Seasonal and Diurnal Variations of Fluorescent Bioaerosol Concentration and Size Distribution in the Urban Environment

A recently introduced fluorescence based real-time bioaerosol instrument, BioScout, and an ultraviolet aerodynamic particle sizer (UVAPS) were used to study fluorescent bioaerosol particles (FBAP) in the Helsinki metropolitan area, Finland, during winter and summer. Two FBAP modes at 0.5–1.5 μm (fine) and 1.5–5 μm (coarse) were detected during the summer, whereas the fine mode dominated in the winter. The concentration and proportion of the coarse FBAP was high in summer (0.028 #/cm, 23%) and low in winter (0.010 #/cm, 6%). Snow cover and low biological activity were assumed to be the main reasons for the low coarse FBAP concentration in the wintertime. Both the fine and the coarse FBAP fraction typically increased at nighttime during the summer. Correlations between the BioScout and the UVAPS were high with the coarse (R = 0.83) and fine (R = 0.92) FBAP. The BioScout showed 2.6 and 9.7 times higher detection efficiencies for the coarse and fine FBAP, respectively, compared to the UVAPS. A long-range transport episode of particles from Eastern Europe increased the fine FBAP concentration by over two orders of magnitude compared to the clean period in the winter, but these FBAP probably also included fluorescent non-biological particles. Correlation analysis indicates that local combustion sources did not generate fluorescent non-biological particles that can disturb fine FBAP counting. The results provide information that can be used to estimate health risks and climatic relevance of bioaerosols in the urban environment.


INTRODUCTION
Bioaerosols such as bacteria and fungal spores can cause adverse health effects for people and animals both in indoor and outdoor environments (Burge and Rogers, 2000;Peccia et al., 2008;Mendell et al., 2011).Atmospheric bioaerosols, usually called primary biogenic aerosol particles (PBAP), consist mainly of bacteria, fungal spores and fragments, pollens, algae and plant debris (Després et al., 2012).PBAPs have been recognized to have important influence on climate, acting as cloud condensation nuclei (CCN) and ice nuclei (IN) and thus contributing cloud formation and precipitation processes (Vali et al., 1976;Bauer et al., 2002Bauer et al., , 2003;;Jaenicke, 2005;Sun and Ariya, 2006;Andreae and transport and abundance of PBAP (Burrows et al., 2009;Heald and Spracklen, 2009;Huffman et al., 2010).Laser induced fluorescence (LIF) based instruments are modern, easy-to-use tools for real-time bioaerosol detection (e.g., Hill et al., 1995;Pinnick et al., 1995;Ho, 2002;Jeys et al., 2007;Pöhlker et al., 2012).The LIF technique is an effective method for detecting biological molecules such as tryptophan, NADH and flavins that are present in microbe cells (Lakowicz, 2006;Hill et al., 2009).Saari et al. (2013) demonstrated that bacterial and fungal spores may be distinguished from each other through their dissimilar fluorescence spectra.The most well-known real-time LIF instrument is the ultraviolet aerodynamic particle sizer (UVAPS; Hairston et al., 1997, manufactured by TSI Inc., St. Paul, Minnesota) that is able to measure both the aerodynamic particle size and the autofluorescence of a single particle.Compared to the UVAPS, our recent study showed that performance of a simple, violet diode laser based bioaerosol detector, BioScout (Environics Oy, Finland), may be even more sensitive in measuring common bioaerosols (Saari et al., 2014).
Recently, LIF based real-time instruments have been used for detection of coarse (> 1.5 µm) PBAP in urban, suburban, desert, tropical rainforest, high-altitude and boreal forest environments (Pan et al. 2008;Gabey et al., 2010;Huffman et al., 2010;Gabey et al., 2011;Huffman et al., 2012;Toprak et al., 2012;Gabey et al., 2013, Huffman et al., 2013;Schumacher et al., 2013).Huffman et al. (2012) applied a UVAPS and a filter sampling technique to measure PBAP size distributions and concentrations in the Amazon rainforest region.They found a good correlation between fluorescent particles (UVAPS) and filter analysis (scanning electron microscopy as well as staining and light microscopy).However, they concluded that only the lower limit of PBAP concentration can be estimated with the UVAPS because some PBAPs cannot be detected due to weak fluorescence.This fluorescent fraction of bioaerosols is usually called fluorescent biological aerosol particles (FBAP).
Note that the fluorescent particle data can also include non-biological particles that have similar fluorescence characteristics, such as cigarette smoke or other combustiongenerated aerosols (e.g., Pinnick et al., 1998;Huffman et al., 2010;Gabey et al., 2011).For the FBAP measurement, these particles form an artifact that needs to be considered when analyzing the data.Huffman et al. (2010) found a high correlation of the number of fine fluorescent particles with the total number of particles, concluding that a large percentage of submicron particles exhibiting fluorescence may have anthropogenic sources.Generally, the combustion generated fluorescent particles are expected to be more abundant in the fine mode.Consequently, the fluorescent fine particles have not been much analyzed.When observed, they have been treated as non-biological particles (Huffman et al., 2010;Gabey et al., 2011).
In this study, we used two LIF based real-time bioaerosol instruments, a recently introduced BioScout and a UVAPS, to study FBAP concentrations and size distributions at urban and suburban residential sites in the Helsinki metropolitan area during winter and summer.To our knowledge, this is the first study wherein the BioScout was used in an outdoor environment and compared to another LIF based instrument.We also demonstrate that comparison between the realtime LIF data and particle mass (PM 2.5 ), black carbon (BC) and nitrogen oxides (NO x ), as well as meteorological data, enables estimation of bioaerosol sources and transportation.Contrary to the accepted custom, we especially extend the analysis also to the fluorescent fine particles.

