Over the last decade, Bogotá has been recognized for its large bike-dedicated infrastructure. The city is currently endowed with more than 450 km of bike-paths and near 600 thousand trips are made by bicycle, accounting for 6% of the total daily trips (Ríos et al., 2015). These figures make Bogotá the Latin-American city with the largest number of kilometers of bike-lines, almost as the double as Sao Paulo, Ciudad de México, and Santiago de Chile (Baumann et al., 2013). The initial motivation for cycling in Colombian cities was primary related to economic reasons. Many people simply could not afford paying the everyday cost of public transportation. Moreover, there has been a relatively recent initiative for encouraging bicycle usage as a sustainable transport mode to face to the critical urban mobility conditions, while fostering healthier life-styles, and increasing the level of physical activity.

However, most of the bike-paths in Bogotá are located alongside busy roads, where cyclists are experimenting air quality conditions that may not be adequate for performing such physical activity. A question arises as to whether biking under such conditions may overcome the benefits of cycling. This is even more of concern, since Bogotá has been listed among the most polluted cities in Latin America (WHO, 2016). Elevated concentrations of pollutants, especially respirable particulate matter (PM₁₀), represent a constant issue (Franco R, 2012). According to the data from the city's air quality monitoring network, in 2013 the annual PM₁₀ average concentration in the city ranged between 32 and 81 μg/m³, in the northern and western regions of the city, respectively (SDA, 2014). In some areas of the city, both WHO reference threshold values (annual–20 μg/m³ and 24-h–50 μg/m³) and national standards (annual–50 μg/m³ and 24-h–100 μg/m³), are regularly exceeded.

This is a matter of concern, since inhalation of particulate matter has been related to various diseases. Many epidemiological studies have provided evidence of association between the exposure to PM and asthma, bronchitis and chronic obstructive pulmonary disease, as well as cardiovascular diseases (van Berlo et al., 2012). Closeness to highly polluted locations, such as heavy traffic streets, reveals to be a risk factor for respiratory complications in vulnerable population, leading to reduction of pulmonary function (Peters et al., 1999; Fritz and Herbarth, 2001; Kim et al., 2004; Calderón-Garcidueñas et al., 2006; Ni et al., 2015).

Much like in other Latin-American cities, transportation is one of the most important sources of air pollution in Bogotá (WMO, 2012). According to the city emission inventory, 65% of PM₁₀ is generated by mobile sources (SDA, 2011), in particular from public transport and freight vehicles using diesel fuel. Although this vehicular category represents <5% of the actual fleet, it emits more than 60% of the PM associated to mobile sources in the city (SDA, 2011). In terms of exposure, mobile sources imply higher negative effects than other sources (i.e., industries), as a consequence of the closer distance between the exhaust pipes and the people in transport-related microenvironments.

In spite of its relevance for public health, the exposure of bike-paths users to air pollution in Bogotá has so far received limited attention and few researches have been conducted. One previous study (Fajardo and Rojas, 2012), drew as a conclusion the need for a more detailed characterization of the environmental conditions of bikers in the city. The main objective of our work is to measure the levels of fine particulate matter (PM_2.5) and black carbon (BC) to which bike-paths users are exposed in Bogotá. In addition, we identified some of the factors that must be considered when designing bike-dedicated infrastructure to minimize users' exposure.

Table 3 shows the descriptive statistics for the total number of valid observations of PM_2.5 and BC. For PM_2.5, more than 75% of the measurements are above the 24-h threshold value of 25 μg/m³ set by WHO, with extreme values beyond 1000 μg/m³. The measurements of BC concentrations raise similar worries (i.e., 33% of the measurements are higher than 25 μg/m³).

Figure 2 shows the histograms for PM_2.5 and BC considering all the samples (for both measurements the rightmost part of the distribution was cut for the sake of visualization). Both distributions of values are exhibiting a long tail, as also indicated by the coefficient of variation >1 (see Table 3). Moreover, the distribution of PM_2.5 shows multimodality, which can be attributed to the presence of multiple emitting sources along the bike-paths.

