Introduction

Air pollution is one of the main problems that the world population is facing nowadays, being responsible for several serious adverse health effects (Maynard 2004; IARC 2013; Hamra et al. 2014). Proof of the magnitude of this problem are the increasing restrictive limits established by specific legislation on air quality and the efforts that these regulations will impose on the reduction of atmospheric pollutant levels in the coming years (European Commission 2008). Still, the concentration that these substances can reach in the air is a subject of high concern, and for that reason, intensive monitoring plans have been carried out by public administrations and researchers (Salvador et al. 2015). Significant advances were achieved in finding out the emission sources of a wide variety of pollutants (either on the gas or the particulate phase) in different locations around the world (Rajsic et al. 2004; Moreno et al. 2006; Tan et al. 2014; Aranda et al. 2015; Rogula-Kozłowska et al. 2015; Salvador et al. 2015), including the area where the present study is developed (Gómez-Carracedo et al. 2015). Although air pollution can have a natural origin, it is well recognised that anthropogenic activities (both traffic vehicle exhaust and industrial emissions) are the main responsible.

An increasing attention has also been focused on particulate matter (PM), its composition and its health effects in the long term (Lippman et al. 2003; Greene and Morris 2006; Kong et al. 2011; Sun et al. 2014; Niu et al. 2015), especially those due to the fine and ultrafine fractions. Regarding their trends, it was reported that episodes were due to industrial emissions, whereas background levels have their origin in vehicle exhaust (Fernández-Camacho et al. 2012; González and Rodríguez 2013). Several exposure-response relationships for particulate matter based on extensive epidemiological studies have been developed (Bell et al. 2004; Samoli et al. 2005; Héroux et al. 2015). However, the high heterogeneity of climate and air pollution status among different areas complicates its site-specific application. Specially, particle composition is a key characteristic that must be addressed when monitoring PM levels. Similar inhalable ambient particle concentrations can significantly differ in composition, causing health risks of very different magnitudes (Moreno et al. 2006). Metals are by far the most analysed compounds in particulate material (Manalis et al. 2005; Moreno et al. 2006; Niu et al. 2015), although some works related with traffic assessment also included organics (Moscoso Pérez et al. 2010; Lawrence et al. 2013; Gallego-Fernández et al. 2015). Specific limit values are established by legislation for those compounds causing acute effects (SO2, ozone, particulate material etc.) (despite chronic effects at high concentrations that were also reported (Abbey et al. 1999)) and for those more prone to cause effects on human health in the long term (metals, benzene, benzo(a)pyrene) (Directive 2008/50/CE; RD 102/2011). The probability of suffering adverse chronic effects due to pollutant exposure can be estimated by risk assessment. This widely used methodology provides adimensional indexes for quantifying risk, presenting a different and opposite approach with respect to the concept of legal limits with specified units. In this sense, there is a lack of an integrative assessment of available data which includes the relevant connections between environment and health and which also considers both acute and chronic exposures to pollutants (Strosnider et al. 2014). To compare and to consider both effects simultaneously should be a key issue within the framework of air pollution control and reduction. Therefore, in the present work, the interpretation of data corresponding to airborne particles (PM10 fraction) is focused beyond a mere comparison with legislation limits. Levels, sources and implications of their composition in the long term were analysed for eight different locations in NW Spain. Besides, the air quality of the studied areas was classified under a holistic approach, including the different effects of several atmospheric pollutants. Limitations of the study and further work to be developed were also discussed.

Materials and methods

Sampling

An extensive sampling of particle-borne metals (Ni, Pb, V, Cd, Se and As) was developed in one rural (Cospeito) and seven urban (A Coruña, Ferrol, Lugo, Pontevedra, Santiago, Vigo and Ourense) locations of NW Spain between June 2008 and January 2010 (Fig. 1). Fifty-six samples were collected in each location (except in Cospeito and Ourense), 28 per seasonal period (cold and warm) and, whenever possible, the same number per day of the week. In the rural environment, 30 samples (15 per season) were collected in total, following the same criterion of the same number of samples per day of the week. Finally, in Ourense, 38 samples (19 of PM10 and 19 of PM2.5) were collected on every other day during the cold period. Cospeito (43° 14′ N, 7° 33′ W), which is located in the district of A Terra Chá (Lugo), was selected as a typical rural site, with neither traffic nor industrial emission sources in the surroundings. PM10 samples were collected using a Digitel DHA-80 high-volume sampler placed in a mobile air quality station belonging to the University Institute of Research in Environmental Studies (University of A Coruña) (Prada-Rodríguez et al. 2014).

