Watch the slidshow

Rongguang Zhu, Lujia Han*, Zengling Yang, Xian Liu
College of Engineering, China Agricultural University，Beijing 100083，P. R. China

1 INTRODUCTION
Feed contaminated with meat and bone meal (MBM) is commonly accepted as the main transmission carrier of the prior responsible for bovine spongiform encephalopathy (BSE). The ban on using processed animal proteins, including MBM as feed ingredient for all farmed animals is an important measure to prevent the spread of transmissible spongiform encephalopathies (TSE). Use of compound fertilizers or composts adulterated with banned MBM in livestock grazing systems may cause potential BSE risk through feed chain. The objective of this study was to demonstrate the feasibility of using NIRS to determinate MBM content in compound fertilizers and composts.

2 MATERIALS AND METHODS

2.1 Preparation of samples
Four kinds of chemical compound fertilizers with different proportion of N, P and K, 15-15-15, 15-10-5, 14-11-8 and 12-8-5 respectively, were purchased in the rural markets in Beijing. 41 animal manure composts with different stuffs, proceecing technics and storage periods, were collected from farms and compost factories in Hebei and Beijing of China. 28 MBM were obtained, 20 were from domestic mills and 8 were imported. All of them were smashed to pass a 1mm sieve. 140 adulterated compound fertilizer samples were prepared in laboratory by mixing the 4 kinds of compound fertilizers with 3 kinds of MBM randomly at different levels of 0.1%-10.0% (w/w). 120 adulterated compost samples were obtained by mixing 41 compost samples with 28 MBM at different levels of 3%-24% (w/w). All of the samples were stored at +4℃ in a refrigerator untill analysis by NIRS.

2.2 NIRS analysis
A NIRS system SPECTRUM ONE NTS (Perkin Elmer, USA) was used for NIRS analysis. Samples were scanned in rotating quartz cell. Each of the samples was scanned 3 times as log 1/R over the wavenumber range 10000 cm-1 to 4000 cm-1 and the average spectrum was recorded prior to analysis.

2.3 Mathematical and statistical analysis
Data analysis was performed using the Spectrum QUANT+ software provided by Perkin-Elmer. Preliminary data evaluation was performed using two parameters, spectral leverage and absolute residual of actual value, to test the outlier. Samples were divided into a calibration set, and a validation set based on concentration grads. For fertilizer group, the number of samples are 91 and 47 for calibration and validation sets, and for compost group, they are 89 and 29, respectively. The NIRS calibration models were developed using partial least squares (PLS1) regression method. Various spectral regions and mathematic pre-
treatment methods, including smoothing, derivative, standard normal variate with detrending (SNV-D) and multiplicative scatter correction (MSC), were applied.
Cross validations were also conducted to select the optimal number of factors and avoid overfitting. The optimum spectral region and pre-treatment methods were selected according to the lowest standard error of prediction (SEP) by full cross validation, and the number of factors recommended by the software was adopted.
Statistical parameters of the calibration results included the coefficient of determination (R2), the standard error of calibration (SEC). For statistical parameters of the validation results, the coefficient of determination (r2), the standard error of prediction (SEP) and the SD/SEP ratio (RPD) were also used to evaluate the calibrations.

3 RESULTS AND DISCUSSION
The spectra of all samples are plotted in Figure1. Table 1 gives the PLS1 analysis results.

Fig.1 The spectra of all samples

Table 1 Statistics of the calibration and validation sets of the PLS1 models

It was showed that the correlation coefficients of calibration (R2) were 0.996 and 0.622 for adulterated compound fertilizers and composts, the correlation coefficients of validation (r2) were 0.988 and 0.722, and the RPD were 8.84 and 1.87 respectively.

4 CONCLUSION
The results indicated that NIRS could be used to quantify the adulteration of banned MBM in compound fertilizers with high prediction accuracy, and be insufficient to determinate the content of MBM in composts for considerably low prediction accuracy.

Keywords:
NIRS, Quantification, MBM, Compost, Compound fertilizer