Interannual Variability of BVOC Emissions in an Alpine City

Terpenoid emissions above urban areas are a complex mix of biogenic and anthropogenic 11 emission sources. In line with previous studies we found that summertime terpenoid emissions in an 12 alpine city were dominated by biogenic sources, but especially at lower temperatures the anthropogenic 13 influences were non-negligible. Inter-seasonal emission measurements revealed consistency for 14 monoterpenes and sesquiterpenes, but a large difference in isoprene between the summers 2015 and 15 2018. Standardized emission potentials for monoterpenes and sesquiterpenes were 0.12 nmol m -2 s -1 and 16 3.0·10 -3 nmol m -2 s -1 in 2015 and 0.11 nmol m -2 s -1 and 3.4·10 -3 nmol m -2 s -1 in 2018, respectively. 17 Observed isoprene emissions were about four times higher in 2018 than in 2015. This factor decreased 18 to 2.3 after standardizing isoprene emissions to 30°C air temperature and photosynthetic active radiation 19 (PAR) of 1000 μmol m -2 s -1 . Based on emission model parameterizations, increased leaf temperatures 20 can explained ~50% of these differences, but standard emission potentials remained higher in 2018, 21 when a heat wave persisted. Potential other reasons for the differences such as emission 22 parameterization, footprint changes, water stress conditions and tree trimming are investigated. 23


Introduction 24
Biogenic and anthropogenic volatile organic compounds (BVOCs, AVOCs) in the atmosphere can 25 contribute to surface air pollution both due to their influence on tropospheric ozone formation and due 26 to their potential to act as precursors for secondary organic aerosol formation (Derwent et al., 1996, 27 Fehsenfeld strength is estimated to be 10 times larger than AVOCs , Piccot et al., 1992. Also, 30 many BVOCs are characterized as highly reactive (Atkinson andArey, 2003, Fuentes et al., 2000), 31 resulting in rapid peroxy radical chemistry important for ozone and ultra-fine particle formation 32 processes (Simon et al., 2020). Of the total global BVOC emissions, terpenes dominate, with 50% 33 attributed to isoprene, 15% to monoterpenes and about 0.5% to sesquiterpenes . 34 In forests with predominating isoprene emissions, isoprene was found responsible for 50-100% of the showed that in a tropical-subtropical metropolis biogenic contributions overwhelmed anthropogenic 56 contributions of isoprene in summer and that biogenic sources started to dominate in all seasons above a 57 threshold temperature of 17-21°C. Whereas all so far cited studies were based on concentration 58 measurements where the influence can be both local and regional and strongly modulated by 59 atmospheric dilution the following studies were based on eddy covariance flux tower sites: At 60 temperatures over 25°C more than 50% of the isoprene flux was found to be biogenic in origin in 61 London h -1 . Rantala et al. (2016) found that 80% of the measured 10 ng m -2 s -1 summer daytime isoprene flux 66 near Helsinki could be contributed to biogenic sources by comparing emissions at low and high 67 temperatures. 68 While there is evidence for urban trees to have positive influence on urban environments such as 69 mitigating the urban heat island effect, sequestering CO 2 and particles as well as acting as storm water 70 interception ( Figure 1A on a 117 1000x1000m map surrounding the site. This will in the following be referred to as the study area.

Eddy covariance fluxes 136
This study focuses on biogenic emissions collected during summer 2015 and summer 2018. As biogenic 137 emissions are strongly light-and temperature-driven, the total available dataset was reduced to daytime 138 hours (9:00-16:00 local time) and mean wind directions from 0°-120°. Data with wind direction from 139 the south and exceeding a wind speed > 10 m/s were excluded as they can be attributed to foehn events. 140 Eddy covariance fluxes were calculated using a MATLAB ® code described by Striednig  and are temperature-and light-dependent coefficients respectively containing current and past 150 (24h and 240h) conditions. In order to investigate relative inter-seasonal changes of isoprene emissions 151 the MEGAN 5-layer canopy model (Guenther et al., 2006) was applied. 152 Monoterpene and sesquiterpene eddy covariance fluxes are known to be purely temperature dependent 153 and can be described as: = 0, * , and = 0, * , where is a temperature 154 dependent factor described by Guenther et al. (1994). 155 In the case of sesquiterpenes, turbulent time scales (100 s) are on the order of chemical time scales as 156 sesquiterpenes can react fast with ozone (typical ozone concentrations were 30 ppbv). The chemical loss 157 was estimated by the following equation: c(t)/c0=exp(-t turb /t chem ) where t turb is the turbulent time scale 158 and t chem the chemical time scale. 159

City tree inventory 160
An inventory of all trees planted by the city municipality is available for the city of Innsbruck, Austria 161 containing location, tree species, diameter at breast height and height. However, this inventory does not 162 include trees from private gardens. Therefore, all accessible trees from private gardens were identified 163 and added to the existing tree inventory. The location of the trees from the city inventory and private 164 gardens in the study area are shown in Figure 1A. Within the study area a total of 1904 registered trees 165 distributed across 129 tree species were counted. A list of the 44 most abundant tree species, where the 166 species count in the study area was 6 or more, is given in Table 1. 167

