Reader Comments

Post a new comment on this article

Commentary upon: Has Alberta oil sands development altered delivery of polycyclic aromatic compounds to the Peace-Athabasca Delta? by R. I. Hall, B. B. Wolfe, J. A. Wiklund, T. W. D. Edwards, A. J. Farwell, and D. G. Dixon

Posted by timoney on 26 Sep 2012 at 21:18 GMT

A pdf of this comment can be accessed at:
http://www.plosone.org/at...


Commentary upon: Has Alberta oil sands development altered delivery of polycyclic aromatic compounds to the Peace-Athabasca Delta? by R. I. Hall, B. B. Wolfe, J. A. Wiklund, T. W. D. Edwards, A. J. Farwell, and D. G. Dixon


Commentary by K. P. Timoney(a) and P. Lee(b)

a Treeline Ecological Research, 21551 Township Road 520, Sherwood Park, Alberta, Canada T8E 1E3

b Global Forest Watch Canada, 10337 146 Street, Edmonton, Alberta, Canada T5N 3A3


This commentary follows the organizational structure of the Hall et al. paper. Comments on specific points are made first, followed by general comments and observations. Quotations from the Hall et al. paper are set in blue font. The writers ask the forbearance of the readers given that the comments were prepared under time constraint in order to meet a publication deadline.


Comments

1. The short title for the paper (Natural vs industrial PACs from Alberta oil sands) is misleading. The authors have not differentiated, nor can their data differentiate, industrial from natural sources of polycyclic aromatic hydrocarbons in sediments. This topic is considered below.

Introduction

2. The citing of reference [12] for the statement that “PAC concentrations are high and comparable in sediments deposited pre- and post-development” is problematic. Fourteen years of industrial development have taken place since the 1998 sediment sample of Evans et al. (2002). There has been much development over the period 1998 to 2012, making its use as a pre- and post-development comparison tenuous. The temporal trends at the two sites reported by Evans et al. were inconsistent.

3. The statement that “As acknowledged by the authors, uncertainty regarding the time interval represented by these samples is substantial, with individual samples potentially encompassing time periods that approach the entire decade of sample collection” is misleading for two reasons. (a) The authors (Timoney and Lee 2011) did not state that the temporal uncertainty in the Regional Aquatic Monitoring Program (RAMP) sediment data was substantial. (b) A recent analysis by RAMP demonstrated clear differences in sediment particle size distributions amongst years at the Athabasca Delta sampling stations. Interannual differences in particle size in RAMP’s PAH samples would not be evident if the samples included sediment from different years or if vertical mixing were a significant issue. RAMP’s sediment samples most likely represent the current year’s deposition ( M. Davies, RAMP, pers. comm., 12 September 2012), not an entire decade as implied by Hall et al.

4. Although the spring freshet may be instrumental as an agent of contaminant dispersal, the relationship between sediment PAH concentration and discharge of the mainstem Athabasca River and its distributaries is not clear. We tested the hypothesis that Athabasca River mainstem and distributary discharge is related to sediment PAH concentrations in the Athabasca River Delta for both annual and May-August discharge and found no significant correlations (Timoney and Lee 2011). The spring freshet is a time of high discharge during which time there may be little deposition of fines across much of the ARD because flow velocities are sufficient to maintain fine sediments in suspension until they leave the delta front and enter Lake Athabasca. At times of high flow, sand and coarse silts may be preferentially deposited and it is believed that these coarser clasts have lower concentrations of PAHs than do fines. In contrast, flooding of a restricted basin at the time of the freshet could result in sedimentation of fines because flow velocities decrease when the river enters an off-channel basin. Therefore the relationships among discharge and sedimentation may be reversed when off-channel and channel sites are compared. Secondly, mobilization of industrial contaminants stored in the snowpack is a different process than is a freshet-driven erosion event.

Study Design

5. A centuries-long highstand of Lake Athabasca is problematic for several reasons (reviewed in Timoney 2013) that lie beyond the scope of the present paper. Suffice to say that historical accounts, maps, and Lake Athabasca hydrometric data do not support a centuries-long highstand, nor do they support Hall et al.’s view of declining Lake Athabasca levels over the period 1940-82. There was a transient drawdown of Lake Athabasca circa the early 1940s, but from the early 1950s to the late 1970s, levels of Lake Athabasca were above normal (Timoney 2013). A more parsimonious explanation for a low flood frequency at PAD 31 during the period would be sediment and/or organic accumulation at the sill of the basin’s levee, a common process in the delta that leads to changes in flooding rates.

6. Headwater capture by Cree Creek resulted in a partial avulsion of the Embarras River into Cree Creek, not Mamawi Creek.

Numerical Analyses
7. The reliance on PAD 31 as the primary PAH sampling basin (discussed below) may be further complicated by its location nearby and downstream of Embarras-Cree Creek avulsion site. The nearby confluence and channel are still evolving due to the changes in flow conditions. Therefore, PAD 31 may receive more locally-eroded sediments than do other basins.
The structuring of the data into flood-prone and not flood-prone periods was not justified. If the authors were focussed on determining the relative contributions of natural and industrial sources, a more defensible subdivision would have been pre- and post-industry. But division of the record into a priori periods was not necessary other than as a statistical convenience and imposed an artificial structure onto the data. A better approach would have to been to analyze the data as simple time series.
The authors placed inordinate emphasis on floods, which is understandable given that their key lake (PAD 31) is located off-channel and depends on flooding for sedimentation. However, flooding is an episodic event that does not occur annually. If a flood does occur, it may carry sediment into the basin for several days (typically). Conversely, sedimentation at the mouths of the Athabasca River channels and in western Lake Athabasca occurs 365 days per year, every year. The small daily incremental addition of PAHs to the delta’s sediment, unrelated to flood events, may prove to be a more important source of PAHs than is episodic flooding. To use a single off-channel basin as a means to estimate the contaminant contribution to the delta is problematic.

Results

8. The statement “Based on geochemical fingerprinting, the main source of the PACs in
the natural oil sands region river sediments is bitumen-rich material in the riverbanks” may be misunderstood. The McMurray Formation geological deposit that is the source of the PAHs is the same deposit that is being exploited by industry along the Athabasca River’s banks. The authors’ predilection to emphasize natural erosion, while failing to differentiate natural from industrially-enhanced erosion, is a recurrent theme in the paper.

9. Similarly, the authors do not examine other sources of PAHs to the delta, nor do their study sites and sampling design seem able to detect them. For example, a Suncor pipeline break in 1970 spilled three million L of oil into the Athabasca River; the oil flowed into Lake Athabasca and was observed there for about six days. In 1982, a spill from Suncor necessitated closure of the commercial fishing season on Lake Athabasca and reportedly caused illnesses among people in Ft. Mackay. The amount of oil and/or bitumen that was deposited in the delta’s sediments was never determined. A considerable volume of PAHs would have been delivered to the delta in these two events that were evidently not detected by Hall et al. The spills were not associated with floods and therefore their study sites may not have sampled the events.

10. The authors state: “During flood-prone periods, PAC composition closely matches that of river-transported sediment originating from bitumen deposits of the McMurray Formation exposed along the riverbanks”. The authors appear to think that PAHs accumulated during flooding at PAD 31 all originate from erosion along riverbanks. Failure to consider how industry can contribute to PAH loading from erosion is unfortunate. The authors then state that: “During periods of reduced flood frequency, higher proportions of unsubstituted PACs (notably N, B, F) identify greater influence of hydrocarbons from fire and catchment vegetation”. Fire data do not support that assertion. According to Wood Buffalo National Park and northern Alberta fire records, 1953 was the largest fire year in the delta and its immediate environs over the period of record (1950 to present). Unfortunately, none of their basins detected the 1953 major fire. Nor did their isolated basin, PAD 18, detect the major 1981 fire (see Timoney 2013 for fire history).

11. Figure 4. The match in the PAH fingerprint for ‘oil sands samples’ and the ‘2007 flood deposit’ indicates that the McMurray Formation is the primary source of the sediment PAHs at the authors’ sites. This is to be expected. It is the geological formation that is being mined and processed in the region. The data do not argue for or against industrial vs. natural sources.

12. Figure 8 presents the crux of the results. PAD 31 indicates no trend in PAH concentrations until recent decades, at which point some of the highest concentrations over the 300-year record are observed. At one extreme, the data could be interpreted to indicate that natural erosion is the sole source for the pattern, which is the view of Hall et al. At the other extreme, the data could be interpreted to indicate that, with the onset and growth of the bitumen industry, PAH concentrations increased. Both interpretations would be over-reaching the data. A more defensible interpretation is that changes in water and sediment inputs from the Athabasca River into this off-channel basin underlie changes in sediment PAH content. The data are mute as to differentiation of natural vs. industrial sources for the riverborne PAHs. This has been the scientific challenge for decades and remains unresolved in this paper.

