Mixture Analysis and Mammalian Sex Ratio Among Middle Pleistocene Mouflon of Arago Cave, France

Abstract

In archaeological studies, it is often important to be able assess sexual dimorphism and sex ratios in populations. Obtaining sex ratio is easy if each individual in the population can be accurately sexed through the use of one more objective variables. But this is often impossible, due to incompleteness of the osteological record. A modern statistical approach to handle this problem is Mixture Analysis using the method of maximum likelihood. It consists of determining how many groups are present in the sample, two in this case, in which proportions they occur, and to estimate the parameters accordingly. This paper shows the use of this method on vertebrate fossil populations in a prehistoric context with implications on prey acquisition by early humans. For instance, the analysis of mouflon bones from Arago cave (Tautavel, France) indicates that there are more females than males in the F layer. According to the ethology of the animal, this indicates that the hunting strategy could be the result of selective choice of the prey. Moreover, we may deduce the presence of Anteneandertalians on the site during spring and summer periods.

The Middle Pleistocene argali (Ovis ammon antiqua) assemblages at the Caune de l'Arago (Tautavel, Pyrénées-Orientales, France): were prehistoric hunters or carnivores responsible for their accumulation?

“…If adults only are considered and young animals are excluded, 88.2% of the population is female. A statistical method applied to limb bones by Monchot (1999) gave the same results. Also, a study of dental eruption and wear suggests that animals were hunted from the end of spring to the end of summer (Rivals, 2004).…”

Sexual dimorphism in limb bones of ibex (Capra ibex L.): mixture analysis applied to modern and fossil data

“…However, we still do not have enough data to know whether the reduced reliability of metapodials (compared with other bones) in calculating sex ratios is a conclusion that applies particularly to ibex, or if it is of wider significance among other caprine or even ungulate species. Although MA has been applied successfully to other bovid species like aurochs (Bos primigenius) (Monchot & Lechelle, 2002;Helmer et al, 2005), moufflon (Ovis ammon antiqua) (Monchot, 1999;Monchot & Lechelle, 2002) and domestic goat (Capra hircus) , as well as to cave bear (Ursus spelaeus) (Quilès & Monchot, 2004), more comprehensive studies are needed to test the general validity of the results emphasised in the present work.…”

“…Prior probabilities represent the proportions of each group in the whole mixture. In the special case of two groups, other parameters which describe the mixture can be computed (Flury et al, 1992;Monchot, 1999Monchot, , 2000Monchot & Lechelle, 2002). The posterior probabilities are monotonic functions that have only one intersection point called the cutoff point y à .…”

“…According to the calculated cut-off point y à , the method gives, for each measurement, an estimation of the proportion of males and females in the samples, as well as the value of the misclassification error e. The misclassification error e is the sum of the probability of classifying all y greater than y à in group 1 knowing that they belong to group 2, plus the probability of classifying all y smaller than y à in group 2 knowing that they belong to group 1. For more details on the mathematical model and on various applications of mixture analysis, readers are referred to Aitkin & Wilson, 1980;Everitt, 1984;Flury et al, 1992;Airoldi et al, 1995;Dong, 1997;Monchot, 1999Monchot, , 2000Monchot & Lechelle, 2002. …”

Documenting the initial appearance of domestic cattle in the Eastern Fertile Crescent (northern Iraq and western Iran)

“…We also apply Gaussian Mixture Analysis to our biometric datasets (using PAST v.1.89 software) in order to address the biometric characteristics of both male and female Bos in our assemblages (Hammer et al, 2001;Monchot, 1999;Vigne, 2015). The PAST software uses an expectation maximization (EM) algorithm to determine the sizes of two component populations (k).…”