Uncertainty in spatially explicit animal dispersal models

Wolf M. Mooij, Donald L. DeAngelis

Research output: Contribution to journalArticle

31 Citations (Scopus)

Abstract

Uncertainty in estimates of survival of dispersing animals is a vexing difficulty in conservation biology. The current notion is that this uncertainty decreases the usefulness of spatially explicit population models in particular. We examined this problem by comparing dispersal models of three levels of complexity: (1) an event-based binomial model that considers only the occurrence of mortality or arrival, (2) a temporally explicit exponential model that employs mortality and arrival rates, and (3) a spatially explicit gridwalk model that simulates the movement of animals through an artificial landscape. Each model was fitted to the same set of field data. A first objective of the paper is to illustrate how the maximum-likelihood method can be used in all three cases to estimate the means and confidence limits for the relevant model parameters, given a particular set of data on dispersal survival. Using this framework we show that the structure of the uncertainty for all three models is strikingly similar. In fact, the results of our unified approach imply that spatially explicit dispersal models, which take advantage of information on landscape details, suffer less from uncertainly than do simpler models. Moreover, we show that the proposed strategy of model development safeguards one from error propagation in these more complex models. Finally, our approach shows that all models related to animal dispersal, ranging from simple to complex, can be related in a hierarchical fashion, so that the various approaches to modeling such dispersal can be viewed from a unified perspective.

Original languageEnglish
Pages (from-to)794-805
Number of pages12
JournalEcological Applications
Volume13
Issue number3
StatePublished - Jun 1 2003

Fingerprint

animal
mortality
modeling

Keywords

  • Dispersal models, uncertainty
  • Dispersal mortality
  • Dispersal success
  • Individual-based modeling
  • Landscape details
  • Likelihood, maximum
  • Managing endangered species
  • Model complexity
  • Population models
  • Random grid-walk models

ASJC Scopus subject areas

  • Ecology

Cite this

Mooij, W. M., & DeAngelis, D. L. (2003). Uncertainty in spatially explicit animal dispersal models. Ecological Applications, 13(3), 794-805.

Uncertainty in spatially explicit animal dispersal models. / Mooij, Wolf M.; DeAngelis, Donald L.

In: Ecological Applications, Vol. 13, No. 3, 01.06.2003, p. 794-805.

Research output: Contribution to journalArticle

Mooij, WM & DeAngelis, DL 2003, 'Uncertainty in spatially explicit animal dispersal models', Ecological Applications, vol. 13, no. 3, pp. 794-805.
Mooij, Wolf M. ; DeAngelis, Donald L. / Uncertainty in spatially explicit animal dispersal models. In: Ecological Applications. 2003 ; Vol. 13, No. 3. pp. 794-805.
@article{cb038e1e85b542479dc83bfd3efcee06,
title = "Uncertainty in spatially explicit animal dispersal models",
abstract = "Uncertainty in estimates of survival of dispersing animals is a vexing difficulty in conservation biology. The current notion is that this uncertainty decreases the usefulness of spatially explicit population models in particular. We examined this problem by comparing dispersal models of three levels of complexity: (1) an event-based binomial model that considers only the occurrence of mortality or arrival, (2) a temporally explicit exponential model that employs mortality and arrival rates, and (3) a spatially explicit gridwalk model that simulates the movement of animals through an artificial landscape. Each model was fitted to the same set of field data. A first objective of the paper is to illustrate how the maximum-likelihood method can be used in all three cases to estimate the means and confidence limits for the relevant model parameters, given a particular set of data on dispersal survival. Using this framework we show that the structure of the uncertainty for all three models is strikingly similar. In fact, the results of our unified approach imply that spatially explicit dispersal models, which take advantage of information on landscape details, suffer less from uncertainly than do simpler models. Moreover, we show that the proposed strategy of model development safeguards one from error propagation in these more complex models. Finally, our approach shows that all models related to animal dispersal, ranging from simple to complex, can be related in a hierarchical fashion, so that the various approaches to modeling such dispersal can be viewed from a unified perspective.",
keywords = "Dispersal models, uncertainty, Dispersal mortality, Dispersal success, Individual-based modeling, Landscape details, Likelihood, maximum, Managing endangered species, Model complexity, Population models, Random grid-walk models",
author = "Mooij, {Wolf M.} and DeAngelis, {Donald L.}",
year = "2003",
month = "6",
day = "1",
language = "English",
volume = "13",
pages = "794--805",
journal = "Ecological Appplications",
issn = "1051-0761",
publisher = "Ecological Society of America",
number = "3",

}

TY - JOUR

T1 - Uncertainty in spatially explicit animal dispersal models

AU - Mooij, Wolf M.

AU - DeAngelis, Donald L.

PY - 2003/6/1

Y1 - 2003/6/1

N2 - Uncertainty in estimates of survival of dispersing animals is a vexing difficulty in conservation biology. The current notion is that this uncertainty decreases the usefulness of spatially explicit population models in particular. We examined this problem by comparing dispersal models of three levels of complexity: (1) an event-based binomial model that considers only the occurrence of mortality or arrival, (2) a temporally explicit exponential model that employs mortality and arrival rates, and (3) a spatially explicit gridwalk model that simulates the movement of animals through an artificial landscape. Each model was fitted to the same set of field data. A first objective of the paper is to illustrate how the maximum-likelihood method can be used in all three cases to estimate the means and confidence limits for the relevant model parameters, given a particular set of data on dispersal survival. Using this framework we show that the structure of the uncertainty for all three models is strikingly similar. In fact, the results of our unified approach imply that spatially explicit dispersal models, which take advantage of information on landscape details, suffer less from uncertainly than do simpler models. Moreover, we show that the proposed strategy of model development safeguards one from error propagation in these more complex models. Finally, our approach shows that all models related to animal dispersal, ranging from simple to complex, can be related in a hierarchical fashion, so that the various approaches to modeling such dispersal can be viewed from a unified perspective.

AB - Uncertainty in estimates of survival of dispersing animals is a vexing difficulty in conservation biology. The current notion is that this uncertainty decreases the usefulness of spatially explicit population models in particular. We examined this problem by comparing dispersal models of three levels of complexity: (1) an event-based binomial model that considers only the occurrence of mortality or arrival, (2) a temporally explicit exponential model that employs mortality and arrival rates, and (3) a spatially explicit gridwalk model that simulates the movement of animals through an artificial landscape. Each model was fitted to the same set of field data. A first objective of the paper is to illustrate how the maximum-likelihood method can be used in all three cases to estimate the means and confidence limits for the relevant model parameters, given a particular set of data on dispersal survival. Using this framework we show that the structure of the uncertainty for all three models is strikingly similar. In fact, the results of our unified approach imply that spatially explicit dispersal models, which take advantage of information on landscape details, suffer less from uncertainly than do simpler models. Moreover, we show that the proposed strategy of model development safeguards one from error propagation in these more complex models. Finally, our approach shows that all models related to animal dispersal, ranging from simple to complex, can be related in a hierarchical fashion, so that the various approaches to modeling such dispersal can be viewed from a unified perspective.

KW - Dispersal models, uncertainty

KW - Dispersal mortality

KW - Dispersal success

KW - Individual-based modeling

KW - Landscape details

KW - Likelihood, maximum

KW - Managing endangered species

KW - Model complexity

KW - Population models

KW - Random grid-walk models

UR - http://www.scopus.com/inward/record.url?scp=0346361812&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0346361812&partnerID=8YFLogxK

M3 - Article

VL - 13

SP - 794

EP - 805

JO - Ecological Appplications

JF - Ecological Appplications

SN - 1051-0761

IS - 3

ER -