PAD 23 indicates a decline in PAH concentrations over the record from 1900 to about 1972 (when the engineered Athabasca River Cutoff was established, leading to a decline in hydraulic connectivity for the basin). Post-Cutoff, the basin became isolated and therefore was not ‘sampling’ potential industrial inputs; PAH concentrations have oscillated without a trend. The data indicate the obvious: that hydraulic connectivity is positively correlated with riverborne PAH inputs. The data are mute as to natural vs. industrial sources for the riverborne PAHs.

PAD 18 indicates overall lower concentrations of sediment PAHs in this isolated basin than in the other two basins, demonstrating the importance of hydraulic connectivity.

The inconsistent sensitivity of the sites to fire is noteworthy. Although the authors ascribe a PAD 23 peak in PAH concentrations to fire in 2010, that conclusion is tenuous. Fire records demonstrate that there were no fires in the delta in 2010. Regionally, 2010 was not a fire-prone year in northern Alberta. The closest fire of any size in 2010 was located 63 km east-southeast of PAD 23. Examination of PM2.5 data for the Fort Chipewyan air quality station indicates a transient spike in PM2.5 concentrations in late July of 2010. Whether this spike would be sufficient to be detected in the PAH record at PAD 23 is unknown. Given a wildfire large enough to be detected in a lake record, it is curious that a similar spike in PAHs would not detected at PAD 31. Lack of time prevents our examination of the PM2.5 record, but such an examination may prove interesting in regard to wind direction, industrial and wildfire particulates, and the lake PAH record.

Discussion

13. The authors begin the discussion with statements that reveal a single-minded focus on nature as the source of the contaminants: “Results from these two sites identify that flooding from the Athabasca River is an important vector that naturally supplies bitumen-sourced PACs to the Athabasca Delta ... In contrast, the increase in flood frequency at PAD 31 that has occurred since the Embarras Breakthrough in 1982 can be used to evaluate the role of industrial contributions of PACs by comparing results from this stratigraphic interval to the pre-1940 frequently-flooded interval ... This comparison identifies that natural processes responsible for delivering bitumen from along the banks of the Athabasca River and its tributaries can account for the PACs most associated with a bitumen origin in the post-1982 sediments of PAD 31. Thus, despite rapid growth of oil sands development during the past 25 years ..., the data reveal no measurable increase in concentration or proportion of river-transported bitumen-associated indicator PACs.”

14. Overall, the data presented in the paper demonstrate the importance of hydraulic connectivity, an inconsistent sensitivity to wildfire events, and large spatio-temporal variability. The latter observation demonstrates the danger of generalizing from a single basin. Conversely, the data tell us that riverine transport is the primary vector of PAHs in sediment, but they do not tell us the relative roles of natural vs. industrially-enhanced inputs of bitumen-bearing materials into the Athabasca River.

15. The view, citing Evans et al. and Hall et al., that flood events are the major vector of PAHs to the delta is not supported by the data (see comment 4). Theoretically, during low flow periods, sediment fines with their associated PAHs would tend to settle in the lower reaches of the ARD and the adjacent delta front and not reach off-channel basins. Conversely, during high flow events, fine sediments tend to stay in suspension. During those times, coarse-grained materials tend to be deposited within channels while fines are carried out to the delta front and into Lake Athabasca; high flows may carry sediments into restricted basins at that time, depending on local conditions. In essence, the relationship between river discharge and PAH sedimentation may be the reverse at off-channel and channel locations. The relationship between flood events and PAH concentrations is more complex than is implied by Hall et al.

16. The authors state that industrial sources of PAHs can be important: “Given the importance of natural processes and industrial sources during river flood stages, monitoring and research programs must begin to sample at locations and times that capture contaminant dispersal during flood events. We propose that a network consisting of a larger number of monitoring sites and broader range of hydrological basin types is critically needed to further evaluate spatial and temporal distributions of PACs in the delta, their sources and their potential toxicity to biota.” In a preceding paragraph the authors admit that their interpretation is based on essentially one basin (PAD 31), and later in the paper the authors recommend that sediment sampling be conducted at 40 basins. This admission is made more cogent by a recent article by most of the same authors (Wolfe et al. 2012) in which they stated: “Our past research has shown that tremendous differences in hydrological conditions exist spatially across lakes of the delta. Therefore, the lake monitoring program must select informative sampling sites that span these gradients so that the data can be scaled-up to the entire delta… We propose that ~60 lakes and ~10 river sites or a representative subset ... be sampled three times a year — soon after ice-off (early- to mid-May), mid-summer (late July), and fall (mid-September) — to capture intra-annual variations in hydrological conditions.” Given that the authors recommend that 40 to 60 basins should be monitored in order to assess spatial and temporal variation, should the results from a single basin be generalized across both time and space to absolve industry of a role in contributing PAHs to the delta? The perspective gained from studying the delta’s ecological and physical dynamics (Timoney 2013) argues that such a generalization is unwarranted.

17. Of the estimated riverine transport of five to nine thousand tonnes per year of bitumen to the Peace-Athabasca Delta, the authors attribute all of the loading to natural erosion. The failure to differentiate natural vs. industrially-enhanced erosion of bitumen substances into the Athabasca River undermines a central conclusion of the paper—that it is nature alone that is the source of sediment PAHs in the delta.

18. The authors’ conclusion that the mean PAC content of sediments at their two sites in the Athabasca Delta (1.55 mg PACs / kg sediment ± 0.51) “is lower than that commonly measured in soils from urban environments in the USA” is interesting in that the delta is not an urban ecosystem. The delta is a Ramsar Wetland of International Importance that is essentially a wild ecosystem driven by natural processes. Comparing the delta to an urban ecosystem rather than to comparable natural ecosystems sets the bar low for what is considered elevated PAH concentrations. Nor is the urban-PA Delta comparison appropriate in that people in urban ecosystems typically do not harvest local mammals, birds, and fishes for a considerable portion of their diet whereas many people in the Peace-Athabasca Delta do so.

19. The authors have made a leap of logic in their statement that ice-jam floods are associated with peak concentrations and proportions of river-transported PAHs. They have demonstrated that sedimentation events in off-channel restricted basins lead to deposition of river-borne PAHs. They have not demonstrated that ice-jam flooding leads to sedimentation across the vast majority of the delta. As noted earlier, whether high flow events lead to deposition of sediment depends upon location within the delta. Secondly, discharge rates can be higher during open water events than during ice-jam events. Thirdly, the spring freshet is not equivalent to ice-jam flooding. The spring freshet occurs annually; ice-jamming is episodic, and flooding resulting from ice-jamming is even more episodic because only a subset of ice-jams result in flooding. In contrast, in the open drainage system, through which the vast majority of the Athabasca River’s flow is directed, sedimentation can occur year-round. See comment 4.

General Observations

20. Kelly et al. (2009) documented the influence of deposition of bitumen industrial particulates upon the concentration of dissolved PAHs in the Athabasca River. They observed a deposition to the snowpack of 11,400 tonnes of particulates, equivalent to 600 tonnes of bitumen, over a four-month period within a 50 km radius of the main industrial facilities. A similar study is needed that examines the contribution of industrial particulates to loading of particulate PAHs which, in water, would be more abundant than the dissolved fraction.

To date, no direct measurements have been made of the relative amounts and proportions of pyrogenic and petrogenic PAHs that are deposited from aerial deposition in the region. Close to the mines, airborne PAHs are likely to be predominantly petrogenic, and far away where dusts have settled, PAHs are likely to be predominantly pyrogenic (H. Namsechi, pers. comm., Alberta Environment and Sustainable Resource Development, 18 September 2012). Currently, an unknown proportion of the particulate PAHs carried downstream by the Athabasca River is derived from industrial sources such as truck dust, wind erosion from the disturbed landscape, stack emissions, fugitive emissions from tailings ponds, mine faces, and other industrial sites, and enhanced fluvial erosion. Of the latter source, it strains credulity to believe that the bitumen industry’s disturbance of 78,000 ha of landscape (as of July 2011) adjacent to the Athabasca River does not have a demonstrable effect on rates of erosion of PAH containing materials into the Athabasca River.

21. In light of the preceding comments and observations, the authors’ conclusion that the results reveal no evidence of industrial contributions of sediment PAHs to the delta is unwarranted. Even if one basin were sufficient to generalize across the entire delta (which it is not), the data for PAD 31 are mute on the topic of industrial vs. natural contributions. We know that riverine transport is an important vector of sediment PAHs. We do not know the relative contributions of nature and industry. The fundamental problem is that natural and industrial processes are ‘sampling’ from the same petrogenic deposits without leaving clear signatures.

