Can Abies alba Needles Be Used as Bio-passive Samplers to Assess Air Quality?

Received: April 22, 2021
Revised: August 5, 2021
Accepted: September 7, 2021

 Copyright The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited.

Cite this article:

Stojanowska, A., Górka, M., Lewandowska, A.U., Wiśniewska, K., Modelska, M., Widory,.D. (2021). Can Abies alba Needles Be Used as Bio-passive Samplers to Assess Air Quality? Aerosol Air Qual. Res. 21, 210097. https://doi.org/10.4209/aaqr.210097

HIGHLIGHTS

• Tree needles are low cost passive samplers and easy to implement.
• Abies alba needles allow identifying sources of aerosol contamination.
• Anthropogenic activities generate ¹³C-enriched aerosols.

Keywords: Bioindicators, Abies alba needles, δ13C, Aerosols

1 INTRODUCTION

Atmospheric particulate matter (PM) is one of the major air pollutants in Europe (EEA, 2019) and carbonaceous aerosols represent one of the major PM components. This carbonaceous fraction is usually subdivided into organic carbon (OC), the light-scattering fraction, and elemental carbon (EC), the light-absorbing fraction (Galindo et al., 2019; Xu et al., 2015). Carbonate carbon (CC) bond to carbonaceous particles is less frequent (Ozdemir et al., 2014). Organic carbon in atmospheric PM can originate from anthropogenic and natural sources and can be divided into two types: primary OC, emitted directly into the air and secondary OC, resulting from the secondary oxidation of VOCs by i.e., O₃, NO₂, OH (Murillo et al., 2013; Kucbel et al., 2016). Primary OC derives from natural biogenic sources (i.e., pollen, bacteria, viruses, several types of fungal spores, plants and fragments of insects) (Schwarz et al., 2008). Secondary extractable OC, which includes VOCs from natural sources and anthropogenic compounds, should also be mentioned (Schauer et al., 2007). Elemental carbon is a primary pollutant created directly from the incomplete combustion of fossil fuels or biomass (Galindo et al., 2019). EC is characterized by a porous structure (Murillo et al., 2013), which facilitates its adsorption of toxic compounds (Lucking et al., 2011; Mills et al., 2011). EC can be transported over thousands kilometers from its source, increasing its environmental impact (Husain et al., 2007). Considering that, carbonaceous aerosols are of much interest when studying air quality. Another issue is analysis of carbon isotopes which can contribute to discrimination of the different sources of aerosols and has now been widely documented (e.g., Aguilera and Whigham, 2018; Górka et al., 2012; Lim et al., 2020; Morera-Gómez et al., 2018; Widory et al., 2004). These studies are based on the analysis of the different carbon isotope systematics: stable carbon isotopes ¹²C, ¹³C (e.g., Aguilera and Whigham, 2018; Morera-Gómez et al., 2020) and/or radiocarbon ¹⁴C (Lim et al., 2020). Among the different sources, C₃ plants (in which the first product of the CO₂ assimilation during photosynthesis is a 3-phosphoglycerate – compound) are characterized by δ¹³C from –34‰ to –24‰ (with an average of –27‰ (e.g., Šturm et al., 2012; Zhang et al., 2019). The major anthropogenic emission sources generate particles with the following typical δ¹³C isotope compositions: (i) coal, between –24.2‰ (poor quality coal) and –25.7‰ (high quality coal), while the corresponding soot has a δ¹³C of –24.5‰ (Górka and Jȩdrysek, 2008); (ii) diesel particles, from –26.5‰ (Widory et al., 2004) to –28.3‰ (Górka and Jȩdrysek 2008); (iii) unleaded fuel particles, from –24.2‰ (Widory et al., 2004) to –26.8‰ (Górka and Jȩdrysek, 2008).

