Hierarchical Models for Computing Inferences About Species Assemblages Subject to Imperfect Detection Completed
USGS is developing strategies to model species assemblages to allow inferences to be made about individual species, local communities of species, or for an entire metacommunity of species - while accounting for errors in species detection during sampling.
The Science Issue and Relevance: Community ecology is largely motivated by a desire to understand observed patterns of variation in abundance or occurrence of species. These pat- terns include variations over space and time and are evident in the contemporary definition of a metacommunity – a set of spatially distinct, local communities that are linked by dispersal of multiple, potentially interacting species. The metacommunity concept occupies a prominent role in ecology and provides a conceptual framework for describing processes involved in the formation and dynamics of species assemblages. Various metacommunity theories are based on differences in the relative contributions of dispersal of species among local communities and interactions of species within local communities. However, metacommunity theories continue to be advanced without much empirical validation. Statistical models are needed to confront these theories with data. These models also may be used to solve problems that arise in conservation biology, such as predicting spatial and temporal changes in biodiversity as a function of natural or man-made changes in the environment.
Methodology for Addressing the Issue: The focus of this research is to develop strategies for modeling assemblages of species in a way that allows inferences to be made for individual species, for local communities of species, or for an entire metacommunity of species while accounting for errors in detection of species during sampling. A hierarchical approach is adopted wherein an observation model, which accounts for the sampling process and for imperfect detection, is augmented with a model of species occurrences or abundances. The latter component is generally tailored to exploit species- or site-level covariates that are thought to be informative of occurrence or abundance and are therefore useful in making predictions.
Future Steps: Hierarchical modeling has been used to analyze the dynamics of a metacommunity of butterfly species, largely as proof of concept. We are currently attempting to add auxiliary information about individual species, such as guild identity or trophic level, to models as a way of incorporating the effects of interactions among species. We are also attempting to make these models accessible to ecologists by providing software.
Related Project(s): Hierarchical Models for Estimating Abundance and Occurrence of Individual Species
USGS is developing strategies to model species assemblages to allow inferences to be made about individual species, local communities of species, or for an entire metacommunity of species - while accounting for errors in species detection during sampling.
The Science Issue and Relevance: Community ecology is largely motivated by a desire to understand observed patterns of variation in abundance or occurrence of species. These pat- terns include variations over space and time and are evident in the contemporary definition of a metacommunity – a set of spatially distinct, local communities that are linked by dispersal of multiple, potentially interacting species. The metacommunity concept occupies a prominent role in ecology and provides a conceptual framework for describing processes involved in the formation and dynamics of species assemblages. Various metacommunity theories are based on differences in the relative contributions of dispersal of species among local communities and interactions of species within local communities. However, metacommunity theories continue to be advanced without much empirical validation. Statistical models are needed to confront these theories with data. These models also may be used to solve problems that arise in conservation biology, such as predicting spatial and temporal changes in biodiversity as a function of natural or man-made changes in the environment.
Methodology for Addressing the Issue: The focus of this research is to develop strategies for modeling assemblages of species in a way that allows inferences to be made for individual species, for local communities of species, or for an entire metacommunity of species while accounting for errors in detection of species during sampling. A hierarchical approach is adopted wherein an observation model, which accounts for the sampling process and for imperfect detection, is augmented with a model of species occurrences or abundances. The latter component is generally tailored to exploit species- or site-level covariates that are thought to be informative of occurrence or abundance and are therefore useful in making predictions.
Future Steps: Hierarchical modeling has been used to analyze the dynamics of a metacommunity of butterfly species, largely as proof of concept. We are currently attempting to add auxiliary information about individual species, such as guild identity or trophic level, to models as a way of incorporating the effects of interactions among species. We are also attempting to make these models accessible to ecologists by providing software.
Related Project(s): Hierarchical Models for Estimating Abundance and Occurrence of Individual Species