Visually, by comparing the map of INDEX2 (Fig. 6) [right] with the one of the distribution of breeds (Fig. 2) [left], it can be seen that the studied breeds seem to be consistently located in regions defined as marginal by the indices.
Well, maybe. Click on the image to see better. But it seems a stretch to me, and the more rigorous logit analysis that the authors subject the data to isn’t exactly overwhelming. If I understand it correctly, the best that a combination of various proxies for marginality can do in predicting the presence of local sheep and goat breeds is 19%. And that’s with breed distribution data which seems to be biased towards marginal areas anyway.
Local sheep and goat breeds are generally argued to be remarkably well adapted to marginal rural areas.
That’s certainly a dominant meta-narrative, and not just for livestock, for agrobiodiversity as a whole. I may even believe it. But not a huge amount of evidence for it here.