ATMOSPHERIC CO₂ CARBON ISOTOPE COMPOSITION IN URBAN AND CLEAN AREAS OF THE NORTHERN ADRIATIC COAST OF CROATIA

Abstract

Over the course of one year (2021), we monitored the carbon isotope composition of atmospheric CO₂ at three locations in Croatia: the Adriatic port city of Rijeka (Cfa climate) and at two rural sites: Gornje Jelenje (Cfb climate) in the vicinity of a main road and clean-air site Parg (Dfb climate). Carbon isotope composition at all sites shows seasonal variation, ranging from –41.3 to 25.2‰ for Δ¹⁴C and from –13.1 to –11.3‰ for δ¹³C. Rijeka systematically has the lowest and Parg the highest Δ¹⁴C, and δ¹³C at the sites are not statistically different one from another. The Δ¹⁴C of leaves of deciduous trees reflect the trend of atmospheric Δ¹⁴C. Based on the assumption that the investigated area is under the influence of two main sources of CO₂: fossil and natural (sea exchange, biosphere, and undisturbed – clean air atmospheric component) the approximate share of fossil CO₂ in total atmospheric CO₂ has been estimated for Rijeka (2.1 ± 1.3%) and Gornje Jelenje (1.0 ± 0.9%). Comparison of our results with the data from European CO₂ sampling stations indicates strong influence of CO₂ from sea and biosphere. Backward trajectories indicate a possibility of Δ¹⁴C_CO2 contribution from distant EU nuclear power plants, but movement of air masses should be considered in more detail to confirm this.

INTRODUCTION

The isotopes of carbon, radioactive ¹⁴C and stable ¹³C (as ¹⁴C/¹²C and ¹³C/¹²C ratios, or Δ¹⁴C and δ¹³C, respectively), are mostly used in global carbon and atmospheric studies, as well as for the study of anthropogenic influence on the environment. Both isotopes originate from natural and anthropogenic sources. Natural ¹⁴C production predominantly depends on the Earth’s geomagnetic field and solar activity and it varied significantly in the past (Muscheler et al. 2005; Miyake et al. 2012; Mekhaldi et al. 2015; Channell et al. 2018; Heaton et al. 2021). Natural ¹⁴C distribution in the Earth’s atmosphere was anthropogenically disturbed in 19th century when a decrease of ¹⁴C concentration in atmospheric CO₂ (Δ¹⁴C_CO2) was recorded in tree rings and correlated with the beginning of the Industrial Revolution and the release of ¹⁴C free CO₂ into the atmosphere due to fossil fuel combustion (Suess 1955). Similarly, δ¹³C of atmospheric CO₂, with pre-industrial value of –7‰, also decreased, since fossil fuels have δ¹³C values around –26‰ (Graven et al. 2020; Keeling and Graven 2021). In the 1950s and 1960s, atmospheric ¹⁴C concentration suddenly increased due to atmospheric nuclear bomb testing. ¹⁴C activity almost doubled in the Northern Hemisphere and reached a peak in 1963 (Nydal and Lövseth 1965), after which it started decreasing due to the Nuclear Test Ban Treaty and CO₂ exchange with other carbon reservoirs, mostly oceans (Levin and Hesshaimer 2000; Hua et al. 2013; Graven et al. 2020; Heaton et al. 2021; Kutchera 2022; Levin et al. 2022). Today, levels of Δ¹⁴C_CO2 have approached pre-bomb values (Heaton et al. 2021; Hua et al. 2022) and further decrease of ¹⁴C concentration is predicted as well as a decrease in δ¹³C values stemming from fossil fuel combustion (Heaton et al. 2021; Hua et al. 2022; Levin et al. 2022). The decrease of Δ¹⁴C_CO2 and δ¹³C_CO2 in the future can be enhanced by the exchange of CO₂ between oceans and the atmosphere (Menviel et al. 2015; Skinner et al. 2017; Keeling and Graven 2021). The oceans may also become a source of ¹⁴C release into the atmosphere due to release of bomb ¹⁴C from the water column (Gao et al. 2019; Levin et al. 2022). Other sources of man-made ¹⁴C, such as various nuclear facilities, can contribute to an increase of Δ¹⁴C_CO2 locally.

