# Capillariid diversity in archaeological material from the New and the Old World: clustering and artificial intelligence approaches

## TL;DR

This study uses new methods to identify ancient capillariid nematode eggs in archaeological samples from Europe and Brazil, revealing insights into historical host-parasite relationships.

## Contribution

The study introduces clustering and artificial intelligence approaches for identifying capillariid eggs in archaeological contexts.

## Key findings

- Capillariid eggs were identified in 10 European and 4 Brazilian archaeological samples, revealing 13 morphotypes.
- Host information allowed species identification in Brazilian coprolites but not in European samples.
- New methodologies identified specific capillariid species in feline, opossum, and bovid coprolites in Brazil.

## Abstract

Capillariid nematode eggs have been reported in archaeological material in both the New and the Old World, mainly in Europe and South America. They have been found in various types of samples, as coprolites, sediments from latrines, pits, or burial. Modern parasitological records show that around 300 species of capillariids have been described in all vertebrate taxa, including humans, making it a very diversified group. The main proposal of this work is to characterize and identify capillariid eggs found in archaeological sites from Europe and Brazil.

A total of 39 samples of archeological sites from Europe, deposited in the paleoparasitological collection of the University Marie & Louis Pasteur, Besançon, France was analyzed. In addition, 80 coprolites from the pre-Colombian archaeological site Gruta do Gentio II, Brazil, deposited in the Paleogenetic Laboratory at Oswaldo Cruz Institute, Oswaldo Cruz Foundation, Rio de Janeiro, were evaluated. Samples were treated according to the protocols of each laboratory and then analyzed under light microscopy. Capillariid eggs were classified according to length, width, plugs, and eggshell sizes, and statistical analysis of the morphometric dataset was performed. Using a reference dataset of specimens provided by both Institutional Collections, three approaches to species identification were applied: discriminant analysis, hierarchical clustering, and artificial intelligence/machine learning.

A total of 10 samples from Europe and 4 from Brazil were positive for capillariid eggs, showing 13 different morphotypes. As European samples were mainly collected from latrines and pits, parasite–host information was absent, and consequently, species identification was impaired. In contrast, the availability of host information rendered the identification of capillariid species for the Brazilian coprolites. The new methodology indicates capillariid species identified on various samples, resulting in the presence of Capillaria exigua (Dujardin, 1845) in feline coprolite, Baruscapillaria resecta (Dujardin, 1845) in opossum, and Aonchotheca bovis (Schnyder, 1906) in bovid, in the Brazilian site, while in European sites, Capillaria venusta (Freitas e Mendonça, 1958), Aonchotheca myoxinitelae (Diesing, 1851), Eucoleus madjerdae (Bernard, 1964), and Baruscapillaria spiculata (Freitas, 1933) were found.

The study provides new results by applying innovative methodologies for parasite identification and gaining insights into the past host (human or animal)/parasite relationships.