*3.1. Soil Characterization*

Composite soil samples of the years 2004 and 2011, from each site, were characterized in terms of pH, conductivity, humidity and organic matter content. The values obtained are shown in Table 1.


**Table 1.** Soil characteristics in two years of sampling.

When comparing both years, the values of pH, organic matter and electrical conductivity are identical. The main difference is found in the humidity values, as they depend directly on the rain on the days before sampling and not on annual precipitation. The paired *t*-test performed between each of the variables and for the two years, showed that all data pairs are statistically significant (*p* = 0.05) with the exception of humidity data.

#### *3.2. Soil Metals Contents*

Metal content values of the sampled soils in the different city sites are presented in Table 2. Overall, and for most metals there is an increase in their content in soils over the years, except for two sites, residential area (RA) and city park (CP). Detailing the analysis of the data for each metal it was found that cadmium levels increased at the entrances (EC1 and EC2) and in the city center (CC), registering however a slight decrease in these places in the last year of sampling. In the urban highway (UH) the levels are stable, showing a decrease over the years in the residential area and in the city park, although being smaller the decreasing trend in these locations. the values for each metal content in each sampled site over the years are shown in Figure 2.

**Figure 2.** Metal content (Pb, Ni, Cr and Cd) in each sampled site over the years of the campaign (2003–2011).


**Table 2.** Metal content (mg/kg) values for each sampling site and each year, minimum (Min), maximum (Max), mean (Mean) values and relative standard deviation of the mean (RSD %). Expanded uncertainty values ( *U*), for 95% confidence level, are shown in the last column.

Chromium levels exhibited, in all sampled sites, an overall increase between 2003 and 2010 and a slight decrease in the last year (2011). Nickel and lead showed a growing trend of their content in the soils over the years except for the residential area and the city park. In these two places the levels remained stable over time. For nickel, however, two sites (CE2 and CC) presented a decrease in 2011.

The decrease in the levels of some metals in the last year of sampling may be directly linked to the decrease in car traffic in the city of Lisbon. In fact, a serious economic crisis began in 2010 in Portugal, which was reflected in the decline of economic activity and in the increase in unemployment with less people moving around the city, either in private or corporate vehicles, or in public transports.

Table 2 also shows the average annual contents for each metal in the city (values in bold), obtained from the average values of each metal in all locations and for each year.

Figure 3 shows an increasing trend in the levels of the various metals in different places over the years except for the last year of sampling and for Cr, Cd and Ni. The average metal values observed over the years of this campaign are close to 0.46 mg/kg for Cd, 44 mg/kg for Cr, 46 mg/kg for Ni and 5.7 mg/kg for Pb. These values represent moderate metal pollution in the soils of Lisbon as also revealed by the calculated environmental and ecological risk indexes presented below. Other authors [38], present identical results to this study observing low levels of these metals in the soils of Lisbon. The city soils can be used

for agriculture, in particular for urban gardens, without significant risk of contamination for the vegetables growing there.

**Figure 3.** Boxplot of the evolution of metal content in soils over the years of the study.

It is presented in Table 3 the mean, minimum and maximum values of the studied metals content from several studies all around the world, which were used to compare with the values obtained in the present work. Except for lead, the average values for each metal shown in Table 2 are within the limits indicated for the crust. Pb has a higher value in the soils of the studied cities, probably due to emissions from burning leaded gasoline throughout many decades. The values of the metal levels in the soils reveal the reality of each city in terms of anthropogenic emissions, namely the existence of manufacturing facilities and the burning of fossil fuels, both from heating and car traffic. Moreover, the baseline values of metals soil content at each site are important but not available in all or in most cases. Additionally, it must be emphasized that the analytical techniques that allow the determination of metal concentration levels in soils, GFAAS and ICP-MS (inductively coupled plasma mass spectrometry) among others, were only developed and implemented in the decades of 1970s and 1980s. At that time, soils already presented reasonable levels of anthropogenic pollution due to many years of industrialization and low environmental protection by state regulation.


**Table 3.** Metal content (mg/kg) mean, minimum (min) and maximum (max) values from different places all around the world.

#### *3.3. Pollution and Ecological Indexes*

In order to estimate the degree of soil pollution and the ecological and environmental risk of Lisbon soils, the most important indexes to assess the level of pollution and ecological risk were calculated: geo-accumulation (*I*geo), contamination factor ( *C*f), degree of contamination ( *C*d), pollution load index (*PLI*), ecological risk factor (*E*r) and global ecological risk (*RI*). The authors had no knowledge or information concerning any past published values for the levels of metals in Lisbon soils in unpolluted conditions.