BioScout
The BioScout (Environics Oy, Finland) uses a 405 nm continuous wave laser diode with 200 mW optical power to excite autofluorescence from individual bioaerosol particles.Both the autofluorescence and the scattering light are collected by an elliptical mirror and focused onto two photomultiplier tubes (PMTs).The autofluorescence is separated from the scattered light using a beam splitter and a long-pass filter with a cut point at 442 nm, and the fluorescence intensity is recorded by a PMT and sorted into 16 channels.The scattering light intensity is used to analyze the optical particle size.The time resolution of the instrument is 1 second.The particle size calibration of the BioScout was conducted against di-octyl sebacate (DOS) and NaCl particles using the UVAPS (TSI Model 3314) and the SCAR (Single Charged Aerosol Reference; Yli-Ojanperä et al., 2010).The operating particle size range was optimized between 0.3 and 5 µm.A laboratory study by Saari et al. (2014) showed that detection efficiency of the BioScout was high for fine bacterial particles (< 1 µm) that may also have a role in the atmospheric fine FBAP mode.Hairston et al. (1997) introduced the first prototype of the UVAPS.The current version of the UVAPS (TSI Model 3014) measures aerodynamic diameter of particles between 0.5 µm and 15 µm with 52 channels.The UVAPS also measures the autofluorescence emission of individual particles using a pulsed 355 nm UV laser with 80 mW maximum power as an excitation source.The autofluorescence is recorded between 430 and 580 nm by a PMT and sorted into 64 channels.In this study, the UV pulse energy and fluorescence PMT gain were set into their default values.The time resolution was adjusted to 5 min in this study.The performance of the UVAPS against common bioaerosols has been reported in several studies (e.g., Agranovski et al., 2003;Kanaani et al., 2007;Jung et al., 2012;Saari et al., 2014).