22. Looking towards the future and declining flows on the Athabasca River and increasing regional aridity, the future of the delta’s aquatic systems is clouded. It has been long known (Water Quality Branch 1973) that during drawdowns, concentrations of sodium and chloride in Lakes Claire and Mamawi can rise to potentially harmful levels. How might evaporative enrichment of a host of compounds in the delta’s lakes affect the biota and the people who depend upon it? The only certainty is that multiple stressors, both natural and anthropogenic, can lead to unforeseen consequences.

23. It is enigmatic that the authors, who are aware of the spatio-temporal variability that is a hallmark of the Peace-Athabasca Delta, should take the extreme view that natural erosion is the only important source of PAHs. The high degree of spatio-temporal variability in the Peace-Athabasca Delta is an over-arching concern when attempting to generalize results from local datasets and has been a constant challenge to all those who hope to understand the delta’s ecology. Those of us who have studied the delta know it is dangerous to generalize across space and time from a single site.

24. The authors have presented an interesting set of data but issues with both the sensitivity and appropriateness of the sites and the interpretation of the data militate against the significance of the study. A review commissioned by Environment Canada (Dowdeswell et al. 2010) recently concluded that many of the region’s monitoring programs were unable to distinguish industrial impacts. The inability to measure industrial impacts was often due to poor sampling design such as insufficient spatial or temporal replication. A similar conclusion may apply to the Hall et al. study.

25. Those who see the world in black and white may welcome the conclusions of the Hall et al. study. But as science has shown time and again, painting ecosystems in black and white fails to capture the behavior of complex systems.

Note: This commentary was prepared without funding in the public interest.

References

Dowdeswell, L., P. Dillon, S. Ghoshal, A. Miall, J. Rasmussen, and J. Smol. 2010.
A Foundation for the Future: Building an Environmental Monitoring System for the
Oil Sands. Environment Canada, Ottawa, ON, Canada. http://www.ec.gc.ca/pollu...
E9ABC93B-A2F4-4D4B-A06D-BF5E0315C7A8/1359_Oilsands_Advisory_Panel_
report_09.pdf.
Evans, M.S., B. Billeck, L. Lockhart, J. P. Bechtold, M. B. Yunker, and G. Stern. 2002. PAH
sediment studies in Lake Athabasca and the Athabasca River ecosystem related to the
Fort McMurray oil sands operations: sources and trends. In C. A. Brebbia (editor), Oil
and Hydrocarbon Spills III, Modelling, Analysis and Control. WIT Press, Southampton.
Boston, Massachusetts. pp. 365-374.
Timoney, K. P. 2013. The Peace Athabasca Delta: Portrait of a Dynamic Ecosystem.
The University of Alberta Press, Edmonton, Alberta. In press.
Timoney, K. P. and P. Lee. 2011. Polycyclic aromatic hydrocarbons increase in Athabasca
River Delta sediment: temporal trends and environmental correlates. Environmental
Science and Technology, published online: dx.doi.org/10.1021/es104375d.
Water Quality Branch. 1973. Report on Water Quality 1971-1972. Section O. In Hydrologic
Investigations: Volume One. Peace Athabasca Delta Project Group. Governments
of Canada, Alberta, Saskatchewan. Ottawa, Ontario.
Wolfe, B. B., R. I. Hall, T. W. D. Edwards, and J. W. Johnston. 2012. Developing temporal
hydroecological perspectives to inform stewardship of a northern floodplain landscape
subject to multiple stressors: paleolimnological investigations of the Peace–Athabasca
Delta. Environmental Reviews 20: 191–210.

Competing interests declared: We have published results that are disputed in the Hall et al. paper (Timoney and Lee 2011).

RE: Commentary upon: Has Alberta oil sands development altered delivery of polycyclic aromatic compounds to the Peace-Athabasca Delta? by R. I. Hall, B. B. Wolfe, J. A. Wiklund, T. W. D. Edwards, A. J. Farwell, and D. G. Dixon

rihall replied to timoney on 30 Sep 2012 at 21:10 GMT

Author Responses to Commentary by K.P. Timoney & P. Lee posted September 26, 2012 at 21:18 GMT

Following the journal’s policy, K. Timoney & P. Lee were invited to provide their Commentary based on the decision by an Associate Editor that findings presented in our research article may be perceived as directly contradicting their previously published study [reference 15 in our article]. Our view is that our study used a very different experimental design from their study to produce data that supported different conclusions about the relative roles of industrial versus natural processes in supplying polycyclic aromatic compounds (PACs) to the environment. In particular, we measured contaminants in samples that pre-dated onset of oil sands development, and utilized chronologically well-constrained cores of floodplain lake sediments rather than surface sediments collected from rivers and distributary channels during 1999-2009. Thus, rather than refuting their study, our view is that our research article has added significant new knowledge that expands upon perspectives developed by their and other studies. Nevertheless, we agreed to follow journal policy and abide by the Associate Editor’s decision. Certainly, debate about environmental consequences of Alberta’s oil sands industry is polarized and the subject of controversy. Opportunity to compare and contrast findings of various studies is therefore necessary and productive.

The policy of PLOS ONE is to invite a signed review from the authors of the disputed work during the review process that leads to a decision of whether a research article is worthy of publication or not. Such a review is considered by an Academic Editor, and, in this case, after we received very positive independent peer reviews. Via this process, Timoney & Lee were invited by an Academic Editor to provide a review of our research article and to add their Commentary. Their prior review included many of the points raised in this Commentary. We provided detailed responses to the Academic Editor to Timoney’s prior review, which the Academic Editor deemed sufficient to base a decision to accept our research article for publication.

Below are our detailed responses to salient comments raised by Timoney & Lee. We apologize to the readers for the length of this document, but their comments are sufficiently detailed as to require accurate and, in places, lengthy responses. Also, we felt it necessary to repeat their comments, with our responses placed immediately below, so that readers do not have to flip back and forth between separate documents. Due to space restrictions we could do this only for the first 12 comments. Numbered citations refer to references in our research article, and the details can be obtained there.


Timoney & Lee Comment #1: The short title for the paper (Natural vs industrial PACs from Alberta oil sands) is misleading. The authors have not differentiated, nor can their data differentiate, industrial from natural sources of polycyclic aromatic hydrocarbons in sediments. This topic is considered below.

Author Response #1:
Our study uses analyses of sediment cores from lakes within the Peace-Athabasca Delta to quantify pre-development levels of PACs transported by natural processes via the air and the Athabasca River. One of the study lakes receives its water and other materials exclusively from precipitation to the lake surface and its small, undisturbed contributing basin. This site allowed us to directly compare levels of PACs transported via the atmosphere pre- versus post-onset of Alberta oil sands development. Two other study lakes received inputs of floodwaters from the Athabasca River prior to oil sands development, which allowed us to establish levels due to natural contributions of PACs via water. One of those sites was flood-prone pre- and post-development of oil sands and allowed us to contrast the concentration and composition of PACs in sediments provided entirely by natural processes (pre-oil sands; 1700 to mid-1900s) with that in sediments deposited after onset of development (post-1982). Given this comparison, we argue our short title (limited to 50 characters) is a reasonable choice.

Timoney & Lee Comment #2. The citing of reference [12] for the statement that “PAC concentrations are high and comparable in sediments deposited pre- and post-development” is problematic. Fourteen years of industrial development have taken place since the 1998 sediment sample of Evans et al. (2002). There has been much development over the period 1998 to 2012, making its use as a pre- and post-development comparison tenuous. The temporal trends at the two sites reported by Evans et al. were inconsistent.

Author Response #2:
The cited sentence was presented in the Introduction with the purpose to develop a balanced set of arguments underlying debate about whether Alberta’s oil sands development is elevating supply of contaminants of concern. We cited studies that have concluded oil sands development is increasing contaminant supply, as well as reference 12 (a study by Evans et al. published in 2002) which concluded the following (P. 6 of the paper), “There was little evidence from the sediment cores of a temporal increase in PAH concentrations, which could be related to the oil sands industry. Data are presented for only the Richardson and Lake Athabasca cores and then for two time periods. In general most PAHs occurred in higher concentrations in the sediment core slices from the 1950s than in recent times. The reasons for this are unclear. Only phenanthrene and anthracene in the Richardson Lake core showed evidence of higher concentrations in recent times.”

This study by Evans et al. (2002) remains the only publication to date that has reported PAC levels from before onset of oil sands development, and thus it remains a highly relevant study. Indeed, no other comparable studies have been published during the 14 years since their samples were collected. Thus, we contend our use of this paper is appropriate, and not problematic.

