Reliability of Different Mark-Recapture Methods for Population Size Estimation Tested against Reference Population Sizes Constructed from Field Data

Annegret Grimm, Bernd GRUBER, Klaus Henle

    Research output: Contribution to journalArticle

    15 Citations (Scopus)
    1 Downloads (Pure)

    Abstract

    Reliable estimates of population size are fundamental in many ecological studies and biodiversity conservation. Selecting appropriate methods to estimate abundance is often very difficult, especially if data are scarce. Most studies concerning the reliability of different estimators used simulation data based on assumptions about capture variability that do not necessarily reflect conditions in natural populations. Here, we used data from an intensively studied closed population of the arboreal gecko Gehyra variegata to construct reference population sizes for assessing twelve different population size estimators in terms of bias, precision, accuracy, and their 95%-confidence intervals. Two of the reference populations reflect natural biological entities, whereas the other reference populations reflect artificial subsets of the population. Since individual heterogeneity was assumed, we tested modifications of the Lincoln-Petersen estimator, a set of models in programs MARK and CARE-2, and a truncated geometric distribution. Ranking of methods was similar across criteria. Models accounting for individual heterogeneity performed best in all assessment criteria. For populations from heterogeneous habitats without obvious covariates explaining individual heterogeneity, we recommend using the moment estimator or the interpolated jackknife estimator (both implemented in CAPTURE/MARK). If data for capture frequencies are substantial, we recommend the sample coverage or the estimating equation (both models implemented in CARE-2). Depending on the distribution of catchabilities, our proposed multiple Lincoln-Petersen and a truncated geometric distribution obtained comparably good results. The former usually resulted in a minimum population size and the latter can be recommended when there is a long tail of low capture probabilities. Models with covariates and mixture models performed poorly. Our approach identified suitable methods and extended options to evaluate the performance of mark-recapture population size estimators under field conditions, which is essential for selecting an appropriate method and obtaining reliable results in ecology and conservation biology, and thus for sound management.
    Original languageEnglish
    Pages (from-to)1-11
    Number of pages11
    JournalPLoS One
    Volume9
    Issue number6
    DOIs
    Publication statusPublished - 2014

    Fingerprint

    Population Density
    population size
    Population
    Conservation
    methodology
    Biodiversity
    Ecology
    Lizards
    Gekkonidae
    Ecosystem
    Tail
    Acoustic waves
    confidence interval
    tail
    Confidence Intervals
    biodiversity
    ecology
    Biological Sciences
    habitats
    sampling

    Cite this

    @article{4cd8fde6144b4e6fa5b2d21ee62c5d09,
    title = "Reliability of Different Mark-Recapture Methods for Population Size Estimation Tested against Reference Population Sizes Constructed from Field Data",
    abstract = "Reliable estimates of population size are fundamental in many ecological studies and biodiversity conservation. Selecting appropriate methods to estimate abundance is often very difficult, especially if data are scarce. Most studies concerning the reliability of different estimators used simulation data based on assumptions about capture variability that do not necessarily reflect conditions in natural populations. Here, we used data from an intensively studied closed population of the arboreal gecko Gehyra variegata to construct reference population sizes for assessing twelve different population size estimators in terms of bias, precision, accuracy, and their 95{\%}-confidence intervals. Two of the reference populations reflect natural biological entities, whereas the other reference populations reflect artificial subsets of the population. Since individual heterogeneity was assumed, we tested modifications of the Lincoln-Petersen estimator, a set of models in programs MARK and CARE-2, and a truncated geometric distribution. Ranking of methods was similar across criteria. Models accounting for individual heterogeneity performed best in all assessment criteria. For populations from heterogeneous habitats without obvious covariates explaining individual heterogeneity, we recommend using the moment estimator or the interpolated jackknife estimator (both implemented in CAPTURE/MARK). If data for capture frequencies are substantial, we recommend the sample coverage or the estimating equation (both models implemented in CARE-2). Depending on the distribution of catchabilities, our proposed multiple Lincoln-Petersen and a truncated geometric distribution obtained comparably good results. The former usually resulted in a minimum population size and the latter can be recommended when there is a long tail of low capture probabilities. Models with covariates and mixture models performed poorly. Our approach identified suitable methods and extended options to evaluate the performance of mark-recapture population size estimators under field conditions, which is essential for selecting an appropriate method and obtaining reliable results in ecology and conservation biology, and thus for sound management.",
    author = "Annegret Grimm and Bernd GRUBER and Klaus Henle",
    year = "2014",
    doi = "10.1371/journal.pone.0098840",
    language = "English",
    volume = "9",
    pages = "1--11",
    journal = "PLoS One",
    issn = "1932-6203",
    publisher = "Public Library of Science",
    number = "6",

    }

    Reliability of Different Mark-Recapture Methods for Population Size Estimation Tested against Reference Population Sizes Constructed from Field Data. / Grimm, Annegret; GRUBER, Bernd; Henle, Klaus.