Emission potentials 168
Literature values of plant-species specific emission potentials of isoprene and monoterpene, in μg 169 compound g -1 dry-weight h -1 standardized to 303.15 K and PAR 1000 μmol m -2 s -1 were assigned to the 170 44 most abundant species in the study area. This includes all tree species with an occurrence larger than 171 6 individuals within the footprint and accounts for ~90% of the total counted trees. Emission potential 172 assignment was based on the detailed work by Stewart et al. (2003) for the overlapping species. Other 173 emission potentials were taken from other literature and if more than one literature value was available, 174 an average was taken. All species, emission potentials and references thereof are shown in Table 1. the average value of 0.1 μg compound g -1 dry-weight h -1 was assigned. 177

Relative IS, MT, SQT emission ratio maps 178
To generate emission ratio maps, the study area was divided into a 100 m by 100 m grid and tree 179 species were counted in each grid tile and multiplied by their emission potential listed in Table 1. The  180 resulting map in units of μg compound g -1 dry-weight h -1 neglects the actual, but unknown, amount of 181 dry leaf weight of each individual tree. 182

Flux footprint, light & temperature conditions 191
The flux footprint of the daytime data at the IAO is shown in Figure 1A (Table 1)

Isoprene flux anomaly 279
The isoprene flux difference measured between the two summers of 2015 and 2018 is shown in Figure 2  280 A and D. Figure 2D represents the isoprene dependence on temperature. (Guenther et al. 1993). Monson 281 and Fall (1989) found isoprene fluxes not only to depend on current temperature conditions, but also on 282 current light conditions. In addition, past 24h and 240h temperature and light conditions play a role (e.g. 283 Guenther et al., 2006). These theoretical temperature and light parameters are plotted vs. the observed 284 isoprene flux in Figure 3A based on the MEGAN big leaf approach (Guenther et al., 2006). Even after 285 including both actual and past temperature and light parameters the difference in isoprene emissions 286 between the two summers could not be resolved and standardized emission factors were still a factor 2.4 287 higher in 2018 than in 2015. Figure 3B shows that the difference was increasing with higher 288 temperature and higher PAR values.

297
In contrast to monoterpene and sesquiterpene emissions, which exhibited comparable emission 298 potentials between the two years and are mainly driven by evaporative emissions from storage 299 reservoirs (e.g. Kesselmeier and Staudt, 1999), it remains a puzzle why the isoprene emission potential 300 was substantially higher in 2018 compared to 2015. As neither actual temperature and light 301 dependencies nor 24h and 240h past temperature and light could explain the observed differences in 302 isoprene fluxes, we investigated the following potential reasons: a) variation in the flux footprint, b) tree 303 trimming, c) water availability/drought, and d) emission parameterization. 304 a) As shown in Figure 1A  Innsbruck: 6 (Karl et al., 2020)) are generally very low (e.g. 5-6 times lower) when compared to 334 measurements over purely vegetated surfaces, and therefore notoriously difficult to interpret. As 335 such we cannot exclude the possibility of processes other than evapotranspiration from city trees 336 contributing to higher water fluxes observed in 2018. An obvious explanation is that a 337 significant water runoff during extensive watering operations resulted in increased evaporation 338 over hot asphalt and other non-vegetated surfaces, leading to higher water fluxes in 2018. Water 339 was also applied to asphalt surfaces more frequently during mornings to minimize the effect of 340 urban aerosol pollution. 341 It is well established that isoprene production in plants can decouple from photosynthesis during 342 periods of drought and can be sustained by alternative metabolic carbon sources (e.g. Bertin & 343 Staudt emissions during drought is not fully unraveled, but has been suggested to represent a response 346 for coping with heat stress (Loreto et al., 1998). Isoprene fluxes were observed to increase 347 during the very early onset of drought conditions. For example, Seco et al. (2015) reported an 348 increase in the ecosystem scale isoprene emission potential about one month before significant 349 changes in pre-dawn leaf water potential were observed, but when CO 2 uptake was already 350 decreasing. Additionally, they observed that the closing of stomata had a bigger effect on CO 2 351 than water fluxes, because gradual increases of vapor pressure deficit during the evening offset 352 reduced leaf conductance. In contrast to isoprene, temperature dependence algorithms accurately predict increases of 376 monoterpenes and sesquiterpenes due to increased temperatures in 2018 compared to 2015 (Fig. 2. B,  377 C). This can be rationalized by the fact that monoterpenes and sesquiterpenes are primarily released 378 from storage pools (Kesselmeier and Staudt, 1999), and temperature dependent changes in vapor 379 pressures of these compounds are accurately predicted between 2015 and 2018.  Branch-level standardized emissions are collected from the literature in Table 1 Table 1

Conflict of Interest 468
There is no conflict of interest.