arXiv subjects

Robert B. O'Hara

Publications and source records attributed to Robert B. O'Hara.

4 recordsLinked to original sources

arXiv2026

Unbiased likelihood estimation of the Langevin diffusion for animal movement modelling

An ongoing challenge in animal ecology is developing movement models that account for the autocorrelation, and often temporal irregularity, in telemetry data. Continuous-time Langevin diffusion models have been proposed to model temporally autocorrelated and irregularly sampled data. However, current estimation techniques obtain increasingly biased parameter estimates as the time between observations increases. In this paper, we propose using Brownian bridges in an importance sampling scheme to improve the likelihood approximation of the Langevin diffusion model. In a series of simulation studies, we showed that our approach effectively removed the bias under various scenarios. We found that the precision of the estimated habitat coefficients increased for data spanning a longer duration at a lower frequency than for shorter, more frequently sampled tracks. This suggests that the model may be well suited for modelling tracking data sampled at a coarser resolution, as is common in datasets collected with older generations of animal tags. We illustrated the application of our model using tracking data from Steller sea lions, \textit{Eumetopias jubatus}. We found that the coefficient estimates converged to values significantly different than those estimated in previous studies, suggesting that bias in conventional estimation methods may meaningfully affect ecological conclusions about habitat preference. Together, these improvements broaden the applicability of Langevin diffusion models, thereby improving ecological insight into habitat selection.

stat.ME↗

arXiv2024

A sequential Monte Carlo algorithm for data assimilation problems in ecology

1. Temporal trends in species distributions are necessary for monitoring changes in biodiversity, which aids policymakers and conservationists in making informed decisions. Dynamic species distribution models are often fitted to ecological time series data using Markov Chain Monte Carlo algorithms to produce these temporal trends. However, the fitted models can be time-consuming to produce and run, making it inefficient to refit them as new observations become available. 2. We propose an algorithm that updates model parameters and the latent state distribution (e.g. true occupancy) using the saved information from a previously fitted model. This algorithm capitalises on the strength of importance sampling to generate new posterior samples of interest by updating the model output. The algorithm was validated with simulation studies on linear Gaussian state space models and occupancy models, and we applied the framework to Crested Tits in Switzerland and Yellow Meadow Ants in the UK. 3. We found that models updated with the proposed algorithm captured the true model parameters and latent state values as good as the models refitted to the expanded dataset. Moreover, the updated models were much faster to run and preserved the trajectory of the derived quantities. 4. The proposed approach serves as an alternative to conventional methods for updating state-space models (SSMs), and it is most beneficial when the fitted SSMs have a long run time. Overall, we provide a Monte Carlo algorithm to efficiently update complex models, a key issue in developing biodiversity models and indicators.

stat.AP↗

arXiv2023

The Point Process Framework for Integrated Modelling of Biodiversity Data

The quantity and types of biodiversity data being collected have increased in recent years. If we are to model and monitor biodiversity effectively, we need to respect how different data sets were collected, and effectively integrate these data types together. The framework that has emerged to do this is based on a point process formulation, with individuals as points and their distribution as a realisation of a random field. We describe this formulation and how the process model for the actual distribution is linked to the data that is collected through observation models. The observation models describe the data collection process and its uncertainties and biases. We provide an example of using these methods to model species of Norwegian freshwater fish, which shows how integrated models can be adapted to the data we can collect. We summarise the modelling issues that arise and the approaches that could be taken to solve them.

stat.AP↗

arXiv2023

Modelling heterogeneity in the classification process in multi-species distribution models can improve predictive performance

1. Species distribution models and maps from large-scale biodiversity data are necessary for conservation management. One current issue is that biodiversity data are prone to taxonomic misclassifications. Methods to account for these misclassifications in multispecies distribution models have assumed that the classification probabilities are constant throughout the study. In reality, classification probabilities are likely to vary with several covariates. Failure to account for such heterogeneity can lead to bias in parameter estimates. 2. Here we present a general multispecies distribution model that accounts for heterogeneity in the classification process. The proposed model assumes a multinomial generalised linear model for the classification confusion matrix. We compare the performance of the heterogeneous classification model to that of the homogeneous classification model by assessing how well they estimate the parameters in the model and their predictive performance on hold-out samples. We applied the model to gull data from Norway, Denmark and Finland, obtained from GBIF. 3. Our simulation study showed that accounting for heterogeneity in the classification process increased precision by 30% and reduced accuracy and recall by 6%. Applying the model framework to the gull dataset did not improve the predictive performance between the homogeneous and heterogeneous models due to the smaller misclassified sample sizes. However, when machine learning predictive scores are used as weights to inform the species distribution models about the classification process, the precision increases by 70%. 4. We recommend multiple multinomial regression to be used to model the variation in the classification process when the data contains relatively larger misclassified samples. Machine prediction scores should be used when the data contains relatively smaller misclassified samples.

stat.AP↗