We appreciate the experimental findings by Falkenberg and coworkers and are glad to see the iron-bearing nerve terminals in a number of other bird species besides homing pigeons. In the results section, the authors interpret the avian X-ray absorption spectra cautiously but then in the discussion section they become set on an interpretation that is not supported by experimental data.

1) Falkenberg et al first write that none of the reference materials matches the avian material completely and that the avian spectra “hint towards iron(III) oxides”. But later they write: “This result clearly shows that the dendrites in the avian beak mainly contain iron(III) oxide and not pure magnetite, as predicted in several papers, where magnetite was assumed to be the only magnetic mineral underlying biological magnetoreception (24,31,45,46).” (Note: ref. 24 and 46 are the same paper).

What the XANES data actually show is that most of the iron in the dendrites is ferric, i.e. iron(III). But the results do not tell yet if the compound in question is an iron(III) oxide, or rather a iron(III) hydroxide, iron(III) oxy-hydroxide, or even iron(III) phosphate because any such ferric compound would fit the near-edge part of the dendrite spectrum between 7115 and 7125 eV. The obvious mismatch between the spectra of maghemite and of dendrite at about 7130 eV (Fig. 7) does not really support the claim made in the abstract (“We also used microscopic X-ray absorption spectroscopy analyses to identify the involved iron minerals to be almost completely Fe III-oxides”).

The point here is: The presence of ferric iron does not automatically make the case for a strongly magnetic compound, and this is what matters most in the context of magnetoreception. Among the iron(III) compounds, only maghemite is strongly magnetic, while other possible iron(III) compounds are weakly magnetic at best and therefore not interesting from the point of view of magnetoreception (although they are interesting from the point of view of biochemistry and biomineralization). The decisive question for the magnetoreception mechanism is therefore: is the compound in question magnetic at all?
As in our comment to the paper by Fleissner et al. 2007 in Naturwissenschaften (see http://arxiv.org/abs/0805...) we would like to call the authors’ attention to the possibility that a lot of ferric iron may reside in the iron-storage protein ferritin, which would contribute to the XANES spectrum as well. We suggest taking into account ferritin when seeking to explain the XANES spectra quantitatively.

Given the suite of possible Fe III compounds, we suggest to use electron diffraction to identify the structure of the compound. Diffraction patterns together with the chemical information will help to identify the mineralogical nature of the compounds.

2) Falkenberg et al write: “This view [ that maghemite may account for most of the iron inside the dendrites] is shared by Tian et al. (49), who performed various physical tests, especially SQUID measurements of isolated beak skin samples and interpreted their data on the magnetic remanence as strong evidence that additionally to magnetite a second magnetic mineral – they proposed maghemite – must be contained in the dendritic system”.

It is interesting to compare this to what was actually written in Tian et al (49): "This suggests that the remanence carriers in the samples are magnetically soft minerals (probably magnetite or maghemite). ... Remanence loss around 120 K on the cooling curves of SIRM300 K and warming curves of SIRM5 K clearly indicates the presence of magnetite in the measured samples (Figures 2 and 3). A significant rapid decay of SIRM5 K in the interval of 5-20 K on both ZFC and FC warming curves (Figure 2) suggests the dominance of superparamagnetic (SPM) particles in the samples. ... The low delta ratios (less than 2) calculated from the thermal demagnetization curves of SIRM5 K in this study do not support a chain arrangement of magnetite crystals as seen in magnetotactic bacteria (Moskowitz et al. 1993; Pan et al. 2005)."

So, since magnetite and maghemite have similar magnetic properties at room temperature (for a given domain state), Tian et al were cautious enough to not assign either mineral to their room temperature remanence curves. But the subsequent low-temperature cycling of the remanence imparted at room temperature clearly showed that magnetite carries a substantial part of the remanence. On the basis of the Moskowitz delta-ratio test, Tian et al inferred that the magnetic properties are not dominated by chains of single-domain magnetite, but this does not mean that single-domain magnetite is absent. It simply means that chains, if present, do not contribute by more than 50% or so to the remanence. To conclude, we find it inappropriate to use the paper by Tian et al as strong support for maghemite.

3) Falkenberg et al write: “The XANES spectra of the dendrites suggest that magnetite is not the main constituent. Thus, care should be taken before using the well established notion of magnetite-based magnetoreception. Rather the term ‘iron mineral-based magnetoreception’ should be used – at least in birds.”

Note that magnetite may still be a major magnetic constituent in case the predominant ferric compounds are nonmagnetic (see point A), which would tally with the magnetic interpretation by Tian et al. (see point B). Apart from that, one has the impression that the authors convey the notion that magnetite and maghemite are altogether different minerals, but in fact they are so closely related to each other that there is a solid-solid solution with magnetite and Fd3m maghemite as end members. Since the two minerals are structurally related and have similar lattice constants, distinction by means of electron diffraction without chemical information is hardly possible (as the authors correctly point out). Also, as the authors acknowledge, maghemite has magnetic properties similar to magnetite (at physiological temperatures). As the authors are aware of all these similarities, one is left wondering why they make such a big deal of the distinction, which to us appears overstated, particularly at this stage where it is not even clear what magnetic compound is biomineralized in birds in the first place.

Along these lines, “iron-mineral based” magnetoreception is not an emerging mechanism, unlike the authors write in their introduction. Magnetite is an iron mineral after all, and the idea has been around since the discovery of biogenic magnetite by Heinz A. Lowenstam in chiton teeth.
“Iron-mineral based” does not sound very specific to a mineralogist, given the plethora of iron-bearing minerals, most of which are not strongly magnetic. Why not call it “magnetite/maghemite based magnetoreception”? This is consistent with the diffraction data available so far. The nanocrystalline particles in the spherical clusters (referred to as “bullets”) in the iron-bearing nerve terminals have a diffraction pattern consistent with randomly oriented magnetite/Fd3m maghemite crystals (Hanzlik et al 2000), while the iron-phosphorous bearing material in the shape-bags (referred to as “platelets”) co-localized with the clusters is amorphous (Fleissner et al 2003) and thus not consistent with either magnetite or maghemite. Although the authors dogmatically rule out magnetite as dominating magnetic constituent (at least in birds), they did not consider the possibility that nanocrystalline magnetite (we are dealing with grain sizes of a few nm in the case of the clusters or “bullets”) may have rapidly oxidized to maghemite during the preparation, because oxygen was not prevented from diffusing into the tissue. It may be worthwhile to subject magnetite nanoparticles from a water based ferrofluid to the same perfusion and fixation protocol as the one applied to the beak tissue. This way, a suitable control sample would be available to test for the influence of oxidation on the spectra.

Last, we have a couple of specific questions:
1) Two spots (pt 1 and pt 3 in Fig. 5) were identified as contamination – does the Titanium map show an anomaly here as well? This would help to narrow down the source of the contamination. If so, it would be interesting to do XANES iron K-edge mapping of the Titanium tools used for tissue preparation in order to see if the Titanium tools have traces of Fe contamination.
2) It is written that the absorption spectra of the dendrites may differ depending on whether they contain only part of a dendrite. How do they differ?
3) Does the section shown in Fig. 1A have the same orientation as the one shown in Fig. 1F?

Michael Winklhofer, Ludwig-Maximilians-University, Munich, Germany
Marianne Hanzlik, Technical University Munich, Germany
Joseph L. Kirschvink, California Institute of Technology, Pasadena, USA