11 Knowledge on the diet is critical to understand the ecology of animal species, and also 12 to design management and conservation strategies. Nonetheless, diet studies of many 13 mammalian carnivores are usually based on indirect analyses (mostly through their 14 faeces) rather than on direct observations of their feeding behavior, which could 15 produce uncertain assignments of the predator species. Here, we tested the hypothesis 16 that differences in the diet between studies are not

more »

... ted with the predator 17 identification method used by comparing results obtained in studies that used either 18 high confidence identification (HCI; i.e. stomach or colon content, or genetic or 19 chemical analyses of faeces) or low confidence identification methods (LCI; i.e. faeces 20 appearance) for jaguars and cougars. We reviewed the literature on diet of these 21 species and 1) assessed if similarity of the diet was related to the method used to 22 identify the species, and 2) the possible consequences that potential misidentification 23 of faeces could have on diet spectrums. Furthermore, we briefly summarized the most 24 reliable knowledge currently available about the diets of both species. Our analyses 25 indicated that the method used for predator identification significantly influences diet 26 similarity, with HCI studies being more similar among them than LCI studies or HCI-LCI 27 studies. Studies based on LCI methods apparently overemphasized the importance of 28 certain prey for both species, whereas other significant prey items were not detected. 29 Although approaches for a reliable identification of predators are expensive and require 30 specialized technicians and equipment, we highlight the need to invert on them in order 31 to accomplish a better ecological understanding of the feeding ecology of carnivore 32 species, which is a key factor to consider in conservation and management plans. 34 Knowledge on the diet of mammalian carnivores is critical to understand predator-prey 35 relationships and species interactions (e. g. Gittleman and Harvey 1982, Bekoff et al. 36 1984 , Sunquist and Sunquist 1989 . However, obtaining data for diet studies through 37 direct observations is difficult for many carnivore species due to their elusive behavior, 38 and most studies are based on the use of indirect methods for the identification of prey 39 consumed, including the analysis of stomach and colon content and, most frequently, 40 the examination of prey remains in carnivore faeces (Mills 1996) . 41 Diet analyses based on faeces requires the previous identification of the producer 42 species, which is often achieved by two general different methods: (1) the inspection of 43 morphology and appearance of faeces in the field and, eventually, other associated 44 evidences such as tracks, photo captures, scrapes, hairs and radio locations; and (2) 45 laboratory genetic or chemical analyses. The identification of the species in the field 46 often implies a high degree of uncertainty and subjectivity since it relies on the 47