It has now clearly been established that PM affects climate, the ecosystems and human health, especially in populated areas where it is present in significant concentrations (Atkinson et al., 2015). Usually, such areas are under continuous control of the influence of anthropogenic sources. However, the rural sites can be also under serious risk in terms of air pollution but, unlike urban or industrial areas, they are usually not monitored. European Union requires that its member countries monitor their PM₁₀ and PM_2.5 emissions as well as their ambient air concentrations (EEA et al., 2018). In order to reach these objectives, active methods, which are based on a dynamic sampling and that require the use of a pumping device by which air passes through either an air sample container or a filter, are generally recommended. These methods are relatively expensive and require specific instrumentation. Hence, in some regions (e.g., mountain or woody) the implementation of this active sampling is often impossible due to the lack of access to electricity, altitude or limited access. The difficulty can be alleviated by the use of living organisms or their products (bioindicators) that are present in the targeted study areas. The use of bioindicators to monitor/assess the quality of different environments offers other advantages such as the possibility of bio-concentrating both organic and inorganic contaminants and of accessing time-integrated information (Conti and Cecchetti, 2001). Similarly to aerosol samples, bioindicators will yield information about the origin of the atmospheric particulate matter, from which the corresponding anthropogenic impact can be evaluated for a given area and time period. Bioindicators have already been proven to be excellent environmental proxies, from which pertinent information can be obtained by studying their carbon isotope compositions (e.g., Brendel et al., 2003; Ciężka et al., 2016). There are many kinds of different bioindicators, for example: lichens (e.g., Kłos et al., 2007; Ciężka et al., 2018; Kłos et al., 2018), mosses (e.g., Kłos et al., 2018; Maxhuni et al., 2016; Shetekauri et al., 2018), spider webs (Bartz et al., 2021; Górka et al., 2018; Stojanowska et al., 2020; Stojanowska et al., 2021). Also tree leaves and needles are considered useful bioindicator (Górka et al., 2020; Lehndorff and Schwark, 2004; Wang et al., 2015; Wuytack et al., 2010) as they have the capacity to accumulate particulate matter on their surface, that is further trapped by being covered with epicuticular wax (Zeisler-Diehl et al., 2018). Tree needles are recording information about their surrounding air quality, such as potentially toxic elements or, indirectly, the carbon isotope compositions of depositing aerosols (e.g., Sensuła et al., 2018; Vladimirovna Afanasyeva and Ayushievna Ayushina, 2019). The fact that the vegetation is highly spatially and temporally distributed makes it easily accessible and inexpensive in biomonitoring approaches (Sawidis et al., 2011). Conifers are generally considered good bio-passive samplers as many of them are evergreen trees, i.e., they do not lose their needles in winter (Mauri et al., 2016). Consequently, monitoring air quality using their needles can be carried out on a full year scale. The needles’ adsorption potential is enhanced by the morphology of the conifers canopies, as they are dense and fine-textured with a high leaf area index (Yang et al., 2015). To our knowledge, no study has ever been conducted on carbonaceous aerosols collected on needles, and in particular focusing on their carbon geochemistry and isotope compositions. With that in mind, our objective was to evaluate the possibility of using Abies alba needles as alternate bioindicators to study air quality, using an approach that coupled the δ¹³C from the needles and the aerosols adsorbed onto their surface with the total carbon (TC), elemental carbon (EC) and organic carbon (OC) contents of the aerosols.

 
2 MATERIALS AND METHODS

 
2.1 Study Area

Samples were collected in the Holy Cross National Park, located in the Holy Cross Voivodship, central Poland (Fig. 1). Several major emission sources are located in the vicinity of the study area (Fig. 1): power stations in Kielce, Połaniec, Jaworzno and Bełchatów; a steel plant in Ostrowiec; sulphur chemical plants in Grzybów and Tarnobrzeg; and a zinc smelter in Miasteczko Śląskie. Road traffic is another major local source of aerosols with an average of 8,500 cars/day driving the national roads of the Holy Cross Voivodship (GDDKiA, 2015). Heating in the Voivodship’s single and multi-family buildings is mostly based on the combustion of poor-quality coal. The air quality in this region is also influenced by transboundary pollutant transport, mostly from the Lesser Poland and Silesian Voivodships, with carbon monoxide, nitrogen oxides and non-methane volatile organic compounds (NMVOC) representing the main transported compounds (Environmental protection program for the Holy Cross Province. Available at: https://bip.sejmik.kielce.pl/142-departament-rozwoju-obszarow-wiejskich-i-srodowiska/4769-program-ochrony-srodowiska-dla-wojewodztwa-swietokrzyskiego-na-lata-2015-2020-z-uwzglednieniem-perspektywy-do-roku-2025.html; accessed September 16, 2019).

Fig. 1. Sampling locations in the Holy Cross National Park. High elevation (> 500 m a.s.l.) locations as well as those located next to roads are indicated. Sampling location number 8 is expected to represent the local “natural” background (center of the National Park, with no roads nearby). Adapted from Ciężka et al. (2018).

2.2 Materials

Abies alba (silver fir) was selected as it represents the most abundant tree in the Holy Cross National Park and its needles are easily accessible. All needle samples were collected in summer 2013 at 20 different locations: 18 in the National Park and 2 in its close vicinity (Fig. 1). At each sampling location, branches from a few trees (typically 5) were collected at a height about 1.5–2 meters above the ground and pooled to both homogenize the needle samples and obtain a sufficient amount of material for the subsequent analysis (generally 15 g). The needles were packed in clean paper bags for transport. To avoid any potential canopy effect between the sampled trees, they were chosen within similar populations. In the laboratory, needles were divided by age classes (6-months old, referred as 2013; 1-year-old, referred as 2012; and older than 1 year, referred as < 2012), weighed and stored in paper bags: 3 samples were thus obtained for each sampling location, for a total of 60 samples. The age division was made taking into consideration the needle growth on branches and the noticeable separation of their individual annual increments. This also allowed to obtain samples with different aerosol collection times that later will permit testing the needle potential and efficiency for collecting aerosols. This protocol for separating needles was adapted from Brendel et al. (2003) and Muukkonen and Lehtonen (2004).

 
2.3 Methods

2.3.1 Samples preparation

Aerosols were washed out of the needles’ surface using distilled water as previous studies demonstrated that this washing agent gives satisfying results (e.g., Al-Alawi and Mandiwana, 2007; Ataabadi et al., 2011; Leśniewicz and Zyrnicki, 2000). In brief, ~15 g of needles were placed into a beaker. 300 mL of distilled water were added and the beaker was gently mixed with a glass rod for 5 minutes. This five-minute mixing was determined by trial and error: longer than 5’ caused a large number of needles to crumble, while a shorter time resulted in a low aerosol yield. The solution was then filtered through a QM-A quartz fiber filter (diameter of 47 mm, porosity 2.2 µm, Whatman) under vacuum. Quartz fiber filters were not baked before collection according to the fact that previous studies from our research group demonstrated that blank filters yielded carbon contents under the CM-CRDS detection limit. After filtration filters were dried in an oven at 45 ± 2°C and stored in a desiccator until further treatment and analysis. The filters were double weighed before and after the filtration using a Vibra HTR-220CE (±0.1 mg, at 23 ± 2°C and relative humidity of 40 ± 5%), yielding the mass of particulate matter adsorbed on the needles. The final PM quantity was expressed as milligrams of aerosols per gram of dry needles.

