Pre-Conference Workshop

Conveners / Organizers:
Jake Lawlor¹, Jennifer Sunday¹, Nikki Moore¹ and Lise Comte²

¹ Department of Biology, McGill University, Montreal, Canada
² Conservation Science Partners, Seattle, USA

The BioShifts database (Comte 2020) is the largest collection of empirical observations of species’ range shifts to date – totaling over 30,000 shift estimates from published literature – and has already been used in several global-scale studies. Since 2020, the BioShifts team has expanded the database, supplementing the initial version with even more shift observations and several derived variables associated with each detected shift that could help to explain the wide variability in estimated range shift rates and directions. In this workshop, we present the “new and improved” version of the BioShifts database, and the BioShiftR R package, which allows for simple access and manipulation of BioShifts data for downstream analyses. Participants in this workshop will be introduced to the database and associated variables (methodological, taxonomic, and environmental), as well as package functions for accessing and subsetting data

Then, participants will separate into groups to hypothesize mechanistic acolnd methodological drivers of observed range shift rates across taxa and locations, and use the R package and any additional databases to lay groundwork for new projects. Expected outcomes of this workshop are wider use and uptake of the tools provided by the BioShifts team, identification of pitfalls or assumptions when using data from the BioShifts database, and new collaborations for future research.

More Details

Conveners / Organizers:
David E. Uribe-Rivera¹ and Wen-Hsi Yang²

¹ CSIRO Environment, Brisbane, Australia
² CSIRO Data61, Brisbane, Australia

The rapid growth of analytical tools and data in ecology poses a key challenge: identifying optimal methods under varying data and analytical constraints. This workshop presents a structured framework to address these challenges through three core themes: modelling species density (hereafter intensity), a latent variable; accounting for observation biases; and leveraging diverse datasets to test analysis robustness. We aim to equip participants with tools and concepts to simulate, test, and benchmark ecological models for robust biodiversity forecasts under global change.

We introduce ecodynsim, an R package for simulating the spatiotemporal dynamics of species intensity and corresponding ecological observations. It enables users to assess ecological models against a known latent process with user-defined spatiotemporal autocorrelation. By generating observations with varying error, bias, and sampling effort, ecodynsim provides a controlled environment for testing model limitations. It can also simulate opportunistic presence-only (PO) data with spatial sampling bias, mimicking real-world sources such as GBIF.

Traditional single-source SDMs suffer from data-type-specific biases, such as those in PO data. We introduce a second package, RISDM, that offers a flexible framework for building integrated species distribution models (ISDMs) from multiple data types. RISDM integrates diverse datasets within a joint-likelihood framework, combining their strengths while mitigating individual weaknesses. Both packages are directly linked: RISDM is designed to integrate three of the four data types simulated by ecodynsim. Within RISDM, the true ecological process (species intensity) is modelled separately from the observation processes associated with each data type. Attendees will use these tools to fit ISDMs and other range-change models to synthetic, controlled, imperfect data—benchmarking model performance and identifying the most stable and robust approaches for ecological inference. Attendees will need basic R knowledge and being familiar with spatial data. The workshop will include time to discuss the usability of these packages for their own research.

More Details