Generally, Δ¹⁴C_CO2 has a seasonal trend with expected maximum concentration during spring/summer mainly due to vertical mixing between the troposphere and stratosphere (Turnbull et al. 2009; Graven et al. 2020). In the autumn and winter seasons Δ¹⁴C_CO2 values decrease due to a lack of atmospheric mixing. Locally, the winter Δ¹⁴C_CO2 minima could be observed in urban (Rakowski et al. 2008; Molnár et al. 2010; Varga et al. 2019) and even clean-air site areas due to fossil fuel burning during heating season (Major et al. 2018). δ¹³C_CO2 also shows seasonality, but for different reason: higher spring/summer values are due to photosynthesis and lower autumn/winter values are due to plant respiration (Turnbull et al. 2009; Graven et al. 2020). Since δ¹³C of fresh plants and respired CO₂ are generally similar to that of the fossil fuel, δ¹³C cannot be used to differentiate bio from fossil carbon. Δ¹⁴C_CO2 is monitored by several long-term stations (Levin and Hammer 2021; Levin et al. 2022) of which there are currently 15 stations active in Europe (ICOS 2022). δ¹³C_CO2 is monitored world-wide by the Scripps CO₂ Program (Keeling et al. 2005; Keeling and Graven 2021) and Global Monitoring Laboratory (NOAA – GML 2023; Vaughn et al. 2010).

We present the establishment of three experimental stations for monitoring Δ¹⁴C_CO2 and δ¹³C_CO2 in the northern coastal area of Croatia: the port of Rijeka and two sites in its surroundings (Gornje Jelenje and Parg). The aim was to determine the influence of fossil fuel combustion on Δ¹⁴C_CO2 and δ¹³C_CO2 values at different locations with the hypothesis that the urban site is affected by fossil fuel uses. It should be emphasized that there is no previous data on carbon isotopic composition of atmospheric CO₂ in the Croatian coastal area and there are scant data from European coastal areas. European monitoring sites included in the ICOS program are mainly continental (Levin et al. 2022; Scripps CO₂ Program 2022) and the stations included in Scripps CO₂ Program, such as Mauna Loa, Hawaii, are situated on Pacific islands or close to the Arctic and Antarctic (Keeling et al. 2005; Keeling and Graven 2021; Scripps CO₂ Program 2022). A few monitoring programs have analyzed carbon isotopic composition of the atmospheric CO₂ above the oceans collected from boats (Dutta et al. 2006; Longinelli et al. 2012; Gao et al. 2019) and isotopic composition of dissolved inorganic carbon (DIC) (Becker et al. 2016; Gao et al. 2019; Scripps CO₂ Program 2023). The Mace Head (MHD) monitoring station situated on west coast of Ireland is exposed to clean marine air nearly 50% of the time but is also under the influence of polluted air masses from the UK and Europe (Yttri et al. 2019).

SAMPLING AND METHODS

Sampling Sites Characterization

The characteristics of all three sampling locations are summarized in Table 1. Rijeka is the third largest city in Croatia and has the largest seaport, with 128,600 inhabitants and a population density of 2900 inhabitants/km². It is located in Primorje-Gorski Kotar County, on the coast of the Kvarner Gulf (Figure 1). The climate of Rijeka is moderately warm and humid with hot summers (Cfa) (Peel et al. 2007; Beck et al. 2018). It is known for the Bora, the cold and dry north-easterly wind (CMHS 2022; Figure 2a) that sometimes gusts at hurricane speeds (160 km/hr). The annual mean air temperature (T) in 2021 was 15.1°C (seasonal variations in Figure 2b), and the total amount of precipitation (P) was 1161 mm (CMHS 2022). The samples were collected in the center of the city (Figure 1d), near the main railway station, 9 m above ground (0 m a.s.l), on the terrace of an apartment.

Table 1 Characteristics of the three sampling locations.

Figure 1 (a) Map of Europe with Croatia highlighted; (b) position of the sampling area in Croatia; (c) enlarged map of sampling points, and (d) micro location of sampling point in Rijeka

Figure 2 (a) Wind roses for Rijeka and Parg in 2021. (b) Air temperatures for Rijeka and Parg (CMHS 2022).

Gornje Jelenje is a saddle located 14 km northeast of Rijeka in the foothills of Risnjak National Park and Snježnik Mountain (Figure 1). The area is covered by forests, and the sampling site is located 500 m away from the main road at 882 m a.s.l.; the samples were collected 2 m above the ground. The climate is moderately warm and humid with warm summers (Cfb) (Peel et al. 2007; Beck et al. 2018). Gornje Jelenje is not a CMHS station and there is no meteorological data for this site.

Parg is a village in Gorski Kotar, 33 km distant from Rijeka. It is located in a wooded, mountainous area far from inhabited zones and roads (Figure 1). The climate is classified as Dfb type (Peel et al. 2007; Beck et al. 2018) with annual mean air temperature in 2021 of 8.4^oC (seasonal variations in Figure 2b), and an annual amount of precipitation 1586 mm with snowfall in the winter period and warm summers. The wind that prevails in the area is SW to SE (CMHS 2022, Figure 2a); samples were collected 2 m above the ground.