To address Timoney & Lee’s second point, the study by Evans et al. does have its limitations in that it measured and compared PAC concentrations in only two sediment core samples from each lake – one from the 1950s and one from the 1990s. This prevents an ability to assess continuous trends over time. Also, the lack of replication means that differences in reported concentrations between the 1950s and 1990s cannot be assessed for statistical significance (i.e., real differences beyond measurement error). And, insufficient presentation of data regarding the dating of the core prevents us from assessing the accuracy of the reported age estimates of the purported 1950s samples. Nevertheless, the data are important and the paper provides informative context regarding the need for our study, because conflicting and inadequate information has continued to generate controversy about environmental effects of oil sands development. Note that phenanthrene and anthracene, the two PAHs which “may have increased” in Richardson Lake (as reported by ref. 12), are ones that our study shows do not have a strong association with bitumen deposits and transport by the Athabasca River. We maintain that our use of the statement to which Timoney & Lee take offense is both accurate and justified to develop the context and need for our study.


Timoney & Lee Comment #3: The statement that “As acknowledged by the authors, uncertainty regarding the time interval represented by these samples is substantial, with individual samples potentially encompassing time periods that approach the entire decade of sample collection” is misleading for two reasons. (a) The authors (Timoney and Lee 2011) did not state that the temporal uncertainty in the Regional Aquatic Monitoring Program (RAMP) sediment data was substantial. (b) A recent analysis by RAMP demonstrated clear differences in sediment particle size distributions amongst years at the Athabasca Delta sampling stations. Interannual differences in particle size in RAMP’s PAH samples would not be evident if the samples included sediment from different years or if vertical mixing were a significant issue. RAMP’s sediment samples most likely represent the current year’s deposition ( M. Davies, RAMP, pers. comm., 12 September 2012), not an entire decade as implied by Hall et al.

Author Response #3:
Our quoted sentence was based on values presented by Timoney & Lee in the Supplementary Information portion of their 2011 research article in Environmental Science & Technology [ref. 15; available on the journal website]. In that Supplementary Information section they provide further methodological details on data handling and assumptions, which states: “Fall sediment samples were collected at sites from depositional reaches in the ARD [= Athabasca River Delta]. Sample depth (Ekman dredge) was 4 to 6 cm. Because the average sedimentation rate in the ARD distributaries is on the order of 0.6-1.2 cm per year (3) to 2.6 cm per year (4), ‘annual’ sediment samples may be composites of the last few years of sediments (essentially, running averages)” [numbers in brackets within the quote refer to sources they cited). Calculations based on these values identify that the individual surface sediment samples captured between 1.5 and 10.0 years of deposition. We view this as substantial temporal uncertainty given that the entire time-series of their data spanned 10 years. Yet, in a linear correlation analysis Timoney & Lee treated those samples as representing the specific calendar year they were collected, which may violate assumptions of correlation analysis. Moreover, if Timoney & Lee [ref. 15] viewed the raw data based on these sediment samples as ‘running averages’ (spanning variable time periods among samples), is it scientifically valid to interpret results of linear correlation tests when these data are further averaged into ‘dataset B’ (mean of 1, 2 or 4 sites per year) and ‘dataset C’ (mean of 2 or 4 sites per year), as they did?

Part b) of this comment relates to information available to Timoney & Lee after our research article was accepted at PLOS ONE and which is not available for others to scrutinize. We contend that the new suggestion (via pers. comm. by M. Davies) that 4-6 cm thick sediment samples collected from distributaries of the Athabasca River Delta by use of an Ekman dredge represent the current year’s deposited sediment remains questionable. Perhaps M. Davies can add a comment with facts to support this statement.


Timoney & Lee Comment #4: Although the spring freshet may be instrumental as an agent of contaminant dispersal, the relationship between sediment PAH concentration and discharge of the mainstem Athabasca River and its distributaries is not clear. We tested the hypothesis that Athabasca River mainstem and distributary discharge is related to sediment PAH concentrations in the Athabasca River Delta for both annual and May-August discharge and found no significant correlations (Timoney and Lee 2011). The spring freshet is a time of high discharge during which time there may be little deposition of fines across much of the ARD because flow velocities are sufficient to maintain fine sediments in suspension until they leave the delta front and enter Lake Athabasca. At times of high flow, sand and coarse silts may be preferentially deposited and it is believed that these coarser clasts have lower concentrations of PAHs than do fines. In contrast, flooding of a restricted basin at the time of the freshet could result in sedimentation of fines because flow velocities decrease when the river enters an off-channel basin. Therefore the relationships among discharge and sedimentation may be reversed when off-channel and channel sites are compared. Secondly, mobilization of industrial contaminants stored in the snowpack is a different process than is a freshet-driven erosion event.

Author Response #4:
We are aware that Timoney and Lee (2011) [=ref. 15] attempted to test the hypothesis of whether there is a relation between sediment PAH concentrations and Athabasca River discharge, but that of course assumes that the sediment PAH concentrations represent the year of discharge that they are being compared to. Because there is substantial uncertainty of the time represented by the sediment samples for reasons that we discuss in the paper, we view that such an attempt to evaluate correlation is beyond the limitations of the RAMP dataset. See also Response #3 above.


Timoney & Lee Comment #5: A centuries-long highstand of Lake Athabasca is problematic for several reasons (reviewed in Timoney 2013) that lie beyond the scope of the present paper. Suffice to say that historical accounts, maps, and Lake Athabasca hydrometric data do not support a centuries-long highstand, nor do they support Hall et al.’s view of declining Lake Athabasca levels over the period 1940-82. There was a transient drawdown of Lake Athabasca circa the early 1940s, but from the early 1950s to the late 1970s, levels of Lake Athabasca were above normal (Timoney 2013). A more parsimonious explanation for a low flood frequency at PAD 31 during the period would be sediment and/or organic accumulation at the sill of the basin’s levee, a common process in the delta that leads to changes in flooding rates.

Author Response #5:
Our 12 years of prior field-based research in the region, which has generated unprecedented, informative and accurate records of hydrological change in the Peace-Athabasca Delta and Lake Athabasca from a dozen sites throughout the delta and adjacent areas of Lake Athabasca, has yielded multiple and independent lines of evidence that are consistently conclusive of the fluctuating delta water levels as presented in our research article. These findings have been reported in numerous papers (refs. 22-31), and synthesized in a recent review paper (ref. 32), all of which have passed rigorous peer-review by top-tier scientific journals. Dr. Timoney, who penned the review of our paper before publication, appears to ignore or dismiss these important contributions to the knowledge of the hydrological dynamics of the delta in favour of selected historical accounts he has read about (and which remain in a book that is scheduled for publication next year, and so remains unavailable to us). The study sites are all remote, and so historical accounts may not always be completely accurate to current names of locations. The geographical locations of features named by early explorers can change with the passage of time. We can counter Timoney’s selection of historical accounts that water levels were not high at the time of early explorers with other historical evidence to the contrary. For example, a key piece of evidence known to us is a feature presently located near the western edge of Lake Athabasca known locally as Potato Island (so named since the early explorers) which is now surrounded by dry land (see: http://mapcarta.com/24567...). But, it was an island within Lake Athabasca when it was so named, and when we have reconstructed that Lake Athabasca water levels were higher than at present (and inundated our study lake PAD 31). This debate is beyond the scope of our paper, as indicated by Dr. Timoney. He may choose to disbelieve and dismiss our published evidence, but the peer-reviewed literature and the currently accepted understanding is based largely on our published scientific data which are consistent with maps based on information gathered by early explorers (see historical maps from the 1800s shown in Figure 2 of Sinnatamby et al. 2010 (Citation: Sinnatamby, R.N., Y.Yi, M.A. Sokal, K.P. Clogg-Wright, T. Asada, S.R. Vardy, T.L. Karst-Riddoch, W.M. Last, J.W. Johnston, R.I. Hall, B.B. Wolfe & T.W.D. Edwards. 2010. Historical and paleolimnological evidence for expansion of Lake Athabasca (Canada) during the Little Ice Age. J. Paleolimnology 43:705–717. DOI 10.1007/s10933-009-9361-4).

The source of evidence for Timoney & Lee’s statement that “from the early 1950s to the late 1970s, levels of Lake Athabasca were above normal” remains unknown to us. Perhaps the source is a study by Meko (2012), which provides a reconstruction of Lake Athabasca water levels during 1801-2000 from ring widths of long-lived trees growing in elevated areas within the delta (Citation: Meko, DM (2006) Tree-ring inferences on water-level fluctuations of Lake Athabasca. Canadian Water Resources Journal 31(4): 229-248). However, a main conclusion (P. 245) is that, “the strength of water-level signal in the current network of tree-ring collections is weak and that uncertainty remains in distinguishing the water-level signal from local precipitation and other correlated hydrologic and climatic variables.” Moreover, water-level declines during 1968-1974 to below average values resulted in more than 40 years of controversy about the environmental consequences of BC Hydro’s Bennett Dam installed on the Peace River in 1968 (Ref. 32). Our 12 years of prior research, based on scientifically rigorous and innovative methods involving analysis of multiple and independent sediment records from lakes spanning a range of elevations and geographical locations within the delta, have provided a detailed and accurate understanding of water-level variations throughout the delta and Lake Athabasca.