Measurements
Two measurement campaigns were conducted at urban and suburban residential sites in the Helsinki metropolitan area.The measurement campaigns took place in winter and summer during the time periods 2.2.-25.2.2012 and 16.6.-22.8.2012, respectively.The winter campaign was performed in a suburban residential area in Kattilalaakso, in Espoo.The summer measurements were placed in an urban residential area, Kallio, beside a green sporting field in Helsinki's downtown area.Both the measurement locations were influenced by local vegetation and human activity.Helsinki is located on the southern coast of Finland and thus the marine climate is strongly present there.The BioScout was used in both campaigns, but the UVAPS was present only during the summer campaign.The instruments were placed in the measurement station, and the aerosol sample was brought via total suspended particle inlet (TSP) on the top of the station during the summer period.In the winter, the BioScout was placed on the roof of the station and the sample was brought through its own inlet, which had a cut point at 10 µm.The sampling height was about 4 meters from the street level in both campaigns.
The supporting PM 2.5 , NO x and BC data was monitored at the measurement stations by the Helsinki Region Environmental Services Authority and Finnish Meteorological Institute.NO x was measured using a chemiluminescence analyzer (model APNA 360, Horiba).PM 2.5 was monitored with a tapered element oscillating system microbalance (TEOM model 1400 AB, Thermo Scientific; summer) or with a Synchronized Hybrid Ambient Real-time Particulate Monitor (SHARP model 5030, Thermo Scientific; winter).BC was measured using a Multiangle Absorption Photometer (MAAP model 5012, Thermo Scientific) with PM 1 cut-off.Meteorological data was observed by the Finnish Meteorological Institute at the Kumpula measurement site in Helsinki.

Data Analysis
The BioScout data were analyzed using MATLAB software.All particles in fluorescence channels 2-16 were classified as fluorescent and particles in channel 1 as nonfluorescent.Non-fluorescent NaCl particles were used to determine the detection limit for fluorescence.The detection limit was adjusted so that less than 1% of NaCl particles reached fluorescence channel 2 or above.The BioScout settings were similar as in the study by Saari et al. (2014).When the fluorescence data are combined with the optical particle size data, fluorescent particle and total particle size distributions and number concentrations can be calculated.The UVAPS data were analyzed using a similar procedure, and all particles in fluorescence channels 3-64 were classified as fluorescent.The UVAPS fluorescence channel 3 was chosen as a limit for fluorescent particles, similarly as in previous atmospheric studies by Huffman et al. (2010Huffman et al. ( , 2012Huffman et al. ( , 2013)), whereas fluorescence channel 2 was used in laboratory studies by Agranovski et al. (2003), Kanaani et al. (2007) and Saari et al. (2014).Pearson's linear correlation coefficients between the different instruments were calculated with one-hour time resolution.

Size Distributions
Averaged FBAP and total particle size distributions over the campaigns are shown in Fig. 1.There were typically two FBAP modes at 0.5-1.5 µm (fine) and 1.5-5 µm (coarse) during the summer.In winter, the fine mode was dominant, and no clear coarse mode was observed.The weather during the winter campaign was typically cold, the average temperature was −3°C, and the ground was covered by snow.We assumed that snow cover and low biological activity were the main reasons for the vanished coarse FBAP in the wintertime.Low FBAP concentrations measured by the UVAPS in wintertime have been reported also at Hyytiälä boreal forest site in Finland in the recent study by Schumacher et al. (2013).The coarse FBAP has been strongly linked to fungal spores that are abundant in this size range as reported also in the previous atmospheric FBAP studies by Huffman et al. (2010Huffman et al. ( , 2012Huffman et al. ( , 2013) ) and by Schumacher et al. (2013).
Origin of the fine FBAP is not clear, but the size range is typical for bacterial spores (Saari et al., 2014).The hypothesis of bacterial presence is also supported in a recent study by DeLeon-Rodriguez et al. (2013), wherein viable bacterial cells were shown to represent, on average, around 20% of the total particles in the 0.25-1 μm size range in the middle-to-upper troposphere.Similar observations were made during three measurement campaigns at Lake Baikal, at an urban site in Mainz and over the South Atlantic Ocean, where around 20% of total particles (D p > 0.2 µm) were detected as PBAPs (Matthias-Maser et al., 1995, 1999, 2000).They suggested that bacterial particle concentration was higher in the urban environment due to the anthropogenic sources.The fine FBAP has also been supposed to come from combustion sources in urban environments (Huffman et al., 2010).This possibility is discussed more below.However, our laboratory studies showed that single bacterial spores are detectable with both the instruments (Saari et al., 2014), so we assume that at least a fraction of the observed fine FBAP are bacteria.
The size distributions measured by the BioScout and the UVAPS were similar but not fully consistent because the UVAPS measures particle aerodynamic size and the BioScout measures particle optical size that depends on a refractive index and absorption of particles.There were also differences in fluorescence sensitivity between the instruments.The fine mode was clearly weaker and barely visible with the UVAPS.The modes have been separated at a cut-off point of 1.5 µm, and concentrations of the fine mode (0.5 < D p < 1.5 µm) and the coarse mode (D p > 1.5 µm) will be analyzed separately below.