Fig. 1
figure 1

Location of the monitoring and sampling stations

Full size image

The collection of ambient particles in the urban locations (except in Ourense) was performed by two Digitel high-volume samplers DHA-80 (Digitel Elektronik AG, Switzerland). This equipment is considered to be equivalent to the requirements of the European standard for sampling PM10 matter (EN 12341:1999). Samples were taken onto high-purity Munktell MK 360 quartz fibre filters (size, 150-mm diameter) for a period of 24 h at a flow rate of 30 m3 h−1. The sampler was loaded with 14 filters, which were changed automatically every 24 h. About 720 m3 of air was pumped through a filter from midnight to midnight (Gallego-Fernández et al. 2015). The stations sited in A Coruña and Vigo sample air under heavy traffic conditions. In fact, the monitoring station of Vigo is located 35 m away to the second busiest street of the city, with a daily vehicle intensity higher than 25,000. These two coastal cities are the biggest ones of Galicia (around 245,000 and 295,000 inhabitants, respectively), with both important industrial activity and port traffic intensity. While in A Coruña, industrial activity is more related with refinery, metal processing and combustion facilities, vehicle parts production and assembly, and shipyard activities are the main ones found in Vigo. Moderate traffic conditions influence the air sampled in the stations sited in Lugo, Santiago, Pontevedra and Ourense, while urban background conditions were monitored in Ferrol. Regarding industrial activities, inner locations (Lugo, Santiago and Ourense) present a moderate industrial activity, with Santiago and Ourense (to a lesser extent) being influenced by wood processing factories. In Pontevedra, the presence of paper mill industry is important. Finally, in Ferrol, a coastal city, mercantile port and shipyard activities could describe the profile of this location. All these equipment belong to the urban subnet of the Galician network monitoring air quality (Xunta de Galicia). In the case of Ourense, both PM10 and PM2.5 fractions were captured by a SWAM 5a Dual Channel Monitor (FAI Instruments) (Prada-Rodríguez et al. 2013).

Analyses

Analyses were carried out according to Moscoso Pérez et al. (2010). The filters used were made of quartz fibre (Munktell) and pretreated at 400 °C for 12 h. Before and after sampling, they are maintained in a conditioned room at 20 °C and 50% relative humidity as described in the standard EN 12341 (1998) for measuring gravimetric PM10. Samples were stored at −18 °C until analysis. Due to the large variety of matrix components of atmospheric particles (organic compounds, oxides and silicates), the use of acid mixtures and a great time of digestion are necessary. Filters were digested following the EN 14902 (2005), and metals were measured by ICP-MS. Samples were digested with 8 mL of concentrated nitric acid (HNO3 (c)) and 2 mL of hydrogen peroxide (H2O2) at 220 °C in closed vessels reinforced with fluorocarbon polymer. Subsequently, the acid-digested samples were allowed to cool to room temperature before being filtered through a Whatman filter no. 40 and carried to a volume of 25 mL with ultrapure deionised water. The acid extracts were frozen until analysed by ICP-MS. Recoveries of As, Pb, Cd, Ni, V and Se ranged from 90 to 100%.

Real-time values of ozone and SO2 were acquired hourly in different stations belonging to the urban subnet of the Galician network monitoring air quality (Xunta de Galicia). The analysers employed the reference methods included in RD 102/2011 of the Spanish government (ultraviolet photometry for ozone and ultraviolet fluorescence for SO2) as measurement techniques. Data were validated and obtained from the regional service of air quality (Xunta de Galicia 2015). Values of ozone and SO2 corresponded to the same sampled days for particle-borne metals in each location.

Data processing

Statistical analysis of data was carried out using the IBM SPSS Statistics 20.0 software. Two multivariate analysis techniques, principal component analysis (PCA) and cluster analysis (CA) were applied to find associations (common emission sources) among the different variables. PCA was developed applying varimax rotation with Kaiser normalisation, while CA was performed using Ward’s method for agglomeration and squared Euclidean distance for measuring the distance among clusters.

