At the last ISEC, in Montpellier in 2014, an informal survey suggested that Methods in Ecology and Evolution was the most cited journal in talks. This reflects the importance of statistical methods in ecology and it is one reason for the success of the journal. For this year’s International Statistcal Ecology Conference in Seattle we have produced a virtual issue that presents some of our best recent papers which cross the divide between statistics and ecology. They range over most of the topics covered at ISEC, from statistical theory to abundance estimation and distance sampling.
We hope that Methods in Ecology and Evolution will be equally well represented in talks in Seattle, and also – just as in Montpellier – some of the work presented will find its way into the pages of the journal in the future.
Without further ado though, here is a brief overview of the articles in our Statistical Ecology Virtual Issue:
Streamlining the Analysis of Data
The first article in our Virtual Issue is taken from our 5th Anniversary Special Feature. In this article, Zuur and Ieno present a 10-step protocol to streamline analysis of data that will enhance understanding of the data, the statistical models and the results, and optimise communication with the reader with respect to both the procedure and the outcomes. Following this protocol will reduce the organisation, analysis and presentation of what may be an overwhelming information avalanche into sequential and, more to the point, manageable, steps.
AIC vs AICC vs BIC
Model selection is difficult. Even in the apparently straightforward case of choosing between standard linear regression models, there does not yet appear to be consensus in the statistical ecology literature as to the right approach. Brewer et al. review recent works on model selection in ecology and subsequently focus on one aspect in particular: the use of the Akaike Information Criterion (AIC) or its small-sample equivalent AICC in our second Virtual Issue article.They find that the relative predictive performance of model selection by different information criteria is heavily dependent on the degree of unobserved heterogeneity between data sets.
Likelihoods for Geolocation
If you keep up-to-date with our Early View articles, you will have seen this paper recently. Basson et al. develop an approach to light-based geolocation that provides the flexibility to integrate movement and behaviour modelling in a novel way. The likelihood surfaces from their approach can be used in any state-space model of animal movement and behaviour, irrespective of whether estimation is by maximum-likelihood or Bayesian methods. They show an example track from a grid-based hidden Markov model applied to light data from a tag deployed on a southern bluefin tuna.
Bayesian Inference
‘Exact Bayesian inference for animal movement in continuous time‘ by Blackwell et al. is the next paper featured. It describes a novel methodology which allows exact Bayesian statistical analysis for a rich class of movement models with behavioural switching in continuous time, without any need for time discretization error. These methods allow exact fitting of realistically complex movement models, incorporating environmental information. They also provide an essential point of reference for evaluating other existing and future approximate methods for continuous-time inference.
Bayesian Movement Model
Regular readers of the blog may remember the next article from a post earlier this year. In ‘A functional model for characterizing long-distance movement behaviour‘ Buderman et al. present a Bayesian movement model that accounts for error from multiple data sources as well as movement behaviour at different temporal scales. The authors apply this model to data from Colorado Canada lynx (Lynx canadensis) and use derived quantities to identify changes in movement behaviour.
Estimating Abundance
N-mixture models have become a popular method for estimating abundance of free-ranging animals that are not marked or identified individually. However, they can uncertainties which limit their usefulness. Chambert et al. introduce a new extension of N-mixture models that accounts for species uncertainty that can be applied to a wide variety of studies and taxa. It should notably help improve investigation of abundance and vital rate characteristics of organisms’ early life stages, which are sometimes more difficult to identify than adults.
Statistical Modelling
In this year’s May issue we published an article from Golding and Purse on Species Distribution Models and Gaussian Processes. The authors propose fitting Gaussian Process (GP) Species Distribution Models using deterministic numerical approximations, rather than Markov chain Monte Carlo methods in order to make GPs more computationally efficient and easy to use. This convenient method for incorporating prior knowledge of the species’ ecology also has strong predictive power. An R package, GRaF, is provided to enable SDM users to fit GP models.
Predictive Accuracy and Variable Selection
Komori et al. propose an asymmetric logistic regression model that uses a new parameter to account for data complexity in our next article. Simulation studies suggest that this approach outperforms a traditional approach in terms of both predictive accuracy and variable selection. It can enhance the applicability of a generalized linear model to various ecological problems using a slight modification, and significantly improves model fitting and model selection.
Optimising Survey Effort
The second last research article in the Statistical Ecology Virtual Issue comes from Moore et al. The authors integrate survey optimisation approaches to construct a model that optimises the allocation of surveillance effort over both space and time. They illustrate their approach by finding the optimal allocation of survey effort over space and time that maximises the expected number of detections of the cascade tree frog (Litoria pearsoniana) in a region.
Latent Variable Models
Last, but certainly not least, we have ‘Using latent variable models to identify large networks of species-to-species associations at different spatial scales‘ by Ovaskainen et al. In this article the authors present a hierarchical latent variable model that partitions variation in species occurrences and co-occurrences simultaneously at multiple spatial scales. Using recent progress in Bayesian latent variable models they implement a computationally effective algorithm that allows you to consider large communities and extensive sampling schemes. You can watch an interview with the author on our YouTube channel.
Variable Importance
The penultimate article in this Virtual Issue is a forum on quantifying variable importance by Giam and Olden. The authors re-evaulated the validity of the sum of Akaike weights (SW) reported in an earlier Methods article by repeating the experiment with a more appropriate benchmark. Despite recent criticisms, their results show that SW is a valid relative variable importance metric.
BORAL
As with many of our issues, we complete our Statistical Ecology Virtual Issue with an Applications article. This applications paper comes from Francis Hui and describes boral. boral is an R package (available on CRAN) for model-based analysis of multivariate abundance data, with estimation performed using Bayesian Markov chain Monte Carlo methods. A key feature of the package is the ability to incorporate latent variables as a parsimonious method of modelling between species correlation.
All of the articles in this Virtual Issue are free for a limited time. You can find the full Virtual Issue on our website.
Methods Blog 