Determination of free amino acids, saccharides, and selected microbes in biogenic atmospheric aerosols – seasonal variations, particle size distribution, chemical and microbial relations

Abstract

Primary biological aerosol particles (PBAPs) play an important role in the interaction between biosphere, atmosphere, and climate, affecting cloud and precipitation formation processes. The presence of pollen, plant fragments, spores, bacteria, algae, and viruses in PBAPs is well known. In order to explore the complex interrelationships between airborne and particulate chemical tracers (amino acids, saccharides), gene copy numbers (16S and 18S for bacteria and fungi, respectively), gas phase chemistry, and the particle size distribution, 84 size-segregated aerosol samples from four particle size fractions (< 1.0, 1.0–2.5, 2.5–10, and > 10 µm) were collected at the SMEAR II station, Finland, in autumn 2017. The gene copy numbers and size distributions of bacteria, Pseudomonas, and fungi in biogenic aerosols were determined by DNA extraction and amplification. In addition, free amino acids (19) and saccharides (8) were analysed in aerosol samples by hydrophilic interaction liquid chromatography–mass spectrometry (HILIC-MS). Different machine learning (ML) approaches, such as cluster analysis, discriminant analysis, neural network analysis, and multiple linear regression (MLR), were used for the clarification of several aspects related to the composition of biogenic aerosols. Clear variations in composition as a function of the particle size were observed. In most cases, the highest concentration values and gene copy numbers (in the case of microbes) were observed for 2.5–10 µm particles, followed by > 10, 1–2.5, and < 1.0 µm particles. In addition, different variables related to the air and soil temperature, the UV radiation, and the amount of water in the soil affected the composition of biogenic aerosols. In terms of interpreting the results, MLR provided the greatest improvement over classical statistical approaches such as Pearson correlation among the ML approaches considered. In all cases, the explained variance was over 91 %. The great variability of the samples hindered the clarification of common patterns when evaluating the relation between the presence of microbes and the chemical composition of biogenic aerosols. Finally, positive correlations were observed between gas-phase VOCs (such as acetone, toluene, methanol, and 2-methyl-3-buten-2-ol) and the gene copy numbers of microbes in biogenic aerosols.