Reassessing atmospheric deposition rates of polycyclic aromatic compounds to the Athabasca River (Alberta, Canada) watershed from oil sands related activities Sierra Rayne a * Keywords: polycyclic aromatic compounds, atmospheric deposition, oil sands, Athabasca River Abstract: In an earlier study (Kelly et al., PNAS, 2009, 106, 22346-22351), spatial patterns for the concentra- tions of particulate matter, particulate polycyclic aromatic compounds (PAC), and dissolved PAC in the snowpack around the Syncrude and Suncor upgrader facilities near the oil sands development at Fort McMurray, Alberta, Canada were determined. A reassessment of the datasets employed in this work yields significantly different depo- sition rates (by up to an order of magnitude) than reported, as well as reveals substantial sensitivity in deposition rate estimates depending on a range of equally valid regression types chosen. A high degree of uncertainty remains with regard to the quantities of particulate matter and PAC being deposited in the Athabasca River watershed from oil sands related activities. In their article, Kelly et al. [1] report on the spatial patterns for concentrations of particulate matter, particulate polycyclic aromatic compounds (PAC), and dissolved PAC in the snowpack around the Syncrude and Suncor upgrader facilities near the oil sands development at Fort McMurray, Alberta, Canada. From their sampling data and subsequent analysis, estimates of total loadings of these contaminants to snowpack over a 4 month period in 2008 were developed for a radius of 50 km around the oil sands upgrading facilities. Based on their data, the authors conduct negative exponential regression analysis and obtain the following equa- tion describing the relationship between distance from the upgrading facilities and concentration of particulate matter: particulates=10.6×e -0.0714x g/m 2 where x is the distance in km from site AR6. In ref. [1], the number of samples is given as 23. A data request from the authors of ref. [1] for their data behind Figure S2 in this work obtained a spreadsheet with 31 datapoints for each of the three variables. It is clear from Figure S2(a) in ref. [1] that this published figure also has 31 datapoints, of which 23 datapoints are within a 50 km radius of the upgraders. However, Fig. S2(a) in ref. [1] does not show an exponential regression with the equation y=10.6×e -0.0714x , as such an equation would have a y-intercept of 10.6. Instead, Fig. S2(a) in ref. [1] shows a regression equation with a y- intercept of between 13 and 14. In the figure below is shown a negative exponential regression with the use of either 23 or 31 datapoints (as obtained from Kelly et al. [1]) from Fig. S2(a) as a black line (y=13.5×e -0.0707x ; r=0.844; both n=23 and n=31 yield the same regression equation, as the datapoints at >50 km distance play negligible role in the regression constant fitting), and the authors’ claimed regression fit with an equation y=10.6×e -0.0714x as a red line. * Corresponding author: Tel/Fax: 1.250.487.0166 Email: sierra.rayne@live.co.uk. a Chemologica Research, PO Box 74, 318 Rose Street, Mortlach, Saskatchewan, Canada, S0H 3E0.