TY - JOUR
T1 - Including unsexed individuals in sex-specific growth models
AU - Minto, Cóilín
AU - Hinde, John
AU - Coelho, Rui
N1 - Publisher Copyright:
© 2018, Canadian Science Publishing. All rights reserved.
PY - 2018
Y1 - 2018
N2 - Sexually dimorphic growth models are typically estimated by fitting growth curves to individuals of known sex. Yet, macroscopically ascribing sex can be difficult, particularly for immature animals. As a result, sex-specific growth curves are often fit to known-sex individuals only, omitting unclassified immature individuals occupying an important region of the age-length space. We propose an alternative whereby the sex of the unclassified individuals is treated as a missing data problem to be estimated simultaneously with the sex-specific growth models. The mixture model that we develop includes the biological processes of growth and sexual dimorphism. Simulations show that where the assumed growth model holds, the method improves precision and bias of all parameters relative to the data omission case. Ability to chose the correct combination of sex-specific and sex-generic parameters is also improved. Application of the method to two shark species, where sex can be ascribed from birth, indicates improvements in the fit but also highlights the importance of the assumed model forms. The proposed method avoids discarding unclassified observations, thus improving our understanding of dimorphic growth.
AB - Sexually dimorphic growth models are typically estimated by fitting growth curves to individuals of known sex. Yet, macroscopically ascribing sex can be difficult, particularly for immature animals. As a result, sex-specific growth curves are often fit to known-sex individuals only, omitting unclassified immature individuals occupying an important region of the age-length space. We propose an alternative whereby the sex of the unclassified individuals is treated as a missing data problem to be estimated simultaneously with the sex-specific growth models. The mixture model that we develop includes the biological processes of growth and sexual dimorphism. Simulations show that where the assumed growth model holds, the method improves precision and bias of all parameters relative to the data omission case. Ability to chose the correct combination of sex-specific and sex-generic parameters is also improved. Application of the method to two shark species, where sex can be ascribed from birth, indicates improvements in the fit but also highlights the importance of the assumed model forms. The proposed method avoids discarding unclassified observations, thus improving our understanding of dimorphic growth.
UR - http://www.scopus.com/inward/record.url?scp=85041326568&partnerID=8YFLogxK
U2 - 10.1139/cjfas-2016-0450
DO - 10.1139/cjfas-2016-0450
M3 - Article
AN - SCOPUS:85041326568
SN - 0706-652X
VL - 75
SP - 282
EP - 292
JO - Canadian Journal of Fisheries and Aquatic Sciences
JF - Canadian Journal of Fisheries and Aquatic Sciences
IS - 2
ER -