Timoney & Lee Comment #6: Headwater capture by Cree Creek resulted in a partial avulsion of the Embarras River into Cree Creek, not Mamawi Creek.

Author Response #6:
It is not clear what is meant by this statement, because our text clearly indicates this feature: “PAD 31 became flood-prone for a second time after the 1982 Embarras Breakthrough event, a natural avulsion that diverted substantial flow from the Athabasca and Embarras rivers into Cree and Mamawi creeks and towards PAD 31, …” [page 3 of the PDF file version of our research article].


Timoney & Lee Comment #7: a) The reliance on PAD 31 as the primary PAH sampling basin (discussed below) may be further complicated by its location nearby and downstream of Embarras-Cree Creek avulsion site. The nearby confluence and channel are still evolving due to the changes in flow conditions. Therefore, PAD 31 may receive more locally-eroded sediments than do other basins.

b) The structuring of the data into flood-prone and not flood-prone periods was not justified. If the authors were focussed on determining the relative contributions of natural and industrial sources, a more defensible subdivision would have been pre- and post-industry. But division of the record into a priori periods was not necessary other than as a statistical convenience and imposed an artificial structure onto the data. A better approach would have to been to analyze the data as simple time series.

c) The authors placed inordinate emphasis on floods, which is understandable given that their key lake (PAD 31) is located off-channel and depends on flooding for sedimentation. However, flooding is an episodic event that does not occur annually. If a flood does occur, it may carry sediment into the basin for several days (typically). Conversely, sedimentation at the mouths of the Athabasca River channels and in western Lake Athabasca occurs 365 days per year, every year. The small daily incremental addition of PAHs to the delta’s sediment, unrelated to flood events, may prove to be a more important source of PAHs than is episodic flooding. To use a single off-channel basin as a means to estimate the contaminant contribution to the delta is problematic.

Author Response #7:
We have added “a)”, “b)” and “c)” to their comment #7 to distinguish three distinctive criticisms of our research article.

a) We have carefully presented the merits and limitations of our use of study lake PAD 31 in our research article. We clearly identify in the last paragraph of our research article that further research should undertake synoptic spatial surveys of surface sediments from delta lakes across the broad range of hydrological conditions and river connectivity to further explore relationships between sedimentary concentrations and composition of PACs provided via the Athabasca River.

Indeed, these comments contrast sharply with those we received from two independent peer-reviewers, who were highly positive about our manuscript and did not identify a methodological problem resulting from the selection of the study sites. For example, one reviewer stated, “I found this article to be well organized and clearly written. Its data are telling and the authors have done a thorough job in analyzing their data and in interpreting their data. It is an impressive study. As they note their main results depend on PAD 31 with [its] record being flood-prone during a time since the oil sands development began. Some follow up studies of other lakes in the Athabasca Delta with flooding during this period are now needed to replicate the results from this study. This article should help justify getting those studies initiated. And from what the authors present, I have no reason to believe that the new studies will show much different from what the current evidence shows. I found nothing in need of revision in the current version of the article.” [Emphasis added by us] The second reviewer added, “Establishing geochemical background for substances of geogenic origin is a very complex and difficult task. This also needs understanding of the role of natural factors in behavior and distribution of chemical substances in the environment. The most important result derived from this study is, in my opinion, recommendation of the sites for further detailed monitoring.”

b) Our structuring of the data into periods when the lakes were flood-prone and not flood-prone, based on multiple and independent paleohydrological evidence developed over several years, is, in our view, a particular strength of the study we report. We did not use only a single off-channel basin to formulate our conclusions. Instead, we used two off-channel basins with contrasting temporal patterns of basin hydrology, and we compared the results with available data from an ‘on-channel’ basin (Richardson Lake from Evans et al, 2002 [= ref. 12 of our research article]). This approach provided an ability to discern that the composition of PACs differs significantly when lakes receive input waters from the Athabasca River compared to when they don’t. And, the approach allowed us to identify the specific PACs, of several dozen that exist in nature and are produced by natural and anthropogenic processes, which have an association with bitumen deposits of the McMurray Formation and are carried by the Athabasca River to the delta during periods of elevated flow. Indeed, periods of elevated flow, particularly during the spring freshet when shoreline erosion is most intense (ice-scour) and accumulated industrial contaminants in snowpacks are released (but also other times during the open-water season when flows are elevated), are the times that have been hypothesized (by Dr. Timoney and others; see refs 10, 11, 15, 20 cited in our research article) as the most important for distributing oil sands contaminants by natural as well as anthropogenic processes. As we highlight in the first paragraph of the Discussion, without knowledge of shifting hydrological conditions at the study sites conflicting and inaccurate interpretations are possible. Indeed, strict time-series comparison of the results from PAD 31 and PAD 23, without consideration of paleohydrological knowledge and context, could have generated entirely opposite conclusions. We contend that our approach was particularly well-suited to identify the compounds delivered by the Athabasca River and address key concerns about effects of oil sands development on contaminant delivery.

c) Sceptics may well continue to disagree with us about whether PAD 31 is well suited to serve as a sensitive recorder of changes in river-borne contaminants. Timoney & Lee’s comment here echoes a point we have discussed with a few colleagues, which postulates that one or more of the five large open-drainage lakes within the Peace-Athabasca Delta could serve as sensitive recorders of changes over time in river-borne contaminants, because they ‘sample’ flows from the Athabasca River and tributaries continuously or near-continuously. But, this argument suffers from a naïveté of the hydrological dynamics and evolution of the delta. Two of the five open-drainage lakes (Claire, Baril) do not receive river flows exclusively from the Athabasca River, and so varying relative contributions of Athabasca, Peace and Birch river flows to these lakes over time will confound attempts to relate sedimentary PAC profiles to oil sands development along the Athabasca River. Furthermore, variations are well known to have occurred over time in the distribution of Athabasca River flow among the distributaries within the Athabasca Delta that also confound attempts to interpret records of PACs in sediments from the other three open-drainage lakes (Mamawi, Richardson, Blanche). Specifically, build-up over time of sediments deposited near the current mouth of the Athabasca River (where it enters Lake Athabasca) has, during the past several decades, been increasingly directing flow away from the current delta mouth and northwards via Cree and Mamawi creeks into Mamawi Lake. This feature has been increasingly routing river-transported sediment (and associated PACs from bitumen) away from Richardson Lake and Blanche Lake and into Mamawi Lake, a feature that is well documents in ref. 27 by our group (Wolfe et al., 2008). Such hydrological changes potentially confound attempts to utilize sediment records from these open-drainage lakes to assess oil sands development on sediment and contaminant delivery (i.e., fluxes) to the downstream delta. Moreover, the open-drainage lakes are shallow (<2 m) and large, resulting in strong influence of wind-induced waves and currents on redistribution of sediments – a feature that will negatively affect the ability to accurately date sediment cores from these lakes and generate reliable profiles of temporal changes in PAC concentration over time. For these reasons, we intentionally avoided the open-drainage lakes to address the goals of our study even though they receive river inflow during the entire year. Instead, we focused the study on two restricted-drainage lakes with differing patterns of flood frequency, and one closed-drainage lake, for which we had prior detailed and robust knowledge of the shifting hydrological conditions, to directly test if the Athabasca River and the atmosphere have been important vectors of oil sands contaminants to the delta even before oil sands development began.

As we develop in our research article, PAD 31 was carefully selected as an extremely informative site to address the goals of the study. Indeed, based on > 6 years of intensive field-based research by our group, PAD 31 is one of the most flood-prone of the restricted-drainage lakes in the Athabasca Delta. High energy flooding (standing waves flowing over the river’s sill) has been observed during spring floods associated with ice-jam events, but the lake also receives floodwaters in the summer following large rain events and associated increased river levels within the lower-Athabasca River catchment [ref. 60]. Unfortunately, knowledge of past hydrological changes is available for only a small number of lakes in the Athabasca delta, because it is very costly to develop. Most of the lakes in the Athabasca Delta are restricted-drainage lakes. Indeed, floodwaters, when they occur, often inundate broad areas of the delta with waters of similar chemical conditions, as we have shown for suspended sediments, nutrients and ions (see refs 25, 60 in our research article). Thus, we anticipate the test results based on our study site (PAD 31; = sample of data) can reasonably be extrapolated to a large number of other flood-prone (or, restricted-drainage) lakes, which are numerically dominant in the delta (statistical population). But, given the potential levels of scrutiny we anticipated this research article could attract (it tackles a socially relevant polarized debate), we were careful to avoid implying that we extrapolate results from PAD 31 to the entire delta. Indeed, a careful read of our research article should readily identify that we never extrapolated the results based on study lake PAD 31 beyond that lake. Certainly, no extrapolation was made based on results from PAD 31 to the entire delta, as accused by this comment.

