90
P. Lazo et al.
main criteria in selecting the number of optimal factors are Kaiser dropping factor
and that of Eigenvalues equal and/or larger than 1. The correct choice of the meaningful number of factors is based on the choice of the model that may describe the
population in factors structure after varimax rotation, such as factor loadings and
elements correlations. It may help to identify the sources of the elements in moss
samples and in the study area. After all, the associations of the elements with high
loadings in the same factor (Reimann et al. 2002) are assumed to be affected by
different characteristics, such as chemical properties of the elements, the geochemical associations of elements in soil and dust, local and long-range transport of the
elements, the inventory of the local emission sources of the elements in the study area
and the previous knowledge of the atmospheric concentrations (Miranda et al. 2015).
The emission inventory of air pollutants could supply the essential information to
understand regional and local emission sources of the pollutants (Qiu et al. 2014).
These are important reasons for linking the metals with their sources of origins or
factors affecting their presence in the study area. Factor loadings (FL) larger than 0.4
(Lazo et al. 2019) are recommended for interpreting each factor. As the variances of
the elements differ significantly, from low variation to very high variation, and the
number of elements under investigation is high by producing a big data of correlation
matrices, the factor loadings (FL) larger than 0.5 were selected for interpreting each
factor (Table 8.1, Figs. 8.1, 8.2, 8.3, 8.4). The model that we applied consists in two
steps.
1st: Factor analysis was applied to the standardized concentration data obtained
as the ratio of the concentration of the element against its respective median
concentration;
2nd: Aiming to avoid the high variability of the concentration data the outlier
sites obtained from the score plot diagram of the original concentration data (Factor
analysis), were excluded.
Six main factors that represent 76% of the total variance were identified from the
standartized data after exluding the outlier points. The associations of metals within
the same factor are as follows:
Factor 1 (F1) and Factor 2 (F2) are the strongest factors representing 41% of the
total variance, 24 and 17% respectively. F1 and F2 are characterized by high loadings
of lithogenic and crustal elements such as Yb, Sc, Ta, Ce, La, Th, Nd, Hf, U, Sm,
Zr, Mn, W, Co and Ti (F1, FL > 0.59), and Al, Li, Sr, V, Fe, Ba and As (F2, FL
> 0.61). These elements are naturally distributed as typical soil elements (Rudnick
and Gao 2003) and crustal materials that may indicate the soil dust as their origin.
The presence of Al in this factor confirms this assumption, since Al compounds
are insoluble and most of the Al found in biological systems comes from soil and
dust contamination (Qarri et al. 2013). Similar associations of the elements were
reported to the 2010 AMS and for other regions of Europe (Harmens et al. 2015) and
Balkan countries (Špiri´ c et al. 2013; Barandovski et al. 2015; Stafilov et al. 2018). It
indicates the origin of the elements from local emission and/or long-range transport
of the pollutants. Spatial analysis plots and GIS maps of FL1 and FL2 data are shown
in Fig. 8.1.
P. Lazo et al.
main criteria in selecting the number of optimal factors are Kaiser dropping factor
and that of Eigenvalues equal and/or larger than 1. The correct choice of the meaningful number of factors is based on the choice of the model that may describe the
population in factors structure after varimax rotation, such as factor loadings and
elements correlations. It may help to identify the sources of the elements in moss
samples and in the study area. After all, the associations of the elements with high
loadings in the same factor (Reimann et al. 2002) are assumed to be affected by
different characteristics, such as chemical properties of the elements, the geochemical associations of elements in soil and dust, local and long-range transport of the
elements, the inventory of the local emission sources of the elements in the study area
and the previous knowledge of the atmospheric concentrations (Miranda et al. 2015).
The emission inventory of air pollutants could supply the essential information to
understand regional and local emission sources of the pollutants (Qiu et al. 2014).
These are important reasons for linking the metals with their sources of origins or
factors affecting their presence in the study area. Factor loadings (FL) larger than 0.4
(Lazo et al. 2019) are recommended for interpreting each factor. As the variances of
the elements differ significantly, from low variation to very high variation, and the
number of elements under investigation is high by producing a big data of correlation
matrices, the factor loadings (FL) larger than 0.5 were selected for interpreting each
factor (Table 8.1, Figs. 8.1, 8.2, 8.3, 8.4). The model that we applied consists in two
steps.
1st: Factor analysis was applied to the standardized concentration data obtained
as the ratio of the concentration of the element against its respective median
concentration;
2nd: Aiming to avoid the high variability of the concentration data the outlier
sites obtained from the score plot diagram of the original concentration data (Factor
analysis), were excluded.
Six main factors that represent 76% of the total variance were identified from the
standartized data after exluding the outlier points. The associations of metals within
the same factor are as follows:
Factor 1 (F1) and Factor 2 (F2) are the strongest factors representing 41% of the
total variance, 24 and 17% respectively. F1 and F2 are characterized by high loadings
of lithogenic and crustal elements such as Yb, Sc, Ta, Ce, La, Th, Nd, Hf, U, Sm,
Zr, Mn, W, Co and Ti (F1, FL > 0.59), and Al, Li, Sr, V, Fe, Ba and As (F2, FL
> 0.61). These elements are naturally distributed as typical soil elements (Rudnick
and Gao 2003) and crustal materials that may indicate the soil dust as their origin.
The presence of Al in this factor confirms this assumption, since Al compounds
are insoluble and most of the Al found in biological systems comes from soil and
dust contamination (Qarri et al. 2013). Similar associations of the elements were
reported to the 2010 AMS and for other regions of Europe (Harmens et al. 2015) and
Balkan countries (Špiri´ c et al. 2013; Barandovski et al. 2015; Stafilov et al. 2018). It
indicates the origin of the elements from local emission and/or long-range transport
of the pollutants. Spatial analysis plots and GIS maps of FL1 and FL2 data are shown
in Fig. 8.1.