Fluorescent Particle Concentrations and Fractions
To compare the response of the instruments, correlation diagrams and averaged values of the total particle and FBAP concentrations measured by the BioScout and UVAPS during the summer period are shown in Fig. 2 and Table 1.Pearson's linear correlation coefficient (R) showed high correlation with both total particles (0.88 and 0.89 for the coarse and fine mode, respectively) and FBAP (0.83 for the coarse mode and 0.92 for the fine mode) between the instruments.Linear fit functions showed slope values close to one for total particles (0.76 and 1.02 for coarse and fine mode, respectively), indicating that concentrations are consistent between the instruments despite the different particle size measurements However, remarkably higher slope values were observed for FBAP (2.6 and 9.7 for coarse and fine mode, respectively).This suggests that the BioScout had 2.6 and 9.7 times higher detection efficiencies for the coarse and fine FBAP, respectively, compared to the UVAPS.Similar results were found also in our laboratory studies (Saari et al., 2014).The reason for the higher detection efficiency of the BioScout is assumed to be high excitation laser intensity and the good fluorescence signal to noise ratio of the instrument.
Averaged coarse FBAP concentrations were 0.010 #/cm 3 in the winter and 0.028 #/cm 3 in the summer as measured by the BioScout.In comparison, the UVAPS coarse FBAP concentration was 0.013 #/cm 3 in the summer period.The BioScout coarse FBAP fraction (FPF) was high in the summer (23%) and low in the winter (6%).The UVAPS coarse mode FPF was 8% in the summer.Surprisingly, the BioScout fine FBAP concentration was higher in the winter (0.13 #/cm 3 ) than in the summer (0.018 #/cm 3 ), whereas the fine FPF were lower in the winter (0.9%) than in the summer (2.9%).The UVAPS fine FBAP concentration was 0.0025 #/cm 3 , and the FPF value was 0.31% in the summer.The BioScout showed higher FPF value for both the coarse and the fine FBAP compared to the UVAPS, which is in line with the laboratory studies of the common fungal spores and bacteria (Saari et al., 2014).
A comprehensive comparison of the averaged PBAP concentrations and fractions between this study and the previous studies is represented in Table 1.All the fluorescence based studies of PBAP coarse mode (measured by BioScout, UVAPS and WIBS) in urban environments are well in line: Concentrations vary from 0.01 #/cm 3 to 0.10 #/cm 3 , and fractions were between 3% and 23% (this study; Huffman et al., 2010;Gabey et al., 2011;Toprak et al., 2012).Interestingly, the results of the off-line staining method showed similar values for coarse PBAP concentration, ranging from 0.026 #/cm 3 to 0.8 #/cm 3 (Chi and Li, 2007;Bauer et al., 2008).