Both pollutants show highly dynamical patterns in its concentrations, with many swift peaks. Figure 3 shows a georeferenced display of PM_2.5 concentrations along the four bike-paths (for one specific experiment). Different levels of each pollutant are represented by using a color scale, were the nuances of red are assigned to the highest concentration levels and those of blue to the lowest ones. Bikers are exposed to highly variable pollution levels during their bike ride.

For those particular experiments we reported in Figure 3, the highest concentrations were recorded in the proximity to the main road crossing (e.g., along bike-path c), but also close to traffic lights and public transportation bus stops along the street (e.g., along bike-path a). The results of this analysis, as well as the proximity of the bike-paths to the street, indicate that the concentrations of air pollutants to which bikers are exposed are highly influenced by vehicular traffic emissions.

Figure 4 presents the average and the 95% confidence interval for PM_2.5 and BC concentrations, classed by workdays and weekends. These charts confirm cyclists are consistently exposed to high concentrations of both pollutants, irrespective of the type of day. For PM_2.5,the average concentration is within 80 and 136 μg/m³ for workdays. These levels are three to five times higher than those suggested by the WHO (if we take the 24 h–25 μg/m³ as a reference value). During weekends, the concentrations of this pollutant ranged between 30 and 72 μg/m³.

The average concentrations of BC were between 16 and 38 μg/m³ during workdays and in the range of 10–32 μg/m³ during weekends. Such air pollution levels are relevant from a public health point of view, considering the positive association of the exposure to particle related pollutants with respiratory and cardiovascular problems (Calderón-Garcidueñas et al., 2003; Kim, 2004; WHO, 2005; Holguin et al., 2007).

We compared the results of this study with the measured concentrations to which bike users are exposed to in other cities of the world. The PM_2.5 levels found in Bogotá are up to one order of magnitude higher than those measured in the proximity of streets along bike-paths in cities such as Minneapolis and Berkeley in the United States (Jarjour et al., 2013; Hankey and Marshall, 2015). They are up to three times higher than the concentrations recorded in European cities (Adams et al., 2001; Zuurbier et al., 2010); in the same range of those measured in Beijing, China (Huang et al., 2012), and lesser than those reported for Indian cities, such as New Delhi (Goel et al., 2015). Similarly, BC concentrations measured in this study were up to two orders of magnitude higher than those observed in North American cities (Jarjour et al., 2013; MacNaughton et al., 2014; Hankey and Marshall, 2015). Table 4 presents a summary of the studies cited above.

In a previous study conducted in Bogotá, Fajardo and Rojas (2012) determined the concentration levels of PM₁₀ in two sites along a bike-path located on a major city avenue. They reported PM₁₀ average concentrations between 77 and 108 μg/m³, and concluded that people who regularly use bikes in Bogotá can inhale doses of particulate matter between 45 and 65% higher than the doses inhaled by people who do not use bikes.

Accounting for the day of the week when each experiment was conducted, and the traffic volume, we can conclude that the dynamics of vehicular traffic during the week has a significant impact on the PM_2.5 and BC concentrations in the proximity of streets. A statistically significant difference exists in pollutant concentrations measured during workdays and weekends at each of the four streets considered (see confidence intervals in Figure 4).

For PM_2.5, in streets such as Carrera 19, the average concentration during workdays is about three times the one measured during weekends, where the vehicular flow is lesser (according to local mobility authorities, Sundays' total public transit traffic is ~50% lesser than in workdays). For BC, streets that bear a higher share of heavy traffic during workdays, such as Calle 26 and Avenida Boyacá, were those with the highest BC concentrations. We compared means of workdays and weekends for each street by a one-way ANOVA test, and in each case we found that the null hypothesis (i.e., mean concentrations are the same) can be rejected with a very small p-value (<5×10⁻⁸). These conclusion has been remarked by other authors (Berghmans et al., 2008; Int Panis et al., 2010; Fajardo and Rojas, 2012). Figure 5 provides a graphical appreciation of the differences in the distribution of the measured pollutants during workdays and weekends.

