Identifying spatially and temporally transferrable surrogate measures of species richness

E. Fleishman, J.D.L. Yen, J.R. Thomson, R. Mac Nally, D.S. Dobkin, M. Leu

    Research output: Contribution to journalArticle

    4 Citations (Scopus)

    Abstract

    We developed a transferable method to identify indicator species and environmental variables that explain considerable variation in species richness. We applied this method to birds and butterflies and conducted novel, rigorous external evaluations of the spatial and temporal transferability of such indicator species. We collected data in the central Great Basin (Lander, Nye, and Eureka Counties, Nevada) and western Great Basin (Mono County, California and Mineral, Douglas, and Lyon Counties, Nevada), United States. We used Bayesian model selection to identify sets of five species that best explained observed variation in species richness of the same or the other taxonomic group in either subregion. We then built random forest models that included only the five
    identified indicator species and externally tested these models with new data from the other subregion (central or western Great Basin) or a later time in the same subregion. We compared the predictive accuracy of indicator species
    only models to that of models based on environmental variables or on both indicator species and environmental variables. In external evaluations, models based on same-taxon indicator species predicted 34–52% of the variation in species richness of birds and 40–70% of the variation in species richness of butterflies. Comparable models based on environmental variables predicted 11–46% of the variation in species richness of birds and 12–67% of the variation in species richness of butterflies. Models based on same-taxon indicator
    species predicted more variation in species richness than those based on environmental variables in seven of eight cases. Our results suggested that the predictive accuracy and spatial and temporal transferability of models
    based on indicator species can exceed that of models based on environmental variables. If mechanistic responses to environmental change are consistent through time, tracking the occurrence of a subset of an assemblage
    during periods of environmental change may allow inference to species richness of the assemblage.
    Original languageUndefined
    Pages (from-to)470-478
    Number of pages9
    JournalEcological Indicators
    Volume84
    DOIs
    Publication statusPublished - 2018

    Cite this

    Fleishman, E. ; Yen, J.D.L. ; Thomson, J.R. ; Mac Nally, R. ; Dobkin, D.S. ; Leu, M. / Identifying spatially and temporally transferrable surrogate measures of species richness. In: Ecological Indicators. 2018 ; Vol. 84. pp. 470-478.
    @article{c963a506b04e4f13871c0554b7013613,
    title = "Identifying spatially and temporally transferrable surrogate measures of species richness",
    abstract = "We developed a transferable method to identify indicator species and environmental variables that explain considerable variation in species richness. We applied this method to birds and butterflies and conducted novel, rigorous external evaluations of the spatial and temporal transferability of such indicator species. We collected data in the central Great Basin (Lander, Nye, and Eureka Counties, Nevada) and western Great Basin (Mono County, California and Mineral, Douglas, and Lyon Counties, Nevada), United States. We used Bayesian model selection to identify sets of five species that best explained observed variation in species richness of the same or the other taxonomic group in either subregion. We then built random forest models that included only the fiveidentified indicator species and externally tested these models with new data from the other subregion (central or western Great Basin) or a later time in the same subregion. We compared the predictive accuracy of indicator speciesonly models to that of models based on environmental variables or on both indicator species and environmental variables. In external evaluations, models based on same-taxon indicator species predicted 34–52{\%} of the variation in species richness of birds and 40–70{\%} of the variation in species richness of butterflies. Comparable models based on environmental variables predicted 11–46{\%} of the variation in species richness of birds and 12–67{\%} of the variation in species richness of butterflies. Models based on same-taxon indicatorspecies predicted more variation in species richness than those based on environmental variables in seven of eight cases. Our results suggested that the predictive accuracy and spatial and temporal transferability of modelsbased on indicator species can exceed that of models based on environmental variables. If mechanistic responses to environmental change are consistent through time, tracking the occurrence of a subset of an assemblageduring periods of environmental change may allow inference to species richness of the assemblage.",
    author = "E. Fleishman and J.D.L. Yen and J.R. Thomson and {Mac Nally}, R. and D.S. Dobkin and M. Leu",
    note = "cited By 0",
    year = "2018",
    doi = "10.1016/j.ecolind.2017.09.020",
    language = "Undefined",
    volume = "84",
    pages = "470--478",
    journal = "Ecological Indicators",
    issn = "1470-160X",
    publisher = "Elsevier",

    }

    Identifying spatially and temporally transferrable surrogate measures of species richness. / Fleishman, E.; Yen, J.D.L.; Thomson, J.R.; Mac Nally, R.; Dobkin, D.S.; Leu, M.

