Application of Principal Component Analysis in Assessment of Relation Between the Parameters of Technological Quality of Wheat Grains Treated with Inert Dusts Against Rice Weevil ( Sitophilus oryzae L . )

Marija Bodroža-Solarov 1, Petar Kljajić2, Gor an Andrić2, Bojana Filipčev1, Olivera Šimurina1, Marijana Pražić Golić2 and Milan Adamović3 1Institute of Food Technology, University of Novi Sad, Bulevar cara Lazara 1, 21000 Novi Sad, Serbia 2 Institute of Pesticides and Environmental Protection, Banatska 31b, 11080 Belgrade, Serbia 3Institute of Technology of Nuclear and Other Mineral Row Materials, Bulevar Franš d’Eperea 86, 11000 Belgrade, Serbia (marija.bodroza@fins.uns.ac.rs) Received: July 8, 2011 Accepted: August 23, 2011


INTRODUCTION
Infestation of stored grains with rice weevils Sitophilus oryzae (L.) (Coleoptera: Curculionidae) is an important quality control problem for food industry due to huge economic losses.Effective insect pest control includes continuous application of chemicals like contact insecticides and fumigants (Hill, 1990;Rees, 2004).Although effective and low-cost, the use of pest-control chemicals exerts many disadvantages such as widespread development of resistance in insects (Korunić et al., 1998;Kljajić and Perić, 2005;Kljajić, 2008) and occurrence of trace residues in food commodities which rises environmental and public health concerns.In order to reduce chemical inputs, investigations and development of alternatives to chemical insecticides have been stimulated.
Interest in suitability of natural zeolite (NZ) and diatomaceous earth (DE) as alternatives to conventional insecticides and fumigants is increasing.Codex Alimentarius Commission (1999) recommends control of insect pests in food commodities and lists zeolite and diatomaceous earth as permitted substances in organic food production and in plant pest and disease control.This has focused efforts on investigating the fumigant action of these materials (commonly designated as inert dusts) for development of a potential new class of safer insect-control agents.
DE is a naturally occurring siliceous mineral compound from marine sediments which consists of microscopic skeletal remains of single-celled algae (phytoplankton) called diatoms.It is composed of amorphous silicon dioxide that is non-toxic to mammals and is registered as a food additive in Canada, the USA, and in many other countries (Korunić et al., 1998).Natural zeolites (NZ) are hydrated aluminosilicates characterized with complex three-dimensional frameworks of silica and alumina tetrahedra linked through shared oxygen atoms.Clusters of tetrahedra form sheetlike or chainlike polyhedral units build up the entire structure by mutual linking (Daković et al., 2007).
There are several research works dealing with the effectiveness of DE (Korunić et al., 1998;Fields and Korunić, 2000;Athanassiou et al., 2005;Vardeman et al., 2007) and NZ (Kljajić et al., 2010) against stored-grain insects.Furthermore, few authors addressed the effects inert dusts treatments exert on the technological quality of grain and flour.It was estimated that around 2% Dryacide (a DE preparation) remains in the flour after treatment (Desmarchelier and Dines, 1987).DE treatments at recommended doses (500-3500 g/t) were found to adversely affect moisture content, flowability, bulk density of treated grains, and hinder handling due to exces-sive dust formation but were not found to diminish the quality of end-products obtained from the treated flour.Korunić et al. (1996) investigated the effect of another commercial DE product, Protect-It ® (at 50 and 300 mg/ kg doses) and showed no significant differences in the protein content, falling number value, sedimentation value and color of wheat flour, but the flour rheological properties changed which was reflected through increase in farinograph development time and increased extensigraph resistance measured at 45 and 90 min.They also found that baking potential of treated flour was not altered.
Multivariate analysis is an analysis of a large number of variables, which enables their examination and quantification, as well as identification of their dependence, i.e. links between a larger numbers of variables.Principal components analysis (PCA) is one of many multivariate analysis methods.This method enables transformation of a large number of variables into a smaller number of latent variables (principal components, PCs) which are not inter-correlated.
These transformed variables represent linear functions of input variables.PCA is a powerful tool for pattern recognition, classification, modeling, and other aspects of data evaluation (Csomos et al., 2002).Also, PCA is a projection method, and dimension reduction of the data can be achieved using a smaller number of principal components than original variables.
The aim of this research is to apply the PCA on a practical example related to treatment of various wheat lots characterized with different endosperm vitreousness and insect infestation status with inert dust preparations and to reveal which treatment showed statistically significant impact on the variability of technological parameters of wheat quality.

Wheat samples
Two types of wheat (Triticum aestivum ssp.vulgare) characterized with different degree of endosperm vitreousness: 17.0% (low-vitreous or mealy grains, LVG) and 81.8% (high-vitreous grains, HVG) were used in the investigation.The material was procured from local producers.

