Evidence for the age and origin of life is sought in the oldest rocks on Earth, which are ductilely deformed and metamorphosed up to amphibolite or granulite facies (see Schopf  for a review). Consequently, preservation of microstructural or microchemical evidence for life requires some form of mineralogical capsule that can protect that evidence from modification during superimposed metamorphic recrystallization with open system behavior. In their recent study, McKeegan et al. (2007) sought graphite inclusions (as possible biogenic relicts) within apatite grains found in siliceous rock sample G91–26 (ANU catalog 92–197) on the island of Akilia (Greenland). The sample is older than 3600 Ma (probably >3830 Ma). This followed an earlier study with the same approach by Mojzsis et al. (1996; our contribution to which was to provide the sample, its petrography, geological context, and the first dating indicating that Akilia rocks might be very ancient, >3830 Ma).
Mojzsis et al. (1996, p. 57) reported “occluded carbon in apatite micrograins from the Akilia island BIF [banded iron formation]” G91–26 as 13C-depleted, with carbon isotopic signatures similar to those characterizing some types of microorganisms. These apatites were thought to have grown very early in the rock's history, thereby leading to the conclusion that this rock contains (the oldest) evidence for life, which had been protected within apatite grains throughout the history of the rock (Mojzsis et al., 1996). Using the same sample, the “frequent” occurrence of these inclusions could not be replicated by Lepland et al. (2005) and Nutman and Friend (2006). Despite a decade having passed since the Mojzsis et al. (1996) paper, neither the actual analytical data nor documentation of the grains analyzed have been presented (not even in the thesis of Mojzsis ), adding to controversy over the results.
McKeegan et al. (2007) try to redress this situation, and we commend them for their very thorough petrographic documentation of a graphite inclusion within a single apatite grain. Aside from that grain, there are only two others with graphite “inclusions” shown in Data Repository item 20071491. These however, do not pass petrographic criteria as genuine inclusions, because one comprises a train of graphite flecks traversing an apatite (along an annealed[?] crack), whereas the other is an invagination of graphite into the side of an apatite (see Figure 3 of Nutman and Friend ). Therefore, McKeegan et al. failed to reproduce the observation of Mojzsis et al. (1996) that graphite inclusions in G91–26 apatites were “frequent.”
Regarding the significance of genuine (rare) graphite inclusions in apatites, McKeegan et al. did not draw attention to papers in the past decade that have improved the understanding of the recrystallization of phosphates during metamorphism. Even in a simple experimental system consisting of quartz and the phosphate monazite, Ayers et al. (1999) demonstrated that small phosphate grains dissolve and are reprecipitated as larger metamorphic grains with progressive metamorphism. In more complex natural metamorphic systems, the dissolution and regrowth of phosphates with increasing metamorphic grade is also well documented (e.g., Williams, 2001; Wing et al., 2002). Therefore phosphates need not necessarily protect pre-diagenetic material from reaction and isotopic exchange with surrounding media, and there is no reason that apatites in the Akilia rocks behaved otherwise. In a paper not cited by McKeegan et al., we (Nutman and Friend, 2006) demonstrated that G91–26 apatites show marked depletion of the heavy rare earth elements. We interpreted this to result from the presence of garnet (a heavy rare-earth-enriched mineral) in the Akilia rocks, occurring mostly in widely spaced discordant veins, and rarely as disseminated grains. We thus concluded that the G91–26 apatites grew in equilibrium with garnet and therefore must be metamorphic in origin. Consequently, any graphite inclusions cannot a priori be regarded as retaining a pre-metamorphic isotopic signature. This graphite could then be the product of several reactions at different times in the history of the rock—recrystallization of (pre-diagenetic) biogenic material is but one possibility.
We (Nutman and Friend, 2006) demonstrated via infrared absorbance spectroscopy that G91–26 apatites contain carbon in the form of carbonate substituting for phosphate and hydroxyl ions. The well-documented analytical spot shown by McKeegan et al. (their Fig. 3D) contained apatite as well as graphite. Due to the mixed-carbon source in this analysis, the isotopic signature for the G91–26 graphite is still not known with confidence. Furthermore, Tumpane and Peck (2006) observed that carbonate-bearing apatites can be 13C-depleted, even when a direct biogenic pedigree for such carbon is unlikely. This further complicates the interpretation of McKeegan et al.'s measurement.
The issues we have raised here undermine the conclusion of McKeegan et al. (2007) that their results support evidence for >3830 Ma life in Akilia sample G91–26.