Temporal Variations
In this section, the FBAP concentrations are analyzed as a function of time and compared to PM 2.5 , NO x and BC concentrations.Concentrations of the coarse and fine FBAP and their FPF values measured by the BioScout as well as PM 2.5 , NO x and BC concentrations in the winter are shown in Fig. 3. Based on the PM 2.5 data from other measurement stations of the Helsinki Region Environmental Services Authority, high concentrations during February 15-19, 2012, were caused by the long-range transportation of aerosol particles, due to prevailing meteorological conditions (Niemi et al., 2009).Back-trajectory analysis of air masses (HYSPLIT Model; Draxler and Rolph, 2014;Rolph, 2014) showed that air flows arrived from Eastern Europe during the LRT episode, and clean air during February 20-28 was coming from the direction of the Atlantic Ocean and Greenland.During the LRT episode, concentration of the coarse mode was only elevated a bit, but the fine FBAP concentration was over two orders of magnitude higher compared to the clean period.PM 2.5 and BC concentrations during the LRT episode were high.The correlation analysis between the variables is discussed more below.
The fine FBAP during the LRT may have been influenced by several anthropogenic sources, including the biomass burning in fireplaces that is popular in Eastern Europe during the cold season.However, bacteria emissions would be also higher in Eastern Europe than in Finland due to temperature and snow cover differences during the winter Smith et al. (2013) reported high concentrations of bacteria taxa in the transpacific plumes from Asian to North.America and showed that long-range transport is possible for bacteria particles.4. Both the coarse and fine FPF were typically increased at night time, which is reported to be a characteristic time for the natural emission of fungal spores (e.g., Huffman et al., 2012).Sometimes, e.g., on June 26, the concentrations were high also in the daytime, which indicates the presence of different sources.The fine FBAP showed some short peaks at nighttime during June 27-29, which indicates the presence of a strong local source.The source may be either natural, such as bacteria, or anthropogenic.Similar trends during the measurement period were shown with both the BioScout and with the UVAPS.The FBAP concentrations were lower with the UVAPS than with the BioScout, and the difference was especially high for the fine mode.This comes from the different bioaerosol detection efficiencies between the instruments, which are also discussed above.

Correlation Analysis
Correlations between the FBAP and total particles measured by the BioScout as well as PM 2.5 , NO x and BC were analyzed in Fig. 5.To our knowledge, this is the first study wherein FBAP concentrations were compared to the PM 2.5 , NO x and BC values.In the winter, correlations were shown separately during the LRT episode and the clean period.
During the LRT, the fine FBAP correlated well with the total particle concentration (R = 0.93), whereas correlation was low between the coarse particles (R = 0.35).Similar results were reported in the previous study by Huffman et al. (2010).High correlations, R coarse = 0.71 and R fine = 0.73, were found between the FBAP and the PM 2.5 during the LRT, indicating high FBAP content in the transported PM 2.5 fraction.Correlations between BC and FBAP were the highest during the LRT episode (R coarse = 0.60 and R fine = 0.65).This indicates that aged BC-containing particles may have influenced the FBAP values acting as nonbiological fluorescent particles.Condensation and cloud processes can increase the size of BC-containing particles in the atmosphere, and therefore, BC also can present in larger particle sizes in LRT aerosols (Niemi et al., 2006).
Correlations between FBAP and total particles were low with both modes (R coarse = 0.15 and R fine = 0.29) during the clean period in winter, indicating random emission of FBAP.Interestingly, coarse particles correlated better in the summer (R coarse = 0.61 and R fine = 0.10).This indicates regular local emission of coarse FBAP (presumably fungal spores) and random fine FBAP emission in summertime.
The results showed that correlations between the FBAP and NO x were low during the all measurement periods (0.01 < R coarse < 0.54 and 0 < R fine < 0.07).This indicates that local vehicle traffic, the dominant NO x source in the Helsinki metropolitan area, did not disturb FBAP counting at these measurement sites.Fine FBAP correlations with BC were low during the clean period in the winter (R = 0.32) and in the summer campaign (R = 0.09).This indicates that local BC related combustion sources did not disturb the fine FBAP analysis during these periods and that most of the fine FBAPs counted by the BioScout were real bioaerosols.The previous studies typically treated fine FBAP as nonbiological particles (Huffman et al., 2010;Gabey et al., 2011), but this study showed that this is not necessarily the case.The types and sources of fine FBAP should be studied more and compared with other analyzing techniques such as microscopy and quantitative polymerase chain reaction (qPCR).