    In: Ecological Indicators, Vol. 84, 2018, p. 470-478.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - Identifying spatially and temporally transferrable surrogate measures of species richness

    AU - Fleishman, E.

    AU - Yen, J.D.L.

    AU - Thomson, J.R.

    AU - Mac Nally, R.

    AU - Dobkin, D.S.

    AU - Leu, M.

    N1 - cited By 0

    PY - 2018

    Y1 - 2018

    N2 - We developed a transferable method to identify indicator species and environmental variables that explain considerable variation in species richness. We applied this method to birds and butterflies and conducted novel, rigorous external evaluations of the spatial and temporal transferability of such indicator species. We collected data in the central Great Basin (Lander, Nye, and Eureka Counties, Nevada) and western Great Basin (Mono County, California and Mineral, Douglas, and Lyon Counties, Nevada), United States. We used Bayesian model selection to identify sets of five species that best explained observed variation in species richness of the same or the other taxonomic group in either subregion. We then built random forest models that included only the fiveidentified indicator species and externally tested these models with new data from the other subregion (central or western Great Basin) or a later time in the same subregion. We compared the predictive accuracy of indicator speciesonly models to that of models based on environmental variables or on both indicator species and environmental variables. In external evaluations, models based on same-taxon indicator species predicted 34–52% of the variation in species richness of birds and 40–70% of the variation in species richness of butterflies. Comparable models based on environmental variables predicted 11–46% of the variation in species richness of birds and 12–67% of the variation in species richness of butterflies. Models based on same-taxon indicatorspecies predicted more variation in species richness than those based on environmental variables in seven of eight cases. Our results suggested that the predictive accuracy and spatial and temporal transferability of modelsbased on indicator species can exceed that of models based on environmental variables. If mechanistic responses to environmental change are consistent through time, tracking the occurrence of a subset of an assemblageduring periods of environmental change may allow inference to species richness of the assemblage.

    AB - We developed a transferable method to identify indicator species and environmental variables that explain considerable variation in species richness. We applied this method to birds and butterflies and conducted novel, rigorous external evaluations of the spatial and temporal transferability of such indicator species. We collected data in the central Great Basin (Lander, Nye, and Eureka Counties, Nevada) and western Great Basin (Mono County, California and Mineral, Douglas, and Lyon Counties, Nevada), United States. We used Bayesian model selection to identify sets of five species that best explained observed variation in species richness of the same or the other taxonomic group in either subregion. We then built random forest models that included only the fiveidentified indicator species and externally tested these models with new data from the other subregion (central or western Great Basin) or a later time in the same subregion. We compared the predictive accuracy of indicator speciesonly models to that of models based on environmental variables or on both indicator species and environmental variables. In external evaluations, models based on same-taxon indicator species predicted 34–52% of the variation in species richness of birds and 40–70% of the variation in species richness of butterflies. Comparable models based on environmental variables predicted 11–46% of the variation in species richness of birds and 12–67% of the variation in species richness of butterflies. Models based on same-taxon indicatorspecies predicted more variation in species richness than those based on environmental variables in seven of eight cases. Our results suggested that the predictive accuracy and spatial and temporal transferability of modelsbased on indicator species can exceed that of models based on environmental variables. If mechanistic responses to environmental change are consistent through time, tracking the occurrence of a subset of an assemblageduring periods of environmental change may allow inference to species richness of the assemblage.

    U2 - 10.1016/j.ecolind.2017.09.020

    DO - 10.1016/j.ecolind.2017.09.020

    M3 - Article

    VL - 84

    SP - 470

    EP - 478

    JO - Ecological Indicators

    JF - Ecological Indicators

    SN - 1470-160X

    ER -