Advanced

Population dynamics and demography - inferences from stochastic models

Knape, Jonas LU (2009)
Abstract
Time series analysis of population abundances is a common approach to making inferences about population dynamics, and especially for estimating density dependence. Traditionally the impact of potential errors in observations or measurements of the abundances on inference has been ignored. Following the increased availability of appropriate software as well as statistical awareness among ecologist, state space modelling explicitly accounting for observation errors is becoming more commonly used.

In paper I we fit a linear state space model to time series of trapping numbers on four species of birds caught during migration. We show that these time series are most likely heavily influenced by local weather conditions. In paper II... (More)
Time series analysis of population abundances is a common approach to making inferences about population dynamics, and especially for estimating density dependence. Traditionally the impact of potential errors in observations or measurements of the abundances on inference has been ignored. Following the increased availability of appropriate software as well as statistical awareness among ecologist, state space modelling explicitly accounting for observation errors is becoming more commonly used.

In paper I we fit a linear state space model to time series of trapping numbers on four species of birds caught during migration. We show that these time series are most likely heavily influenced by local weather conditions. In paper II I use numerical evaluations to show that the precision of density dependence estimates obtained by fitting linear state space models depends critically on how strong the density dependence is. The precision can be increased if independent estimates of the observation error variance are available. If such independent estimates are inaccurate they may however induce bias. This is examined in paper III. In paper IV we numerically evaluate how estimates of environmental effects are affected by observation errors. If observation errors are ignored, effects of covariates may be biased if the covariate is temporally autocorrelated.

Unobserved individual heterogeneity can cause problems in inferences about individual level vital rates. Individual heterogeneity is also of interest in its own right, e.g. in studies of life-history trade-offs. In paper V we analyse data on hatching success in a population of individually marked great reed warblers (Acrocephalus arundinaceus). We show that hatching success was reduced if individuals were related, as measured by pedigree data, and that some females performed consistently worse than the majority. Any differences between males were on the other hand not discernible. In paper VI we analyse individually based data on survival and breeding performance in an island population of silvereyes (Zosterops lateralis). We fit models of breeding success and survival including both age effects and individual random effects. We found senescence in both breeding success and survival but no strong indications of any correlations between individual breeding success and survival. We also show that the type of age effect included in such models may heavily influence estimates of within individual correlations. (Less)
Please use this url to cite or link to this publication:
author
supervisor
opponent
  • Prof. Dennis, Brian, University of Idaho, Moscow, USA
organization
publishing date
type
Thesis
publication status
published
subject
publisher
Department of Ecology, Lund University
defense location
Blå Hallen, Ekologihuset, Sölvegatan 37, Lund
defense date
2009-02-06 10:00
ISBN
978-91-7105-289-6
language
English
LU publication?
yes
id
cbf105c7-37e5-4997-989d-326caf0b2649 (old id 1277541)
date added to LUP
2009-01-14 08:32:16
date last changed
2016-09-19 08:45:14
@phdthesis{cbf105c7-37e5-4997-989d-326caf0b2649,
  abstract     = {Time series analysis of population abundances is a common approach to making inferences about population dynamics, and especially for estimating density dependence. Traditionally the impact of potential errors in observations or measurements of the abundances on inference has been ignored. Following the increased availability of appropriate software as well as statistical awareness among ecologist, state space modelling explicitly accounting for observation errors is becoming more commonly used. <br/><br>
 In paper I we fit a linear state space model to time series of trapping numbers on four species of birds caught during migration. We show that these time series are most likely heavily influenced by local weather conditions. In paper II I use numerical evaluations to show that the precision of density dependence estimates obtained by fitting linear state space models depends critically on how strong the density dependence is. The precision can be increased if independent estimates of the observation error variance are available. If such independent estimates are inaccurate they may however induce bias. This is examined in paper III. In paper IV we numerically evaluate how estimates of environmental effects are affected by observation errors. If observation errors are ignored, effects of covariates may be biased if the covariate is temporally autocorrelated.<br/><br>
 Unobserved individual heterogeneity can cause problems in inferences about individual level vital rates. Individual heterogeneity is also of interest in its own right, e.g. in studies of life-history trade-offs. In paper V we analyse data on hatching success in a population of individually marked great reed warblers (Acrocephalus arundinaceus). We show that hatching success was reduced if individuals were related, as measured by pedigree data, and that some females performed consistently worse than the majority. Any differences between males were on the other hand not discernible. In paper VI we analyse individually based data on survival and breeding performance in an island population of silvereyes (Zosterops lateralis). We fit models of breeding success and survival including both age effects and individual random effects. We found senescence in both breeding success and survival but no strong indications of any correlations between individual breeding success and survival. We also show that the type of age effect included in such models may heavily influence estimates of within individual correlations.},
  author       = {Knape, Jonas},
  isbn         = {978-91-7105-289-6},
  language     = {eng},
  publisher    = {Department of Ecology, Lund University},
  school       = {Lund University},
  title        = {Population dynamics and demography - inferences from stochastic models},
  year         = {2009},
}