Sampling Methods

Sampling for atmospheric CO₂ Δ¹⁴C and δ¹³C analyses was performed from January 2 to December 28, 2021. Atmospheric CO₂ for Δ¹⁴C analyses was sampled by static absorption method on saturated NaOH in plastic pads as a one-month integrated sample. Absorption occurs under highly alkaline conditions (Dietzel 1997) accompanied by intensive isotope fractionation and the δ¹³C of absorbed CO₂ (in form of Na₂CO₃) is not representative for atmospheric CO₂ (Krajcar Bronić et al. 2006). Na₂CO₃ samples were transferred from sampling pads into plastic bottles and thoroughly sealed to avoid isotopic exchange between sample and the atmosphere. For δ¹³C_CO2 analyses, grab air samples were collected by 50 mL plastic syringe and transferred to a 12 mL Labco glass ampoule fitted with a gas-tight septum, flushed using two needles (input and output needle) and filled with air under pressure (Kanduč et al. 2021). Three ampoules were filled per location: at 2 m above ground at Parg and Gornje Jelenje, and 9 m above the ground in Rijeka. These samples were taken at the same sampling heights as the samples for the Δ¹⁴C analyses and stored at standard atmospheric conditions until analysis. Sampling was performed between 2:00 and 5:00 pm.

The leaves of deciduous trees were collected on August 30, 2021, directly from a plane tree (Platanus) in Rijeka and maple trees (Acer) in Parg and Gornje Jelenje.

Measurement Methods

The majority of the CO₂ samples and leaf samples were measured in the form of benzene in a Liquid Scintillation Counter Quantulus 1220 (Horvatinčić et al. 2004; Krajcar Bronić et al. 2023). CO₂ was released from Na₂CO₃ by HCl (18%) in an inert N₂ atmosphere and cryogenically purified. The obtained CO₂ was converted to benzene in vacuum synthesis line (Horvatinčić et al. 2004). The leaf samples were carbonized at 650°C and converted to benzene (Horvatinčić et al. 2004; Krajcar Bronić et al. 2023). Accelerator mass spectrometry (AMS) was used in the cases when the sample quantity of absorbed CO₂ was not enough for benzene synthesis. Samples were prepared in the form of graphite targets: CO₂ from Na₂CO₃ was evolved using acid (4% HCl) and reduced to graphite with Zn reduction on Fe powder (Sironić et al. 2013). Graphite samples were measured at the Center for Applied Isotope Studies, University of Georgia (Cherkinsky et al. 2010). Anthracite and SRM 4990C (OxA II) were used as background and reference samples, respectively, for both techniques. All ¹⁴C activities were corrected to the date of sampling and defined as Δ¹⁴C (Mook and van der Plicht 1999; van der Plicht and Hogg 2006).

δ¹³C_CO2 from the air samples was determined on the Europa Scientific 20-20 mass spectrometer with the continuous flow IRMS ANCA-TG preparative module at the Jožef Stefan Institute, Ljubljana in Slovenia. Reference material with a known value of δ¹³C_CO2 (–35.4 ± 0.2‰) calibrated to VPDB (Vienna Pee Dee Belemnite) was used. Additionally, as a working standard, natural gas from emission sources in Sicily with a known δ¹³C_CO2 (–0.87 ± –0.20‰) determined at the Istituto Nazionale di geofisica e vulcanologia, manufacturer: gtp-gas.it, (homepage: https://gtp-gas.it/diassido-liquidi-criogenici-i/9) was used. Two-point normalization was performed for CO₂. 0.01 mL of CO₂ (reference gas and working standards) was transferred to Labco ampoules. The whole collected sample was used for δ¹³C_CO2 determination. Because concentrations of CO₂ in air are low, optimal integral peak area (liquid nitrogen cryo trap) was used. Stable isotopes are reported in the δ notation (Coplen 1996) relative to VPDB and measurement uncertainty was between ±0.3 and ±0.7‰.

Estimation of the Fossil CO₂ Component

We estimated the fossil CO₂ share in the Rijeka and Gornje Jelenje sampling sites assuming that Parg as a clean-air site had ¹⁴C signal only from natural sources (CO₂ free from fossil fuel influence, biospheric and sea CO₂). Since the shares of the CO₂ natural components were unknown, we assumed that Rijeka and Gornje Jelenje differed from Parg only in fossil CO₂ share, i.e., we approximated biospheric, sea and fossil-free CO₂ components as a background (Levin et al. 2003). This simplification enabled the estimation of fossil CO₂ contribution in Rijeka and Gornje Jelenje from a simple two component mixing relation:

$${\rm{a = - 1000 \times X + b \times (1}} - {\rm{X)}}$$

where “a” represents Δ¹⁴C_CO2 from Rijeka or Gornje Jelenje, “b” represents the background Δ¹⁴C value, –1000‰ is Δ¹⁴C of fossil CO₂, and “X” is the share of fossil CO₂. Such mixing relations are used when data on CO₂ concentration are not available (Piotrowska et al. 2019). As a background Δ¹⁴C values, data from JFJ (Jungfraujoch, Switzerland) station are usually used (Levin et al. 2008). In this paper the Δ¹⁴C_CO2 values from Parg were used as background, and additional calculation was made using the JFJ data from 2020 as data for entire 2021 were not available.