    In: PLoS One, Vol. 9, No. 6, 2014, p. 1-11.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - Reliability of Different Mark-Recapture Methods for Population Size Estimation Tested against Reference Population Sizes Constructed from Field Data

    AU - Grimm, Annegret

    AU - GRUBER, Bernd

    AU - Henle, Klaus

    PY - 2014

    Y1 - 2014

    N2 - Reliable estimates of population size are fundamental in many ecological studies and biodiversity conservation. Selecting appropriate methods to estimate abundance is often very difficult, especially if data are scarce. Most studies concerning the reliability of different estimators used simulation data based on assumptions about capture variability that do not necessarily reflect conditions in natural populations. Here, we used data from an intensively studied closed population of the arboreal gecko Gehyra variegata to construct reference population sizes for assessing twelve different population size estimators in terms of bias, precision, accuracy, and their 95%-confidence intervals. Two of the reference populations reflect natural biological entities, whereas the other reference populations reflect artificial subsets of the population. Since individual heterogeneity was assumed, we tested modifications of the Lincoln-Petersen estimator, a set of models in programs MARK and CARE-2, and a truncated geometric distribution. Ranking of methods was similar across criteria. Models accounting for individual heterogeneity performed best in all assessment criteria. For populations from heterogeneous habitats without obvious covariates explaining individual heterogeneity, we recommend using the moment estimator or the interpolated jackknife estimator (both implemented in CAPTURE/MARK). If data for capture frequencies are substantial, we recommend the sample coverage or the estimating equation (both models implemented in CARE-2). Depending on the distribution of catchabilities, our proposed multiple Lincoln-Petersen and a truncated geometric distribution obtained comparably good results. The former usually resulted in a minimum population size and the latter can be recommended when there is a long tail of low capture probabilities. Models with covariates and mixture models performed poorly. Our approach identified suitable methods and extended options to evaluate the performance of mark-recapture population size estimators under field conditions, which is essential for selecting an appropriate method and obtaining reliable results in ecology and conservation biology, and thus for sound management.

    AB - Reliable estimates of population size are fundamental in many ecological studies and biodiversity conservation. Selecting appropriate methods to estimate abundance is often very difficult, especially if data are scarce. Most studies concerning the reliability of different estimators used simulation data based on assumptions about capture variability that do not necessarily reflect conditions in natural populations. Here, we used data from an intensively studied closed population of the arboreal gecko Gehyra variegata to construct reference population sizes for assessing twelve different population size estimators in terms of bias, precision, accuracy, and their 95%-confidence intervals. Two of the reference populations reflect natural biological entities, whereas the other reference populations reflect artificial subsets of the population. Since individual heterogeneity was assumed, we tested modifications of the Lincoln-Petersen estimator, a set of models in programs MARK and CARE-2, and a truncated geometric distribution. Ranking of methods was similar across criteria. Models accounting for individual heterogeneity performed best in all assessment criteria. For populations from heterogeneous habitats without obvious covariates explaining individual heterogeneity, we recommend using the moment estimator or the interpolated jackknife estimator (both implemented in CAPTURE/MARK). If data for capture frequencies are substantial, we recommend the sample coverage or the estimating equation (both models implemented in CARE-2). Depending on the distribution of catchabilities, our proposed multiple Lincoln-Petersen and a truncated geometric distribution obtained comparably good results. The former usually resulted in a minimum population size and the latter can be recommended when there is a long tail of low capture probabilities. Models with covariates and mixture models performed poorly. Our approach identified suitable methods and extended options to evaluate the performance of mark-recapture population size estimators under field conditions, which is essential for selecting an appropriate method and obtaining reliable results in ecology and conservation biology, and thus for sound management.

    U2 - 10.1371/journal.pone.0098840

    DO - 10.1371/journal.pone.0098840

    M3 - Article

    VL - 9

    SP - 1

    EP - 11

    JO - PLoS One

    JF - PLoS One

    SN - 1932-6203

    IS - 6

    ER -