Repository files navigation

The packageeDNAjoint is useful for interpreting observations frompaired or semi-paired environmental DNA (eDNA) and traditional surveys.The package runs a Bayesian model that integrates these two data streamsto jointly estimate parameters like the false positive probability ofeDNA detection and expected catch rate at a site. Optional modelvariations allow inclusion of site-level covariates that scale thesensitivity of eDNA sampling relative to traditional sampling, as wellas estimation of gear scaling coefficients when multiple traditionalgear types are used. Additional functions in the package facilitateinterpretation of model fits.

Check out apaper about thepackage in Methods in Ecologyand Evolution.

The most stable version ofeDNAjoint can be found on Cran:

install.packages("eDNAjoint")

You can also install the development version ofeDNAjoint fromROpenSci:

install.packages("eDNAjoint",repos="https://ropensci.r-universe.dev")

The main functionality ineDNAjoint is the use ofjoint_model() thatwill fit the model to data. Further functions likejoint_summarize()anddetection_calculate() can be used to help with model fitinterpretation.

This example fits the joint model to data from paired, replicated eDNAqPCR and seine sampling observations of endangered tidewater gobies(Eucyclogobius newberryi) from a study by Schmelzle and Kinziger(2016). The following variation of the joint model includes site-levelcovariates that scale the sensitivity of eDNA sampling relative totraditional sampling.

library(eDNAjoint)data(goby_data)# run the joint model with two covariatesgoby_fit<- joint_model(data=goby_data,cov= c("Filter_time","Salinity"),family="poisson",p10_priors= c(1,20),q=FALSE)

And then this model fit can be accessed to do things like summarize theposterior distribution for the probability of a false positivedetection,$p_{10}$:

# summarize p10 posteriorjoint_summarize(goby_fit$model,par="p10")#>      mean se_mean    sd  2.5% 97.5%    n_eff Rhat#> p10 0.003       0 0.001 0.001 0.007 13482.83    1

Or to find the number of eDNA samples and traditional survey samplesnecessary to detect presence of the species at a given expected catchrate:

# find the number of samples necessary to detect presence with 0.9 probability# at the mean covariate values, if the expected catch rate (mu) is 0.1, 0.5, or# 1 individuals/traditional survey unit.detection_calculate(goby_fit$model,mu= c(0.1,0.5,1),cov_val= c(0,0),probability=0.9)#>       mu n_traditional n_eDNA#> [1,] 0.1            24     14#> [2,] 0.5             5      4#> [3,] 1.0             3      2

You can find much more detailed examples of the functions ineDNAjointand the model underlying the package in thepackagevignette.

UsingeDNAjoint in a manuscript? Please consider citing either themanuscript about the package:

Keller, A.G., & Kelly, R.P. (2025). eDNAjoint: An R package forinterpreting paired or semi-paired environmental DNA and traditionalsurvey data in a Bayesian framework.Methods in Ecology and Evolution,00, 1–9.https://doi.org/10.1111/2041-210X.70000

Interested in contributing to this package? Seesome notes oncontributing.

Please note that eDNAjoint is released with aContributor Code ofConduct. By contributing to thisproject you agree to abide by its terms.

Keller, A.G., & Kelly, R.P. (2025). eDNAjoint: An R package forinterpreting paired or semi-paired environmental DNA and traditionalsurvey data in a Bayesian framework.Methods in Ecology and Evolution,00, 1–9.https://doi.org/10.1111/2041-210X.70000

Keller, A.G., Grason, E.W., McDonald, P.S., Ramon-Laca, A., Kelly, R.P.(2022). Tracking an invasion front with environmental DNA.EcologicalApplications. 32(4): e2561.https://doi.org/10.1002/eap.2561

Schmelzle, M.C. and Kinziger, A.P. (2016). Using occupancy modelling tocompare environmental DNA to traditional field methods forregional-scale monitoring of an endangered aquatic species.MolecularEcology Resources. 16(4): 895-908.https://doi.org/10.1111/1755-0998.12501