RESULTS AND DISCUSSION

Results of Δ¹⁴C_CO2 are shown in Figure 3a, while numerical individual data are presented in Table SM1 of supplementary material. For all three locations the data show similar seasonal patterns (Figure 3), with significant correlation with temperature values (Rijeka r = 0.58, Parg r = 0.70, p < 0.05; Figure 3c). Δ¹⁴C_CO2 mean values (Table 2) show that Gornje Jelenje represents an approximate middle value (–10 ± 8‰) between Rijeka (–22 ± 12‰) and Parg (0 ± 12‰). Tree leaves should reflect Δ¹⁴C_CO2 values from the spring/summer period (March–September). This is true for the tree leaves from Gornje Jelenje and Parg: their Δ¹⁴C values correspond to Δ¹⁴C_CO2 values for respective periods (Table 2). The leaves from Rijeka show a higher, but statistically insignificant (z = 0.65; Table SM3) Δ¹⁴C value compared to the Δ¹⁴C_CO2 values determined in Rijeka for the spring/summer period (Table 2), and it is more similar to the Δ¹⁴C_CO2 from winter season. This could be explained by the altitude difference and coastal position of Rijeka, where the budding period generally starts earlier than in the mountain area. The altitude Δ¹⁴C effect has been reported in tree leaves from higher altitudes (Sakurai et al. 2013), but higher Δ¹⁴C values were found at above 3000 m a.s.l. The trend of a Δ¹⁴C increase from the lowest altitude sampling point in Rijeka, to the highest, through Gornje Jelenje to Parg observed in samples of tree leaves, as well as in respective Δ¹⁴C_CO2 values (Table 2) is the most likely caused by distance from the expected fossil fuel and marine reservoir carbon sources and climatic differences between sites, rather than altitude effect. The increase in Δ¹⁴C_CO2 between Rijeka and Gornje Jelenje to Parg (from –22‰, –10‰, to 0‰; Table 2) much better fits the increase in air distance from Rijeka (15 km to Jelenje and 32 km to Parg), than the increase in elevation (0 m a.s.l., 882 m a.s.l., and 863 m a.s.l.; Table 1).

Figure 3 (a) Integrated monthly Δ¹⁴C values at the Rijeka, Gornje Jelenje, and Parg stations for 2021, and for leaves sampled in August 2021 at the respective stations; (b) Δ¹⁴C difference of atmospheric CO₂ for Rijeka and Gornje Jelenje sites compared to the clean-air site Parg; (c) Δ¹⁴C vs. monthly temperature.

Table 2 Characteristic carbon isotopic composition of samples from the investigated area, 2021. Individual data are presented in Tables SM1 and SM2.

The observed decrease of Δ¹⁴C_CO2 from north to south (clean-air to urban site) could be caused by fossil fuel influence in Rijeka, but this signal is not clear throughout the whole year. In January, the coldest month in 2021 (Figure 2b), Δ¹⁴C_CO2 values are identical for all three sites (Figure 3a; Table SM1), although it is expected that Δ¹⁴C_CO2 would be lower in Rijeka due to heating with natural gas and oil. In February, when atmospheric temperatures start to increase, we observed a decrease of Δ¹⁴C_CO2 in Rijeka and an increase in Gornje Jelenje and Parg. In March, Δ¹⁴C_CO2 values decreased at all three locations, with the largest difference between Rijeka and Parg (Figure 3b; Table SM3). Starting from May when all three locations had same value, Δ¹⁴C increased peaking in September (Rijeka and Parg), and October (Gornje Jelenje), showing an increasing difference between Parg and the other two locations (Figure 3b). The minimum Δ¹⁴C_CO2 value was detected in Rijeka in October, but this cannot be explained solely by start of heating season, because November and December temperatures were lower, and Δ¹⁴C values higher than in October (Figure 2b). Rijeka is the main Croatian seaport and a busy tourist destination, so fossil fuel emissions due to heavy traffic should be expected during summer months, but the Δ¹⁴C values were almost identical to those from Gornje Jelenje. During the tourism season in 2021, there were no COVID travel restrictions in Croatia and the number of visitors in the Kvarner area was nearly the same as in 2022 (CBS 2022).

