Review by Peer 297 On Manuscript
Review by Peer 297 On Manuscript
Review by Peer 297 On Manuscript
PEQ = 4.3 / 5
Introduction
The manuscript builds on a sizeable data set of feather samples collected over a wide region. The
data clearly demonstrate that there is large isotopic variation among the feathers collected within a
given breeding region.
Based on the preliminary information provided (L. 64-68, 99-100; ████ are highly mobile), the
collected feathers cannot be treated as of 'known origin', which unfortunately renders all
geographical analyses flawed, because they assume that samples are of known origin. The authors
acknowledge that they do not know when a collected feather was moulted (1, or 2 years previously,
L. 135-137), but they fail to acknowledge that they also do not know where the feathers were
grown. They state that "isotopic signatures in the feathers of birds returning to breed at a given ██
should reflect the local environment on the breeding grounds in which that feather was grown" (L.
138-140). The feathers collected in a given ███ can therefore only be considered of 'known
A Peerage of Science peer review. Reproduced with permission. Copyright belongs to the reviewer.
1
origin' if the birds ██ display 100% breeding site fidelity for at least 2 years. This critical
assumption underpins the entire geographic assignment analysis of this study. Unfortunately, this
assumption is almost certainly violated. If this assumption would be met then the goal of the study
(to find out whether birds disperse between breeding regions) would be moot.
At each colony (or in each region) the feathers collected consist of an unknown mixture of birds
that moulted in various regions in the 1-2 years previously (the only exception may be ███ ).
Treating the isotope ratios of all these feathers as representative of the ███ where they were
collected will lead to bias, because any local isotopic signal is contaminated by the inclusion of
feathers that were actually grown elsewhere, and may therefore have a different isotopic signature.
In L. 227-228 the authors describe actions they took to improve assignments by using only
'correctly assigned' feathers. This statement indicates that the authors are aware that their feather
collections at any given ██ are a mixture of feathers of different origins, and thus must be
separated into 'local' and 'immigrant' feathers. However, the posterior 'correct' assignment will not
offer a robust and unbiased way to do that: if all feathers were used to train the model initially, then
the feathers that are 'correctly assigned' are those that reflect the mixture of isotope values from
local and immigrant feathers, and not necessarily the true local feathers. Thus, it seems both circular
and unreliable to use the 'correct assignments' to identify (and exclude) the immigrant feathers.
The lack of truly 'known origin' feathers also renders the latitudinal gradient analysis problematic
(L. 196-201). The authors use GLMMs to test for latitudinal gradients in d2H, d13C and d15N in
isotope ratios, but such a gradient analysis is only valuable if the feathers are of known origin. In
addition, these analyses seem very speculative for d15N and d13C, for which in my opinion no
plausible biological hypothesis exists why there should be a latitudinal gradient in ████ feathers.
The large variation in feather isotope ratios in each region is valuable and informative, and would
be indicative of birds from various origins even without the sophisticated geographic assignment
analysis. A geographic assignment analysis would only be possible with isotope data of 'known
origin'. Unless there are certain characteristics in the collected feathers that could be used to
unambiguously determine their origin, disentangling the 'locally grown' from the 'immigrant'
feathers for each of the ███ will be challenging (or impossible). A far simpler analysis that would
overcome the problem of not having truly 'known origin' samples would be to pool all feathers ██,
and conduct a hierarchical cluster analysis (e.g. see [1] and [2]) to split the data set into 5 isotopic
clusters (tentatively representing the 5 regions, under the assumption that they differ isotopically -
an assumption that is more likely to be valid than the current assumption that feathers were grown
locally). Based on the number of represented clusters in each region, inference could be drawn
about the number of other regions from where breeders 'immigrated' to a given target region. The
downside of this analysis is that it would be difficult to unambiguously associate isotope clusters
with a given region (with the possible exception of ██ ).
A Peerage of Science peer review. Reproduced with permission. Copyright belongs to the reviewer.
2
Even if isotope clusters could not be unambiguously assigned to geographic regions, there should
be supporting evidence to suggest that regions may differ isotopically (such as those outlined for the
colony in ███ , L. 343-351). In conjunction with that assumption, the results (which would
presumably assign feathers from each geographic region to >1 isotopic cluster) could be used to
argue that ████ do indeed disperse between breeding regions, and thus support the findings from
genetic studies which indicate a ██████. This simplified analysis would provide more credible
evidence than a geographic assignment analysis that relies on an almost certainly invalid
assumption.
High dispersal that ████████ has been demonstrated in other species as well. For example,
King Eiders (Somateria spectabilis) do not show any spatial genetic structure across much of the
Arctic [3], which is most likely due to the high dispersal of breeding males. King Eiders exhibit
diffuse migratory connectivity, and because males pair up with females on wintering grounds and
follow philopatric females back to breeding areas [4-7], there is a high rate of dispersal between
different breeding populations.
References
[1] Kelly J.F., Johnson M.J., Langridge S.& Whitfield M. (2008) Efficacy of stable isotope ratios in
assigning endangered migrants to breeding and wintering sites - Ecol Appl. 18:568-76.
[2] Oppel S., Pain D.J., Lindsell J., Lachmann L., Diop I., Tegetmeyer C., Donald P.F., Anderson
G., Bowden C.G.R., Tanneberger F.& Flade M. (2011) High variation reduces the value of feather
stable isotope ratios in identifying new wintering areas for aquatic warblers in West Africa - J
Avian Biol. 42:342-54.
[3] Pearce J.L., Talbot S.L., Pierson B.J., Petersen M.R., Scribner K.T., Dickson D.L.& Mosbech
A. (2004) Lack of spatial genetic structure among nesting and wintering King Eiders - Condor.
106:229-40.
[4] Oppel S., Powell A.N.& Dickson D.L. (2008) Timing and distance of King Eider migration and
winter movements - Condor. 110:296-305.
[5] Phillips L.M.& Powell A.N. (2006) Evidence for wing molt and breeding site fidelity in King
Eiders - Waterbirds. 29:148-53.
[6] Oppel S.& Powell A.N. (2008) Assigning king eiders to wintering regions in the Bering Sea
using stable isotopes of feathers and claws - Mar Ecol Prog Ser. 373:149-56.
[7] Mehl K.R., Alisauskas R.T., Hobson K.A.& Kellett D.K. (2004) To winter east or west?
Heterogeneity in winter philopatry in a central-arctic population of King Eiders - Condor. 106:241-
51.
A Peerage of Science peer review. Reproduced with permission. Copyright belongs to the reviewer.