The Spatial Distribution Assessment of Particulate Matter by Biomagnetic Monitoring Using Phoenix dactylifera Leaf Samples and Azimuthal Dust Samplers in Kuwait

This study employs multi-magnetic parametric methods as proxies to measure particulate matter (PM) concentration and spread in Kuwait. It examines the reliability of biomonitoring receptors in the assessment of atmospheric air quality through the utilization of passive biomonitoring methodology using cleaned and non-cleaned Phoenix dactylifera leaves and active biomonitoring through the application of dust samplers in the study area. Four radial sampling areas are located at 2, 6, 10, 14 km from Kuwait’s city center with 10 sampling degree points selected from each radial area, and the closest palm tree in the vicinity to the preselected sampling point with a height of 4 m were sampled. Using a compass, the 4 azithumal points were pin pointed on the selected tree and a 2 × 2 cm dust sampler was attached to each direction at a height of 2 m. The dust sampler was made of clear plastic paper attached with double sided tape. Magnetic susceptibility and Saturation Isothermic Remanent Magnetization (SIRM), Natural Remanent Magnetization (NRM), Hard Isothermal Remanent Magnetization (HIRM), Soft Isothermal Remanent Magnetization (SOFT), HIRM%, soft IRM% and s-ratio were determined for P. dactylifera and dust samplers. Magnetic parameters were mapped to assess the spatial variation of air quality in Kuwait and the values between dust samplers and P. dactylifera. Results indicate that the highest magnetic concentration values for NRM and SIRM for P. dactylifera occurred near Kuwait bay and that the majority of the samples contain ferromagnetic minerals with magnetite most likely from anthropogenic sources. The results of the interpolation models for P. dactylifera and dust samplers as well as the overall mean for dust samplers distinguished short-term PM deposition and concentration and how it is impacted by wind direction in comparison to P. dactylifera which identifies long-term pollution impacts pin pointing PM sources and hotspots.


Introduction
Particulate matter (PM) is a detrimental threat to the environment and the health of residents in urban areas. It is associated with a variety of respiratory and cardiovascular diseases such as: asthma, lung cancer, heart attacks and respiratory infections [1] [2]. Ambient airborne particulate matter (PM) is defined by Chen et al. (2016) as airborne particles which are grouped as coarse, fine, and ultrafine particles (UFPs) with aerodynamic diameters within 2.5 to 10 μm (PM 10 ), and <2.5 μm (PM 2.5 ), and <0.1 μm (PM 0.1 ), respectively. High concentrations of PM are associated with industrial emissions, roadways, construction, road dust, traffic emissions and its concentrations are influenced by humidity, wind speed, direction and turbulence [3] [4] [5]. Wind events can mobilize and transport surficial soils over a range of distances depending on their mass, with large particles travelling at short distances and having a higher rate of sedimentation in comparison to small particles [6]. These airborne PM derived from surficial soils can be traced to its source or point of origin as it has a geochemical signature and contaminant loadings from the original source material [6].
The highest amounts of dust emissions are Australia, Arabia, China and Central Asia, and the Sahara regions [7]. The Arabian Peninsula generates approximately 11% of the global airborne mineral dust mass [8]. Therefore, environmental and health risks are higher in Kuwait especially as it is an arid, semi-arid region where uptake and erosion/leaching processes are less pronounced and frequent sand and dust storms are encountered [6]. The number of dusty days and daily PM10 concentration are rising further as a result of climate change, desertification, urbanization land use change and anthropogenic influences [6] [9] [10] [11]. Dust storms occur as a result of turbulent winds raising large quantities of dust from surfaces, that can reach up to 6000 µgm −3 in severe events [10] [12]. Aerosols, particulates and pollutants are easily transferred across the peninsula by dust storms, chemically made of SiO 2 (60%), Al 2 O 3 (14%), Fe 2 O 3 (7%), CaO (4%), MgO (2.6%) and K 2 O (2.4%) and are mineralogically dominated by quartz, magnetite/hematite and carbonates [9] [13].
PM source appointment in Kuwait presents a challenge as it is a small country neighboring several countries that are also engaged in the petroleum and gas industry and chemical manufacturing that are possibly contributing sources of  [14]. There are several studies that have tackled the topic of air quality in Kuwait covering; source appointment and mineralogical compositions of dust [14] [15]; PTE contamination and enrichment in soil [6], spatial and seasonal variations of the PM10 [16]; vegetation and soil conditions [17] and dust fallout [18] [19]. These traditional studies on characterization of suspended PM and assessment of air quality in Kuwait dust have been primarily based on geochemical and mineralogical methods, while magnetic properties of suspended dust have yet to be examined.
As well as the majority of pollution exposure data is often sourced from low partial-resolution networks of monitoring stations that are incapable of capturing fine scale variations in PM concentration and/or particle size across diverse urban environment at pedestrian exposure height [20]. The geochemical and magnetic properties of transported materials, and their inter-relationships may be affected due to changes in the composition of transported materials from near-surface wind fields hindering the identification of the source material [21].
The application of environmental magnetism as a proxy for atmospheric pollution levels enables the identification of magnetic components, the determination of PM transport, daily atmospheric PM concentration, relative contribution at a given site, atmospheric heavy metals provenance, their distribution and localization in addition to being proxy underlying environmental processes [22] [23] [24]. Biomonitoring of air quality through the utilization of plant leaves and bark has been proven to be more economical and time effective at larger and more diverse spatial scale set-ups, especially in unforeseen releases and accidental situations [25] [26]. Plants have the capacity to accumulate PM, they capture PM through dry and wet deposition onto leaf surface areas by gravity, diffusion and turbulent transfer giving rise to impaction and interception and root uptake [27] [28] [29]. Though, PM capture is affected by specific species characteristics such as: trichome density, leaf surface size, leaf hydrophobicity, chemical composition and structures of epicuticular waxes and plant height [30].
Two types of biomonitoring passive and active biomonitoring can be applied.