A screening-level human health risk assessment was developed as a first approximation to quantify the probability of adverse effects due to particle-borne metals in NW Spain. The EPA guidelines (US EPA 1992 and US EPA, US Environmental Protection Agency, 2009) were employed to estimate carcinogenic and non-carcinogenic risks due to ingestion and inhalation exposure pathways. Parameters and equations to estimate hazard quotients (HQs) and cancer risk (CR) for each metal can be consulted in the SI. In multiple chemical risk evaluations, the concept of hazard index (HI) can be applied to similarly acting pollutants (US EPA 1986 and 1989). Therefore, a cumulative approach is justified in the present risk evaluation. Rather than assuming a probabilistic distribution for each exposure parameter in order to assess uncertainty and variability, discrete worst-case values were selected (US EPA 2011). This approach was based on the difficulty of associating detailed activity patterns to general population. An epidemiological study recording relevant values for exposure times and frequencies for receptors living in the area of study would be necessary to develop a second-tier assessment. For the same reason, different age groups were not considered in this evaluation. Therefore, and providing a different perspective, the range (minimum–maximum) and the mean values of risk estimates depended only on the metal concentration variation during the sampling period.

Finally, multicriteria analysis based on the analytical hierarchy process (AHP) methodology (Saaty 2008) was applied to evaluate the relative air quality among the different areas. Air quality was defined by two decision criteria, chronic and acute effects, which were represented by five different indicators. HI and CR estimated for particulate-borne metals (both adimensional) accounted for chronic exposure, while PM10 (μg m−3), ozone (μg m−3) and SO2 (μg m−3) daily levels were selected to address acute exposure. Average annual values of these indicators were normalised to the same unifying decision scale, which was set at a length of 8. Data were uniformly distributed as follows: index 1 to 7 included values from the minimum to the safety threshold for each compound. Index 8 included those values equal or higher than these limit values established by legislation (Table S1). Apart from the indexed data, value sets were also needed. Value sets accounted for the preferences given to the different indicators and are obtained by making pairwise comparisons among them. With this information, a combined air quality indicator was calculated by summing the product of the indexed value and its preference within each subcategory of the hierarchy. Higher values of this indicator mean worse air quality. Results considering different value sets were evaluated to identify which indicators were the most influential.

Results and discussion

Levels and sources of particle-borne metals

A summary of descriptive statistics of metal concentrations in ambient particles for the studied locations is shown in Table 1. Mean values for all pollutants were below both objective levels and lower limits defined by current legislation relative to human health protection due to exposure to atmospheric particle-borne metals (Directive 2008/50/CE). Therefore, at first glance, it could be stated that the air quality of the area did not present any serious pollution problem. In fact, average values in both urban and rural areas were comparable to other locations (Querol et al. 2007). On the other hand, the presence of metals in the airborne particles of rural locations suggests medium-range transport from traffic or industrial sources.

Table 1 Mean, minimum and maximum (between the parentheses) values in ng m−3 of metal concentrations in ambient particles (PM10) for the different locations
Full size table

Possible emission sources of particle-borne metals were investigated according to the location by PCA. The results of the rotated component matrix can be seen in Table 2. Two components with eigenvalues higher than 1 were extracted in each location. Se was excluded from this analysis, since values below the detection limit were obtained in the majority of cases. Se values <LOD were set at half of this limit for calculating values present in Tables 1 and 3. The most populated locations (A Coruña and Vigo), both with important ports, presented the same pattern of metal aggregation: a first component representing Ni, V and As and a second one associating Cd and Pb. The contribution to variance, although slightly higher for component 1, was around 40% for both components and in both locations. Lugo and Ourense presented a similar trend, except for As, which was more strongly associated to Cd and Pb in a first component explaining 42–46% of variance, while Ni and V were clearly associated in component 2. Ni was isolated in Ferrol (21.6%) and Santiago (30.7%), although in this, the latter was partially combined with V. The opposite behaviour was found in Pontevedra, where Ni presented a partial contribution between the two components, while V was only found in component 2. In this location, the first component could be associated to the emission of a close industrial facility. Finally, Ni was not included in the analysis in the rural location (Cospeito) since all the obtained values were below the detection limit. Globally, it can be observed that Cd and Pb were always associated in one component, while Ni and V were usually associated in another different component.