Moreover, we emphasize that our findings identify a need for detailed monitoring at more sites. And, we present a section at the end of the Discussion that addresses this and other key limitations of our study, and recommendations for future improvements. This text includes comparison of past PAC concentrations measured in our study lakes, delivered by short-lived periodic flood events, with concentrations reported for the same time periods in Richardson Lake (by reference 12), which receives continuous inflow from the Athabasca River. The close correspondence of petrogenic PACs in these lakes is suggestive that the periodic floods are a major contributor to annual loading of petrogenic PACs to the delta lakes. As the two independent peer-reviewers identified, our study does identify key sites for effective ongoing and future monitoring of contaminants supplied via the Athabasca River, and our research article should lead to more studies of other lakes within the delta.


Timoney & Lee Comment #8: The statement “Based on geochemical fingerprinting, the main source of the PACs in the natural oil sands region river sediments is bitumen-rich material in the riverbanks” may be misunderstood. The McMurray Formation geological deposit that is the source of the PAHs is the same deposit that is being exploited by industry along the Athabasca River’s banks. The authors’ predilection to emphasize natural erosion, while failing to differentiate natural from industrially-enhanced erosion, is a recurrent theme in the paper.

Author Response #8:
The statement referred to here (last paragraph on page 6 of the PDF version of our research article) contains text that is specific to findings reported from independent studies conducted by other scientists and that we cited here (rather than an interpretation by us of our data or the data from others). The cited studies analyzed PACs in samples of oil sands and river sediments that were obtained from locations deemed by those authors as unaffected by Alberta oil sands development. Perhaps Timoney & Lee are under the mistaken impression that we were referring to our 2007 Athabasca Delta flood deposit? But, the statement they identify clearly does not refer to our 2007 Athabasca Delta flood deposit, which we consistently and clearly refer to as the ‘2007 Athabasca Delta flood deposit’ throughout the manuscript (to avoid this sort of possible confusion). Instead, the goal here was to compare the composition of PACs in the 2007 Athabasca Delta flood deposit we obtained with the composition of PACs in sediments from natural oil sands region sediments at upstream sites near the McMurray Formation that, as stated by the authors of those studies, were unaltered by anthropogenic activities.


Timoney & Lee Comment #9: Similarly, the authors do not examine other sources of PAHs to the delta, nor do their study sites and sampling design seem able to detect them. For example, a Suncor pipeline break in 1970 spilled three million L of oil into the Athabasca River; the oil flowed into Lake Athabasca and was observed there for about six days. In 1982, a spill from Suncor necessitated closure of the commercial fishing season on Lake Athabasca and reportedly caused illnesses among people in Ft. Mackay. The amount of oil and/or bitumen that was deposited in the delta’s sediments was never determined. A considerable volume of PAHs would have been delivered to the delta in these two events that were evidently not detected by Hall et al. The spills were not associated with floods and therefore their study sites may not have sampled the events.

Author Response #9:
In earlier responses to their comments we have addressed many of the issues related to the suitability of lake PAD 31 to provide a useful record of anthropogenic contaminant supply, because it is a restricted-drainage (or, off-channel) lake that only receives river-borne materials when it is actively flooded. It is true that our study sites cannot detect materials from industrial spills when they pass the lakes at a time they are not being flooded. However, in Response #7, we identify that even open-drainage (or, online) lakes have limitations that will render sediment-based analyses ineffective at identifying such spills, even though they receive the contaminants. PAD 31 is amongst the most frequently flooded of all lakes in the Athabasca Delta and so its interpretable sediment record has distinct value, which we have attempted to present in a balanced approach in our research article. Moreover, Timoney & Lee and others (cited refs. 10, 11, 15, 20) have identified that river flood stages, especially the spring freshet, but also summer floods, are particularly important times when pulses of anthropogenic contaminants from oil sands development become transported to the Athabasca River and dispersed downstream. Indeed, residues of oil spills often accumulate near and above the water line for some time until more elevated flood stages cause their transport downstream. So, our offline lakes do have the potential to receive materials from oil spills and other industrial processes, even if they are not sampled in ‘real time’ of the associated releases. Accumulated contaminants in the winter snowpack, as shown by ref. 10, certainly have high potential to become delivered to PAD 31. As indicated in Response #7, we included a section entitled ‘Limitations’ at the end of the Discussion that succinctly addresses this topic (that PAD 31 does not sample the river continuously), as well as other key limitations of our study and recommendations for future improvements (see the second paragraph of this sub-section).

Timoney & Lee provide some details of two specific oil sands spills. PAD 31 would not have received contaminants from the spill in 1970, as this lake had closed-drainage hydrology at that times (pre-Embarras Breakthrough). The 1982 spill might have been recorded, depending on when the spill occurred.


Timoney & Lee Comment #10: a) The authors state: “During flood-prone periods, PAC composition closely matches that of river-transported sediment originating from bitumen deposits of the McMurray Formation exposed along the riverbanks”. The authors appear to think that PAHs accumulated during flooding at PAD 31 all originate from erosion along riverbanks. Failure to consider how industry can contribute to PAH loading from erosion is unfortunate. b) The authors then state that: “During periods of reduced flood frequency, higher proportions of unsubstituted PACs (notably N, B, F) identify greater influence of hydrocarbons from fire and catchment vegetation”. Fire data do not support that assertion. According to Wood Buffalo National Park and northern Alberta fire records, 1953 was the largest fire year in the delta and its immediate environs over the period of record (1950 to present). Unfortunately, none of their basins detected the 1953 major fire. Nor did their isolated basin, PAD 18, detect the major 1981 fire (see Timoney 2013 for fire history).

Author Response #10:
We have added “a)”and “b)” to their comment #10 to distinguish two distinctive criticisms of our research article.

a) This statement synthesizes information on the composition of PACs in sediments from PAD 31 deposited during flood-prone periods both before and since onset of oil sands surface mining and processing. PAC composition was remarkably similar both pre- and post-development, and comparable to bitumen deposits obtained from non-impacted areas of the McMurray Formation exposed in riverbanks based on information obtained from refs. 10 and 44. PACs in samples deposited in PAD 31 prior to 1967 were not affected by oil sands development, as industrial development only began in 1967. PACs in samples deposited prior to about 1900 were likely unaffected by any human activities.

b) Study lake PAD 18 is the site that contributed information about contaminants delivered via the atmosphere, including the role of fires. The sampling resolution for the core from PAD 18 (on average 6.1 years per sediment sample analysed for PACs) was not designed to be able to identify individual fire events. Instead, the study focuses on assessing longer term patterns of change and trends (at sub-decadal resolution) to address the stated objectives (to assess baseline pre-development levels of PAHs and natural variability, compare pre- versus post-development levels of PAHs delivered via the atmosphere). Thus, it is not a surprise to us that the data do not pick up fire events of particular years, such as 1953 and 1981 highlighted by Timoney & Lee. Our data do show that during the past 100 years, the 1950s corresponded with rather high concentrations of PAHs at PAD 18. The early 1980s do not reflect occurrence of large fluxes of PACs from fires, as Timoney suggests occurred (but the evidence for this contention remains unavailable in a promised 2013 publication). But, not all fires in Alberta will deposit PACs to PAD 18 and the delta region, and not all fires produce the same amounts of PACs. So, reasons why low PAC concentrations occurred during the 1980s remain unknown (and outside the scope of this study). Timoney & Lee did not provide the context of what time period they are basing their comparisons on (available only in a 2013 publication not yet available), but the data are likely in government agency records that do not extend back through the 1700s and 1800s when we show fire-supplied PACs were larger and corresponded with a regionally arid climatic period known as the Little Ice Age at a time prior to extensive forest clearance and fire suppression by humans.

In a recently published paper by our group (ref. 60 [= Wiklund et al., 2012]), we show that the sediments at PAD 18 (analyzed at the same temporal resolution as in our research article) possess a reliable record of long-distance metal contamination via the atmosphere, which corresponds well with other sites throughout temperate and subarctic regions of North America. This paper shows patterns of changes in lead, arsenic, antimony and mercury that are typical of 20th century industrial airborne metal emissions in North America. The trends can be explained by post-1920 increases in long-distance atmospheric transport of industrial emissions, followed by decreases arising from the installation of emission control devices on coal-fired power plants and metal smelters after 1950, and the phase-out of leaded gasoline in the 1970s. Similar to the patterns of PACs shown in our research article, the Wiklund et al. (2012) study shows metal deposition has declined since the late 1960s, despite onset and expansion of oil sands development. Wiklund et al. (2012) conclude that oil sands development is not increasing the supply of airborne metal contaminants to the Peace-Athabasca region – a finding that is consistent with the PAC data presented in our research article.


