The actual paper presents the issue and conclusions much better than this news article. The authors address these concerns, although I’m not entirely convinced that they’ve demonstrated causation. In particular, the lack of ejective consonants in languages of the Tibetan plateau poses a problem for their conclusion. After all, that is the highest elevation region of the world, and one would think that if this were a phenomenon shaping the development of languages that a move toward using ejectives would occur there. I’d also like to see a more robust analysis of the data to show that the correlation is not simply a matter of massaging the data (e.g., defining “high elevation” in a way that achieves a coherent result).
From the conclusion:
The evidence we have adduced clearly supports the following conclusion: Languages spoken at higher altitudes are significantly more likely to rely on ejective phonemes. This conclusion is supported by data from every major world region. Languages with ejectives are generally absent in vast linguistically dense regions at low elevations. Despite the fact that the great majority of languages do not utilize ejective phonemes, however, languages with ejectives are quite common in regions of high altitude. This distribution does not owe itself simply to the influence of language families or the homogenizing effects of particular linguistic regions. Ejectives have spread across languages in numerous areas in high elevation zones. This is not to suggest that languages with ejectives do not occur in areas far from high elevation zones. There are twelve such cases in our data sample. What is striking is that in those cases ejectives have not spread to surrounding languages. In contrast, in areas of high altitude ejectives are in numerous cases a regional feature, and these articulatorily complex sounds have spread across many languages of distinct linguistic stocks in such areas. Interestingly, on the Tibetan plateau and in adjacent regions, ejectives are altogether lacking in our data set. Given that no ejectives seem to exist in the region, ejectives have obviously not been able to spread within the area. This region would be more exceptional, given the distribution of languages with ejectives observed elsewhere, if only one or a few such languages were observed in the region.
[Edited to provide a stable URL (I hope). PLOS ONE seems determined not to allow people to link to their articles. Citation: Everett C (2013) Evidence for Direct Geographic Influences on Linguistic Sounds: The Case of Ejectives. PLoS ONE 8(6): e65275. doi:10.1371/journal.pone.0065275]