1. IntroductionThe idea of “mammoth megasites” during the Gravettian andEpigravettianhaslongbeenacontroversialtopicamongresearchers,with debate over whether they are the result of natural deaths orspecialized human hunting (Haynes, 1991; Soffer et al., 2001;Svoboda et al., 2005). From the same time period there is alsomuchdisagreementaboutthepresenceofpossibledomesticateddogremains, representing initial domestication event(s) across Eurasia.In her recent paper, Shipman (2015) considers these two topics,suggestingtheproliferationofmammothremainsfromanumberofEurasian sites is the result of human hunting, while also suggestingthe hunting of these mammoths was facilitated by the assistance ofsome form of early domesticated dog (referred to as “wolf dogs”).While we strongly support additional research into the issues ofmammoth megasites and initial dog domestication, the evidencepresented by Shipman fails to support the conclusions made. Here,we present issues with the taphonomic considerations, mammothage profiles, and the proposed domesticated canid hypothesis.2. Taphonomic considerationsBefore the age-at-death profiles from mammoth teeth can beused for meaningful statements about behavior, the overall contextof theseassemblages must,andshould, be considered.Taphonomicand site formation history has not been discussed by Shipman insufficientenoughdetailtoascertainthetimingandfunctionofeachof these sites, especially since many appear to have had serveddifferent functions (Bosch, 2012; see also Smith, 2015). Defleshedmammoth bones might have been gathered as raw material fortools andornaments,orasbuildingmaterial(e.g.Sofferetal.,2001;Niven, 2006), as has been suggested for Predmosti and Milovice(Musil, 1968; Brugere and Fontana, 2009). For example, extensivegeoarchaeological work has shown that the Berelekh mammothbone bed, which Shipman has suggested may be an example of‘repeated, non-selective hunting’ (page 3), actually consists of aseries of natural deposits unrelated to the Berelekh archaeologicalsite, though the human occupants later exploited the raw materialas a resource (Pitulko, 2011; see also Pitulko et al., 2014).The mammoth age profiles and open-air exposure of these sitessuggest that faunal assemblages from these localities reflect time-averaged palimpsests, perhaps resulting from single kill eventsand a repeated accumulation of faunal material at these locations(e.g. Bosch, 2012; Smith, 2012). Haynes (1991) discussed variationbetween age profiles, particularly between palimpsest and livepopulations, noting that even though two profiles may appearsimilar, their “meaning” may reflect the opposite. Alongside thetime-averaged quality of many of the faunal assemblages, Shipman(2015) does not provide a structured and detailed taphonomicanalysis in support of the model proposed by the mammoth ageprofiles. The paper has not considered the depth and range of timerepresentedateachofthesesitesandwhetheranalogousanalyticalunits are actually being compared. Failing to contextualize ageprofiles from teeth alongside the survival of postcranial elementsprevents a more accurate assessment of either selective transportordestruction,orboth.Whilequantityofpost-cranialmaterialmaybe reduced (compared to dental fragments), these portions couldprovide useful taphonomic data about seasonal use of sites andwhether the mammoth assemblage may have suffered some formofpreorpost-depositionaldisturbance.Thoughmammothremainsundoubtedly dominate these sites, a useful comparative perspec-tive could be gleaned from considering the other fauna from thesesites; in particular the proportion of canids to other carnivores.Indeed, given the importance of “wolf dogs” for this model, it iscurious that no information was provided in terms of the propor-tion and distribution of carnivore modifications. Without data ontaphonomic agents, including human and carnivore modifications,it is difficult to compare and contrast the assemblages in themanner presented by Shipman.3. Mammoth age-at-death profilesAs Shipman states, statistical evaluation of the data onmammoth age-at-death profiles is needed, however there seem tobe some issues with the raw data and statistical tests presented inthe original paper. No raw data is provided for the numbers of