Apart from burning of fossil fuels, the observed Δ¹⁴C_CO2 fluctuations can be explained by several possible processes. Δ¹⁴C_CO2 values at all three stations simultaneously started increasing in May, which can be attributed to a penetration of stratospheric air masses enriched with ¹⁴C into the troposphere (Levin et al. 2010, 2022; Graven et al. 2012; McDonald et al. 2019; Heaton et al. 2021); this vertical mixing does not occur during the autumn/winter season, resulting in lower Δ¹⁴C_CO2 values. Wind directions in the investigated areas can also help in data interpretation. Rijeka is under the influence of cold northeast wind that blows from the mountains, i.e., from the direction of Parg and Gornje Jelenje (Figure 2a). Parg is under strong influence of southerly winds, i.e., from the coast toward mountains (Figure 2a). Wind directions imply strong mixing of air masses above Rijeka and Parg, with the influence of the marine air. Due to Marine Reservoir Effect (Faivre et al. 2015) and to the strong vertical mixing processes, especially pronounced in the northern Adriatic during the winter (Vilibić and Supić 2005), surface dissolved inorganic carbon (DIC) is ¹⁴C depleted, which is reflected in atmospheric CO₂ (Skinner et al. 2017; Graven et al. 2020; Keeling and Graven 2021); this corresponds to lower Δ¹⁴C_CO2 values in the autumn/winter period. During summer stratification, Δ¹⁴C_DIC values are generally higher due to the still present ¹⁴C from the bomb peak in marine water column (Fallon et al. 2003; Gao et al. 2019; Guilderson et al. 2021). The presence of bomb peak residue is exhibited in shells from the northern Adriatic (Δ¹⁴C = 25‰ in 2013; Peharda et al. 2019). This corresponds to higher Δ¹⁴C_CO2 during the spring/summer period. The especially high Δ¹⁴C_CO2 level at Parg could be attributed to the influence of this ¹⁴C enriched marine CO₂ that can be transported by strong S and SW winds (Figure 2a), and increased CO₂ outgassing due to high summer temperatures (Figure 2b). Mediterranean influence during the summer has been shown on isotopic composition of precipitation (Vreča et al. 2006; Krajcar Bronić et al. 2020), as well as from backward trajectories for July and September (Stein et al. 2015; Rolph et al. 2017; NOAA – HYSPLIT 2023). The photosynthesis/respiration ratio also has a certain role in Δ¹⁴C_CO2 dynamics, but it complicates the interpretation of our results as photosynthetic activity during spring/summer that overrides respiration should decrease Δ¹⁴C_CO2 due to CO₂ consumption, and the respiration, with no photosynthesis (autumn/winter) should increase Δ¹⁴C_CO2 due to CO₂ production. Nevertheless, photosynthesis vs. respiration could explain the Δ¹⁴C_CO2 drop in March at all three locations, the most intensive of which was in Rijeka. This corresponded to the observed Δ¹⁴C value of tree leaves from Rijeka and start of budding season earlier compared to the two mountain sites Gornje Jelenje and Parg. Maximum values of Δ¹⁴C_CO2 in September (Rijeka and Parg) and October (Gornje Jelenje) could be due to soil respiration and CO₂ release into the atmosphere (Trumbore 2000).