Study Area
Kuwait is a small country covering an area of 17,818 km 2 Figure 1). Sampling occurred from the 26 th January to the 8 th February 2019. A wind rose indicating wind speed and direction for the sampling duration ( Figure 2).

Passive Biomonitoring
The species sampled in this study is P. dactylifera, it was selected as it is widely used and distributed as an ornamental street tree or in parks in Kuwait. Tree species with a similar height of 4 m were selected, to ensure that they have simi-

Active Biomonitoring
Dust was collected using a sampler that was made using a clear plastic paper, cut

Magnetic Measurements
Magnetic measurements were conducted at the laboratory of the Departments of Bioscience Engineering of the University of Antwerp, Belgium. Dried leaf samples and dust samplers were first wrapped in A4 plastic cling wrap and placed in a 6.7 cm 3 plastic cube container. Initially the magnetic susceptibility of the samples was measured at room temperature using a Barrington MS2B magnetic susceptibility magnetometer (Barrington Ltd., U.K.). Samples were first tested for low frequency (χ LF ) and then high frequency (χ H ) enabling frequency dependent susceptibility χ FD (%) to be calculated. However, χ FD (%) could not be calculated for dust sampler as readings were too low for χ LF .
Samples were then measured for natural remnant magnetization (NRM).

Spatial Distribution Mapping
The spatial variation of magnetic parameters and difference between azithumal directions for leaf and dust samplers were illustrated using geostatistical interpolation using Kriging, with a power of 2 to indicate the variance of magnetic concentration and grain size based on the distance on the direction and distance of two sampling locations, using ArcGIS (10.6.1) (Environmental System Research Institute (ESRI), Redlands, Canada). It models the trend as a linear function with independent external variables it derives the local mean (trend) of the dependent magnetic variables enabling the assessment of values at unobserved locations [39]. Kriging produces unbiased predictions with minimal variance based on the principle that variables are regionalized and utilizes the spatial structure of the data [40]. The overall mean of dust sampler of each magnetic parameter was inputted to compare passive biomonitoring techniques to active biomonitoring.

Active and Passive Biomonitoring
P. dactylifera samples for (W) values were lower than (NW) values for NRM, SIRM and s-ratio. There was correlation between NRM and radius for P. dactylifera (p = 0.001). Overall radial distances there was a difference between mean s-ratio, SIRM and NRM for P. dactylifera (W) and P. dactylifera (NW) was 0.  The average s-ratio for dust samplers for overall radial distances is 0.96, with s-ratio increasing as distance increases (p = 0.005). Overall NRM for dust samplers       (Figure 4).

