Agglomerative hierarchical clustering analysis of twenty-six rice samples analysed by instrumental neutron activation analysis and other techniques

Abstract

As part of the ongoing work on the safety and security of foods grown locally and imported into Jamaica, twenty five rice brands available on the Jamaican market and one sample of locally grown brown rice were collected and analysed for 36 elements using a combination of instrumental neutron activation analysis, total reflection X-ray fluorescence, flame atomic absorption spectrometry and inductively coupled plasma optical emission spectrometry. Quality control for each of the techniques was provided by analysis of reference materials and inter-comparisons of common elements. Agglomerative hierarchical clustering analysis performed on the raw data set identified white, parboiled and brown rice as the major clusters. The element and cluster median concentration in μg/g (for polished, parboiled and brown respectively) for each of the three clusters were Al (8.7, 11.3, 14.9), As (0.20, 0.21, 0.15), Br (12.1, 0.9, 3.4), Ca (60.4, 409.1, 112.5), Co (0.1, 0.09, 0.11), Cr (0.11, 0.10, 0.11), Cs (0.012, 0.004, 0.018), Eu (0.004, 0.004, 0.003), Cu (2.2,2.7, 3.6), Fe (13, 17, 20), K (894,1,815,2,538), La (0.009, 0.014, 0.015), Mg (340, 504, 1,303), Mn (9.7, 14.9, 28.5), Mo (0.7, 1.1, 0.8), Na (5.1, 9.9, 16.9), P (936, 2,328, 3,670), Rb (4.5, 6.1, 7.3), S (1,088,1,308, 1,321), Sb (0.004, 0.006, 0.005), Sc (0.001, 0.002, 0.002), Se (0.09, 0.17, 0.16), Sm (0.001, 0.003, 0.001), Sr (2.0, 1.9, 2.9), V (0.036, 0.041, 0.058), Zn (17.9, 15.3, 22.8). The use of agglomerative hierarchical clustering has proven to be a rapid method for the analysis of the dataset, correctly clustering rice by process that is, polished, parboiled and brown, with sub-clusters that further identified location, cultivars and an adulterated sample. The data validated here has been used to estimate the elemental contribution of rice to the Jamaican diet.