In the absence of background values, one of the solutions often used is to consider the estimated values for the earth's crust as a background. The crust values are global values that are intended to represent the average composition of the crust based not on global and systematic measurements but on values resulting from the combination of several studies. There is obviously an over-representation of countries and regions in which studies were carried out, in comparison to less studied areas. In the specific case of Lisbon, the values proposed for the crust are higher than some of the metal contents found. This shows that the values of the crust cannot be used as reference values in this study. Table 4 presents some values from literature taken as reference values for the concentration of metals in soils and earth's continental crust.

It should also be noted that the soils of Lisbon have a reasonable geological diversity [62] including regions with volcanic soils, alluvion soils, medium Cretacic and from Miocenic periods. The sampling sites CC, RA and UH are in a Miocenic formation (sand, clay and limestone). The CP and CE1 are in an antique volcanic zone and the CE2 site is located in a calcaric cambisols zone. However, the reasonable antiquity of Lisbon as a city (more than twenty centuries) and therefore a recipient of anthropogenic pollution for many centuries, as well as the frequent works that take place in cities involving soil movement, prevent the unambiguous geological classification of the soils of each of the studied areas.


**Table 4.** Typical metal content in earth's continental crust and in soil (mg/kg).

To overcome the limitation of not knowing the background values and being unable to use the crust values method above mentioned, an established statistical method was used to estimate background values for metals in the various locations studied in the city of Lisbon. On this subject, several authors [63,64] emphatically refer to the impossibility of obtaining exact or even reasonably values for the background values of metals (and other pollutants) in soils and other environmental compartments. Among the estimation methods referred by Reimann [65], the iterative 2s-technique was chosen because it is the method that best adapted to the available amount of data. This method is also referred by the authors as the one that presents the most reasonable degree of approximation to real values [66]. The obtained values are shown in Table 5 and were used to calculate the pollution indexes and ecological indexes.

**Table 5.** Calculated background values (mg/kg) for each sampling site according to the iterative 2s-technique, proposed in [65].


Results of *<sup>I</sup>*geo calculated values are presented in Figure 4. Ni shows the highest values followed by Cr and Pb. Ni exhibits moderate levels of contamination at all sites except in residential area (RA) and the city park (CP), both classified as "uncontaminated". Higher values of *<sup>I</sup>*geo for Ni were found in the soils next to the city entrance 2 (CE2). For Pb and Cr the *<sup>I</sup>*geo index have classifications between "uncontaminated" and "moderately contaminated". Regarding the values of *<sup>I</sup>*geo for Cd, this metal reveals an "uncontaminated" contamination class in UH, RA and CP and an "uncontaminated to moderately" class in the city entrances (CE1, CE2) and at the city center (CC).

**Figure 4.** *<sup>I</sup>*geo values calculated for each metal (Pb, Ni, Cr and Cd) in each sampled site.

The values for the Contamination factor (*C*f) are shown in Figure 5. Nickel presents a "very high contamination" level in three sites, city entrances (CE1, CE2) and city center (CC). The residential area (RA) and the city park (CP) present values in the "moderate contamination" level but very close of the "low contamination" level. Pollution by Cr and Pb also presents some places with a "very high contamination" level, namely in urban highway (UH). It should be noted that contamination by these metals in the city park and residential area is also reasonable. Cd has low values for this index with only one of the city entrances (CE2) showing a "moderate contamination" level.

**Figure 5.** Contamination factor (*C*f) values for each metal, in each sampling site. Insert: values of degree of contamination (*C*d) and pollution load index (*PLI*) with the values obtained for each sampling site and the global *PLI* calculated for the city of Lisbon.

A global view of pollution in the city of Lisbon can be given by the indexes that include the contribution of different metals: degree of contamination (*C*d) and pollution load index (*PLI*). The *C*d and *PLI* indexes are important because they considered all the analyzed metals in the soil and not only an individual metal. The insert graphs in Figure 5 show the results for these two indexes. The values of *C*d and *PLI* have a quite similar profile for the various studied locals. These global indexes reveal a grea<sup>t</sup> accumulation of metals in the city entrances (CE1 and CE2) and in the urban highway (UH) and a little less in the city center (CC). The residential area (RA) and the city park (CP) were the less polluted places. The *PLI* calculated for Lisbon, according to Equation (5), presented a value of 1.6 which represents a "moderate pollution" value for the whole city.