Generally, Δ¹⁴C_CO2 values from urban Rijeka site are the lowest (average value –22 ± 12‰, Table 2) compared to ICOS stations, while values from Gornje Jelenje (average Δ¹⁴C_CO2 is –10 ± 8‰, Table 2) are comparable to STE (Steinkimmen, Germany, average Δ¹⁴C_CO2 is –9 ± 6‰, Kubistin et al. 2022a), KRE (Křešín u Pacova, Czechia, average Δ¹⁴C_CO2 is –8 ± 3‰, Marek et al. 2022). It must be emphasised that clean-air site Parg has higher Δ¹⁴C_CO2 values (average 0 ± 12‰, Table 2) compared to other EU monitoring stations, but due a huge Δ¹⁴C_CO2 fluctuation during the year, average value does not show a clear difference. The peak Δ¹⁴C_CO2 value of 24.8 ± 6.4‰ was observed at Parg in September, and the highest value among the EU stations was 2.6 ± 1.8‰ measured in September at SVB (Svartberget, Sweden, data available until September 2021, Marklund et al. 2022). The ICOS station that could be regarded as most similar to Parg and with data available for the whole of 2021 is KRE, but the maximum Δ¹⁴C_CO2 value determined for this site was –3.5‰ in August. Other ICOS stations with data available for the whole year are STE, PAL (Pallas, Finland; without data for December; Hatakka 2022) and OPE (Observatoire pérenne de l’environnement, France; without data for December, Ramonet et al. 2022). The maximum Δ¹⁴C_CO2 values were –0.2‰ measured in August at STE, 1.3‰ measured at PAL in September, and –2.78‰ at HTM (Hyltemossa, Sweden, Heliasz and Biermann 2022), also in September. There are not data for the whole year from the JFJ, but data for 2019 and 2020 show maxima Δ¹⁴C of 6.1 ± 1.5‰ in September and 5.4 ± 1.7‰ in July, respectively (Emmenegger et al. 2022). Data from Mace Head (MHD) would be interesting as this is European maritime station, but Δ¹⁴C_CO2 values are available only until March 2016. However, a comparison of MHD data from 2015 with the available data for 2015 from JFJ station, show higher summer Δ¹⁴C_CO2 values at MHD (maximum was 23‰ in August, while maximum Δ¹⁴C_CO2 for JFJ was 17.4‰ in February). These two values could imply summer ¹⁴C sea CO₂ influence onto the atmosphere at MHD, and a similar explanation might be applicable to high summer Δ¹⁴C_CO2 values at the Parg site. Parg has a Δ¹⁴C_CO2 range for January–June and October–December, of –18.2‰ to 4.0‰, while between July and September these values are considerably higher within the range from 10.1‰ to 24.8‰. To the best of our knowledge there is no source of anthropogenic ¹⁴C in the area, since the contribution of Δ¹⁴C from the only nuclear facility in a radius of 100 km from the investigated sites (NPP Krško, Slovenia, NE direction) disappears 15 km from the facility (Krajcar Bronić et al. 2017). Backward trajectories (Stein et al. 2015; Rolph et al. 2017; NOAA – HYSPLIT 2023) indicate travel of air masses from distant areas and that high Δ¹⁴C_CO2 signals could be transferred from EU nuclear power plants. Investigation of atmospheric pollution in Croatia showed that more than 70% of pollutants have been transferred from other European countries by air masses (Špoler Čanić et al. 2009). However, much more investigation is needed on this topic. As we discussed earlier, apart from seasonal influence of ¹⁴C enriched marine CO₂, the high Δ¹⁴C_CO2 values from Parg could be the result of CO₂ originated from organic matter decomposition. The Parg area is covered mainly by coniferous trees and fallen leaves may be several years or decades old, so the CO₂ resulting from decomposition has higher Δ¹⁴C values compared to the atmosphere of deciduous forest area (Trumbore 2000).

The approximate shares of calculated fossil CO₂ (Table SM5) were 2.1 ± 1.3% at Rijeka and 1.0 ± 0.9% at Gornje Jelenje (yearly averages). The maximum shares were calculated for Rijeka in October (4.5%) and for Gornje Jelenje in September (2.6%) and the minimum, even negative, values of –0.4% (Rijeka) and –0.3% (Gornje Jelenje) in January (Table SM5). The values were also lower compared with data for January from majority of ICOS stations, except the STE (Δ¹⁴C_CO2 is –15.0‰) and KIT (Karlsruhe, Germany, Δ¹⁴C_CO2 is –22.3‰, Kubistin et al. 2022b) while other stations had values between –12.9 and 5.6‰. Although the Δ¹⁴C_CO2 difference between Rijeka and Parg, and Gornje Jelenje and Parg were non-significant for January (z = 0.37 and 0.31, respectively, Table SM3), negative shares of fossil CO₂ values indicate the influence of ¹⁴C depleted CO₂ at the Parg station, probably transported by SSW wind from the Adriatic Sea (Figure 2a). It is worth mentioning that backward trajectories for January 2021 for three investigated sites as well as for STE and KIT indicate air masses transport from the North Atlantic, detailed investigation on this subject is necessary as well as the ¹⁴C analyses of sea DIC to determine the influence of CO₂ form the sea. The calculation with JFJ data as a background (from 2020) give very similar yearly averages for both Rijeka and Gornje Jelenje (Table SM5) without negative values in January, which also implies a source of ¹⁴C depleted CO₂ in the Parg site during winter.

The comparison between clean-air sites and industrial areas based on tree ring ¹⁴C analyses show higher fossil CO₂ shares in urban European areas compared to Rijeka and Gornje Jelenje: in south-west Slovakia 3.4 ± 1.5% (Kontul’ et al. 2020), southern Poland 3.6 ± 0.3% (Rakowski et al. 2005, 2008; Pazdur et al. 2007), and 5.8% in Gliwice, Poland, which is regarded as one of the most fossil CO₂ affected region (Piotrowska et al. 2019). Samples from the Debrecen area, in Hungary, indicate smaller fossil CO₂ contribution in tree leaves (0.9 ± 1.2%) compared to grasses (2.5 ± 2.5%), with higher values in the vicinity of busy crossroads (4.7 ± 0.7% and 9.6 ± 0.7% for tree leaves and grasses, respectively, Varga et al. 2019). Dilution of fossil CO₂ observed in Debrecen at sites away from busy crossroads may be used as an explanation for our data from Rijeka and Gornje Jelenje. Rijeka, although an urban and touristic place, has many city parks that clearly help to remove produced CO₂. This could be valid for the Gornje Jelenje area as well, where the vicinity of the road is heavily forested and therefore rife with photosynthetic activity and CO₂ consumption.

