Cristin-resultat-ID: 1891440
Sist endret: 26. april 2021, 14:13
NVI-rapporteringsår: 2021
Resultat
Vitenskapelig artikkel
2021

Efficient estimation of large-scale spatial capture–recapture models

Bidragsytere:
  • Daniel Turek
  • Cyril Pierre Milleret
  • Torbjørn Ergon
  • Henrik Brøseth
  • Pierre Dupont
  • Richard Bischof
  • mfl.

Tidsskrift

Ecosphere
ISSN 2150-8925
e-ISSN 2150-8925
NVI-nivå 1

Om resultatet

Vitenskapelig artikkel
Publiseringsår: 2021
Volum: 12
Hefte: 2
Artikkelnummer: e03385
Open Access

Importkilder

Scopus-ID: 2-s2.0-85101499528

Klassifisering

Vitenskapsdisipliner

Basale biofag

Beskrivelse Beskrivelse

Tittel

Efficient estimation of large-scale spatial capture–recapture models

Sammendrag

Capture–recapture methods are a common tool in ecological statistics, which have beenextended to spatial capture–recapture models for data accompanied by location information. However,standard formulations of these models can be unwieldy and computationally intractable for large spatialscales, many individuals, and/or activity center movement. We provide a cumulative series of methodsthat yield dramatic improvements in Markov chain Monte Carlo (MCMC) estimation for two examples.These include removing unnecessary computations, integrating out latent states, vectorizing declarations,and restricting calculations to the locality of individuals. Our approaches leverage the exibility providedby the nimble R package. In our rst example, we demonstrate an improvement in MCMC efciency (therate of generating effectively independent posterior samples) by a factor of 100. In our second example, wereduce the computing time required to generate 10,000 posterior samples from 4.5 h down to ve minutes,and realize an increase in MCMC efciency by a factor of 25. These approaches can also be applied gener-ally to other spatially indexed hierarchical models. We provide R code for all examples, an executable web-appendix, and generalized versions of these techniques are made available in the nimbleSCR R package. Markov chain Monte Carlo; Mark–recapture; nimble; sampling efficiency; spatial capture–recapture

Bidragsytere

Daniel Turek

  • Tilknyttet:
    Forfatter
    ved Williams College

Cyril Pierre Milleret

  • Tilknyttet:
    Forfatter
    ved Miljøvitenskap og naturforvaltning ved Norges miljø- og biovitenskapelige universitet

Torbjørn Håkan Ergon

Bidragsyterens navn vises på dette resultatet som Torbjørn Ergon
  • Tilknyttet:
    Forfatter
    ved Centre for Ecological and Evolutionary Synthesis ved Universitetet i Oslo

Henrik Brøseth

  • Tilknyttet:
    Forfatter
    ved NINA terrestrisk økologi ved Norsk institutt for naturforskning

Pierre Dupont

  • Tilknyttet:
    Forfatter
    ved Miljøvitenskap og naturforvaltning ved Norges miljø- og biovitenskapelige universitet
1 - 5 av 7 | Neste | Siste »