Multiscale occupancy modeling of environmental DNA using the R package EDNAOCCUPANCY
Presence-absence surveys are commonly used to estimate the spatial distribution of a species. However, during a field survey some organisms, especially rare or elusive species, can be missed; increasing the chance of error in the data collected. To combat the potential error, researchers will regularly visit sites and collect more data that will later be analyzed using occupancy models. Occupancy models are used by researchers to estimate the true occupancy of a species and accounts for imperfect detection of organisms in a study. Typically, occupancy models are two levels, however, when using environmental DNA (eDNA) due to the sample processing methods a third level is added to the occupancy model. This project developed an R package to fit three level models.
R package creation supports cutting edge science
Sources of eDNA include skin cells, mucus, eggs, urine, and feces that are shed from species within an ecosystem. Surveys using eDNA are designed to accommodate both spatial and temporal differences at each of the sampling locations. This is essential since each sample taken at this location may not contain eDNA from every organism located in that area. The amount of eDNA in each sample is depended on a few factors, including the source of eDNA, degradation or transport of eDNA, and the size of the sample.
After the samples are collected, the presence of eDNA is assessed by amplifying the eDNA in each of many independent subsamples using polymerase chain reaction (PCR) chemistry. Meaning all eDNA surveys include at least three nested sampling levels consisting of:
- location within study area,
- eDNA samples collected from each location, and
- subsamples taken from each eDNA sample.
Current modeling methods using R while working with eDNA samples have some faults including:
- they are focused on presence or absence of a specific species,
- the models are not able to be extrapolated out to unsampled locations,
- there are a few formulation errors (e.g., incorrect algorithm), and
- the models were limited to studies without covariates.
Due to these faults, researchers developed the R package EDNAOCCUPANCY to provide a way to complete Bayesian, multiscale occupancy models with or without covariates.
Further details and methods can be found in Dorazio and Erickson 2017.
This project is complete and collaborated with the USGS Wetland and Aquatic Research Center.
References:
Dorazio, R. M., & Erickson, R. A. (2017). EDNAOCCUPANCY: An r package for multiscale occupancy modelling of environmental DNA data. Molecular Ecology Resources, 18(2), 1-13. https://doi.org/10.1111/1755-0998.12735
eDNAoccupancy: An R package for multi-scale occupancy modeling of environmental DNA data
eDNAoccupancy
Presence-absence surveys are commonly used to estimate the spatial distribution of a species. However, during a field survey some organisms, especially rare or elusive species, can be missed; increasing the chance of error in the data collected. To combat the potential error, researchers will regularly visit sites and collect more data that will later be analyzed using occupancy models. Occupancy models are used by researchers to estimate the true occupancy of a species and accounts for imperfect detection of organisms in a study. Typically, occupancy models are two levels, however, when using environmental DNA (eDNA) due to the sample processing methods a third level is added to the occupancy model. This project developed an R package to fit three level models.
R package creation supports cutting edge science
Sources of eDNA include skin cells, mucus, eggs, urine, and feces that are shed from species within an ecosystem. Surveys using eDNA are designed to accommodate both spatial and temporal differences at each of the sampling locations. This is essential since each sample taken at this location may not contain eDNA from every organism located in that area. The amount of eDNA in each sample is depended on a few factors, including the source of eDNA, degradation or transport of eDNA, and the size of the sample.
After the samples are collected, the presence of eDNA is assessed by amplifying the eDNA in each of many independent subsamples using polymerase chain reaction (PCR) chemistry. Meaning all eDNA surveys include at least three nested sampling levels consisting of:
- location within study area,
- eDNA samples collected from each location, and
- subsamples taken from each eDNA sample.
Current modeling methods using R while working with eDNA samples have some faults including:
- they are focused on presence or absence of a specific species,
- the models are not able to be extrapolated out to unsampled locations,
- there are a few formulation errors (e.g., incorrect algorithm), and
- the models were limited to studies without covariates.
Due to these faults, researchers developed the R package EDNAOCCUPANCY to provide a way to complete Bayesian, multiscale occupancy models with or without covariates.
Further details and methods can be found in Dorazio and Erickson 2017.
This project is complete and collaborated with the USGS Wetland and Aquatic Research Center.
References:
Dorazio, R. M., & Erickson, R. A. (2017). EDNAOCCUPANCY: An r package for multiscale occupancy modelling of environmental DNA data. Molecular Ecology Resources, 18(2), 1-13. https://doi.org/10.1111/1755-0998.12735