HMF, invertase and diastase activity of swiss and imported honeys. Analysis with logistic regression

Honey freshness and heat damage are generally determined by measuring hydroxymethylfurfural (HMF) content and invertase and diastase activity. The objective of the study is to find a single function combining these variables, by which honey freshness and heat damage can be elucidated. HMF, invertase and diastase activity data from the control activity of the Swiss canton laboratories were submitted to normal and logistic regression statistical analysis: 88 blossom- and 52 honeydew honeys of Swiss origin and 162 blossom and 48 honeydew honeys from foreign origin. The values for each parameter are given in table 1. Swiss honeys have lower HMF values and higher invertase and diastase activities. The data are graphically represented in figure 1. The HMF distribution, expressed by the kurtosis value is not normal, which can be also seen on fig. 2. The compliance of the honey to the different norms regarding HMF, diastase and invertase activity is given in table 2. Logistic regression, generally applied also on data with non normal distribution, was applied on the data. The logistic regression of individual parameters is summarised in table 3. Blossom honeys of Swiss and foreign origin can be best separated by the HMF content. The differences between the parameters of honeydew honeys of Swiss and foreign origin are less pronounced. Logistic regression analysis on all parameters showed that a better differentiation than with single parameters could be achieved between the mostly fresh and unheated Swiss honeys and the longer stored and exposed to more heat foreign honeys (table 4, figure 3). Three models were determined for all honeys (table 4): Model 1 ©¯1 = b0 + b1 DN + b2 IN + b3 HMF DN: diastase number, IN: invertase number HMF in mg/kg In this model all coefficients bi were significant at the 5 % level, if the model was applied on both blossom and honeydew honeys. The coefficient for the diastase activity (b1) was not significant for the blossom honey data, thus leading to model 2: ©¯2 = b0 + b2 IN + b3 HMF For honeydew honeys the HMF coefficient (b3) was not significant, leading to model 3: ©¯3 = b0 + b1 DN + b2 IN The results of the logistic regression show that the important parameters for the determination of freshness and of heat damages of blossom honeys are HMF and invertase, while for honeydew honeys only diastase and invertase are important. In figure 3 the models 2 and 3 are applied on the data of the blossom and the honeydew honeys. An individual h value for blossom and honeydew honeys based on the models two and three can be introduced for the judgment of honey freshness, storage and heat damages. In table 5 values for ©¯ values for fresh and unheated honeys on one side and commercial honeys on the other, are proposed, both for blossom and for honeydew honeys.