2.3.2 Carbon stable isotope analysis (δ¹³C) of aerosol and needle samples

δ¹³C isotope compositions were determined by Cavity Ring-Down Spectroscopy (CRDS) coupled to a combustion module. For filter samples, a punch was cut using an IVA Analysentechnic puncher. Needle samples were homogenized by grounding them in a clean fine grit mill. Both the size of the punch and the mass of grounded needles were adjusted during each run to ultimately obtain ~2000–4000 ppm of CO₂ generated by the combustion process. For the carbon stable isotope analysis, the filter punch or the needle powder was placed into a tin capsule that was sealed and analyzed using a Picarro G2201-i CRDS analyzer coupled with a CM (combustion module). Carbon isotope compositions were expressed as δ¹³C values, which represent the relative difference expressed in per mil (‰) between the isotope ratio of the sample and that of the Pee Dee Belemnite (PDB) standard:

δ¹³C values were calibrated using certified standards NBS-19 (δ¹³C = +1.95‰), NBS-18 (δ¹³C = –5.014‰), IAEA CO-8 (δ¹³C = –5.764‰), USGS-24 (δ¹³C = –16.049‰) and USGS-40 (δ¹³C = –26.389‰). Analyses were made on 2 to 4 replicates and the associated analytical error on the measured δ¹³C was < 0.2‰.

 
2.3.3 Total carbon, organic carbon and elemental carbon quantification in aerosols

Total carbon (TC), elemental carbon (EC) and organic carbon (OC) were determined using a Sunset Laboratory Dual-Optical Carbonaceous Analyzer at the University of Gdansk (Division of Marine Chemistry and Environmental Protection, Institute of Oceanography). We followed the EUSAAR2 protocol, owing to the optimal maximum temperature obtained at the end of the first stage, which corresponded to 650°C, and ensured that only 2.5 ± 2.4% of the elemental carbon was combusted during the first stage of analysis (Cavalli et al., 2010). A 1.5 cm² rectangular piece of the quartz fiber filter was placed into a quartz oven, where it was analyzed for both OC and EC. The method's limit of detection (MLD) was 0.3 µg cm^–2 (n = 72) for both OC and EC, while the analytical error was < 6% for EC and < 10% for OC (with a 99% confidence interval). All carbon results were corrected for blanks: 3.0 µg cm^–2 for OC while no correction was necessary for EC as blanks were under the MLD. Apart from automatic calibration (internal standard: 5.0% methane in equilibrium with analytically pure He), which takes place at the end of the second stage of the analysis, an external standard was analyzed (99% analytically pure sugar solution) after every 10–15 samples. The average analytical error of the external calibration was 4.5%. Additionally, an inter-laboratory comparison using an Elemental Analyzer Instruments NC 2500 NC at the Université du Québec a Montréal (UQAM) was performed. The agreement between the two methods was confirmed by a high Pearson correlation coefficient value for total carbon (r > 0.9). More details can be found in our previous publications (Lewandowska et al., 2018; Wiśniewska et al., 2019).

 
2.3.4 Isotope mass balance calculation

We performed an Isotope Mass Balance (IMB) to estimate the respective contributions of the following 2 endmembers on the final δ¹³C_aerosols we measured: contamination (i.e., external) and biogenic (i.e., needle). Contributions were calculated using the following equation:

where δ¹³C_aerosols, δ¹³C_needles and δ¹³C_source are the δ¹³C of the aerosols, needles and of the corresponding aerosol emission source. “a” is the percentage contribution of the contaminant source and “b” the one of particles resulting from the eventual inclusion of needle debris during the sample preparation, when needles are washed with distilled water.

 
2.3.5 Statistical analysis

We used the Kruskal-Wallis test to highlight differences between dust concentrations with respect to the different needle age classes. To discriminate groups, multiple pairwise-comparisons using the Dunn's test were applied. These median tests are the non-parametric equivalent of ANOVA and allow to evaluate differences between more than 2 variables. Test were performed on the 60 samples using the computing environment R (R Core Team, 2013; additional software packages tidyverse, ggpubr, rstatix and ggplot2 were used). Multivariate Principal Component Analysis (PCA) was performed on 5 standardized parameters measured in dust (for 2012 needles): dust concentration, TC, EC, OC and δ¹³C. The objective of the PCA (Statistica 13 software) was to identify the interrelated factors characterizing the processes controlling the dust carbon chemical and isotope compositions. Transformation factor loads matrix (varimax rotation; Table 3) and factor scores values (Fig. 9) were calculated (Cattell, 1966; Drever, 1997; Johnson, 1978). As it was demonstrated by Manly (1998) PCA requires a limited number of useful variables and it helps to describe relations amongst them. The PCA results were considered reliable due to the high number of samples (58) analysed, the observed normal distribution (for δ¹³C) or close to normal or lognormal for the other data, and the correct conditioning of the matrix as well as eigenvalues close to 1 for the parameters. The Surfer 10.0 software was used to create the isoline maps (Figs. 3, 5, and 7), applying an ordinary kriging method.