CONCLUSIONS
Two fluorescence based real-time instruments were used to measure bioaerosol concentrations and size distributions in the two different case studies in urban environments during winter and summer.Two FBAP modes at 0.5-1.5 µm (fine) and 1.5-5 µm (coarse) were detected during the summer, whereas the fine mode dominated in the winter.Correlations between the BioScout and the UVAPS were high with the FBAP coarse mode (R = 0.83) and fine mode (R = 0.92).The BioScout showed 2.6 and 9.7 times higher detection efficiencies for the coarse and fine FBAP, respectively, compared to the UVAPS.This is the first study, wherein the BioScout was used in outdoor environment and compared to another LIF based instrument.
The coarse FBAP concentrations and fractions were the highest in summer (0.028 #/cm 3 , 23%) and the lowest in winter (0.010 #/cm 3 , 6%).Snow cover and low biological activity were assumed to be the main reasons for low coarse FBAP concentration in the wintertime.Both the fine and the coarse FBAP fraction typically increased at night during the summer; that is characteristic for natural fungal spore emission.Various factors suggest that at least a fraction of the observed fine FBAP were bacteria.
The long-range transport episode of particles from Eastern Europe increased the fine FBAP concentration over two orders of magnitude compared to the clean period in the winter, but these FBAP probably also included fluorescent non-biological particles.The potential sources of fluorescent fine particles during the long-range transport episode might be bacteria and/or anthropogenic emissions such as biomass burning.Low correlations between FBAP and NO x and BC indicate that local traffic emissions and biomass burning are not considerable sources for fluorescent particles at the measurement sites.The previous studies typically treated fine FBAP as non-biological particles, but this study showed that this is not necessarily the case, and fine FBAP should be studied more and compared with other analyzing techniques such as microscopy and qPCR.The results provide information that can be used to estimate health risks and climatic relevance of bioaerosols in urban environments.

Fig. 1 .
Fig. 1.Averaged fluorescent (left) and total (right) particle size distributions during the winter and summer periods measured by the BioScout and the UVAPS.The cut-off point between the modes is shown at 1.5 µm.

Fig. 2 .
Fig. 2. Correlation diagrams of the total (N TOT ; left) and fluorescent particle (N FL ; right) concentrations measured by the BioScout and UVAPS during the summer period.Black circles represent the coarse mode (D p > 1.5 µm), and grey triangles represent the fine mode (D p < 1.5 µm).Pearson's linear correlation coefficients (R) and linear fit functions (y = kx) are shown in diagrams.

Fig. 3 .
Fig. 3. Concentrations of PM 2.5 , NO x , BC and fluorescent coarse (solid) and fine (dashed) modes and their fractions (N FL /N TOT ) measured by the BioScout during the long-range transport episode (LRT; February 15-19) and clean condition in the winter of 2012.

Fig. 4 .
Fig. 4. Time series of fluorescent particle concentrations and averaged daily variations of the fluorescent particle fractions (N FL /N TOT ) for the coarse mode (D p > 1.5 µm; above) and fine mode (D p < 1.5 µm; below) measured by the BioScout and the UVAPS during the summer campaign of 2012.

Fig. 5 .
Fig. 5. Correlation scatter diagrams of fluorescent (N FL ) coarse (D p > 1.5 µm) and fine (D p < 1.5 µm) particle and total particle concentrations (N TOT ) measured by the BioScout, PM 2.5 , NO x and BC during the winter and the summer periods.Pearson's correlation coefficients (R) are shown in diagrams.

Table 1 .
The comparison between the results of this study and the previous studies.Coarse and fine PBAP represent supermicrometer and submicrometer bioaerosol particles, respectively.PBAP concentrations are shown as (#/cm 3 ) and fractions as (%).