Correlation between Magnetic Parameters and Substrates
For dust sampler radial distance, sampling degree and azithumal direction ac-  Table 1). The first factor accounted for 22.6% of the variance relative to the total variance in all the variables. The second factor 17.6%, the third factor 15.6%, the fourth factor 14.9% and the fifth factor 7.9% (Table 1).
Finally, the rotated factor matrix provided partial correlations between magnetic parameters and sampling material that are necessary to interpret the factors (

Distribution Pattern and Source Identification through Mapping
The predicted surfaces spatial distribution of magnetic parameters and inter-parametric ratios indicating the concentration, mineralogy, and grain size of magnetic minerals present in Kuwait for the duration of the study is illustrated in Figures 9-15. There were some similarities in the spatial distribution patterns        [43]. Regardless, the standardization of (W) P. dactylifera in comparison to dust sampler will vary as some concentrations of elements in plant tissues can be technically affected by; the soil or substrate they are growing on, interspecies differences, vegetation effects, altitude and proximity to the source [41]. The washing procedure offers one great advantage which is the elimination of soil-relation effect [42]. Aboal [49]. Low HIRM% values for (NW) P. dactylifera and dust samplers average at 3.77% and 6.57%, an indication of fairly low concentration of hematite/goethite. This data is comparable to modern African soil samples from Niger, Morocco and Tunisia sampled as potential dust sources that displayed a HIRM 100% between 8% and 32% [46] [59] [60]. This implies that low coercivity multidomain (MD) ferromagnetic minerals (magnetite) dominate the magnetic properties of PM, while incomplete anti-ferromagnetic minerals make up a low proportion [61]. Natural sources are reflected through hard IRM%, this is possibly due to the formation of hematite favoring drier climates with low water content and high temperatures owing to the dehydration ferrihydrates [62] [63] [64]. The presence of ferrimagnetic mineral components dominating the samples is further confirmed as s-ratios were close to 1 reaching almost to unity over all samples, as low s-ratio is identified as haematite, high s-ratio is greigite [65]. A relationship between radial distance and NRM for P. dactylifera was observed, with NRM values dropping as radial distance increased, moving away from the highly urbanized city center and the coast towards more suburban, industrial areas. The relationship between magnetic intensity and land use corres-

Azithumal Direction
The effect of azithumal direction was tested using the dust samplers to test the effect of azithumal sampling degree at each sampling location impacted magnet-  [67]. Dust events in the Middle East are determined to have relatively weak wind speeds that can reach 10 ms causing dust-in-suspension to predominantly to occur [68]. These findings concede with Hofman et al. (2013) results in the assessment of the spatial distribution of particulate matter of deposited particles on tree crowns in an urban street canyon that determined that air circulation caused azimuthal effects and that it is directly influenced by street architecture, as individual tree crowns indicated an azimuthal leaf SIRM variation (p = 0.0446) depending on their position in a street canyon [69]. Hofman et al. (2013) found that all edge trees that were exposed to intense air circulation were determined to have the lowest SIRM values at the windward side of the tree crown (NW-SW), whilst the highest leaf SIRM values were at the leeward sides of the tree crown (NE, SE) as changing airflows increase of impaction and interception of particles causing azimuthal SIRM variations [69]. This is further confirmed by a study conducted by Zhang et al. (2008) on Chinese willow (Salix matsudana) tree ring cores, that suggests that magnetic particles enter into the xylem during the growing season as tree ring cores facing the pollution source displayed higher SIRM than other sides of the tree [70]. It must be noted further research needs to conduct on the comparison of azithumal effects between P. dactylifera (active) and dust sampler (passive) biomonitoring effects over a long duration of time to see if the amount of deposited magnetic particles differed at similar azithumal degrees or whether they reach equilibrium.