CO₂ for the determination of δ¹³C was sampled as a grab sample at the beginning and at the end of the sampling period for Δ¹⁴C_CO2 determination (Figure 4; Table SM2). δ¹³C_CO2 values have similar seasonal trends for all locations and δ¹³C values between all three sites are significantly correlated (R-GJ: r = 0.66, p = 0.025; R-P: r = 0.72, p = 0.011; P-GJ: r = 0.72, p = 0.011). Of the three sites, only the clean-air site Parg is significantly correlated to temperature (r = 0.77, p < 0.02), implying the predominant influence of natural processes at Parg (photosynthesis with higher δ¹³C values and organic matter decomposition with lower δ¹³C values). A decreasing trend of δ¹³C values from Parg, through Gornje Jelenje to Rijeka is barely visible, and is the most distinguished in minimal values between Parg and the two other sites (Table 2; Table SM4). Although the mean δ¹³C values are not significantly different, seasonal fluctuation show differences. Compared to Parg, δ¹³C_CO2 in Rijeka is lower during spring, summer and autumn, but statistically significantly only in September by –1.1‰ (z = 2.6) (Figure 4b; Table SM4). δ¹³C_CO2 values at Gornje Jelenje are also lower compared to Parg within the same period as Rijeka, but none of these differences are significant. It is interesting that both, Rijeka and Gornje Jelenje, have significantly higher δ¹³C_CO2 for 1‰ (z = 2.4 and 2.6, respectively; Table SM4) compared to Parg in March which corresponds to Δ¹⁴C_CO2 values drop (Figure 3a) explained earlier as CO₂ photosynthesis uptake. The minima δ¹³C values were in January (Parg and Rijeka) and December (Gornje Jelenje) and maxima in September (Parg and Gornje Jelenje) and June (Rijeka) (Figure 4a; Table SM2). The spring and summer δ¹³C maxima reflect CO₂ consumption and discrimination of ¹³C due to photosynthesis. Higher summer values at Parg and Gornje Jelenje compared to Rijeka may be attributed to geographical differences between sites: Parg belongs to the Dfb climate type, Gornje Jelenje to the Cfb while Rijeka to the Cfa climate. The described differences mean that photosynthetic activity at Parg and Gornje Jelenje begins later than in Rijeka. This trend is observed in Δ¹⁴C values of atmospheric CO₂ and tree leaves. Decreasing of δ¹³C values from June to the end of the year could be connected to fossil fuel combustion from more intensive traffic in Rijeka, but contribution of soil respiration cannot be neglected and were confirmed by the maximum Δ¹⁴C value in Rijeka in September.

Figure 4 (a) δ¹³C in grab samples of atmospheric CO₂ at three locations; (b) δ¹³C differences for the Rijeka and Gornje Jelenje sites and the clean-air Parg site; (c) δ¹³C vs. temperature.

The decrease in δ¹³C values at Parg and Gornje Jelenje starts later (in September) reflecting a shift in seasonal activity (start of respiration and decrease in photosynthetic activity). High late summer/autumn Δ¹⁴C_CO2 values (Figure 3a) at Parg and Gornje Jelenje confirm soil respiration influence.

The δ¹³C_CO2 and Δ¹⁴C_CO2 values of each investigated site do not correlate to each other, however Parg shows the most similar behavior of δ¹³C_CO2 and Δ¹⁴C_CO2 values in year (Figures 3a and 4a). This corroborates the assumption that Parg is influenced mostly by natural processes. The δ¹³C and Δ¹⁴C differences relative to Parg (Figures 3b and 4b; Table 2) show that Rijeka and Gornje Jelenje are lower than Parg in both values throughout nearly the whole year. All three sites are affected by strong winds and the transport of atmospheric CO₂ depleted in ¹³C from the sea towards the land, which contributes to the decrease in δ¹³C_CO2 during late autumn/winter period (Figure 4a). The possible sea CO₂ influence are also confirmed by lower winter Δ¹⁴C_CO2 values (Figure 3a). However, there are no data for seasonal δ¹³C and Δ¹⁴C values of DIC from the Adriatic Sea necessary for interpretation of this influence. Additionally, the interpretation of CO₂ sources on the basis of δ¹³C is more complicated due to the lack of δ¹³C differentiation between δ¹³C_CO2 value derived from fossil fuel combustion and from plant respiration (C3 plants dominate in the Kvarner Gulf) and the lack of data for year 2021 for January and February.

