Testing the accuracy of population viability analysis

M. A. McCarthy, H. P. Possingham, J. R. Day, A. J. Tyre

Research output: Contribution to journalArticle

52 Citations (Scopus)

Abstract

Despite the growing use of population viability analysis (PVA), the predictions of these models rarely have been tested with field data that were not used in initially developing the model. We review and discuss a suite of methods that may be used to test the predictive ability of models used in PVA. In addition to testing mean predictions, appropriate methods must analyze the probability distribution of the model predictions. The methods we discuss provide tests of the mean predictions, the predicted frequency of events such as extinction and colonization, and the predicted probability distribution of state variables. We discuss visual approaches based on plots of observations versus the predictions and statistical approaches based on determining significant differences between observations and predictions. The advantages and disadvantages of each method are identified. The best methods test the statistical distribution of the predictions; those that ignore variability are meaningless. Although we recognize that the quality of a model is not solely a function of its predictive abilities, tests help reduce inherent model uncertainty. The role of model testing is not to prove the truth of a model, which is impossible because models are never a perfect description of reality. Rather, testing should help identify the weakest aspects of models so they can be improved. We provide a framework for using model testing to improve the predictive performance of PVA models, through an iterative process of model development, testing, subsequent modification and re-testing.

Original languageEnglish (US)
Pages (from-to)1030-1038
Number of pages9
JournalConservation Biology
Volume15
Issue number4
DOIs
StatePublished - Aug 15 2001

Fingerprint

population viability analysis
viability
Testing
testing
prediction
probability distribution
Probability distributions
methodology
model uncertainty
statistical distribution

ASJC Scopus subject areas

  • Ecology, Evolution, Behavior and Systematics
  • Ecology
  • Nature and Landscape Conservation

Cite this

Testing the accuracy of population viability analysis. / McCarthy, M. A.; Possingham, H. P.; Day, J. R.; Tyre, A. J.

In: Conservation Biology, Vol. 15, No. 4, 15.08.2001, p. 1030-1038.

Research output: Contribution to journalArticle

McCarthy, M. A. ; Possingham, H. P. ; Day, J. R. ; Tyre, A. J. / Testing the accuracy of population viability analysis. In: Conservation Biology. 2001 ; Vol. 15, No. 4. pp. 1030-1038.
@article{c79eb88645424000b621b0888ce3f553,
title = "Testing the accuracy of population viability analysis",
abstract = "Despite the growing use of population viability analysis (PVA), the predictions of these models rarely have been tested with field data that were not used in initially developing the model. We review and discuss a suite of methods that may be used to test the predictive ability of models used in PVA. In addition to testing mean predictions, appropriate methods must analyze the probability distribution of the model predictions. The methods we discuss provide tests of the mean predictions, the predicted frequency of events such as extinction and colonization, and the predicted probability distribution of state variables. We discuss visual approaches based on plots of observations versus the predictions and statistical approaches based on determining significant differences between observations and predictions. The advantages and disadvantages of each method are identified. The best methods test the statistical distribution of the predictions; those that ignore variability are meaningless. Although we recognize that the quality of a model is not solely a function of its predictive abilities, tests help reduce inherent model uncertainty. The role of model testing is not to prove the truth of a model, which is impossible because models are never a perfect description of reality. Rather, testing should help identify the weakest aspects of models so they can be improved. We provide a framework for using model testing to improve the predictive performance of PVA models, through an iterative process of model development, testing, subsequent modification and re-testing.",
author = "McCarthy, {M. A.} and Possingham, {H. P.} and Day, {J. R.} and Tyre, {A. J.}",
year = "2001",
month = "8",
day = "15",
doi = "10.1046/j.1523-1739.2001.0150041030.x",
language = "English (US)",
volume = "15",
pages = "1030--1038",
journal = "Conservation Biology",
issn = "0888-8892",
publisher = "Wiley-Blackwell",
number = "4",

}

TY - JOUR

T1 - Testing the accuracy of population viability analysis

AU - McCarthy, M. A.

AU - Possingham, H. P.

AU - Day, J. R.

AU - Tyre, A. J.

PY - 2001/8/15

Y1 - 2001/8/15

N2 - Despite the growing use of population viability analysis (PVA), the predictions of these models rarely have been tested with field data that were not used in initially developing the model. We review and discuss a suite of methods that may be used to test the predictive ability of models used in PVA. In addition to testing mean predictions, appropriate methods must analyze the probability distribution of the model predictions. The methods we discuss provide tests of the mean predictions, the predicted frequency of events such as extinction and colonization, and the predicted probability distribution of state variables. We discuss visual approaches based on plots of observations versus the predictions and statistical approaches based on determining significant differences between observations and predictions. The advantages and disadvantages of each method are identified. The best methods test the statistical distribution of the predictions; those that ignore variability are meaningless. Although we recognize that the quality of a model is not solely a function of its predictive abilities, tests help reduce inherent model uncertainty. The role of model testing is not to prove the truth of a model, which is impossible because models are never a perfect description of reality. Rather, testing should help identify the weakest aspects of models so they can be improved. We provide a framework for using model testing to improve the predictive performance of PVA models, through an iterative process of model development, testing, subsequent modification and re-testing.

AB - Despite the growing use of population viability analysis (PVA), the predictions of these models rarely have been tested with field data that were not used in initially developing the model. We review and discuss a suite of methods that may be used to test the predictive ability of models used in PVA. In addition to testing mean predictions, appropriate methods must analyze the probability distribution of the model predictions. The methods we discuss provide tests of the mean predictions, the predicted frequency of events such as extinction and colonization, and the predicted probability distribution of state variables. We discuss visual approaches based on plots of observations versus the predictions and statistical approaches based on determining significant differences between observations and predictions. The advantages and disadvantages of each method are identified. The best methods test the statistical distribution of the predictions; those that ignore variability are meaningless. Although we recognize that the quality of a model is not solely a function of its predictive abilities, tests help reduce inherent model uncertainty. The role of model testing is not to prove the truth of a model, which is impossible because models are never a perfect description of reality. Rather, testing should help identify the weakest aspects of models so they can be improved. We provide a framework for using model testing to improve the predictive performance of PVA models, through an iterative process of model development, testing, subsequent modification and re-testing.

UR - http://www.scopus.com/inward/record.url?scp=0034906377&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0034906377&partnerID=8YFLogxK

U2 - 10.1046/j.1523-1739.2001.0150041030.x

DO - 10.1046/j.1523-1739.2001.0150041030.x

M3 - Article

AN - SCOPUS:0034906377

VL - 15

SP - 1030

EP - 1038

JO - Conservation Biology

JF - Conservation Biology

SN - 0888-8892

IS - 4

ER -