Comment on “Abrupt environmental change in Canada's northernmost lake inferred from fossil diatom and pigment stratigraphy” by Dermot Antoniades et al.
This is a commentary on DOI:10.1029/2007GL030947
 Antoniades et al. [2007; hereinafter referred to as A07] present an analysis of a 19-cm lake sediment core from Ward Hunt Island from northern Canada. They conclude that a significant and abrupt change in the aquatic communities occurred in this lake during the past 2 centuries that was unusual with respect to the previous 8ka. In the paper, and especially in quotes to the press (e.g., http://www.cbc.ca; http://www.sciencedaily.com; http://scitizen.com) the authors conclude that their results, along with those from Smol et al.  provide evidence for amplification of climates with latitude due to human caused global warming, an event unique in the past 8 ka. However, there are a number of problems in these data and flaws in the author's interpretations that render their results questionable. Their conclusions do not hold up to scrutiny as they do not concord with other information we have about climate variability in the Arctic. I have four major comments on the data, and several on the interpretation.
1. Sample Collection
 Details are lacking about the sample collection, but are critical to the interpretation of the data. The depth of the water where the core was collected was not provided. It was apparently collected from the moat zone; the maximum depth is reported at 5 metres, but the ice thickness is 4.0 – 4.1 metres. If scuba divers did manage to squeeze under the ice, how did they avoid disturbing the sediment? However, if collected from shallow water, this means the entire water column and sediment are frozen much of the year.
 Sampling Arctic sediments, especially frozen sediments from shallow lakes, is fraught with problems [Nichols, 1967]. It is possible that most of the sediment accumulation was simply scoured away or slumped into the deeper portion during warm periods when the lake did open. Freezing a wet sediment core within a core tube may disturb the stratigraphy as the frozen water expands. Finally, why did the authors collect the core by scuba divers, with all the attendant problems? Core collection methodology in the Arctic is well established [Gajewski and MacDonald, 2004], and by using a Livingstone-type sampler the authors could have collected a longer core with wider diameter and cleared up some of the ambiguities caused by this short sequence. In their reply [Antoniades et al., 2008, hereinafter referred to as A08], they speculate that the sediments are undisturbed, but provide no data to justify their assertions.
2. Dating of the Sediments
 The authors attempted to date the sequence using 210Pb, 237Cs and 14C dating methods. Insufficient information was given about the 14C date; some provided in A08, some still lacking (the actual depth of sediment dated, e.g., 18–19 cm; the bedrock of the area).
 Without this information, we cannot make critical sense of this date, which is extremely suspect, and the information they provided in A08 simply confirms that we should regard the chronology with caution. Experience with radiocarbon dating in the Arctic indicates that it is necessary to be cautious in interpreting chronologies [Gajewski et al., 1995; Wolfe et al., 2004] especially with only one date for an 8000 year sequence. In many Arctic cores, there are frequent reversals that can only be identified by obtaining multiple dates. Bulk sediment can easily be several thousand years too old [Gajewski, 1995; Gajewski et al., 1995].
 Several alternative sedimentation models could be proposed for this site: extremely slow accumulation due to an unproductive lake, intermittent sedimentation, or one or more hiatus in the core. Cores collected from shallow water frequently contain hiatus. But we have no way to chose among these alternatives, given the lack of information about the core.
 A07 report that the 210Pb and 137Cs dating was also problematic, as is common in Arctic sediments. However, the authors are unduly confident in their acceptance of the 137Cs peak, which they admit was “weak”. There is no explanation of why there is no Pb deposition but there is Cs. Without a secure chronology, there is little that can be interpreted from this core.
 There is one conclusion we can make from the radiocarbon date. If enough organic matter was being deposited such that a radiocarbon date could be obtained from the deeper sediments, then the lake should have been open at this time in order to accumulate organic matter from the watershed or lake.
3. Pigment Data
 The conclusion is based on an increase in pigments in the upper 2 cm. However, there is neither indication of instrument error nor apparently any replication. I realize that it is not common in paleoclimate work to provide error bars. However, A07 note that studies of pigments are relatively rare, so it would be helpful to see if the small peaks in the upper sediments are reproducible. The sense given in review articles [Leavitt and Hodgson, 2004] is of the potential of these studies, but that consideration must be taken of problems and possibilities of error, especially when interpreting the uppermost sediments. Given the lack of experience in pigment studies in Arctic environments, perhaps the methodology should be explored using more typical sites to check for potential problems.
4. Diatom Analysis
 Since diatoms disappear below 2 cm, they conclude this is due to continuous ice cover. However, diatom-free zones are frequently found in Arctic lake sediment cores. In some lakes, diatoms are missing in the sediments, although nearby lakes contain them [Smith, 2002]. Several alternatives can be proposed: there are few diatoms in the lake or they are being re-dissolved in the water or on the sediment surface before burial, especially given the long residence time of the surface sediment in contact with the water.
