Annual exposure to PAHs in urban environments linked to wintertime wood-burning episodes

Polycyclic aromatic hydrocarbons (PAHs) are organic pollutants in fine particulate matter 20 (PM) long known to have mutagenic and carcinogenic effects, but much is unknown about the importance of local and remote sources tofor PAH levels observed in population-dense urban environments. A yearlong sampling campaign in Athens, Greece, where more than 150 samples were analyzed for 31 PAHs and a wide range of chemical markers were used in combination, was combined with Positive Matrix Factorization (PMF) to constrain the temporal variability, sources and carcinogenic risk associated with 25 PAHs. We findIt was found that biomass burning (BB), a source mostly effectivepresent during wintertime intense pollution events (observed for 18% of measurement days in 2017), leadled to wintertime PAH levels 7 times higher than in other seasons and was responsibleas important for annual mean PAH concentrations (31%) comparable to those fromas diesel/oil (33%) and gasoline (29%) sources. The contribution of non-local sources, although limited on an annual basis (7%), was increased 30 during summer, becoming comparable to that of local sources combined. The fraction of PAHs (12 members that were included in the PMF analysis) that was associated with BB iswas also linked to increased health risk compared to the other sources, accounting for almost half the annual PAH carcinogenic potential of PAHs (43%). This can result in a largerlarge number of excess cancer cases due to BB-related high PM levels and urges immediate action to reduce residential BB emissions in urban 35 areas facing similar issues.


PAH contribution to carcinogenic risk
The "toxic equivalence factor" (TEF) approach was used to estimate the carcinogenic potency of measured PAHs, in which the toxicity of each member is expressed using BaP as reference (Taghvaee et al., 2018): where is the concentration (ng m -3 ) and is the Toxicity Equivalent Factor of each member (Bari et al., 2010;Nisbet and LaGoy, 1992). The lifetime excess cancer risk (ECR) from inhalation was estimated as follows:  (Table S3), that have identified similar seasonal profiles, although not with such pronounced winter-summer differences. Moreover, this study reports the highest mean annual BaP concentrations at a background location in the GAA in over two decades (Marino et al., 2000). In the few studies in the area that have compared traffic with background sites, there appears to be an important roadside enhancement of PAH levels. Therefore, it is noteworthy that the present, urban background, BaP annual 180 mean concentration is comparable to the mean BaP concentrations reported at 21 sites in the GAA by a study of annual duration in 2010-2011, which however included 7 high-traffic locations (Jedynska et al., 2014). This likely indicates an increase of urban background levels in Athens, with implications for the population's exposure. BCbb maxima, indicating a predominance of biomass burning over other combustion sources (Liakakou et al., 2020). These high levels are considered to be mainly driven by emissions and not changes in mixing-layer height (Liakakou et al., 2020). suggesting that biomass burning has a more limited effect in their particle-phase concentrations, compared to higher MW members.

215
To explore the various sources that drive the variability of PAHs, the correlations of ΣPAHs with specific tracers is examined (Table S4). Given the very high correlations of ΣPAHs with BaP ( Figure S4), The results indicate a significant impact of BB emissions on PAH levels during nighttime IPEs. The ratio of ΣPAHs to levoglucosan during these events was 53% lower than in daytime, when biomass burning and therefore levoglucosan concentrations decrease. This indicates an important activity of additional

245
PAH sources such as vehicular traffic and is consistent with past studies in Greece (Saffari et al., 2013).
The levoglucosan/(mannosan+galactosan) ratio can also indicate whether solid fuels used for heating are either "aged" (e.g. aged wood or lignite) producing more levoglucosan, or more fresh (Saffari et al., 2013). The calculated value close to 6 for both daytime and nighttime samples suggests that solid fuels used for residential heating in Athens are mainly associated with fresh firewood and their type has not 250 changed since 2013, when a similar wintertime ratio was reported in Athens (Fourtziou et al., 2017).
Finally, the ratio of levoglucosan to mannosan can be indicative of wood type, with hardwood (e.g., olive, oak, beech) producing ratios around 14-15, while softwood (e.g., pine) gives lower ratios, around 4 (Schmidl et al., 2008). In our study, this ratio ranged from 7.3 tο 9.7 (Table S5) consistently used for residential heating over the recent years.

PMF modeling and source characterization
Solutions with 3-8 factors were examined, with the four-factor solution deemed as the most physically meaningful (a more detailed presentation of the selected solution can be found in Section S5 of the 260 supplement). The four identified factors are presented below; an extended description of their validation is provided in Section S5 of the Supplement.
The first factor was attributed to biomass burning (BB). Levoglucosan, a well-established BB signature marker, is almost exclusively associated with this factor ( Figure 2). The factor is also characterized by important loadings in 5-6 ring PAHs, a feature that has been reported in BB source profiles of studies in  of July-August.
The third factor was associated with emissions from diesel/oil combustion and is characterized by an increased abundance of lower MW members (Shirmohammadi et al., 2016;Zheng et al., 2017). It  Given this, and that PAHs are subject to oxidative aging and removal during atmospheric transport (Galarneau, 2008;Ravindra et al., 2008) it is likely that the non-local factor not only includes transboundary aerosols, but also partially-aged aerosol from a less extended spatial scale (e.g. from energy production using fossil fuels in continental Greece or from emissions from marine oil combustion clusters during the study period converge to the north of the GAA before arriving in the Athens basin (see also the discussion in the following section).