A direct comparison of δ¹³C values with other locations in Europe in 2021 is not possible because there are no numerical data available. However, graphical data from GML indicate that all three investigated Croatian sites have lower δ¹³C values, ranging from −13.1 ± 0.7‰ to –10.5 ± 0.3‰. For same year, data from HUN (Hegyhatsal, Hungary), and HPB (Hohenpeissenberg, Germany) vary between –8.0‰ during the summer and –9.5‰ during the winter (numerical data were not available, so data were read from the graph from NOAA – GML 2023, Vaughn et al. 2010). The δ¹³C data for JFJ station for 2017 has a mean value of –11.4 ± 0.7‰ (Levin and Hammer 2021). The difference in δ¹³C_CO2 from 2017 to 2021 at Mauna Loa, Hawaii, USA (Scripps CO₂ Program 2022) is only 0.06‰ (–8.54 ± 0.02‰ in 2017, –8.60 ± 0.02‰ in 2021) so we may approximate δ¹³C in 2021 for JFJ with the one from 2017. There is no significant difference between δ¹³C at all three investigated sites and approximate values from JFJ as z values between JFJ and Rijeka, Gornje Jelenje and Parg are 0.24, 0.11, and 0.01 respectively (Table SM6). Similar results of δ¹³C_CO2, from –18.0‰ to –6.4‰, with an average of –11.7‰ were obtained in the Velenje Basin (9 locations) in 2011 and were explained by the influence of the largest thermal power plant in Slovenia (Kanduč 2015).

CONCLUDING REMARKS

This paper presents the results of δ¹³C and Δ¹⁴C values of atmospheric CO₂ at three experimental sites over a one-year period. The aim was to distinguish CO₂ sources between the clean-air site Parg, the Gornje Jelenje site (in Risnjak National Park, in the vicinity of a main arterial road), and the port city of Rijeka, as Croatia’s tourism and seaport center of the northern Adriatic. The obtained results showed the trend of descending Δ¹⁴C_CO2 from Parg through Gornje Jelenje towards Rijeka (from 0 to –22‰). A similar trend in Δ¹⁴C values was observed in leaves samples collected at the same locations, with decrease from Parg to Rijeka (2 ± 3‰ to –25 ± 7‰, respectively). Based on the assumption that the investigated area is under the influence of three sources of CO₂ (fossil, sea/air exchange and biosphere) and that the difference between Parg and other the two sites is only due to fossil CO₂, the approximate share of fossil CO₂ in atmospheric CO₂ was determined for urban Rijeka site (2.1 ± 1.3%) and Gornje Jelenje (1.0 ± 0.9%) using a Parg as the background level.

Δ¹⁴C values showed a seasonal pattern with lower autumn/winter values and higher spring/summer values, especially at Parg. Seasonality is expected due to natural influences. The Parg summer Δ¹⁴C values reached rather high peaks, not observed in any other EU monitoring station. A possible explanation for the high summer and autumn values is CO₂ enriched with ¹⁴C transported from Adriatic Sea and released by soil respiration. The influence of anthropogenically enriched ¹⁴C sources cannot be totally excluded, due to the possibility of large air mass transportation from distant nuclear power plants.

There was no significant difference among mean δ¹³C_CO2 values: –11.4 ± 0.7‰ Parg, –11.5 ± 0.8‰ Gornje Jelenje and –11.6 ± 0.7‰ Rijeka. The values are generally quite low, and very similar to those measured at JFJ in 2017. δ¹³C values for the station nearest to Croatia, Hegyhatsal, Hungary and Hohenpeissenberg, Germany for 2021 are significantly higher and vary between –8.0‰ during the summer and –9.5‰ during the winter. δ¹³C values showed an increase in spring and summer due to photosynthetic discrimination of isotopically lighter CO₂ toward isotopically heavier CO₂. Decrease of δ¹³C during winter is possible due to a ¹³C depleted CO₂ transported from the sea by a strong wind, fossil fuel influence estimated from Δ¹⁴C and to soil respiration. However, differentiation of CO₂ sources on the basis of δ¹³C values is not as clear as for Δ¹⁴C values as fossil fuel signal is the same as the fresh C3 biomass signal.

The significant differences in δ¹³C between European clean-air site Jungfraujoch and Mauna Loa indicates that it is very important to compare climatologically similar sites, whenever possible.

At the moment, none of the aforementioned possibilities can be clearly distinguished due to the short investigation period. However, it was shown that, in even a such a small experimental area as this one, investigation of isotopic composition of CO₂ has great potential. For future research, clarification of sources of CO₂ in atmosphere other phases (DIC(aq) – CO₂(g)) of carbon cycling (e.g., vegetation, ocean) should be simultaneously traced. The impact of distant ¹⁴C sources from nuclear power plants should also be explored in more detail using transport models.