 
2.3.6 Scanning electron microscope (SEM) analysis

SEM analysis, on a randomly-selected subset of samples, was conducted in order to investigate i) if the our protocol (i.e., rinsing with distilled water) was efficient at washing out the aerosols deposited at the surface of the needle samples and ii) to verify that the surface of the needles had not been damaged by the subsequent stirring, which may have impacted the δ¹³C analysis. Fig. 2 reports the SEM images of the adaxial surface of the several-years-old samples (< 2012), before (Fig. 2(A)) and after washing (Fig. 2(B)). The comparison between the two images confirms that aerosols were successfully washed out without damaging the surface of the needles although we acknowledge that some small quantities of needle fragments may still have been collected concomitantly with the aerosols.

Fig. 2. SEM images (A) before and (B) after washing out with distilled water. The needle sample shown here corresponds to the age class < 2012.

 
3 RESULTS

 
3.1 Aerosol Concentrations and TC, OC and EC Values

Concentrations of aerosols adsorbed on the needle samples are reported in Table 1. The highest concentrations were on average 4.6 ± 2.4 mg g^–1 and corresponded to the oldest samples (< 2012; > 1 year). In contrast, the lowest concentrations (average of 1.7 ± 0.7 mg g^–1) corresponded to samples collected in 2013, the shortest exposure time of our study (< 1 year). The aerosol concentrations measured for the year 2012 ranged from 1.8 to 5.5 mg g^–1, with an intermediate average value of 3.2 ± 1.2 mg g^–1.

In the case of TC, EC, and OC the results also varied depending on the age of the samples. The average TC results were as follows: 1.48 ± 0.64 mg g^–1, 1.28 ± 0.53 mg g^–1, 0.72 ± 0.18 mg g^–1 for < 2012, 2012, and 2013 respectively. The respective OC averages of 1.43 ± 0.62 mg g^–1 for samples older than 2012, 1.23 ± 0.54 mg g^–1 for 2012 and 0.70 ± 0.17 mg g^–1 for 2013 were noted (Table 1). The concentrations of EC were very low, revealing 0.05 ± 0.04, 0.06 ± 0.02 and 0.03 ± 0.01 mg g^–1 for < 2012, 2012, and 2013 respectively.

 
3.2 Aerosol and Needle Carbon Isotope Compositions

The samples marked as older than 2012 were characterized by δ¹³C_aerosols equal to –29.5 ± 1‰ and δ¹³C_needle equal to –32.1 ± 1.3‰. On the other hand, one-year old samples (2012) revealed averages of δ¹³C_aerosols and δ¹³C_needle equal to –29.3 ± 1.1‰ and –32.1 ± 1.3‰, respectively. The samples with the shortest exposure time yielded δ¹³C_aerosols equal to –29.7 ± 1‰ and δ¹³C_needle equal to –31.9 ± 1.3‰ (Table 1).

 
4 DISCUSSION

 
4.1 Concentrations of Adsorbed Aerosols

It seems that results of concentrations of adsorbed aerosols are correlated to their age, i.e., exposure time to ambient air, with the shorter exposure times yielding lower PM concentrations (Table 1). The increasing trend observed for the aerosol concentrations on the surface of the needles confirm that they can be used as passive samplers for characterizing air quality. Previous studies have shown that plants in general play a significant role in removing atmospheric particulate matter (e.g., Nowak et al., 2013) due to their capacity to accumulate PM, which depends on the properties of the leaf surfaces (Wang et al., 2015). Leaf and needle surfaces consist of cuticles and an epicuticular wax, the outer surface of the cuticle (Buschhaus and Jetter, 2011; Zeisler-Diehl et al., 2018) that protect plants from water loss, UV light and abiotic stress factors (Zeisler-Diehl et al., 2018). The structure of the needle surface facilitates the adsorption of aerosols and once deposited they are not easily removed by wind or rainfall. Moreover, aerosols, deposited on the needle surface can block the suprastomatal chambers (Teper, 2009). This may lead to i) a decrease in the photosynthesis rate by reducing sunlight availability, ii) an increase in the leaf temperature triggered by changes in the surface optical properties, and iii) a modification of gas diffusion into and out of the leaves (Prajapati and Tripathi, 2008; Squires, 2016).

As a result, the basic functions of the plant are disrupted and the damaged fragments become susceptible to fungal infections (Teper, 2009). If needle samples are covered by fungi this may ultimately limit their potential to adsorb aerosols.

Additionally, the concentrations of adsorbed aerosols in our study area varied greatly, seemingly depending on the proximity to anthropogenic sources (Fig. 3). Ciężka et al. (2018) previously studied air quality in the Holy Cross National Park using passive samplers (NO₂ and SO₂ concentrations) and bioindicators (heavy metals in Hypogymnia physodes). The authors concluded that air quality in our study area is impacted by emissions from local and regional anthropogenic sources (fossil fuel combustion; Ciężka et al., 2018): i) the impact of gaseous pollutants increased during heating season; ii) higher NO₂ concentrations were observed along roads; iii) higher SO₂ concentrations were observed at higher elevations; and iv) the highest Zn, Pb and Cu concentrations in bioindicators were recorded at high altitude and were linked to long-range transport. Our results are showing similar trends, independently of the age of the needles (Fig. 3). Still, Fig. 3 shows that aerosol concentrations were higher in the older needle samples and seem positively related to their age (i.e., exposure time).

