Assessment and statistical modelling of airborne microorganisms in Madrid

Highlights

• High temperature favours Actinobacteria/Proteobacteria relative abundance ratio.

• Microbial relative abundance correlates with PM₁₀ and O₃ in some cases.

• Proteobacteria show a positive correlation with pollen.

• Generalized Additive Models (GAM) able to capture non-linear relationships.

• Coefficient of determination up to 0.76 for microbial relative abundances predictions.

Abstract

The limited evidence available suggests that the interaction between chemical pollutants and biological particles may intensify respiratory diseases caused by air pollution in urban areas. Unlike air pollutants, which are routinely measured, records of biotic component are scarce. While pollen concentrations are daily surveyed in most cities, data related to airborne bacteria or fungi are not usually available. This work presents the first effort to understand atmospheric pollution integrating both biotic and abiotic agents, trying to identify relationships among the Proteobacteria, Actinobacteria and Ascomycota phyla with palynological, meteorological and air quality variables using all biological historical records available in the Madrid Greater Region. The tools employed involve statistical hypothesis contrast tests such as Kruskal-Wallis and machine learning algorithms. A cluster analysis was performed to analyse which abiotic variables were able to separate the biotic variables into groups. Significant relationships were found for temperature and relative humidity. In addition, the relative abundance of the biological phyla studied was affected by PM₁₀ and O₃ ambient concentration. Preliminary Generalized Additive Models (GAMs) to predict the biotic relative abundances based on these atmospheric variables were developed. The results (r = 0.70) were acceptable taking into account the scarcity of the available data. These models can be used as an indication of the biotic composition when no measurements are available. They are also a good starting point to continue working in the development of more accurate models and to investigate causal relationships.

Introduction

Mortality and morbidity are associated with poor air quality through a breadth of respiratory causes, including cardiovascular diseases and lung cancer (WHO, 2016). Up to 8.9 million excess deaths are attributed to long-term exposure to outdoor PM_2.5 (particulate matter with an average aerodynamic diameter of up to 2.5 μm) alone (Burnett et al., 2018). Moreover, the contribution of air pollution to premature mortality could double by 2050 (Lelieveld et al., 2015). Air pollution has significant impacts all over the world, including Europe. Over 500000 premature annual deaths are attributed to population exposure to PM_2.5, NO₂ (nitrogen dioxide) and O₃ (ozone) in Europe (EEA, 2019). Further reductions or ambient concentrations of these pollutants would bring considerable health benefits in major urban areas including Madrid (Izquierdo et al., 2020).

In addition to traditional abiotic air pollutants (PM, NO₂, SO₂ (sulphur dioxide), O₃, etc.), the presence of airborne biological particles is gaining interest within the air quality field since they also pose a risk for human health. Many fungal spores present in the air (Cladosporium spp, Epiccocum spp., Alternaria spp., Aspergillus spp.) cause allergic asthma (Vacher et al., 2015), which in combination with pollen allergies affect several million people around the world (Bousquet et al., 2008). Although the knowledge in the field remains insufficient to reach conclusions, current evidence suggests that the interaction between chemical pollutants and biological particles would intensify respiratory diseases such as asthma, allergic rhinitis and bacterial infection (Baldacci et al., 2015), and also induce changes in lung microbiota (Yang et al., 2019).

Accordingly, the European Union has incorporated new documents about pollen and fungal spores sampling normalization in ambient air because of their association with allergy (European Technical Specification note CEN/TS 16868:2019; and British adaptation BS EN 16868:2019). Furthermore, pathogenic bacteria such as Legionella pneumophila (Legionnaires’ disease), Mycobacterium tuberculosis (tuberculosis) or Haemophilus influenzae (pneumonia, meningitis) are transmitted via aerosols. Moreover, recent studies have also stated that pollen grains, another major biotic component of the air, not only may reduce the immune response to other pathogens like viruses (Gilles et al., 2020) but also act as important carriers of bacteria and endotoxins (bacterial antigens), promoting the allergic sensitization to pollen (Obersteiner et al., 2016; Oteros et al., 2019).

While pollen concentrations are daily monitored in the urban areas to publish alerts related to allergy for the population, airborne fungi and bacteria are not routinely analysed in most cities. The processing and identification of airborne microorganisms requires expertise, it is time-consuming and costly (especially for bacteria). New technologies based on DNA sequencing (Next-generation sequencing/high-throughput sequencing) have enabled a more accurate identification of these entities from environmental samples. Thus, specific works have characterized the microbial component (bacteria and/or fungi) in the urban atmosphere (Bertolini et al., 2013; Bowers et al., 2013; Du et al., 2018; Fan et al., 2019a; Gandolfi et al., 2015; Lee et al., 2017; Lu et al., 2018; Núñez et al., 2019; Uetake et al., 2019; Yamamoto et al., 2012; Yan et al., 2016, 2018; Zhen et al., 2017). These studies consistently point out that biological diversity present in the air is wider than previously thought, but no conclusive relationships have been stated with the abiotic variables, mostly because of the large number of factors involved (local sources, microenvironments, seasonality, meteorology, long-distance transport, etc.) and the lack of long-term studies (Núñez et al., 2016; Zhai et al., 2018). They frequently correlate chemical pollutants, environmental and meteorological factors with the microbial communities, providing general tendencies. For instance, Wei et al. (2020) discovered that a moderately polluted atmosphere promoted the microbial growth due to higher nutrients availability. Some works also attempted to develop predictive models, mostly using Multiple Linear Regression for fungi (Molina et al., 1998; Pyrri and Kapsanaki-Gotsi, 2017; Recio et al., 2012), or more sophisticated non-linear algorithms for bacteria (Fan et al., 2019b; Yoo et al., 2018).

The main objective of the present work, part of the AIRTEC-CM (urban air quality and climate change integral assessment) research project, is to obtain a wider vision of the atmospheric pollution in Madrid (Spain) by discovering potential relationships among abiotic and biotic pollutants and environmental variables. In this first approach, we made use of all the available observations of biotic agents in Madrid (Proteobacteria, Actinobacteria, Ascomycota relative abundance) and records of the main allergenic pollen types: “Cupressaceae”, “Olea”, “Platanus”, “Poaceae”) in addition to meteorological and air quality data. We performed a number of statistical analyses, including clustering techniques to identify significant relationships. Finally, we fitted Generalized Additive Models (GAMs) that could be used to predict biotic components relative abundance in the absence of conventional DNA analyses, usually unavailable.