Ooops!
Though in their 2011 paper, Fig. 1, one of their serrirostris samples is closer to fabalis than to the other serrirostris samples, and another sister to all three, suggesting that serrirostris is paraphyletic with respect to fabalis and rossicus. That would be a good reason for justifying conspecifity.
Yes I saw that - this is the haplotype tree and haplotype data does feature in a different format in the earlier paper. In neither paper are the percentage probabilities of the nodes in the rossicus/serrirostris/fabailis part of the tree given, and in fact in the 2011 paper, it explicitly states that these are only given where they're greater than 50%, so almost all of the probabilities in this part of the tree can be taken to be <50% i.e. a fairly flaky tree. At least that's my understanding - I'm happy to be corrected if my understanding of the statistics surrounding branching support for these kinds of tree isn't sophisticated enough to understand what they're are saying.
The mtDNA tree in the 2008 paper, on the other hand, gives 99% probability for the split between fabalis and rossicus/serrirostris (more than the 97% for Pink-footed vs all the Beans). So if I had to weigh up the totality of the genetic evidence, I'd say on balance it favours (though doesn't conclusively prove) reciprocal monophyly between Taiga & Tundra breeders.
But ... if you add this to the Sangster & Oreel data on morphology, lack of intermediates, reproductive isolation, ecology, diel activity patterns, differences in wintering and breeding ranges (I'm doing this from memory so there's no doubt more stuff that I've forgotten) you've got a weighty case for a Taiga/Tundra split.
If Ruokonen et al had come up with some evidence which cast doubt on Sangster & Oreel's findings in all (or even some of) those other areas, then I think their argument would deserve a bit more credit. But they appear to be saying "because not all of our molecular data gives a clear-cut picture for a split, these taxa MUST be conspecific".
It's generally accepted that genetic data shouldn't be used in isolation (as gene trees aren't necessarily taxa trees), so they're on dodgy ground there to start with. And, as I argued in my letter in BB earlier in the year, in the light of evidence that the mid-20th Century lumping spree (which is where the bean lump originated) generated vast numbers of lumps-in-error, to refuse to split on the grounds that one strand of evidence doesn't support a conclusion to split when all other lines of evidence do, is - I can't remember exactly how I phrased it in BB but as this is Birdforum - bonkers). Loaded dice and all that.