Here is a link to another article on dog food protein source contamination that was cited by the article above:
Identification of undeclared sources of animal origin in canine dry foods used in dietary elimination trials - Ricci - 2013 - Journal of Animal Physiology and Animal Nutrition - Wiley Online Library. BTW, I find this article more valid than the one cited above because they used microscopic analysis of bone fragments to determine the species present, as well as DNA analysis. AND their main issue is with foods containing protein sources not on the list of ingredients, not with foods that don't appear to contain protein sources that ARE on the list of ingredients.
Unfortunately, this article doesn't name the brands that were tested, but they found that 10 out 12 foods tested contain protein sources not listed on the label. Now THAT is problematic for dogs with allergies, because these were supposed to be limited ingredient foods:
Summary
Failure to respond to commercial limited antigen diets can occur in dogs kept on a dietary trial for the diagnosis of adverse food reaction (AFR). The aim of this study was to assess twelve canine dry limited antigen diets (eleven novel protein diets and one hydrolysed diet) for potential contamination by ingredients of animal origin not mentioned on the label. The validity of the two methods adopted for the detection of such food antigens was also evaluated. Each dietary product was analysed by microscopy analysis using the official method described in Commission Regulation EC 152/2009 with the aim of identifying bone fragments of different zoological classes (mammalian, avian and fish) and by
polymerase chain reaction (PCR) for the identification of DNA of animal origin. Discrepancies between the results obtained by PCR and/or microscopy analysis and the ingredients listed on pet food packages were found. Only in two pet foods did the results of both analyses match the ingredients listed on the label. In the remaining ten samples, microscopy detected bone fragments from one or two unpredicted zoological classes, revealing avian fragments in six of ten samples followed by those of fish in five of ten and mammalian fragments in four of ten. In two samples, microscopy analysis identified a contamination that would have otherwise passed unobserved if only PCR had been used. However, PCR confirmed the presence of all the zoological classes detected by microscopy and also identified the DNA of an additional unexpected zoological class in two samples. Dogs might fail to respond to commercial limited antigen diets because such diets are contaminated with potential allergens. Both PCR and microscopy analysis are required to guarantee the absence of undeclared animal sources in pet foods.
Before ruling out AFR, a novel protein home-made diet should be considered if the dog is unresponsive to a commercial regimen.