Fig. 3. Concentrations of aerosols deposited on needle samples. (A) 2013 samples, (B) 2012 samples, (C) < 2012 samples.

The lowest aerosol concentrations were observed in 6-months old samples (2013; Fig. 3(A)). The samples were collected during the summer 2013, implying that in these 6-months old samples the corresponding adsorbed aerosols were mainly emitted by natural sources or/and resulted of long-range transport, but were not linked to emissions from home-heating. The one-year-old samples (2012) revealed the highest aerosol concentrations at sampling locations 4, 12 and 13 (Fig. 3(B)). These same sampling locations were also characterized by the highest aerosol concentrations for samples with the longest exposure time (Fig. 3(C)). This may result from transboundary transport: sampling location 12 is the highest elevation while sampling location 4 is located in the close vicinity of an inhabited area. Pairwise comparisons using the Dunn’s test showed that among the needle groups, the < 2012 and 2013 group and the 2012 and 2013 groups were significantly different (Fig. 4).

 Fig. 4. Distribution of the aerosol concentrations as a function of the needle age.

4.2 Aerosol and Needle Carbon Isotope Compositions

Our results show that aerosols adsorbed on the needles were enriched in ¹³C (i.e., higher δ¹³C), especially for samples collected close to inhabited areas (3, 4, 6), at the station located outside the park (20) as well as at the high-elevation ones (11, 12 and 15) (Fig. 5). For this latest group of samples, the increase in the δ¹³C may be associated to transboundary contamination. South-western winds dominate in the study area (Olszewski et al., 2000) and bring air pollutants from other Voivodships (especially from the Lesser Poland and the Silesian ones). This hypothesis is comforted by the fact that the highest δ¹³C we measured corresponded to locations in the southern part of the park (around sampling location 15 and further south; Fig. 5).

Fig. 5. Carbon isotope compositions (δ¹³C) of the deposited aerosols. (A) 2013 samples, (B) 2012 samples, (C) < 2012 samples.

Variations of the δ¹³C_needle were as follows: i) for the oldest samples (< 2012) carbon isotope compositions varied between –34.4 and –29.4‰, ii) 2012 samples varied between –34.5 and –29.4‰ and iii) those of 2013 between –34.1 and –29.7‰. However, while we observed small variations in the average δ¹³C_needle values (around ±0.2‰, i.e., similar to the analytical error), differencing the different sampling years was not possible. In fact, the δ¹³C_needle variations seemed more correlated to the sampling location than to exposition duration. Isotope variations generally result from numerous factors. One of them could be the Suess effect, which corresponds to a change in both the ¹³C and ¹⁴C ratios following the emissions of large amounts of fossil-fuel derived CO₂, although its influence is not fully understood (Sensuła et al., 2018). The δ¹³C difference between needles of different ages from the same sampling locations did not exceed 1‰ (SM Table 1). This may be explained by the fact that even if the time of sampling was the same (summer 2013) for all needle samples, the duration of their exposure to air pollution was different. Consequently, it was difficult to compare our results with previous similar studies, especially with that of Sensuła et al. (2018) who studied yearly collected needles (i.e., no variation in the duration of exposure to air pollution). Precipitations and humidity are two other important factors that control the δ¹³C of plants. Cuna et al. (2007) suggested that a higher amount of precipitations leads to a ¹³C-depletion (i.e., lower δ¹³C) of the plant δ¹³C_org. This comes from the fact that under humid conditions ¹²C-enriched components are preferably accumulated, inducing a negative correlation between the δ¹³C of the plant and the prevailing relative humidity (Cuna et al., 2007). Air humidity and soil moisture also control the δ¹³C values of tree rings (e.g., Gagen et al., 2007; McCarroll and Pawellek, 2001; Pazdur et al., 2013). Cuna et al. (2007) showed that elevation also impact carbon isotope compositions, showing that lichens collected at high elevations were enriched in ¹³C, and linked it to the increase in the atmospheric ozone concentrations with altitude: the excess of ozone induces a decrease in photosynthesis, leading to a ¹³C-enrichment. Ozone levels in Poland may vary depending on the time of year (GIOŚ, 2019). Finally, the canopy effect can also impact the δ¹³C_needle, with the most negative values supposed to be measured close to the ground surface (Merwe and Medina, 1991). However, here, in order to cancel the canopy effect within each sampling location, we selected and sampled trees growing in groups within a similar population. To summarize, many distinct factors can influence the carbon isotope compositions of needles, making the interpretation of the δ¹³C_needle values difficult. Here, while we did not observe a significant time-related isotope discrimination, the location of collection seems nevertheless to influence the measured δ¹³C_needle.

As discussed, the analysis of the δ¹³C of needles and aerosols may bring critical information about the corresponding contamination sources. But we first had to assure that the collected aerosol samples were not largely contaminated by crushed needles during the preparation protocol, which would artificially shift the δ¹³C value we are attributing to the aerosols. We hypothesized that δ¹³C of the deposited aerosols is different from that of needles. Additionally, we postulated that higher differences between δ¹³C_aerosols and δ¹³C_needle indicate a higher contribution of anthropogenic aerosols. Finally, we used a Isotope Mass Balance to estimate the respective contributions of the different contributing emission sources (Table 2).