Timoney & Lee Comment #11: Figure 4. The match in the PAH fingerprint for ‘oil sands samples’ and the ‘2007 flood deposit’ indicates that the McMurray Formation is the primary source of the sediment PAHs at the authors’ sites. This is to be expected. It is the geological formation that is being mined and processed in the region. The data do not argue for or against industrial vs. natural sources.

Author Response #11:
Similar to the above comment and response. The composition of PACs in sediments of PAD 31 deposited during a flood-prone period prior to as well as after onset of oil sands development closely matches that of the 2007 Athabasca Delta flood deposit and the ‘oil sands’ samples, which suggests natural erosional processes have been a main process delivering PACs to PAD 31 and they continue to be. This feature is well developed and clearly explained in our research article.


Timoney & Lee Comment #12: a) Figure 8 presents the crux of the results. PAD 31 indicates no trend in PAH concentrations until recent decades, at which point some of the highest concentrations over the 300-year record are observed. At one extreme, the data could be interpreted to indicate that natural erosion is the sole source for the pattern, which is the view of Hall et al. At the other extreme, the data could be interpreted to indicate that, with the onset and growth of the bitumen industry, PAH concentrations increased. Both interpretations would be over-reaching the data. A more defensible interpretation is that changes in water and sediment inputs from the Athabasca River into this off-channel basin underlie changes in sediment PAH content. The data are mute as to differentiation of natural vs. industrial sources for the riverborne PAHs. This has been the scientific challenge for decades and remains unresolved in this paper.

b) PAD 23 indicates a decline in PAH concentrations over the record from 1900 to about 1972 (when the engineered Athabasca River Cutoff was established, leading to a decline in hydraulic connectivity for the basin). Post-Cutoff, the basin became isolated and therefore was not ‘sampling’ potential industrial inputs; PAH concentrations have oscillated without a trend. The data indicate the obvious: that hydraulic connectivity is positively correlated with riverborne PAH inputs. The data are mute as to natural vs. industrial sources for the riverborne PAHs.

c) PAD 18 indicates overall lower concentrations of sediment PAHs in this isolated basin than in the other two basins, demonstrating the importance of hydraulic connectivity.

d) The inconsistent sensitivity of the sites to fire is noteworthy. Although the authors ascribe a PAD 23 peak in PAH concentrations to fire in 2010, that conclusion is tenuous. Fire records demonstrate that there were no fires in the delta in 2010. Regionally, 2010 was not a fire-prone year in northern Alberta. The closest fire of any size in 2010 was located 63 km east-southeast of PAD 23. Examination of PM2.5 data for the Fort Chipewyan air quality station indicates a transient spike in PM2.5 concentrations in late July of 2010. Whether this spike would be sufficient to be detected in the PAH record at PAD 23 is unknown. Given a wildfire large enough to be detected in a lake record, it is curious that a similar spike in PAHs would not detected at PAD 31. Lack of time prevents our examination of the PM2.5 record, but such an examination may prove interesting in regard to wind direction, industrial and wildfire particulates, and the lake PAH record.

Author Response #12:
We have added “a)”, “b)”, “c)” and “d)” to their comment #12 to distinguish four distinctive criticisms of our research article.

a) This possible source of confusion is exactly why our research article focussed on identifying and comparing only the PACs which have an association with bitumen and transport by the Athabasca River to the study lakes. The use of these PACs (Figure 7) rather than all of them is particularly instructive to underpin our comparison of pre- versus post-onset of oil sands development. Indeed, our research article communicates that there are no statistically significant differences in the sedimentary concentrations and proportions of the river-transported bitumen-associated indicator PACs at PAD 31 during the flood-prone pre-1940 and post-1982 intervals (Figure 7ab), which supports the conclusion that we cannot detect measureable increases in the concentration or proportion of these PACs at lake PAD 31 during the time when oil sands industrial development has occurred.

b) Timoney & Lee are correct on this point. We never used the data from PAD 23 to compare PAC transport pre- and post-onset of oil sands development, because an engineered channel excavation reduced river connectivity of this at the same time when oil sands development occurred. Instead, we used only the pre-oil sands development period to gain appreciation that natural processes transport PACs to the delta. It is a key component of the study.

c) We agree with this statement. This feature helps illustrate that natural erosion of the riverbanks and transport by the Athabasca River are indeed important vectors of PAC transport to the delta.

d) The information in this comment is speculative. We collected the cores from PAD 23 in September 2010, following a fire located 63 km away. We can’t definitively identify if this fire affected PAC content of the uppermost sample.


Author Response #13:
This comment merely provides an extensive quote of text from our research article. We cannot detect a question or comment that deserves addressing. Insinuation that we had “a single-minded purpose” is speculative and not based on evidence. Indeed, the statement is not true and it contravenes PLOS ONE’s regulations for good practice for Comments. Specifically PLOS ONE requires that “Unsupported assertions or statements should be avoided. Comments must be evidence-based, not authority-based” and “Discussions should be confined to the demonstrable content of articles and should avoid speculation about the motivations or prejudices of authors.”

Author Response #14:
We feel that we have already addressed all these points of criticism in our responses to earlier comments.

Author Response #15:
We feel that our research article has accurately portrayed a comprehensive understanding of the processes delivering river-transported sediment to floodplain lakes of the Athabasca Delta, as required to assess the veracity of our conclusions. Certainly, dynamics of flood events are complex, including the sorting of sediments of various grain sizes. Nevertheless, in our research article we have provided a significant advance in understanding of PAC transport to delta lakes by floodwaters. Indeed, this comment by Timoney & Lee aligns very well with the features presented in Figure 6 of our research article, where relationships between sediment organic matter content and river-transported bitumen-associated PAC content are shown to differ between river sediments and restricted-drainage lake sediments. [Note that the data from the river sites in this graph are the same RAMP data used by Timoney & Lee in ref. 15]. In the rivers, sediment organic content is related to grain size: coarser sediments tend to have lower organic content. In river sediments, hydrological variations tend to influence grain size and associated concentrations of hydrophobic organic contaminants such as PACs. Thus, for reasons outlined in this comment by Timoney & Lee, their analysis of the river sediment data [in ref. 15] should have accounted for such grain-size variations by using common methods such as those presented in ref. 57.

Author Response #16:
This comment confuses two different issues. One issue is about how representative results of analyses from a sediment core study (= paleolimnological study) from one lake can be of the entire floodplain landscape. We addressed this concern in our responses to earlier comments. We certainly advocate that further paleolimnological studies at additional sites in the delta and along the Athabasca River upstream of the delta would be helpful to develop a more comprehensive understanding.

The second issue is about how many sampling sites you need to monitor in real-time to assess relationships between river connectivity and the composition and concentration of PACs in recently deposited lake sediment samples (= contemporary study), and to develop long-term monitoring records able to detect trends into the future. This second set of questions would require around 40 or more lakes to span the hydroecological gradients in the delta, and should be based on methods developed in ref. 25. We remain unsure why Timoney & Lee have confused the sample size needed for contemporary monitoring versus paleolimnological studies. They are distinctive and require different numbers of sample sites to address different questions.

Author Response #17:
Findings we report from our study, based on sediments delivered to the study lakes before onset of Alberta oil sands development, clearly identify that natural river processes have been (and continue to be) an important source of bitumen-associated PAC delivery to the delta. In the paragraph that Timoney & Lee refer to here (second paragraph of P. 15 of the PDF file version), we actually present information contained in publications provided by others (not by us), which use independent methods to provide important data about how substantial the natural erosion processes are in delivering sediment, bitumen and associated PACs to the delta. Text in this paragraph merely uses values provided in publications by others that report rates of natural erosion of sediment and bitumen along the Athabasca River downstream of Fort McMurray in areas purported by those authors to be unaffected by oil sands industrial activities. We believe this is a relevant approach which was not driven by an unwarranted fixation on natural erosion. Quantification of erosion rates by Alberta’s oil sands development is an enormous task, well outside the scope of our study. To our knowledge, data are not available that we could use to quantify industrial contributions to erosion.

