Bayesian reference condition models achieve comparable or superior performance to existing standard techniques

J. A. Webb, Elise King, Trefor Reynoldson, Mark Padgham

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Despite the existence of many approaches to reference-condition modeling, Bayesian statistical methods have not been used. We assessed whether a hybrid approach that combined features of existing reference-condition approaches with Bayesian model fitting and assessment of test sites could provide superior results to existing established methods. We used 4 Bayesian models of increasing complexity to develop and test referencecondition models for 5 biotic endpoints across 3 data sets. Our best models were comparable or superior to standard approaches (Benthic Assessment of Sediment, Australian River Assessment System) using the same data. Those of our models with the simplest endpoint (species richness) performed best. On average, those models with the simplest model structures also performed best, but differences in performance among models of different complexity were small. All models performed poorly at detecting the lower levels of simulated impact in the test data. However, these impacts were small relative to the variation among validation sites and consequent predictive uncertainty of the models. The Bayesian approach to reference-condition modeling shows promise as an alternative to existing methods. It also has advantages in terms of the ease of interpretation of model outputs. However, for the approach to be relevant, further development work should be driven by a perceived need to revise standard methods used by management agencies
Original languageEnglish
Pages (from-to)1272-1285
Number of pages14
JournalFreshwater Science
Volume33
Issue number4
DOIs
Publication statusPublished - 2014

Fingerprint

methodology
endpoints
model uncertainty
model test
testing
modeling
species richness
statistical analysis
method
species diversity
sediments
rivers
river
sediment
test

Cite this

Webb, J. A. ; King, Elise ; Reynoldson, Trefor ; Padgham, Mark. / Bayesian reference condition models achieve comparable or superior performance to existing standard techniques. In: Freshwater Science. 2014 ; Vol. 33, No. 4. pp. 1272-1285.
@article{ed7122808a5c4287a1bbc943d38843d9,
title = "Bayesian reference condition models achieve comparable or superior performance to existing standard techniques",
abstract = "Despite the existence of many approaches to reference-condition modeling, Bayesian statistical methods have not been used. We assessed whether a hybrid approach that combined features of existing reference-condition approaches with Bayesian model fitting and assessment of test sites could provide superior results to existing established methods. We used 4 Bayesian models of increasing complexity to develop and test referencecondition models for 5 biotic endpoints across 3 data sets. Our best models were comparable or superior to standard approaches (Benthic Assessment of Sediment, Australian River Assessment System) using the same data. Those of our models with the simplest endpoint (species richness) performed best. On average, those models with the simplest model structures also performed best, but differences in performance among models of different complexity were small. All models performed poorly at detecting the lower levels of simulated impact in the test data. However, these impacts were small relative to the variation among validation sites and consequent predictive uncertainty of the models. The Bayesian approach to reference-condition modeling shows promise as an alternative to existing methods. It also has advantages in terms of the ease of interpretation of model outputs. However, for the approach to be relevant, further development work should be driven by a perceived need to revise standard methods used by management agencies",
keywords = "Bayesian hierarchical model, Macroinvertebrate, OpenBUGS, Principal components analysis, R, Random effects model, Reference condition model",
author = "Webb, {J. A.} and Elise King and Trefor Reynoldson and Mark Padgham",
year = "2014",
doi = "10.1086/678949",
language = "English",
volume = "33",
pages = "1272--1285",
journal = "Journal of the North American Benthological Society",
issn = "2161-9565",
publisher = "The Society for Freshwater Science",
number = "4",

}

Bayesian reference condition models achieve comparable or superior performance to existing standard techniques. / Webb, J. A.; King, Elise; Reynoldson, Trefor; Padgham, Mark.

In: Freshwater Science, Vol. 33, No. 4, 2014, p. 1272-1285.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Bayesian reference condition models achieve comparable or superior performance to existing standard techniques

AU - Webb, J. A.

AU - King, Elise

AU - Reynoldson, Trefor

AU - Padgham, Mark

PY - 2014

Y1 - 2014

N2 - Despite the existence of many approaches to reference-condition modeling, Bayesian statistical methods have not been used. We assessed whether a hybrid approach that combined features of existing reference-condition approaches with Bayesian model fitting and assessment of test sites could provide superior results to existing established methods. We used 4 Bayesian models of increasing complexity to develop and test referencecondition models for 5 biotic endpoints across 3 data sets. Our best models were comparable or superior to standard approaches (Benthic Assessment of Sediment, Australian River Assessment System) using the same data. Those of our models with the simplest endpoint (species richness) performed best. On average, those models with the simplest model structures also performed best, but differences in performance among models of different complexity were small. All models performed poorly at detecting the lower levels of simulated impact in the test data. However, these impacts were small relative to the variation among validation sites and consequent predictive uncertainty of the models. The Bayesian approach to reference-condition modeling shows promise as an alternative to existing methods. It also has advantages in terms of the ease of interpretation of model outputs. However, for the approach to be relevant, further development work should be driven by a perceived need to revise standard methods used by management agencies

AB - Despite the existence of many approaches to reference-condition modeling, Bayesian statistical methods have not been used. We assessed whether a hybrid approach that combined features of existing reference-condition approaches with Bayesian model fitting and assessment of test sites could provide superior results to existing established methods. We used 4 Bayesian models of increasing complexity to develop and test referencecondition models for 5 biotic endpoints across 3 data sets. Our best models were comparable or superior to standard approaches (Benthic Assessment of Sediment, Australian River Assessment System) using the same data. Those of our models with the simplest endpoint (species richness) performed best. On average, those models with the simplest model structures also performed best, but differences in performance among models of different complexity were small. All models performed poorly at detecting the lower levels of simulated impact in the test data. However, these impacts were small relative to the variation among validation sites and consequent predictive uncertainty of the models. The Bayesian approach to reference-condition modeling shows promise as an alternative to existing methods. It also has advantages in terms of the ease of interpretation of model outputs. However, for the approach to be relevant, further development work should be driven by a perceived need to revise standard methods used by management agencies

KW - Bayesian hierarchical model

KW - Macroinvertebrate

KW - OpenBUGS

KW - Principal components analysis

KW - R

KW - Random effects model

KW - Reference condition model

UR - http://www.scopus.com/inward/record.url?scp=84937579356&partnerID=8YFLogxK

UR - http://www.mendeley.com/research/bayesian-reference-condition-models-achieve-comparable-superior-performance-existing-standard-techni

U2 - 10.1086/678949

DO - 10.1086/678949

M3 - Article

VL - 33

SP - 1272

EP - 1285

JO - Journal of the North American Benthological Society

JF - Journal of the North American Benthological Society

SN - 2161-9565

IS - 4

ER -