Distribution Pattern and Source Identification through Mapping
Interpolation of data has enabled more insight into the spatial distribution and behaviors of pollutants in the area as it included all input points including the outliers. However the magnetic maps may not be true spatial representatives of the local concentrations as previous studies that have used kriging to map PM have observed that by mapping large geographical areas with large distances between monitoring sites to have reduced reliability in the predicted values [71].
As seen on the maps there were some similarities in the spatial distribution's trends of soft IRM, HIRM, hard IRM% and s-ratio between active and passive biomonitoring methods and the overall mean of dust samplers. Passive biomonitoring offers more information on long term deposition rates and pinpoints hot spots and PM sources in comparison to active biomonitoring that offers more short-term PM deposition accumulation rate and behaviors in relation to meteorological conditions during the sampling time [54]. The azithumal direction in which the dust samplers were placed gave distinct results this may be a result of varying infrastructure across the area as well as building height and wind direction (Figures 11-13). The magnetic outcomes for dust samplers follow a similar pattern to the wind trajectories. The highest concentration appeared in Kuwait City as well near the vicinity of industrial areas, ports and power plants and desalination plants and their surrounding area. As industrial hotspots and dense residential areas with high vehicular traffic tend to be exposed to higher PM 2.5 concentrations [72]. The prevailing wind direction during the sampling was northwest, the decrease in SIRM and NRM values and spread in maps can be attributed to the prevailing wind direction especially when comparing dust samplers to P. dactylifera (Figure 9 & Figure 10). The transportation, dispersal and dilution of air pollution are actively impacted by local winds, high amounts of air pollutants particularly from pointary or linear sources that produce high proportions emissions, to receptors or samplers at distinct wind directions [67]. Al-Nassar et al. (2005) determined that wind power density (WPD), that is defined as the frequency distribution of wind speed and the dependence of wind power on air density and the cube of the wind speed, expressed in Watt per square meter (W/m 2 ), in Kuwait is lower at the bay and increases outwards into open flat desert areas in the northern, northwestern and southern parts of the country, such as Al-Wafra, Al-Taweel and Umm Omara, indicating that wind direction has a significant effect on daily magnetic mineralogizes and grain size [73]. Prevailing wind direction is seen to impact, magnetite concentration to a range of 0.24% -0.89% at distance of 500 m away from cement plants, while decreasing to a concentration of 0.02% to 0.16% at a distance of 3 km or more [74]. As well as parallel MS profiles measured perpendicular to road surfaces at a distance of 20 m on either side of a road have been determined to be a reflection of clear symmetry to prevailing winds [75]. A study on pollution utilizing roadside plant leaves in an Indo-Burma hot spot region found that plants showed higher magnetic concentrations in sites with poor roads that experience heavy traffic, street dust load and tall buildings that tend to concentrate pollutants as there is lower rate of PM dispersal [76]. As wind flow patterns can be skewed in cities with individual or groups of buildings, governing the interaction between the flow and buildings in the vicinity and local air pollutant dispersion [77]. Further research can address these limitations through the application of 3D spatial visualization that incorporates seasonal and meteorological analytics that can better help in the examination of local air circulation and its role in the dispersion and transport of PM throughout urban infrastructure.
The highest magnetic concentration values of NRM and SIRM for P. dactylifera occur near Kuwait bay, with the majority of urban infrastructure and PM sources located at the bay, the proximity of Kuwait Bay mandates the effects land and sea breeze impacting air flow and PM dispersion. In a previous study con- to be similar to North of Kuwait, while the central and southern dust samples were distinctive by their 'hard', haematite-like behavior [78]. This could be as a result of an association between lithogenic origins of soil, organic matter content and certain magnetic properties of urban street dust that may vary due to lithogenic origins depending on varying locations within an area and roads [79] [80].
The maps clearly demonstrate that dust samplers although time consuming are suitable method in the determination of short-term PM pollution distribution patterns that are heavily influenced by prevailing wind directions, while P. dactylifera offers corresponding information towards determining long term PM pollution patterns, hot spots and sources. The reliability of these predictions would thoroughly be improved in further research through the addition of meteorological influences to the predicted values to better discriminate magnetic particles dispersal patterns [72].

Conclusion
The outcome of this study validates the utilization of both P. dactylifera and dust samplers as proxies for PM pollution. The outcome of magnetic parameter data of both the passive and active biomonitoring techniques was similar in certain aspects in that the magnetic properties of Kuwait's PM P. dactylifera and dust samplers were both dominated by ferrimagnetic minerals (magnetite) from anthropogenic sources with low coercivity in the multidomain (MD) range and at the superparamagnetic/stable single domain (SP/SD) border with a grain size of (>10 -15 µm) and (80 nm -~10 -15 µm) that may be influenced by the prominence of quartz, calcite in resuspended soil. Quartz, calcium carbonate, feldspar and calcite in Kuwaits dust and P. dactylifera leaf material have a diamagnetic behavior that may have resulted in very low magnetic susceptibility values. Similar results between both techniques were found for radial distance moving away from the city center and sampling degree. Radial distance affected NRM and HIRM for P. dactylifera and NRM for dust samplers. The analyses on the effects of azithumal direction of dust samplers indicated that wind direction influenced greatly, specifically in the northwest direction for the study period.
The influence of wind direction, the position of Kuwait's bay and urban infrastructure was greatly highlighted through the application of simple kriging enabling interpolation to predict the spatial distribution of magnetic mineral concentrations and grain sizes in other sites of interest that were clearly better observed when mapped yet were not apparent in the statistical analysis. Geomagnetic spatial maps are proven to be a less tumultuous task and more economically efficient in comparison to geochemical analysis, particularly on a grand scale.