[R-pkgs] PVAClone: new package for population viability analysis

[Peter Solymos]

Dear UseRs!

We are pleased to announce the release of our new package 'PVAClone'.

The 'PVAClone' package implements Population Viability Analysis (PVA)
methodology using data cloning. The data cloning algorithm by Lele et
al. (2007, 2010) is employed to compute maximum likelihood estimates
of the state-space model parameters and the corresponding standard
errors, heavily capitalizing on JAGS, dclone and dcme packages (see
Solymos 2010).

The main components of the package include estimation of univariate
population growth models, model selection and extinction risk
estimation:

* Model Estimation *
- computes maximum likelihood Estimation of the univariate population
growth models, both in the presence or absence of observation error;
- population time series with missing observations are also accommodated;
- population growth models: Gompertz, Ricker, theta-logistic and the
generalized Beverton-Holt;
- models can also be fitted by fixing a subset of model parameters to
a priori values of interest;
- observation error is incorporated via the general state-space
modeling framework.

* Model Selection *
- we implement Ponciano et. al.'s (2009) data cloned likelihood ratio
algorithm (DCLR) to compute likelihood ratios for comparing
hierarchical (random effect) models;
- this feature allows comparison of any two nested or non-nested
state-space models fitted using the Model Estimation procedure above
- for instance one can compare the state-space Generalized
Beverton-Holt model with a logistic model, even when observations are
missing;
- the underlying function is pva.llr can also be called repeatedly to
compute profile likelihood of a parameter of interest.

* Extinction Risk Estimation (under development) *
- data cloning based frequentist prediction of latent variables in a
general hierarchical model (Lele et al. 2010) is used to forecast
future abundance time series;
- a large number of future population trajectories are generated under
the observed data and estimated model parameters;
- these are then used to estimate various extinction metrics (see
Nadeem and Lele 2012).

Feedback, bug reports and feature requests are welcome!

Khurram and Peter

--
Khurram Nadeem
knadeem at math.ualberta.ca
University of Alberta

Péter Sólymos
solymos at ualberta.ca
University of Alberta


References

Nadeem, K. & Lele, S.R. 2012. Likelihood based population viability
analysis in the presence of observation error. Oikos, early online

Ponciano, J. M. et al. 2009. Hierarchical models in ecology:
confidence intervals, hypothesis testing, and model selection using
data cloning. Ecology 90: 356--362.

Lele, S. R. et al. 2007. Data cloning: easy maximum likelihood
estimation for complex ecological models using Bayesian Markov chain
Monte Carlo methods. Ecol. Lett. 10: 551--563.

Lele, S. R. et al. 2010. Estimability and Likelihood Inference for
Generalized Linear Mixed Models Using Data Cloning. J. Am. Stat.
Assoc. 105: 1617--1625.

Solymos, P. (2010): dclone: Data Cloning in R. The R Journal, 2(2): 29--37.