Banca de QUALIFICAÇÃO: ALLANA LÍVIA BESERRA PAULINO

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
STUDENT : ALLANA LÍVIA BESERRA PAULINO
DATE: 21/02/2025
TIME: 09:00
LOCAL: Google Meet
TITLE:

Bayesian Isotope Assignment with Different Probability Distributions in the “AssignR” Package


KEY WORDS:

Isotope; AssignR; Probability Distributions; Geographic Assignment.


PAGES: 14
BIG AREA: Ciências Agrárias
AREA: Agronomia
SUMMARY:

This study analyzes how different probability distributions influence the isotopic assignment process using the ‘assignR‘ package. The approach adopted was Bayesian, with adaptations to the ‘pdRaster‘ function, which generates posterior probability density maps. While ‘assignR‘ originally employs the normal distribution to infer geographic origin, the modifications allowed for the application of other continuous distributions — Cauchy, Logistic, Laplace, and Student’s t — selected for their parametric and graphical similarities to the normal distribution. The analysis was based on two datasets: (1) deuterium isotopes (D) from the Loggerhead Shrike bird species, available within the package itself, and (2) an additional database containing carbon (δ13C) and nitrogen (δ15N) isotope information, analyzed separately, from cannabis (Cannabis sativa) samples seized in São Paulo. To evaluate the accuracy of the probability distributions, a quality analysis method based on the quantiles of the assigned coverage area was used. In tests with deuterium isotopes, the Logistic distribution showed the best fit. For carbon and nitrogen isotopes, the Laplace distribution demonstrated the best performance, indicating its higher predictive quality. The results highlight that the application of different distributions can impact isotopic assignment and suggest that alternative approaches to the normal distribution may increase the precision of geographic inference models.


COMMITTEE MEMBERS:
Externo à Instituição - DEOCLÉCIO JARDIM AMORIM - USP
Externo à Instituição - CARLOS ANTONIO ZARZAR - UFOPA
Externo à Instituição - JOÃO PAULOS SENA SOUZA - UNIMONTES
Presidente - ***.221.388-** - LUIZ ANTONIO MARTINELLI - USP
Interno - TIAGO ALESSANDRO ESPINOLA FERREIRA
Notícia cadastrada em: 20/02/2025 16:47
SIGAA | Secretaria de Tecnologias Digitais (STD) - https://servicosdigitais.ufrpe.br/help | Copyright © 2006-2025 - UFRN - producao-jboss12.producao-jboss12