As our old samples (< 2012) were covered by high amount of organic particles and those of 2013 represented only the vegetative season, the IMB was only run for the 2012 needles and their aerosols, as they were exposed to air contamination during both vegetative and heating season. We also ran the IMB considering two distinct scenarios for the source of air contamination: i) the emission source is local and mostly correspond to road traffic in the park. Its carbon isotope composition was calculated considering the respective particle contributions of typical Wroclaw road traffic (Central Statistical Office, 2013) and δ¹³C (–28.3‰ for diesel and –26.8‰ for unleaded; Górka and Jȩdrysek 2008: δ¹³C_source= 2/3 δ¹³C_diesel + 1/3 δ¹³C_unleaded = –27.8‰), ii) the emission source corresponds to home heating with a δ¹³C of –24.5‰ (Górka and Jȩdrysek, 2008).

Previous studies have shown that using bioindicators in environmental monitoring impacts the size of the aerosols collected. Teper (2009) demonstrated that for pine needles most of the particles collected are < 10 µm as larger ones are possibly washed out by rain. On the other hand, Freyer (1978) showed that, as deposition velocity is higher for fine or ultra-fine particles, conifers preferentially collect the coarse fraction. Moreover, Bartz et al. (2021) concluded that the fine fraction presents higher physical and chemical heterogeneities as it is generally connected to anthropogenic emissions. Ultimately, we can hypothesize that the fraction of fine aerosols collected on our pine needles was less abundant, rendering the identification of their emission sources difficult.

Results (Table 2) indicated that if we consider that calculated contributions > 70% are significant, road traffic (referred as the a1 value) had the strongest influence on sampling points located close to inhabited areas (3, 4, 6; Fig. 1), elevated areas (11, 15), at the sampling location outside the National Park (20) and at location 13. Also, for all samples the contribution of road traffic was dominant, the contribution of home heating being always secondary. Moreover, in scenario 2 the calculated a₂ values were almost always lower than the b₂ ones, which indicated a higher influence of particles from crushed needles. Sample 13, located in a residential area and very close to a busy road, identified significant contributions from both road traffic and home heating, and was the only sampling location where these contributions (a₁ and a₂) were higher than those of crushed needles (b₁ and b₂).

Results also confirmed our hypothesis that the isotope difference (∆) between the two isotope compositions (δ¹³C_aerosol – δ¹³C_needle) is a good proxy of the anthropogenic impact. In the Holy Cross National Park, the highest ∆ values were observed at sampling locations 13, 16, 6 and 9 with carbon isotope fractionations of 5.6‰, 3.8‰ 3.7% and 3.5‰, respectively. Location 9 is in the immediate vicinity of a busy road. For sampling locations 6, 13 and 16 local home heating and road traffic were the most probable sources of contamination. To a lesser extent, lower but still significant ∆ values were also observed at highly elevated sampling locations, which may indicate that their δ¹³C was controlled by air masses transported from distant areas or higher ozone concentrations.

 
4.3 Relationships between δ¹³C_aerosols and δ¹³C_needle

Fig. 6 reports the δ¹³C variations of the total carbon between the needles and the deposited aerosols. Results show that deposited aerosols samples were all enriched in ¹³C compared to the needles they were deposited on, which indicates that the collected aerosol particles were not dominated by needle-derived fragments.

Fig. 6. Relationship existing between the δ¹³C of needles and that of deposited aerosols.

Needle samples with several years of exposition were expected to yield the lowest δ¹³C values as their longer exposure time and subsequent accumulation of aerosols are propitious to fungal growth, that will ultimately impact their carbon isotope compositions. Meanwhile, one-year-old needle samples (2012, i.e., that were exposed to air pollution during growing and heating seasons) were expected to record the highest δ¹³C values resulting from the incorporation of ¹³C-enriched aerosols from road traffic and home heating (δ¹³C > –24.5‰). Finally, 6-months-old needle samples (2013; i.e., that were not exposed to emissions from heating sources) were expected to present δ¹³C values mostly corresponding to road traffic. Results from Fig. 6 show that only samples from location 3 follow this trend. Needle samples older than 2012 exhibited the greatest fungal growth, similar to the findings of Teper (2009), who suggested that this class of needles (i.e., older than one year) should not be used as isotope bioindicators due to their elevated amount of natural particles that ultimately modifies the final δ¹³C.

 
4.4 Total Carbon (TC), Organic Carbon (OC) and Elemental Carbon (EC) Concentrations

The analysis of individual components in aerosols may bring more constrains on the respective sources of pollution involved within the study area. The average carbon content in our aerosol samples was 40.54 ± 0.09%, a relatively high value that may indicate that local air pollution results of coal combustion for heating purposes in the industry and communal sectors as well as of PM transport from surrounding areas. High inputs of carbon in PM has already been widely documented in Poland (e.g., Zimnoch et al., 2020; 31.3–67.5% of TC in TSP) but also in Spain (e.g., Galindo et al., 2019; 26% of TC in PM₁₀ and 48% of TC in PM₁), Czech (Kucbel et al., 2016; 26–39% of TC in PM₁₀) and even in more remote areas such as China (e.g., Xu et al., 2015; ~20% of TC in PM_2.5 and ~30% of TC in PM_0.25). While the highest TC concentrations (in SM Table S1) were observed in samples older than 2012 (from 0.65 mg g^–1 (location 11) to 2.68 mg g^–1 (location 14), average of 1.48 ± 0.64 mg g^–1), the lowest ones were measured in samples from 2013 (from 0.33 mg g^–1 (location 11) to 1.01 mg g^–1 (location 15), average of 0.72 ± 0.18 mg g^–1; Table 1). Samples from 2012 presented intermediate contents (from 0.58 mg g^–1 (location 12) to 2.37 mg g^–1 (location 13), average of 1.28 ± 0.53 mg g^–1; Table 1). The fact that the highest TC concentrations were measured in the oldest samples probably results of a longer exposure time to aerosol deposition. On the opposite, the year 2013 when the lowest TC concentrations were obtained, lasted only the period of the vegetative season and thus was expectedly only impacted by emissions from road traffic. For the years encompassed by our needle samples the Regional Inspectorate for Environmental Protection (WIOŚ) reported negligible variations in the levels of atmospheric PM for the Holy Cross Voivodship area, with levels that did not exceed the EU guidelines (WIOŚ, 2012, 2013, 2014). Hence, we can conclude that aerosol accumulation on needles is linked to the duration of exposure rather than to any increase in the pollution level over the same period, in agreement with Teper (2009), who also showed that the mass of the particles deposited on needles increases with their age.

