Using prior information to build probabilistic invasive species risk assessments

Jeffrey Diez, Philip Hulme, Richard Duncan

    Research output: Contribution to journalArticle

    12 Citations (Scopus)

    Abstract

    Understanding why some introduced species become naturalized and invasive whereas others do not is a major focus of invasion ecology. Invasive species risk assessments address this same question, but are not typically based on the results from recent ecological studies. Applying results fromthe ecological literature to risk assessment is difficult, in part because there are no general explanations of invasion likelihood across taxa. Most ecological studies are also specific to a particular region and it is unclear whether outcomes in one region will necessarily apply to another. Here we show how a hierarchical Bayesian statistical framework can make better use of ecological studies for applied risk assessments. We focus on three key opportunities afforded by these models: (1) the ability to leverage information from one region to form prior expectations for other regions about which little is known, (2) the ability to quantify uncertainty of predictions, and (3) flexibility to incorporate within-group heterogeneities in probabilities of naturalization. We illustrate these principles using a case study where we predict the probability of plant taxa naturalizing in New Zealand and Australia, showing how prior information can be particularly valuable when data are limited. As more studies document invasion patterns around the world, a framework that can formally incorporate prior information will help link the accumulating data on species introductions to risk assessments.
    Original languageEnglish
    Pages (from-to)681-691
    Number of pages11
    JournalBiological Invasions
    Volume14
    DOIs
    Publication statusPublished - 2012

    Fingerprint

    invasive species
    risk assessment
    naturalization
    introduced species
    uncertainty
    case studies
    ecology
    prediction

    Cite this

    @article{07dfd6f7cea84e40abfe1fad4e74fbf3,
    title = "Using prior information to build probabilistic invasive species risk assessments",
    abstract = "Understanding why some introduced species become naturalized and invasive whereas others do not is a major focus of invasion ecology. Invasive species risk assessments address this same question, but are not typically based on the results from recent ecological studies. Applying results fromthe ecological literature to risk assessment is difficult, in part because there are no general explanations of invasion likelihood across taxa. Most ecological studies are also specific to a particular region and it is unclear whether outcomes in one region will necessarily apply to another. Here we show how a hierarchical Bayesian statistical framework can make better use of ecological studies for applied risk assessments. We focus on three key opportunities afforded by these models: (1) the ability to leverage information from one region to form prior expectations for other regions about which little is known, (2) the ability to quantify uncertainty of predictions, and (3) flexibility to incorporate within-group heterogeneities in probabilities of naturalization. We illustrate these principles using a case study where we predict the probability of plant taxa naturalizing in New Zealand and Australia, showing how prior information can be particularly valuable when data are limited. As more studies document invasion patterns around the world, a framework that can formally incorporate prior information will help link the accumulating data on species introductions to risk assessments.",
    keywords = "Invasive species, Risk assessment, Bayesian, Prior information.",
    author = "Jeffrey Diez and Philip Hulme and Richard Duncan",
    year = "2012",
    doi = "10.1007/s10530-011-0109-5",
    language = "English",
    volume = "14",
    pages = "681--691",
    journal = "Biological Invasions",
    issn = "1387-3547",
    publisher = "Springer",

    }

    Using prior information to build probabilistic invasive species risk assessments. / Diez, Jeffrey; Hulme, Philip; Duncan, Richard.

    In: Biological Invasions, Vol. 14, 2012, p. 681-691.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - Using prior information to build probabilistic invasive species risk assessments

    AU - Diez, Jeffrey

    AU - Hulme, Philip

    AU - Duncan, Richard

    PY - 2012

    Y1 - 2012

    N2 - Understanding why some introduced species become naturalized and invasive whereas others do not is a major focus of invasion ecology. Invasive species risk assessments address this same question, but are not typically based on the results from recent ecological studies. Applying results fromthe ecological literature to risk assessment is difficult, in part because there are no general explanations of invasion likelihood across taxa. Most ecological studies are also specific to a particular region and it is unclear whether outcomes in one region will necessarily apply to another. Here we show how a hierarchical Bayesian statistical framework can make better use of ecological studies for applied risk assessments. We focus on three key opportunities afforded by these models: (1) the ability to leverage information from one region to form prior expectations for other regions about which little is known, (2) the ability to quantify uncertainty of predictions, and (3) flexibility to incorporate within-group heterogeneities in probabilities of naturalization. We illustrate these principles using a case study where we predict the probability of plant taxa naturalizing in New Zealand and Australia, showing how prior information can be particularly valuable when data are limited. As more studies document invasion patterns around the world, a framework that can formally incorporate prior information will help link the accumulating data on species introductions to risk assessments.

    AB - Understanding why some introduced species become naturalized and invasive whereas others do not is a major focus of invasion ecology. Invasive species risk assessments address this same question, but are not typically based on the results from recent ecological studies. Applying results fromthe ecological literature to risk assessment is difficult, in part because there are no general explanations of invasion likelihood across taxa. Most ecological studies are also specific to a particular region and it is unclear whether outcomes in one region will necessarily apply to another. Here we show how a hierarchical Bayesian statistical framework can make better use of ecological studies for applied risk assessments. We focus on three key opportunities afforded by these models: (1) the ability to leverage information from one region to form prior expectations for other regions about which little is known, (2) the ability to quantify uncertainty of predictions, and (3) flexibility to incorporate within-group heterogeneities in probabilities of naturalization. We illustrate these principles using a case study where we predict the probability of plant taxa naturalizing in New Zealand and Australia, showing how prior information can be particularly valuable when data are limited. As more studies document invasion patterns around the world, a framework that can formally incorporate prior information will help link the accumulating data on species introductions to risk assessments.

    KW - Invasive species

    KW - Risk assessment

    KW - Bayesian

    KW - Prior information.

    U2 - 10.1007/s10530-011-0109-5

    DO - 10.1007/s10530-011-0109-5

    M3 - Article

    VL - 14

    SP - 681

    EP - 691

    JO - Biological Invasions

    JF - Biological Invasions

    SN - 1387-3547

    ER -