Table 2 Total variance explained and matrix of principal component analysis
Full size table
Table 3 Hazard quotients for considered metals according to the location
Full size table

Ni and V usually have a common source in the combustion of heavy or residual fuel oil, like those produced in power stations or oil refinery facilities (Pakkanen et al. 2001; Jang et al. 2007). Thus, Ni and V could also be usually associated to ship traffic emissions (Zhao et al. 2013; Querol et al. 2007; Healy et al. 2009). These metals are preferentially found in the fine fraction PM <2.5 of particles (Jang et al. 2007; Zhao et al. 2013) and can have a common origin in this fraction with As (oil combustion) (Negral et al. 2009), as it happens in A Coruña and Vigo. Although Ni and V can be partially associated with these sources in coastal locations with high ship traffic and presence of refinery facilities, the contribution of other industrial point sources and vehicular traffic was also present. This occurred not only in A Coruña and Vigo but also and especially in less populated inner locations (Lugo and Ourense). In fact, V/Ni ratio was always below 2 in Lugo and Ourense, whereas the ratio associated with oil combustion is higher than this value (Murphy et al. 2009; Moreno et al. 2010). Ni isolated in component 2 in Ferrol can be associated with the presence of an important shipyard in this location (coarse particles) (Negral et al. 2009). In Santiago, Ni concentration in PM10 increases with NE wind direction (MeteoGalicia 2015). Considering NE as a reference, 3 km upwind is located an important wood processing facility, which is classified within the first positions in the ranking of Ni emissions in NW Spain (Spanish Government 2015). On the other hand, it seems that Cd and Pb have a common emission source in all the monitored areas. These metals have been related together with automobile emissions, brake wear use and fuel combustion (Moreno et al. 2006; Sha and Shaheen 2010; Song and Gao 2011). However, the association of both metals is still not clear (Canepari et al. 2008). More information on metal distribution within particle fractions and a more in-depth analysis along with meteorological parameters would be necessary to support these conclusions.

CA was developed for those locations where high concentrations of one or more metals were observed (Santiago and Pontevedra). CA was applied to cases (samples) instead of variables (metals), in order to identify episodic events and to complement the results obtained by PCA. Resulting dendrograms for the two locations are available in the SI (Figs. S1 and S2). Samples were distributed among three main groups in Santiago. Group 1 included a 7% of the samples which presented significantly high Ni values (24–40 ng m−3) and intermediate values of the other metals (0.08 ng m−3 of Cd, 10.35 ng m−3 of Pb, 7.21 ng m−3 of V and 0.14 ng m−3 of As). Groups 2 (15%) and 3 (76%) included high- and low-moderate concentrations of all metals, respectively. Finally, an outlier sample with an extremely high Pb concentration (42.18 ng m−3) and intermediate values of the remaining metals was identified, being indicative of a discrete point pollution event. In Pontevedra, samples in group 1 (5%) presented the highest As concentrations (2.5–7.2 ng m−3), high concentrations of Pb (10.4 ng m−3) and Cd (0.19 ng m−3) and moderate levels of Ni (6.91 ng m−3). Group 2 (18%) included high concentrations of Cd, moderate/high concentrations of Pb and moderate/high concentrations of As. V was high (although moderate when compared to other cities) in the samples aggregated in group 3 (21%), while group 4 included those cases with moderate to low values of all metals (55%).