Figure 6 shows scatter plots for the pairwise concentrations of BC and PM_2.5–BC measured during workdays and weekends for the bike–paths along Carrera 19 and Calle 26. There is not graphical evidence of significant correlation between the two measurements. This could be explained by the simultaneous co-existence of different emitting sources. Similar results have been obtained for the other two bike-paths (data not shown).

The charts at the top and at the right of the scatter plots in Figure 6 are providing the distribution of the measured concentrations. The distribution for PM_2.5–BC for Calle 26 shows a very pronounced bimodality, both during workdays and weekend days. This important avenue of Bogotá is very wide, with large stretches of fluid vehicular traffic. However, it has several intersections with mayor confluent streets, where traffic jams are constant. Such diverse conditions may be a cause of the prominent bimodality.

It is also important to remark the impact that the usage of paths located on narrow, closed geometry streets may have on bikers' level of exposure to air pollution. Even though the total vehicular traffic along the narrowest street considered in this study (Carrera 11) is about four times lesser than the one in the largest and open geometry one (Calle 26), the concentrations of PM_2.5 and BC (as measured during workdays) were practically equivalent. Even if several other elements are still to be analyzed at a finer level, these results suggest that the geometry of the streets can significantly affect air pollutant dispersion, increasing levels of exposure for bike-path users.

While promoting the bicycle as an alternative transport mode in Bogotá, it would be significant to stress the importance of reducing air pollution exposure levels of cyclists. It appears advisable to consider using secondary, low traffic level streets, for new bike dedicated infrastructure, as well as physical barriers in heavy-traffic roads were existing bike-paths are located. Also, it is urgent to actively promote emission control systems on mobile sources such as particulate filters, as well as to encourage the massive use of face filter masks among bike users. This fact clearly indicates the need of more accurate studies on the interrelations between mobility strategies and their effects on public health.

It is well-documented that DustTrak and Aethalometer monitors readings may diverge from the standard PM determination methods (Ramachandran et al., 2000; Yanosky et al., 2002; Wallace et al., 2011). The readings of photometers such as DustTrak monitors may overestimate the concentration of airborne particulate matter compared to gravimetric techniques (Gorner et al., 1995; Ramachandran et al., 2000; Tasić et al., 2012). We previously reported results from specific experiments carried out to compare and validate the two methods for mass concentrations measurement in Bogotá (Franco et al., 2009, 2013). Even though those local comparisons were only conducted to assess PM₁₀ concentrations, we believe that under our study conditions, DustTrak monitors provided representative information on the range of PM concentration. In addition, other authors (Wallace et al., 2011) have shown that continuous monitors such as DustTrak and Aethalometer can provide reliable information for environmental studies.

The authors are aware of the representativeness limits of the results in terms of the number of experiments and bike-paths addressed. However, they believe this pilot study can provide useful data for an improved understanding of air quality conditions for bikers in Bogotá and in other Latin American cities.

For the very first time for Bogotá, we quantified in this study PM_2.5 and BC concentrations at bike-paths located alongside major streets. Bicycle is a sustainable mode of transport and it has important health benefits for the population. However, the pollution levels indicate the existence of low air quality conditions and show that cyclists are exposed to concentrations that largely exceed the threshold values established as potentially dangerous for human health. We consider these results provide useful insights to increase the appreciation of this problem in Latin American cities and to contribute with technical elements that lead to the inclusion of air quality variables when designing and planning sustainable urban mobility infrastructures.

JF, JS, and IM were involved in field work design. JF and JS were in charge of the data collecting process. JF and IM were in charge of the data processing process. JS contributed in structuring the paper, and JF and IM wrote the final version. All authors supported the experiments, critically commented, and revised the manuscript.