Building matrix population models when individuals are non-identifiable

Carlos Hernandez-Suarez, Paula Medone, Carlos Castillo-Chavez, Jorge Rabinovich

Research output: Contribution to journalArticle

Abstract

Matrix Population Models (MPM) are among the most widely used tools in ecology and evolution. These models consider the life cycle of an individual as composed by states to construct a matrix containing the likelihood of transitions between these states as well as sexual and/or asexual per-capita offspring contributions. When individuals are identifiable one can parametrize an MPM based on survival and fertility data and average development times for every state, but some of this information is absent or incomplete for non-cohort data, or for cohort data when individuals are not identifiable. Here we introduce a simple procedure for the parameterization of an MPM that can be used with cohort data when individuals are non-identifiable; among other aspects our procedure is a novelty in that it does not require information on stage development (or stage residence) times, which current procedures require to be estimated externally, and it is a frequent source of error. We exemplify the procedure with a laboratory cohort dataset from Eratyrus mucronatus (Reduviidae, Triatominae). We also show that even if individuals are identifiable and the duration of each stage is externally estimated with no error, our procedure is simpler to use and yields the same MPM parameter estimates.

LanguageEnglish (US)
Pages13-17
Number of pages5
JournalJournal of Theoretical Biology
Volume460
DOIs
StatePublished - Jan 7 2019
Externally publishedYes

Fingerprint

Matrix Models
Population Model
Population
Triatominae
Reduviidae
Residence Time
Fertility
Ecology
Parameterization
Life Cycle Stages
Life Cycle
Life cycle
Likelihood
Research Design
Model-based
ecology
duration
Estimate

Keywords

  • Life-history traits
  • Matrix models
  • Non-cohort data
  • Non-identifiable individuals
  • Parameter estimation
  • State-frequency data

ASJC Scopus subject areas

  • Statistics and Probability
  • Modeling and Simulation
  • Biochemistry, Genetics and Molecular Biology(all)
  • Immunology and Microbiology(all)
  • Agricultural and Biological Sciences(all)
  • Applied Mathematics

Cite this

Building matrix population models when individuals are non-identifiable. / Hernandez-Suarez, Carlos; Medone, Paula; Castillo-Chavez, Carlos; Rabinovich, Jorge.

In: Journal of Theoretical Biology, Vol. 460, 07.01.2019, p. 13-17.

Research output: Contribution to journalArticle

Hernandez-Suarez, Carlos ; Medone, Paula ; Castillo-Chavez, Carlos ; Rabinovich, Jorge. / Building matrix population models when individuals are non-identifiable. In: Journal of Theoretical Biology. 2019 ; Vol. 460. pp. 13-17.
@article{9a0473d94703481e8356262ffee8b277,
title = "Building matrix population models when individuals are non-identifiable",
abstract = "Matrix Population Models (MPM) are among the most widely used tools in ecology and evolution. These models consider the life cycle of an individual as composed by states to construct a matrix containing the likelihood of transitions between these states as well as sexual and/or asexual per-capita offspring contributions. When individuals are identifiable one can parametrize an MPM based on survival and fertility data and average development times for every state, but some of this information is absent or incomplete for non-cohort data, or for cohort data when individuals are not identifiable. Here we introduce a simple procedure for the parameterization of an MPM that can be used with cohort data when individuals are non-identifiable; among other aspects our procedure is a novelty in that it does not require information on stage development (or stage residence) times, which current procedures require to be estimated externally, and it is a frequent source of error. We exemplify the procedure with a laboratory cohort dataset from Eratyrus mucronatus (Reduviidae, Triatominae). We also show that even if individuals are identifiable and the duration of each stage is externally estimated with no error, our procedure is simpler to use and yields the same MPM parameter estimates.",
keywords = "Life-history traits, Matrix models, Non-cohort data, Non-identifiable individuals, Parameter estimation, State-frequency data",
author = "Carlos Hernandez-Suarez and Paula Medone and Carlos Castillo-Chavez and Jorge Rabinovich",
year = "2019",
month = "1",
day = "7",
doi = "10.1016/j.jtbi.2018.10.014",
language = "English (US)",
volume = "460",
pages = "13--17",
journal = "Journal of Theoretical Biology",
issn = "0022-5193",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - Building matrix population models when individuals are non-identifiable

AU - Hernandez-Suarez, Carlos

AU - Medone, Paula

AU - Castillo-Chavez, Carlos

AU - Rabinovich, Jorge

PY - 2019/1/7

Y1 - 2019/1/7

N2 - Matrix Population Models (MPM) are among the most widely used tools in ecology and evolution. These models consider the life cycle of an individual as composed by states to construct a matrix containing the likelihood of transitions between these states as well as sexual and/or asexual per-capita offspring contributions. When individuals are identifiable one can parametrize an MPM based on survival and fertility data and average development times for every state, but some of this information is absent or incomplete for non-cohort data, or for cohort data when individuals are not identifiable. Here we introduce a simple procedure for the parameterization of an MPM that can be used with cohort data when individuals are non-identifiable; among other aspects our procedure is a novelty in that it does not require information on stage development (or stage residence) times, which current procedures require to be estimated externally, and it is a frequent source of error. We exemplify the procedure with a laboratory cohort dataset from Eratyrus mucronatus (Reduviidae, Triatominae). We also show that even if individuals are identifiable and the duration of each stage is externally estimated with no error, our procedure is simpler to use and yields the same MPM parameter estimates.

AB - Matrix Population Models (MPM) are among the most widely used tools in ecology and evolution. These models consider the life cycle of an individual as composed by states to construct a matrix containing the likelihood of transitions between these states as well as sexual and/or asexual per-capita offspring contributions. When individuals are identifiable one can parametrize an MPM based on survival and fertility data and average development times for every state, but some of this information is absent or incomplete for non-cohort data, or for cohort data when individuals are not identifiable. Here we introduce a simple procedure for the parameterization of an MPM that can be used with cohort data when individuals are non-identifiable; among other aspects our procedure is a novelty in that it does not require information on stage development (or stage residence) times, which current procedures require to be estimated externally, and it is a frequent source of error. We exemplify the procedure with a laboratory cohort dataset from Eratyrus mucronatus (Reduviidae, Triatominae). We also show that even if individuals are identifiable and the duration of each stage is externally estimated with no error, our procedure is simpler to use and yields the same MPM parameter estimates.

KW - Life-history traits

KW - Matrix models

KW - Non-cohort data

KW - Non-identifiable individuals

KW - Parameter estimation

KW - State-frequency data

UR - http://www.scopus.com/inward/record.url?scp=85054679026&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85054679026&partnerID=8YFLogxK

U2 - 10.1016/j.jtbi.2018.10.014

DO - 10.1016/j.jtbi.2018.10.014

M3 - Article

VL - 460

SP - 13

EP - 17

JO - Journal of Theoretical Biology

T2 - Journal of Theoretical Biology

JF - Journal of Theoretical Biology

SN - 0022-5193

ER -