Interpretation of risk indexes and uncertainty limitations

The estimated risk indexes for each metal (HQ) and the total hazard index in the locations studied can be seen in Table 3. Values shown corresponded to average, minimum and maximum values obtained from each sample rather than providing risk estimates based on the 95th percentile concentrations. This approach allows to extract more information on a yearly basis, and it was selected due to the number of samples obtained in each location. Average values of the HI were in the 1E−01 order of magnitude, with the lowest value obtained in the rural location Cospeito (1.96E−01), while Santiago and Ourense presented the highest values (5.80E−01 and 5.20E−01, respectively). Regarding maximum HI, Santiago (2.52E+00), followed by Ferrol and Vigo, showed concerning values of this parameter. Nonetheless, all the remaining locations (except the rural one) also presented maximum HI higher than 1. As in the present work, HI exceeding the safety limit when assessing the risk due to airborne metals is commonly reported (Díaz and Dominguez 2009; Hu et al. 2012; Sun et al. 2014; Niu et al. 2015). The particle-borne metal which in general mostly contributed to the HI in the different locations was Ni, followed by As, V and Pb. Cd and Se contributed in a very low percentage to non-carcinogenic total risk (around 3% as the maximum). Ni contributions higher than 60% were obtained in Ferrol, Lugo and Santiago, while for the other locations, it ranged between 35 and 52%. No significant differences between risk indexes obtained for PM10 and PM2.5 in Ourense were observed. This result is expected due to the similar metal concentration found in both fractions (Table 1). Although PM10 and higher fractions (TSP) have been employed in risk evaluations (Shi et al. 2011; Hu et al. 2012), it is well recognised that fine particles (PM2.5) can more easily penetrate the lungs until reaching the alveolar region and be more efficiently absorbed by the human body (Schaumann et al. 2004; Kreyling et al. 2006; Avino et al. 2016). Besides, some authors found that anthropogenic pollutants are more present in the ultrafine-nanofraction of particles (Lin et al. 2005), while others found crustal and anthropogenic metals in both PM10 and PM2.5 fractions (Negral et al. 2008). In the scientific literature, there are available from empirical to pharmacokinetic models of different complexities to estimate lung deposition rates as a function of particle size (Isaacs et al. 2012). Recent particle-borne metal health risk assessments considering simplified deposition rate models showed that decreased daily intakes are estimated, indicating in some cases a significant percentage of reduction (Luo et al. 2014; Niu et al. 2015). According to the available data, a worst-case approach considering the inhalable and further deposited fraction equal to 1 was assumed. In this case study, neither size distribution within a particle fraction nor particle fractions were monitored. These data are key factors for the estimation of reliable deposition efficiencies in order to develop a more in-depth risk assessment, which was not the main objective of this work. An additional assumption leading to overestimation has to be mentioned. The bioavailable fraction of metals in the particulate matter depends on many factors: the metal itself, the particle size, the source and the core composition of the particle, and even it depends on the extraction method employed (Canepari et al. 2010; Sun et al. 2014). Regarding metal bioavailability of different particle matter fractions, similar values have been reported in TSP and PM2.5 (Hu et al. 2012). Thus, an important source of uncertainty is associated with this parameter. A conservative average particle bioavailability factor of 0.7 was employed as an attempt to roughly encompass differences in particle composition. Estimated bioavailability may range from 0.2 in metals like Cr and Ni to 0.8 in Pb, Cd and As (Canepari et al. 2010; Hu et al. 2011 and 2012; Sun et al. 2014). Taking into account the high contribution of Ni to the HI, the value selected for this factor could also favour the overestimation of the HI. However, the monitoring campaign carried out in Ourense indicated that the content of PM2.5 within the PM10 fraction ranged between 61 and 94%, with an average value of 73%, this indicating an important contribution of fine particles. On the other hand, there is not an agreement in toxicological data for assessing chronic exposure to Pb, since some of the health effects associated with lead exposure occur at blood lead levels as low as to be essentially without a threshold (US EPA 2004). Although the US EPA has not established neither a reference concentration (RfC) to assess inhalation exposure nor a reference dose (RfD) for ingestion exposure (ATSDR 2007), HQ of Pb by the ingestion pathway was estimated using a value proposed by RIVM (Baars et al. 2001). The influence of the exposure pathway (ingestion and inhalation) varied among locations. In general, the contribution of inhalation to HI ranged between 60–70% in all locations, except in Pontevedra (54%) and Cospeito (49.5%), where a higher influence of As via ingestion was observed.

Average values of CR (Table 4) due to Ni, Cd and As exceed the limit value of 1.00E−06 in Ourense, whereas in the other locations was within the order of 1.00E−07. Maximum values were higher than the acceptable threshold, except in the rural location. As, followed by Ni, were the most contributing metals to CR, in variable proportions depending on the location. Cd presented a low influence on CR, reaching a maximum value of 10% in A Coruña. Ni showed the highest influence in Santiago (54%) and the lowest in Pontevedra (14%). As showed significant contributions in all the studied locations, reaching values higher than 80% in Pontevedra. Pb was not included in the CR assessment due to uncertainties in the inhalation unit risk (IUR), although it is classified as probably carcinogenic to humans (IARC 2004).

Table 4 Cancer risk (CR) for Ni, Cd and As and incremental lifetime cancer risk (ILCR) depending on the location
Full size table

Despite the approach and limitations of the estimated results, it has to be mentioned that only six particle-borne metals have been considered to evaluate total risk. Taking into account the existing influence of anthropogenic sources in the studied area, assuming that other potentially hazardous metals like Co, Cr, Zn, Mn or Hg might be present in airborne particles, is logical. These metals would contribute to increase the HI and/or the incremental lifetime cancer risk (ILCR), not to mention other known toxic compounds like PAHs. Therefore, further monitoring would be necessary to elucidate the magnitude of total risk.