The presence of OC in our samples could be associated with biogenic particles, volatile biogenic compounds from natural sources or with secondary processes taking place in the atmosphere (e.g., oxidation by O₃, OH, NO₃, NO₂; e.g., Kucbel et al., 2016). The highest OC concentrations were observed in samples older than 2012, were slightly lower in 2012 and the lowest ones in 2013 (Table 1). As dust particles damage the needle wax layer, they inherently block stomata that damages needles’ conditions, and ultimately lead to the intensification of fungal growth on their surface (Bačić et al., 1999). The intensity of this fungi development is positively related to the needle exposure time to air pollutants (Teper, 2009). While we did not monitor the presence of fungi, Teper (2009) proved that a long exposure time can lead to the fast development of fungi and hence to an eventual increase in the OC concentrations. The largest OC difference we observed between the oldest and the youngest samples however corresponds to an exposition time that was too short to allow for an abundant development of fungi. Hence, in the oldest samples high OC may be associated with a large amount of biogenic particles accumulated over time on the surface of the needles. For the oldest samples (< 2012) the OC concentrations varied between 0.61 (location 11) and 2.65 mg g^–1 (location 14), average of 1.48 ± 0.64 mg g^–1; in 2012, from 0.47 (location 12) to 2.32 mg g^–1 (location 13), average of 1.23 ± 0.54 mg g^–1; and in 2013 from 0.33 (location 11) to 0.94 mg g^–1 (location 15), average of 0.70 ± 0.17 mg g^–1. Similarly to what we observed for the TC concentrations, the maximum OC values, for the three different periods, were measured for the exact same samples: sampling locations 14, 13, 15 for < 2012, 2012 and 2013, respectively. This may indicate that within the total carbon pool, organic carbon (OC) represents the major reservoir.

On the other hand, elemental carbon (EC) concentrations did not exceed 0.2 mg g^–1, which indicates that EC contributed slightly to TC. This might be explained by the fact that EC is mostly contained in the fine fraction of the aerosols (Zhang et al., 2020) that is less easily deposited compared to the coarse fraction. While diesel road traffic has been identified elsewhere as one of the major sources of EC (e.g., Bond et al., 2013; Yamagami et al., 2019) our highest EC concentrations were recorded during the heating season (one-year old samples; Fig. 7(B)), when emission from road traffic is not expected to dominate. During the heating season the generally high carbon concentration in the atmosphere are related to low temperatures and meteorological conditions that cause poor dispersion conditions that lead to the condensation of semi volatile species (Galindo et al., 2019). Hypothesizing that i) diesel was a major source of EC and ii) the yearly road traffic was somewhat constant in the region, it was expected that EC concentrations would be similar over the different sampling periods. Results also indicated that EC concentrations differed between sampling location, and moreover seemed to increase with the needle exposure time to air pollution (Fig. 7). The highest EC concentrations were measured in aerosols deposited on needles corresponding to sampling locations 4 and 12, in needles older than 2012 (Fig. 7(C)). Sampling location 12 is on a peak, hinting that the high EC concentrations may be associated to long-distance air masses transport. For sample 4, the contamination seemed to be related to home heating from the nearby habitations, although road traffic may also be a non-negligible source of contamination. For samples with shorter exposure times (Figs. 7(A) and 7(B)) the highest EC concentrations were observed at sampling points 15 and 12, both located on peaks that may again indicate the impact of transboundary pollutants (i.e., incomplete fossil fuel combustion) from nearby industrial plants. Coal combustion from home heating may also contribute to the EC budget, but delineating its impact zone is somewhat difficult at this point.

Fig. 7. Elemental carbon concentrations at the surface of needle samples. (A) 2013 samples, (B) 2012 samples, (C) < 2012 samples.

Fig. 8 reports the covariations between δ¹³C_TC and EC in aerosols deposited on the surface of our needle samples. Fields A, B and C delimit samples that were collected close to residential buildings (orange arrows, 3 and 4) and at elevated locations (blues arrows, 11 and 15) for different groups of needle samples (2013, 2012 and older than 2012, respectively).

Fig. 8. Relationship between δ¹³C_TC and elemental carbon (EC) in deposited aerosols. Fields A (2013), B (2012) and C (< 2012) correspond to samples collected close to residential buildings and at the Park’s peaks. Fields D and E represent points located close to roads (9, 10). Arrows represent changes in the carbon isotope compositions between years 2012 and 2013 (orange for residential areas, blue for peaks). The typical δ¹³C ranges for potential anthropogenic sources are taken from the literature.

