Publications des agents du Cirad


Prediction by near infrared spectroscopy of the biochemical composition of various raw materials used in the organic fertiliser industry

Thuriès L., Bastianelli D., Davrieux F., Bonnal L., Oliver R.. 2007. In : Burling-Claridge G.R. (ed.), Holroyd S.E. (ed.), Sumner R.M.W. (ed.). Near infrared spectroscopy : Proceedings of the 12th International Conference, Auckland, New Zealand, 9th - 15th April 2005. Chichester : IM Publications, p. 781-785. International Conference on Near Infrared Spectroscopy. 12, 2005-04-09/2005-04-15, Auckland (Nouvelle-Zélande).

Composting and recycling agro-food wastes reduce their volume and can be a solution for waste management. Organic fertilisers utilisation may lead to several beneficial aspects as improving soil fertility. Composts and organic fertilisers may originate from very different sources of raw materials, and their elaboration may also vary according to the producers. In order to evaluate more accurately their potential quality (carbon and nitrogen transformations when added to soils), it is useful to investigate the relationships between their characteristics and their transformations. The fibre content estimated by Van Soest fractions, proved to be a good indication of the potential capacity of C and N transformations. Nevertheless, a complete fibre analysis (5 fractions) is expensive and time-consuming. This is why a programme was set up to evaluate the potential of NIRS to assess the quality of raw materials used in compost and organic fertilisers elaboration. The raw materials originated from (i) industrially pre-processed plant residues: wet and dry grape skins, de-oiled grape pips, coffee cake, de-fatted cocoa bean, cocoa skin, olive pulp, maize cob, barley straw, rice hulls, rapeseed cake, soybean cake; and (ii) and tropical plant residues samples, as parts of trees, shrubs, crops, cover crops, potentially utilisable in composting. The parameters measured were Lignin (LIG, n=122), Organic Matter (OM, n=322) and Total Nitrogen Kjeldahl (TN, n=300). Due to the heterogeneity of fresh materials, samples were dried (40°C) ground (<1mm sieve) and scanned on a NIRS 6500 (Foss NIRSystems) in ring cups. Spectra were corrected with SNVD 2,5,5 (WIN-ISI) mathematical treatment and calibrations were performed using a modified partial least square regression (mPLS, WIN-ISI). All the parameters varied widely, because of the diversity of the raw materials. The equations had Standard Errors of Calibration (SEC) of 3.03; 0.78; 0.16 g 100 g-1 d.m., and determination coefficients (R2) of 0.96; 0.93; 0.92 for LIG, OM, and TN respectively. Standard Errors of Cross Validation (SECV) were 3.51; 0.97; 0.18 g 100 g-1 d.m., for LIG, OM and TN respectively. The corresponding SD/SECV ratios were equal to or above 3, (RPDcv= 4.6; 3.1; 3.0) the model developed here could be thus considered as satisfactory. The precision of the models obtained is sufficient for a global characterisation of samples. It is concluded that the major parameters used in quality characterisation of compost and organic fertilisers can be valuably estimated by NIRS, and that this tool can be applied to quality control in routine.

Mots-clés : compost; matière organique; spectrométrie; fertilité du sol

Documents associés

Communication de congrès

Agents Cirad, auteurs de cette publication :