Workshops

Pre-Conference Workshops

* Please note that day-long workshops include lunch; and all workshop participants will have access to light refreshments on break times. Registration check-in area and food service will be in the lobby of Johnson Hall.

Some workshops are in Mary Gates Hall (MGH), while others are in Johnson Hall (JHN). These 2 buildings are next to each other.


Sunday 26 June

Workshop A:  Spatially-explicit capture-recapture concepts and methods

Instructors: David Borchers, Ben Stevenson
0900-1700, Sunday
$100.00
Location: MGH 234

Spatially explicit capture-recapture (SECR) methods have come a long way in the last decade. Not only do they provide a powerful means of estimating density and abundance, they now also allow researchers to model species distribution and animal movement in heterogeneous environments, and to model changes over time. They accommodate binary capture data, count frequency data and time-of-detection data, detection by physical trapping or remote device (camera traps for example), or no detection of animals themselves at all, but detection of evidence of their presence (via hair snares, dung surveys or acoustic surveys, for example). There are also methods for situations in which not all individuals are identifiable.

This workshop will introduce participants to SECR methods ranging from the most basic to the most recently developed. Online material will be provided in advance to help those new to SECR to develop an understanding of the basics prior to the workshop, so that there is time in the workshop to cover more recently-developed SECR methods, and time for participants to conduct some example analyses in the workshop. (Some prior knowledge of R is desirable for this.)

Participants should bring their own laptop computer.

Enquiries can be addressed to David Borchers at dlb@st-andrews.ac.uk

Workshop B:  Bayesian Decision Theory and Model Selection

Instructors: Mevin Hooten, Perry Williams, and Trevor Hefley
0900-1700, Sunday
$70.00
Location: MGH 251

Decision theory is a fundamentally important aspect to both statistical inference and decision making in practice.  Similarly, model selection has proven to be important in ecological science.  Bayesian methods have become popular for providing statistical inference in all aspects of ecology, but they have been particularly useful in complicated data settings as well as spatial and dynamic ecological process modeling.  In this workshop, we will demonstrate how both decision theory and model selection have formal connections to Bayesian statistics.  We will briefly review the fundamentals of Bayesian statistics, then introduce the concept of decision theory, and finally cover the myriad of approaches that can be used to perform model selection in the Bayesian setting.  These topics will be covered mainly in lecture format, while demonstrating ecological examples and computer code using R and JAGS.  Participants will achieve the greatest learning if they have a familiarity with statistical modeling and Bayesian methods in general, as well as basic R programming background.

 

Monday 27 June

Workshop C:  Analysis of occupancy data using hidden Markov models and E-SURGE

Instructors: Olivier Gimenez, Rémi Choquet and Roger Pradel
0900-1700, Monday
$100
Location: JHN 111

Occupancy models allow estimating and inferring about species occurrence while accounting for false absences (or imperfect species detection). The aim of this 1-day workshop is to demonstrate how occupancy models can be formulated as hidden Markov models (HMM), then implemented in program E-SURGE.

In brief, HMM operates by distinguishing i) the state process that captures the Markovian dynamic of the actual but latent states (e.g., sites being occupied or not), from ii) the observation process that consists of observations (e.g., detection/non-detection data) made from these underlying states. E-SURGE makes the building of complex models easy based on a language describing the structure of models in a compact and user-friendly way. Besides, E-SURGE uses advanced numerical algorithms that produce fast and reliable results. The HMM approach is highly versatile and, whenever coupled with E-SURGE, allows building almost any occupancy model within a single framework.

The content of this workshop is aimed at providing the participants with some theory and practice of the analysis of detection/non-detection data using HMM. The workshop will be a combination of lectures and computer lab exercises with program E-SURGE. After a brief introduction to E-SURGE and HMM, we will present single-season models, possibly incorporating heterogeneity via finite mixtures, accommodating with multiple states (breeding/non-breeding, low/high population size) and dealing with multiple species (coexistence, predator-prey interactions). We will also consider multiple-season (dynamic) models, possibly incorporating the effect of site-level covariates and incorporating heterogeneity via random effects. Depending on the attendees’ preferences, we will consider other models (e.g., occupancy models with species misidentification).

If you have any question, contact Olivier (olivier.gimenez@cefe.cnrs.fr). Further information is available in “Gimenez et al (2014), Fitting occupancy models with E-SURGE: hidden Markov modelling of presence–absence data. Methods in Ecology and Evolution, 5: 592–597″ (available from http://goo.gl/oTv4qw) and associated Wiki http://occupancyinesurge.wikidot.com/).

Workshop D:  Flexible programming with BUGS models: Using NIMBLE for MCMC and beyond

Instructor: Perry de Valpine
1300-1700, Monday
(Note recent start time change. Lunch available 11:45 for those who pre-ordered.)
$50
JHN 175

This workshop will provide a comprehensive introduction to the NIMBLE package (R-nimble.org). NIMBLE adopts and extends the BUGS language for writing statistical models but provides a new implementation in R via compiled C++.  NIMBLE also provides a system for writing programs to use BUGS models in arbitrary ways, making it easy to write new algorithms for general hierarchical models. This makes it possible to customize the choice of MCMC samplers for a particular model, write new MCMC samplers, simulate from BUGS models, write sequential Monte Carlo (particle filter) methods, combine methods in new ways, extend the BUGS language with new distributions and functions, and (by the time of this workshop) use derivatives. The NIMBLE model programming language is an enhanced subset of R that is compiled via C++, so it is easy for R users to pick up. In this workshop we will cover how to use BUGS models for basic MCMC and particle filters, how to customize these, and how to write new functions to implement your own methods. Prior to the workshop we will poll participants about particular application areas of interest, such as capture-recapture, occupancy, data augmentation, spatial models, state-space models, integrated population models, nonparametric models, or others.

Workshop E:  Spatial Statistical Models for Stream Networks

Instructors: Jay Ver Hoef, Erin Peterson
0900-1700, Monday
$150
MGH 251

We recently developed some new spatial statistical models for data on stream networks (SSNMs) and this course covers the basic theory, including parameter estimation with proper uncertainty bounds, mapping, and designing monitoring programs. We also demonstrate their use with free analytical software (STARS ArcGIS toolset; SSN package for R Statistical Software), including model fitting, simulation and visualization. SSNMs account for network topology (i.e., flow direction, stream size, tributary confluences) and offer significant improvements over many traditional statistical techniques (e.g., regression, mapping) that were developed originally for terrestrial applications. The SSNMs develop novel spatial autocorrelation formulations among observations and are applicable to common types of stream data (e.g., water quality attributes, biological surveys, habitat conditions) using parametric distributions (e.g., Gaussian, binomial, Poisson), including generalized linear models. General classroom discussions of spatial statistical methods are included.


Please note:  ISEC 2016 Conference Registration fees do not include any workshops. Also, there are a limited number of seats in each workshop.

Link to registration for excursions, workshops and/or the main ISEC2016 conference.  People may sign up for Workshops Only, but excursions are available only when choosing to register for the full conference.