Integrating exposure assessment criteria

Four different classification value sets were considered to perform an integrated exposure assessment: equal, long term, short term and worst case. The equal value set weights all criteria equally, although those indicators included within chronic criteria have a greater impact on the results because their indexed values were higher. Long and short-term data sets give preference to chronic and acute effect criteria, respectively. The worst-case data set was defined giving more importance to PM10 among acute criteria and CR within the chronic category. Detailed preferences and weights of the different criteria within each classification data set can be seen in Table S3 of the SI. On the other hand, air quality ranking results according to the four classification sets for the considered locations are shown in Fig. 2.

Fig. 2
figure 2

Relative air quality ranking of the considered locations depending on the value set

Full size image

Regardless the value set selected, Ourense and Vigo were always the two cities with worse air quality. However, a significant difference in the magnitude of the air quality index can be observed depending on the classification set. Thus, the value of this index in the worst-case set was around 1.5 times higher than in the acute set, since Ourense and Vigo presented high values of both CR and PM10 indicators and moderate values in the remaining ones. The difference between these two cities varied from 0.4% for the acute set to 10.7% in the long-term scenario. Whereas chronic criteria have more influence in the air quality of Ourense, acute indicators have a higher impact in the index of Vigo. Pontevedra occupied different positions depending on the classification set: from the third position when the worst-case value set was used to the fifth when the short term was selected. The different rankings of this location depend on the weight given to the CR indicator, which had a very large value for this city, due to the concentrations of particle-borne As. On the other side, Ferrol had a worse relative condition if the acute value set was considered, due to the impact of ozone. Although the total index of Ferrol barely varied (from 3.66 when applying the short-term data set to 4.46 in the worst case), its relative condition with respect to Ourense changed from a 9 to a 46% better (Fig. 2). Santiago, which had moderate to strong values for all the indicators, was ranked in the third position when both equal and chronic value sets were considered and fourth in the other cases.

A Coruña and Lugo were always classified in the latest positions, indicating a good air quality in comparison with the other urban areas. A Coruña had a higher PM10 concentration value, while chronic criteria were slightly higher in Lugo. Lugo always presented an air quality index around a 50% better than Ourense, while in A Coruña, air quality was a 26–55% better. This city achieved the best value under the long-term scenario when compared with the other locations, which means lower metal concentrations of the most toxic metals. Finally, the rural area (Cospeito) presented an air quality index of approximately 50% better in relation with the best urban location. If maximum instead of average metal concentration values were considered, the global air quality index would be higher (especially for the long-term and worst-case sets) and the ranking of the considered locations might change.

Concluding remarks

Data of different air pollutants, both in the gas (ozone and SO2) and in the particulate phase (Ni, Pb, V, Cd, Se and As), were employed to provide a comprehensive assessment of air quality in different locations of NW Spain. Special attention was focused in the estimation of health effects in the long term due to particle exposure, considering particle composition as the main influencing factor. Results showed that average HI were lower than the safety limit in all locations, although maximum values were higher than 1 (except in the rural location). Regarding CR, only one inner location (Ourense) presented average values exceeding the recommended threshold. A preliminary analysis of the composition between particle fractions (PM10 and PM2.5) showed that, in general, there are no significant differences. However, a more detailed study covering all the locations would be necessary to obtain more accurate conclusions. Ni and As were the metals most contributing to the risk estimates, being this influence due to their concentration levels and toxicity. However, the contribution of the remaining metals should not be neglected in terms of added toxicity to particles. In that sense, source apportionment indicated the influence of industrial emissions. Especially, the area of Santiago and Pontevedra were affected by pollution episodes that, although short, increased the average concentration values and, consequently, the magnitude of risk indexes. Investigating the frequency of these events would be a further step to perform a higher-tier risk assessment. On the other hand, it has to be mentioned that the shorter sampling period in Ourense might influence the results, since it is possible that average values in the present sampling were higher than those obtained if a longer period of time including more background samples was considered. Finally, the integrative assessment of the different effects caused by several types of pollutants provided a global vision of air quality in the studied locations. In general, the values obtained in the context of the defined indexing (e.g. 8 as maximum) indicated a good to moderate air quality in all areas.