 In other sites, diatoms disappear after a short depth in cores. A possible interpretation is that are being dissolved in the sediment. Many Arctic cores consist of a thin layer of brownish and organic-looking sediment at the sediment-water interface, followed by a transition to clayey sediment. The abruptness of this transition observed in many lake sediments would lead to the results reported by A07.
 In yet other sites, diatoms appear and disappear through the core. The reasons are not clear [Podritske and Gajewski, 2007]: dissolution, permanently ice-covered sites, dilution by incoming sediments. The dissolution can be rapid, appearing as an abrupt change. One aid to interpretation would be to plot diatom concentrations. This may enable us to determine if the appearance of diatoms in the upper sediments of the Ward Hunt Lake core is really due to colonization of the lake by the diatom, or simply due to a small number of valves that have not yet been dissolved.
 Thus, an alternate conclusion to that of A07 is the following:
 1. The short core collected using non-standard methods could have many problems, including hiatus, disturbance, slumping, etc.
 2. Since the core is essentially undated, no real conclusions can be made about the stratigraphy or climate variations. Insufficient information was presented in the original paper, making evaluation impossible.
 3. The changes in the diatom communities could be small, since concentrations were not reported. There could, in fact, be no environmental change at all but only dissolution in deeper sediments.
 4. The changes in pigments can also be explained by diagenesis.
 The authors note that their conclusions are similar to those of Smol et al. ; not surprising, since this study is also based mostly on very short cores. They further conclude that the changes in Ward Hunt Lake predated similar changes in nearby Alert [Antoniades et al., 2005]; however, any comparison of this sort is premature. More importantly, the interpretations of A07 are at odds with results from most proxy records from across the Canadian Arctic, which show evidence of a warm early Holocene and an alternation between warm and cool periods, synchronous with millennial-scale climate variability observed elsewhere.
 Evidence from across the Arctic shows large climate changes before the current warming [Gajewski and Atkinson, 2003]. A pollen diagram from northernmost Greenland, around 600 km to the east [Fredskild, 1989] showed warmer conditions prior to 3300 14C BP. During the early and mid Holocene, extensive peat deposits accumulated on the Fosheim Peninsula, whereas accumulation ceased in the late Holocene [Garneau, 2000]. The Agassiz ice core shows extensive summer melting prior to 6ka [Fisher et al., 1995]. Driftwood and whalebone records indicate alternating cold and warm periods across the Canadian Arctic during the Holocene [Dyke et al., 1996, 1997]. Pollen and diatom records show a warm early Holocene across the Canadian Arctic, including Ellesmere Island [Hyvärinen, 1985; Gajewski, 1995; Smith, 2002, Wolfe, 2003; Wolfe and Smith, 2004; LeBlanc et al., 2004; Finkelstein and Gajewski, 2007; Podritske and Gajewski, 2007; Peros and Gajewski, 2008].
 In summary, climate variability of several scales is reported from dozens of sites of the Canadian Arctic that have records extending beyond the past 200 years. What are the synoptic mechanisms that would cause large sections, if not all of the Canadian Arctic and Greenland to be warmer during several millennia of the early to mid Holocene, but this one small area not to be affected? Did the changes in ocean currents and sea ice that affected the Canadian Archipelago [Dyke et al., 1997] have no impacts in the region of Ward Hunt Island? My issues with this study are not due to an “insistence on the homogeneity of past climates across the entire Arctic Archipelago”, as A08 claim, and as should be clear from the reference list above. However, as the three references they provide so clearly show, cores that indicate a lack of climate variability before AD1850 are very short, in almost all cases undated or poorly dated; or with poor temporal or taxonomic resolution [see Podritske and Gajewski, 2007]. In fact, a careful reading of their references does show variability in the various proxy records, which the authors chose to ignore, due to their preconceived notions of Holocene climate variability.
 A more parsimonious explanation is that the core of Antoniades et al.  does not extend through much of the Holocene, does not record past climates, and is probably only recording local diagenesis or disturbance of the sediment. The problem is that the conclusions of A07 and Smol et al.  are based almost exclusively on very short, poorly-dated sediment sequences that span only the last couple of hundred years. It is impossible to claim that the results of the past 200 years are “unusual” with respect to the past if the record is only 200–300 years long, or that they are “abrupt” if there is no historical context defining the difference between “abrupt” and “gradual” variations. Given the problems with the data and analysis of A07 and the contrast of these results with what is known about climate change of the Canadian Arctic, I propose that the conclusions of Antoniades et al  be rejected.
 Our research is funded by the Natural Sciences and Engineering Research Council of Canada and the Canada Foundation for Climate and Atmospheric Sciences.