Body mass estimations in extinct caviomorphs (Ctenohystrica, Hystricognathi) from cheek tooth measures: importance of predictor, reference sample and method

Caviomorph rodents present an astonishing variation in sizes both nowadays and in the past. The objective of this work was to estimate the body mass of 32 extinct caviomorphs. Moreover, we seek to analyse the influence on the estimations of the: i. selected predictors (i.e., measurements on cheek teeth); ii. used reference sample comparing three distinct datasets linked to the representativeness of sizes among herbivores; and iii. employed method constructing simple and multiple regressions, which consider phylogeny and not. The results indicated that the best dental structures to predict body mass usually were the second upper and lower molars, the upper and lower rows and first upper molar. Including ungulates improved the predictions for living large rodents, while it was not the case for the living caviomorphs and small rodents. Thus, different sets of best models were proposed depending of the size range to which belongs the extinct caviomorph considered. OLS had better performance in most cases. Multiple regressions, associating dental length and width, usually improved the mean of the prediction errors with respect to simple models. Despites a probable important redundancy between these variables, independent information would be thus sufficient to improve the error of prediction. Our estimations from dental dimensions for large and giant chinchilloids are within the range of values previously obtained using a cranial variable recently considered as a robust body size proxy. For the small extinct caviomorphs, our estimations showed more deviation with respect to the body mass values available in literature.