The obtained values for these different pollution indexes (*I*geo, *C*f, *C*d and *PLI*) follow the expected behavior, showing that the city entrances and the urban highway had the highest pollution load, followed by the city center. On the other hand, the residential area and the city park had lower pollution values.

The calculated values for the ecological risk factor (*E*r) and global potential ecological risk (*RI*) are shown in Figure 6.

**Figure 6.** Values of the ecological risk factor (*E*r) and the global potential ecological risk (*RI*).

The *E*r index presents higher values of risk for cadmium resulting from the greater toxic response factor attributed to this metal. The values of *E*r for cadmium are higher for the CE2 site with values classified as "considerable ecological risk". For the remaining locations, the ecological risk is moderate and is even considered as "low ecological risk" for the RA and CP locations. For Ni and Pb, the values of *E*r vary between "low" and "moderate ecological risk". For Cr the values of this index are also classified as "low ecological risk" but with globally lower values than Ni and Pb.

The RI values are shown in the insert graph of Figure 6. Moreover, for this risk factor that takes into account the various metals, CE2 is the site with the highest risk but only classified as "moderate ecological risk". The entrance to the city (CE1), the urban highway (UH) and the city center (CC) have a "low" but close to "moderate ecological risk". The city park (CP) and residential area (RA) locations have the lowest values of this ecological risk index.

Recent work [67] shows the usefulness of the various indexes for the classification of soil pollution. The use of indices makes possible to compare the soil pollution from different locations based on their initial state. However, the authors stress the necessary precaution in the use of soil pollution indexes. Other factors than the initial and final metal contents are relevant, including the geological structure of the soil that affects pH, a determining factor in the bioavailability of the elements.

Despite the fact that the present study was focused on these four dangerous metals, there is for sure contamination by other metals in the soils of Lisbon. This contamination certainly increases the pollution and composite ecological indexes, namely *C*d, *PLI* and *RI* as these indexes consider the contribution of different metals and reflect the accumulated health risk of all metals in the soils. In fact, the metals considered are the most toxic and those that must be monitored in the soil according to the European Directive [68], except for mercury and arsenic.

#### *3.4. Data Correlation and Principal Component Analysis*

Significant correlations with *p* = 0.01 were obtained for the following pairs of metals: (Ni, Cr), (Ni, Cd), (Ni, Pb) and (Pb, Cd). In other words, Ni is positively correlated with all

other metals and in addition to that, only the correlation between Pb and Cd is verified. The pairs (Cr, Pb) and (Cr, Cd) do not present a significant correlation.

For the principal component analysis (PCA), the levels of metals in the soil in each location of the city and in each year were considered. It was intended to verify the existence of a relationship between metals, that is, variations in soil metal concentrations that exhibit similar behaviors, in time or in different places in the city. From the data, three main principal components were extracted that explain 78% of the total variance of the results. The first factor (PC1) is responsible for 58% of the variance, PC2 for 11% and PC3 for 8%.

Figure 7 shows the data obtained for PCA analysis, PC1 *vs*. PC2. The grouping of sites with a high correlation in Cr, Ni and Pb concentrations is evident. These sites are located at the entrances to the city, establishing a clear link between the levels of these metals in the soil and vehicle traffic. In addition to those locations, the city center and the urban highway are also part of this set. The third factor (PC3), a factor that explains only a small variance of the global system, may explain the conjoint variation of Pb and Ni in the city entrances and in the urban highway.

**Figure 7.** Principal component analysis plot of the soil samples (PC1 and PC2).

Authors such as Imperato et al. [69], Andersson et al. [70], Meza-Montenegro et al. [71], Pant and Harrison [72] and Moaref et al. [73] stressed out that the origin of anthropogenic pollution in cities without large industrial facilities (like Lisbon) is mainly due to car traffic. The principal source of lead was, during the past, the leaded fuels. Nowadays lead is still present in soils despite its drastic reduction due to the abolition of added lead gasoline. The main sources of Cr and Ni are the vehicle braking system and tire wear. However, these two elements also have a strong relationship with parent material. Several studies have detected this trend [40]. So, for Cd and Ni there will be two source contributions, the natural one due to the lithogenic contribution and the anthropogenic one, originated from car traffic pollution.