Inert dusts
Various samples of inert dusts were used in the experiment: 1) an inert dust based on natural zeolite (NZ) originating from Serbia and processed at the Institute for Technology of Nuclear and Other Mineral Materials in Belgrade; 2) two dusts based on diatomaceous earth (DE S-1 and DE S-2, originating from Serbia) and 3) registered product Protect-It ® (Hedley Technologies Inc., Canada).

Grain treatment with inert dusts
Inert dusts were applied on 0.5 kg wheat samples in five replicates, using doses that were found optimal for efficient storage pest control in preliminary experiments 1) natural zeolite (NZ), 1.0 g/kg, 2) diatomaceous earths (DE S-1 and DE S-2), 0.75 g/kg, and 3) as standard DE product registered worldwide (Protect-It ® ), 0.2 g/kg.
In separate tests with insect infestation, inert dusts were applied on two types of wheat in the same manner as described above.After 21 days of incubation, the samples were sieved to separate insects.After total seven weeks of incubation, sieving was repeated to remove the progeny in F 1 generation.The samples were then placed into plastic bags and put into refrigerator for 24 h.Next day, sieving was repeated again to remove the rest of possibly remained insects.Sieving was conducted using sieves 7/64'' for insect removal only whereas dust was returned to initial grain mass by mixing for one minute.The sieved samples were stored in plastic bags at room temperature until further examination.Kernel vitreousness was determined by visual inspection where vitreous kernels appear glassy and translucent whereas non-vitreous kernels appear starchy and opaque (ICC, No 129).Test weight was determined using Schopper scale.Measurement of test weight was repeated five times for each sample.Protein contents were determined according to ICC approved methods No 105/2.Content of SiO 2 in wheat was determined grav imetrically according to official methods (31,1972).Rheological properties of wheat flour samples were analyzed in a Brabender farinograph according to method ICC No 115/1, and Brabender extensigraph according to method No ICC 114/1.

Statistical analysis (Principal Component Analysis -PCA)
The algorithm of PCA can be found in standard chemometric material (Oto, 1999).Descriptive analysis of the data and the PCA were performed using the software package STATISTICA 10.0.In summary, PCA decom-poses the original matrix into several products of multiplication into loading (parameters of quality) and score (different treatment with inert dusts) matrices.Parameters of grain quality are taken as variables (column of the input matrix) and different treatment with inert dust as mathematical-statistical cases (rows of the matrix).
The number of factors retained in the model for proper classification of the data from Table 2 was determined by application of Kaiser's and Rice's rule (1974).Therefore, two components having eigenvalues >1 were used for further analysis.PCA yields two PCs explaining 74.2% of the total variance in the data.Loading values (i.e.correlation coefficients) higher than 0.700 were marked throughout Table 3 in boldface type.
Projection of the variables in the factorial plane (Table 3) indicates that the variables vitreousness (0.948); test weight (0.932); resistance to extension (0.920); FQN (0.913) most contributed to the first PC indicator (which accounted for 55.61% of the variability), and thus to the total variability of the basic set.The second PC indicator (which accounted for 18.93% of the variability) is contributed most by the content of SiO 2 (-0.653) and extensigraph area (-0.762).The loadings plot of components in factorial 2D plane shows that the highest contribution to the description of the second PC indicator was due to SiO 2 content as it forms the smallest angle with PC2 loading axis.The content of SiO 2 was higher in those samples treated with inert dusts but more damaged by insects (Figure 1).
Factor coordinates of individual observations (Figure 2) indicate that the total variability of the first component is influenced mostly by the non-treated low vitreous samples without insect infestation (Control) (-3.64) and those infested and treated with Protect-It ® (-3.51).
Factor coordinates also indicate that the total variability of the second component was influenced mostly by the low vitreous samples without insect infestation (3.68) and high vitreous non-treated samples (3.41).Figure 2 shows that according to the first two PCs, the grain lots of low vitreous wheat (infested and noninfested) treated with the preparations are similar to each other.Also, there is similarity within the treated wheat lots of high vitreous wheat (infested and non-infested).The first PC distinguishes between the lots of low and high vitreous wheat grains which suggest that the inert dust treatments produce different effect regarding test weight, FQN and resistance to extension in dependence to endosperm vitreousness.Two points (sample 6 and 16) appeared as outliers: control samples of low and high vitreous wheat infested with insects (non-treated with inert dusts).These samples were characterized with the poorest technological quality as they were most damaged by insects.
Principal component analysis (PCA) of data set was able to distinguish among the various treatments of wheat lots.It was revealed that inert dust treatments produce different effects depending on the degree of endosperm vitreousness.The best predictor of technological quality is endosperm vitreousness and the most affected parameters of technological quality are test weight, resistance to extension, flour quality number, and extensigraph area.46008).The authors would like to express their gratitude to the company AgroChem MAKS from Zagreb, Croatia, for providing samples of product Protect-It ® .

Table 1 .
Experimental design and sample designationInert dust treatment with and without insect Sample

Table 2 .
Technological quality parameters of wheat grain lots

Table 3 .
Results of Principal Component Analysis for wheat quality parameters in Grain sample treated with different inert dusts: Varimax Rotated Principal Component Loadings