Source contributions
Average source contributions to PAHs and TC are presented in Figure 3 for  Regarding local sources, the annual contribution of biomass burning source ΣPAHs is amplified compared to TC (31% vs 12%, Figures 2a, 2b), ever so more when assessing only the contributions to 360 carcinogenic PAHs (36%, Figure 3c). The large impact of biomass burning on long-and short-term exposure becomes more evident considering that it essentially is a source active only in wintertime and manifests mostly during IPEs (18% of measurement days in 2017).
The other two local sources (diesel/oil and gasoline) accounted for a combined 62% of ΣPAHs ( Figure   3b), highlighting the importance of urban vehicular emissions on a long-term basisbut not during IPEs.

365
Even though the participation of diesel cars is minimal in the passenger fleet in Athens (<10%, the vast majority being Euro 5 and 6 vehicles), the contribution of the diesel/oil factor is at least comparable to the gasoline factor for ΣPAHs (and much larger for TC), indicating that it is probably emissions from elsewhere. The effect of non-local sources is greatly reduced (Figure 3b) when examining contributions to ΣPAHs (7%) against those of the three local sources (29-33%). Figure 4a shows that non-local daily 380 contributions remain lower than 20% during winter but regularly exceed 50% during summer.
Non-local contributions are intimately linked with regional transport, therefore, an air mass trajectory cluster analysis can help understand its variability and origin. Four major air mass source regions are identified for the GAA during the study period, using 96-h back-trajectories: The Black Sea area (with a frequency of 43%), Northern Greece/Balkans (32%), Western Europe (20%) and Eastern Europe (5%). When examining total PAH levels per trajectory cluster, associated synoptic circulation patterns can affect the intensity of local emission sources (e.g. during wintertime cold fronts that may lead to increased BB), so this impedes a fully unbiased assessment. The source apportionment results were utilized here 395 instead to remove the effect of local sources and associate trajectory clusters more directly with nonlocal sources. Figure 4b shows that the non-local contribution of the four clusters to ΣPAH concentrations during the non-winter period is practically the same (differences from the mean within 10%). For the winter period it is observed that the contribution from Eastern Europe is very small compared to the other clusters. to non-local sources show limited seasonal variability, meaning that increased atmospheric degradation during the warm period plays a minor role.

Contributions to carcinogenic potency and risk assessment
The  (Table S6). Figure S7 shows the members that contribute the most to BaPeq during the full measurement period; 50% of the annual BaPeq is  (Table S7) (Table S6) estimated for the annual period were equal to 0.58 × 10 -6 (OEHHA method) and 45.73

435
× 10 -6 (WHO method). These estimates approached or exceeded 10 -6 , thought to be a threshold above which carcinogenic risks become not acceptable (EPA, 2011

Conclusions
Domestic biomass burning is identified as a considerable source of carcinogenic PAHs in one of the most 445 populated regions of the Mediterranean. Overutilization of wood burning for domestic heating during the economic recession in Greece, persists even today despite the improved economy, leading to a significant increase in annual urban background levels of ΣPAHs and BaP, with respect to the period preceding the recession. The local biomass burning source, that is present almost exclusively during the winter period, emerges as the most important contributor to carcinogenic toxicity of ΣPAHs (43% on an annual basis).

450
Therefore, wintertime exposure is seen as responsible for the largest part (76%) of the estimated excess lifetime cancer risk. This large wintertime enhancement in 2017 can be mostly attributed to a few nighttime episodes (19 events in 105 days with measurements), revealing a disproportional impact of residential BB emissions but also an opportunity for targeted intervention measures. Given this, and the extended usage of biomass burning throughout Europe (e.g., France, Germany, Ireland and the UK),

455
European action and policies aimed at the regulation of biomass burning emissions are immediately needed in order to achieve considerable benefits for public health.
Sources related to local road transport were found responsible for the major part of ΣPAH concentrations where diesel penetration in the passenger fleets is much higher, could be rather useful to study the relative contributions and trends of diesel and gasoline vehicle contributions.
Non-local sources had a relatively small contribution to ΣPAHs level and toxicity but their relative 465 contribution during the warm period of the year becomes comparable with that of the local sources. It should be noted that oxidized PAHs products, which can be considerably more toxic than parent PAHs, may revise the relative importance of non-local to local sources, although this remains to be explored in future studies.
We have shown that a comprehensive observation dataset, combined with receptor modeling and back-

470
trajectory analysis provides powerful insights on the source apportionment and contributions to the health risks from PAH exposure. Despite the large body of work to date on PAHs, similar studies are surprisingly scarce in Europe and the US, so we hope our study will motivate urgently needed followups in other urban environments.

Data availability 475
Data are available upon request, by the corresponding authors

Competing interests
The authors declare that they have no conflict of interest.