietary

fibres in oilseed rape: physiological and antinutritional effects in rats of

isolated IDF and SDF added to a standard diet. J. Anim. Physiol. An. N. 66,

69―79.

Font, R., M. del Río, J. M. Fernández, and A. de Haro, 2003: Acid detergent

fiber analysis in oilseed Brassicas by near-infrared spectroscopy. J. Agric.

Food Chem. 51, 2917―2922.

5

García-Ciudad, A., B. García-Criado, M. E. Pérez-Corona, B. R. Vázquez

De Aldana, and M. A. Ruano-Ramos, 1993: Application of near-infrared

reflectance spectroscopy to chemical analysis of heterogeneous and

botanically complex grassland samples. J. Sci. Food Agric. 63, 419―426.

Goering, H. K., and P. J. Van Soest, 1970: Forage fiber analysis; apparatus,

reagents, procedures, and some applications. USDA-ARS Agric. Handbook

No. 379. U.S. Gov. Print. Office, Washington DC.

Marten, G. C., G. E. Brink, D. R. Buxton, J. L. Halgerson, and J. S.

Hornstein, 1984: Near infrared reflectance spectroscopy analysis of forage

quality in four legume species. Crop Science 24, 1179―1182.

Michalski, K., P. Ochodzki, and B. Cicha, 1992: Determination of fibre,

sulphur amino acids and lysine in oilseed rape by NIT. In: I. Murray, and I.

A. Cowe (eds), Making Light Work: Advances in Near Infrared

Spectroscopy. VCH Verlagsgesellschaft, Weinheim (Federal Republic of

Germany) and VCH Publishers, New York, 333―335.

Shenk, J. S., and M. O. Westerhaus, 1996: Calibration the ISI way. In: A.

M. C. Davies, and P. C. Williams (eds), Near infrared spectroscopy: The

future waves. NIR publications, Chichester, 198―202.

6

Table 1. Calibration and cross-validation statistics for ADF in the different presentation forms (n= 98) (% dry wt)

Calibration Cross-validationEquation SECc R2cd SD/SECVe R2cvf SECV/SELis/10 eqaa 0.50 0.85 2.13 0.80 2.30is/1 eqab 0.61 0.77 1.91 0.73 2.57

aintact seed 10 ml seed volume equation; bintact seed 1 ml seed volume equation; cstandard error of calibration; dcoefficient of determination in the calibration; eratio of the standard deviation of the reference data to standard error of cross-validation; fcoefficient of determination in the cross-validation

7

Figure 1. Cross-validation scatter plot for ADF belonging to the intact seed 10 ml seed volume model.

6 7 8 9 10 11 12 13 146

7

8

9

10

11

12

13

14

acid

det

erge

nt fi

ber (

NIR

S) (

% d

ry w

t)

acid detergent fiber (laboratory) (% dry wt)

8