Incorporating intraspecific variation into dental microwear texture analysis

Abstract

Dental microwear texture analysis (DMTA) quantifies microscopic scar or wear patterns left on teeth by different foods or extraneous ingested items such as grit. It can be a powerful tool for deducing the diets of extinct mammals. Here we investigate how intraspecific variation in the dental microwear of macropodids (kangaroos and their close relatives) can be used to maximize the dietary signal inferable from an inherently limited fossil record. We demonstrate significant intraspecific variation for every factor considered here for both scale-sensitive fractal analysis and International Organization for Standardization surface texture analysis variables. Intraspecific factors were then incorporated into interspecific (dietary) analyses through the use of Linear Mixed Effects modelling, incorporating Akaike's Information Criterion to compare models, and testing models through independent cross-validation. This revealed that for each DMTA variable only a small number of intraspecific factors need to be included to improve differentiation between species. Including specimen as a random factor accounted for stochastic inter-individual variation, and facet , incorporated effects of sampling location. Intraspecific effects of ecoregion, microscope, tooth position and wear were often but not universally important. We conclude that models of microwear data that include intraspecific variation can improve the resolution of dietary reconstructions.

Dental microwear texture analysis along reptile tooth rows: complex variation with non-dietary variables

“…Isolated teeth of extant and extinct taxa, especially species that exhibit low degrees of tooth heterodonty, should therefore only be sampled if their original locations in the tooth row are known. Robust sampling strategies will need thorough appreciations for the study taxon or taxa, and for the research question(s) being asked [ 34 , 36 ]. Dietary reconstructions, for example, could focus on teeth that are less affected by non-dietary variables, and investigations into mechanical and behavioural differences in feeding could focus on teeth that are maximally affected by non-dietary variables.…”

“…With respect to tooth shape, DMTA is not underpinned by assumptions of close relationships between the morphology and inferred functions of teeth [ 5 , 9 , 54 ]. However, few studies have explicitly taken tooth shape into consideration when investigating intraspecific microwear differences [ 36 , 55 ], and none have done so for taxa with non-occlusal dentitions. As a taxonomic grade, reptiles exhibit varying levels of tooth heterodonty.…”

“…Fewer studies, however, have explicitly investigated intraspecific tooth microwear texture differences within tooth rows and the influence of endogenous non-dietary variables on microwear textures [29][30][31][32][33][34][35][36]. Non-dietary variables include: (i) tooth position, (ii) bite force, and (iii) tooth shape.…”

Paleoecology of the Rhinocerotidae (Mammalia, Perissodactyla) from Béon 1, Montréal-du-Gers (late early Miocene, SW France): Insights from dental microwear texture analysis, mesowear, and enamel hypoplasia

“…General Linear Mixed Models (GLMM) were applied to our data. They were built using a R code modified from Arman et al (2019), based on the package lme4 (Bates et al, 2015), and adapted to each tested response variable. For DMTA, response variables were the five DMTA parameters (epLsar, Asfc, FTfv, HAsfc9, and HAsfc81) and factors were: specimen (random factor), species, tooth (e.g., first molar, fourth premolar), position (upper or lower), side (left or right), cusp (protocone, protoconid, hypoconid), and facet (grinding or shearing).…”

“…ScoreA: extended mesowear (refined categories for CS and OR) giving scores between 1 and 17, based on the work of Winkler and Kaiser (2011); ScoreB: conversion of the classic original scoring from Fortelius and Solounias (2000) as proposed by (Rivals et al, 2007) into a score ranging from 0 to 3; Ruler: mesowear ruler developed by Mihlbachler et al (2011), giving scores from 0 to 6. as this was the principal interest of our analyses. We tested few interactions (e.g., Species x Facet for microwear) in the models as most factors were considered independent and to avoid unnecessarily complex models (models with interaction effects rarely selected after preliminary analysis; Arman et al, 2019). We created additional sets of models by adding a different factor (e.g., ontogeny, tooth, category, side, wear, cusp) to the second model (species + specimen) and selected the model that produced the lowest AIC score (Akaike's Information Criterion) as the base for the next set of models.…”