Although we previously discussed the possibility that samples older than 2012 may be impacted by the accumulation of organic particles over time, results indicated statistically significant (p < 0.05) differences between 6-months-old samples (2013) and one-year-old samples (2012) (Fig. 8). Results in Fig. 8 show that for some needle samples we can observe noticeable differences between the different needles generations. For example, samples from locations 3 and 4 showed a combined increase in their respective EC concentrations and δ¹³C_TC from 2012 to 2013 that can be interpreted as an increasing impact of emissions from home heating. For samples 11 and 15 (elevated areas) the signal probably corresponds to aerosol contamination transported from nearby urbanized areas.

Samples 9 and 10 have characteristics in Fig. 8 that are similar to those of fields D and E, representative of emissions from road traffic. Our results suggest that differences are only noticeable at sampling locations dominated by a single source of pollution. Interestingly, the δ¹³C_TC values are increasing with age at location 10. As location 10 is closer to residential buildings, this trend might be explained by inefficient coal combustion, as an accumulation of coal soot (δ¹³C = –24.5‰) will increase the carbon isotope composition. In turn, samples at location 9 were collected right next to a busy road. Hence, the influence of diesel particles (–28.3‰) or unleaded fuel particles (–26.8‰) were clearly identified. For both fields D and E δ¹³C_TC differences were observed between samples from 2012 and 2013: one-year-old samples (2012) were characterized by slightly more positive values than 6-months-old samples (2013); –30.2‰ (location 9), –29.0‰ (location 10) and –31.0‰ (location 9), –29.6‰ (location 10), respectively. This decrease in δ¹³C between 2012 and 2013 may be explained by the fact that in 2013 the heating season signal was not preserved. Most terrestrial plants (C₃ plants) are characterized by δ¹³C values ranging from –34‰ to –24‰ (Šturm et al., 2012). Thus, the δ¹³C of samples that only reflect the growing season will be more negative than those exposed for longer periods of time. Here, we observed more positive δ¹³C values resulting of fossil fuel or biomass combustion in samples exposed for one or more years.

Additionally, to confront our conclusions we conducted a Principal Component Analysis. Results showed that the first 2 factors explained 84% (determined by a Scree test) of the observed variations among the variables: 58% for Factor 1 and 26% for Factor 2 (Table 3). The remaining 16% constituted a random noise that is not interpretable using this technique. Factor 1 positively correlates aerosol, TC and OC concentrations, while Factor 2 is controlled by EC concentrations and δ¹³C. As Factor 2 yields a positive correlation between δ¹³C and EC concentrations, it can be associated with the combustion processes whereas Factor 1 represents all other processes.

Fig. 9 reports the factor scores obtained for each sample along Factors 1 and 2. While factor scores ranging between –1 and +1 indicate an average intensity for the controlling process, values > +1 indicate that the process is dominating and inversely a value < –1 indicates that the process is not contributing. Results (Fig. 9) show that Factor 2, associated to combustion processes, dominates for samples collected on peaks (sampling locations 12 and 15) or high altitudes (sampling location 11). This is also true for samples collected nearby inhabited areas (sampling locations 3, 4 and 6). However, factor scores for Factor 2 in samples collected close to roads did not yield high values, which may indicate that among the combustion processes associated to this factor, road traffic is not the most contributing.

Fig. 9. Factor scores obtained for the one-year old needle samples at the different sampling locations.

 
5 CONCLUSIONS

Here, we investigated and demonstrated that Abies alba needles are reliable bioindicators to identify sources of aerosol contamination for a given area. These bio-passive samplers help to avoid the use and maintenance of expensive automatic samplers in inaccessible areas where air quality monitoring is required. Here, we are proposing that, in areas where they grow, fir needles are a good air quality monitoring tool. This applies especially to mountains and woody areas where using specialized equipment is not possible or very difficult due to altitude, the lack of access to electricity or a limited access for people. Still, this method requires that the accumulated mass of particles deposited on the surface of the needles be quantified, as an excess can block the suprastomatal chambers and alter the carbon isotope signal. Using the example of the Świetokrzyski National Park (Holy Cross Voivodship, Poland) our results show that the highest δ¹³C values were observed at locations that were: a) at high elevations, b) near inhabited areas and c) outside the park. Using a PCA approach we identified that combustion processes mostly control their carbon contents. Our study also demonstrates that one-year-old needles are the most reliable samples. While part of the aerosols deposited on the needles can be removed by wind or rain, combining carbon stable isotopes and geochemical analysis provides a reliable assessment of the aerosol mass and greatly helps identifying the major sources of air pollution. Our results identified emission from home heating, road traffic and transport from nearby urbanized areas as the major sources of pollution in the ŚNP. Finally, we do not recommend using needles older than one year due to a high accumulation of organic particles that inherently alters the information that could be retrieved about obtained air pollution. The analysis of TC/OC/EC and of SEM images suggest that for these older samples, interpretation is rendered difficult and imprecise. At the same time, it still arises the need for further studies, especially inter-comparisons with reference methods.

 
ACKNOWLEDGEMENTS

This research was supported by the University of Wroclaw grants 0401/1017/18 and 501/0410/2988/18. The authors would like to thank Monika Ciężka, Anna Łubek and Agnieszka Winiszewska for helping with the field work and sample collection. We wish to express sincere gratitude to prof. Gregory D. Hoke for helping to improve the manuscript.