Author Response #18:
The main reason for the comparison we made here is that we are concerned our findings (that the Athabasca River has long supplied substantial loads of bitumen to the delta – to the estimated order of an Exxon Valdez crude oil spill every 4-8 years) could be taken out of context and raise fears in the general public that humans and biota have long been exposed to extremely high and dangerous levels of PACs in the delta. However, based on information we have generated in this and other publications (e.g., ref. 32) the delta remains by-and-large a natural or pristine environment (it is protected within a National Park, the hydrology of the delta continues to be regulated mainly by natural factors (climate)). Thus, some type of context is needed, and we think the comparison is helpful in this regard. For reasons detailed below, we cannot compare our data against thresholds for toxic effects, which is one logical way we thought of to provide context. However, much of the general public (and readership of PLOS ONE) lives in urban environments, and so comparison to levels in urban soils will mean something to urban-based decision makers and many others. Thus, we feel our comparison is helpful and justified, though we do appreciate the aboriginal and rural perspective that urban folk do not rely as heavily on hunting and fishing for their nutrition. Secondly, we feel that we can make the comparison between our data and data presented in the two papers we cited for PAC levels in urban soils. This is because the study by reference 65 (Mauro et al., 2006) analysed 43 PACs (including alkylated PACs) which included most of the same compounds measured in our study. Their study included 319 samples collected across 29 urban areas from 3 US states. The study by reference 64 (Bradley et al., 1994) analysed only 17 EPA priority PAHs (from three cities in New England, N =62), but these 17 compounds were all measured by the methods we used in our study. So, because the total PAC concentrations they reported for the 17 PACs exceed the total PAC concentrations we reported for those 17 compounds plus many more compounds, our statement remains correct.

We did consider comparing the concentration of PACs in lake sediments to published thresholds for toxic effects for biota. Indeed, this was recommended to us by Dr. Timoney in an earlier review prior to the Academic Editor’s decision to accept our research article for publication in PLOS ONE. However, the goal of this particular paper was to examine the sources and loadings of PAHs to sites within the delta. And, our next phase of research intends to examine the toxicity of these sediments in a risk assessment framework. In the current research article, we decided that attempting to compare our sediment PAC concentrations with published PAH effects data would be reckless given the importance of these findings to people populating the region of study. One of the greatest challenges when comparing sediment or effects "total or summed" PAH concentration data between studies is the EXTREME INCONSISTENCY in the number of individual PAH compounds that were analyzed as part of the so called "total or summed" PAHs for any given study. For example, Dr. Timoney had suggested we compare our sediment concentrations to a paper by Myers et al. (2003. Citation: Myers MS, Johnson LL, Collier TK (2003) Establishing the causal relationship between polycyclic aromatic hydrocarbon (PAH) exposure and hepatic neoplasms and neoplasia-related liver lesions in English Sole (Pleuronectes vetulus). Human and Ecological Risk Assessment: Vol. 9, No. 1, pp. 67-94), but the list of PAHs analyzed is not available. A quick search of other research presented from the same institution (NOASS/NMFS) suggests that in general most of the reports likely measure mainly 16-18 PAHs, mostly US EPA priority PAHs. These priority PAHs are parent compounds derived from combustion. However, it is well understood that petroleum-related PAHs, which are the main concern in the delta, are dominated by alkylated PAHs and other polycyclic aromatic compounds such as dibenzothiophenes (collectively referred to as PACs). In our study, a total of 52 PACs (this includes both parent and alkylated PAHs and S-containing compounds) were analyzed for the determination of sediment concentrations, and 43 of them occurred at concentrations above the detection limits. It would be reckless to attempt to compare our sediment total concentrations of 43 PACs with studies reporting thresholds based on ‘totals’ that have only measured a fraction of the relevant PACs present in petroleum impacted sediments (i.e., Myers et al. 2003). Their thresholds would be much higher if they considered the full suite of PAH/PACs as included in our study!

Author Response #19:
See our responses #4 and #7.

Author Response #20:
a) We agree that similar paleolimnological studies of contaminant trends should be performed on closed-drainage lakes at varying distances and directions from oil sands operations in the Ft. McMurray area to map out the spatial and temporal footprint of contaminants delivered via the air. To our knowledge, a collaborative project between Environment Canada and an academic researcher is underway in this regard (D. Muir, Environment Canada, pers. comm. September 27, 2012).

In the context of our study and the contaminant load delivered via the Athabasca River to the Athabasca Delta, it remains uncertain how much of the estimated 600 tonnes of bitumen accumulated in a 4-month snowpack within a 50-km radius of the oil sands main facilities actually become mobilized to the river. This remains a key unknown for future research. Data from ref. 62 identify that natural erosion of bitumen exposures in the river banks contributes ~5,000 to 9,000 tonnes of bitumen per year. Our study shows that the industrial load has not led to an increase in the concentration and composition of bitumen-associated river-transported PACs to flood-prone lake PAD 31 compared to pre-industrial times.

b) Certainly, we agree that more research and data are needed to ‘fingerprint’ sources of contaminants in the region.

However, the last sentence of their comment (“ it strains credulity …”) appears to us to contravene PLOS ONE’s guidelines of Good Practice for Comments. Specifically PLOS ONE requires that “Unsupported assertions or statements should be avoided. Comments must be evidence-based, not authority-based.” And that “Arguments based on belief are to be avoided. For example the assertion, "I don't believe the results in Figure 2." must be supported”. Here, Timoney & Lee appears to convey their belief that given the substantial area of land affected by oil sands mining, there must be increases in erosion into the Athabasca River and that any study which does not detect these changes has structural problems. Our approach was to address as objectively as possible the question, “has Alberta oil sands development altered delivery of PACs to the delta?” using carefully selected sites and rigorous scientific methods that allow the hypothesis to be tested (both via the Athabasca River and the atmosphere). Indeed, the initial peer reviewers of our paper found our study to be impressive and that we analyzed and interpreted our data thoroughly.

Importantly, our analyses of the data from PAD 31 (specifically the results of the independent-samples t-test) do not support their contention that the oil sands industry has altered the types of erosion to be detectable 200 km downstream, because the concentration and proportion of the bitumen-associated river-transported PACs in sediments deposited in lake PAD 31 during flood-prone periods have not increased significantly post-oil sands development compared to pre-development.

Author Response #21:
We feel we have adequately addressed these issues within our research article and in the above responses.

Author Response #22:
No response is required. We are not in a position to speculate about the how evaporation in the future will play a role on contaminant effects on biota.

Author Response #23:
As demonstrated in a comprehensive review article (ref. 32), during the past 12 years of active research we have develop amongst the most comprehensive understanding of the complexities and dynamics of processes that have shaped and continue to shape the Peace-Athabasca Delta. Despite the hydrological complexity, we have been able to carefully select floodplain lakes for paleolimnological analyses which have been highly informative of processes playing out across the delta.

Here, we remind readers that we have addressed in response #7c the issue of the role that PAD 31 (a single lake) has played in our research article, including that we did not extrapolate the findings at this lake to the entire delta, as Timoney & Lee have indicated.

Author Response #24:
We’ll leave it to the readers to critically assess our study and research article on their own. We are the first to provide critically needed information on baseline contaminant levels and changes over time spanning centuries before oil sands development and several decades since onset of development in a region where there is considerable public debate and controversy. We paid careful attention to the study design, including important knowledge of hydrological variations, which are key to begin to decipher relative roles of natural versus industrial processes.

Author Response #25:
This comment appears to us to stray from PLOS ONE’s guidelines of Good Practice for Comments. Specifically PLOS ONE requires that “Unsupported assertions or statements should be avoided. Comments must be evidence-based, not authority-based” and “Discussions should be confined to the demonstrable content of articles and should avoid speculation about the motivations or prejudices of authors.”

The genesis of our research article began 12 years ago, when we collected our first sediment core from an oxbow lake known locally as “Horseshoe Slough”. We obtained a 3.6-metre long sediment core from this frequently flooded lake that possessed alternating lighter and darker bands of sediment. We discovered the darker bands are deposited by flood events and the lighter bands during non-flood periods (refs. 24, 32). Indeed, the flood-frequency record we generated from this sediment record closely matched a flood record based on Traditional Knowledge and historical archives (see refs. 24, 32). A 1-cm thick black oily layer occurred near the base of this core, which is bitumen-rich sediments deposited by a flood event in the 1820s from the Peace River. With this discovery, we became overtly aware of the complex dynamics of this delta, because it told us that ecologically important flood events have long been depositing naturally-occurring contaminants to the delta (like the Athabasca River, the Peace River also traverses surface deposits of bitumen, though a much smaller area). In recent years, polarized debate has developed around oil sands environmental consequences. With regards to the Peace-Athabasca Delta, the debate can be broadly characterized as industry and government contending that the contaminants are natural in origin, while local communities and aboriginal and environmental organizations arguing that pollution emitted by the oil sands development is threatening the health of humans and ecosystems. With the knowledge we had from Horseshoe Slough, we knew natural processes played a role and that our scientific skillset could begin to address the relative roles of natural and industrial processes. Our intent was to explore the shades of grey between the black and white polarization of the debate. Our research article presents, we feel, a balanced approach to communicate the outcome of our enquiry.

Again, we’ll leave it to the readers to decide for themselves.

No competing interests declared.