A new modelling framework for fast Bayesian inference in large occupancy data sets by Alex, Emily, Eleni and Byron just published in Biometrics.
The model accounts for spatio-temporal autocorrelation and gives robust inference on species presence, as evidence by the simulation results in the paper and the two case studies on one common and widespread and one rare species, using records from the Butterflies for the New Millennium database, producing occupancy indices spanning 45 years.
Alex Diana, Eleni Matechou, Jim E. Griffin, Yadvendradev Jhala, Qamar Qureshi
Capture-recapture (CR) data and corresponding models have been used extensively to estimate the size of wildlife populations when detection probability is less than 1. When the locations of traps or cameras used to capture or detect individuals are known, spatially-explicit CR models are used to infer the spatial pattern of the individual locations and population density. Individual locations, referred to as activity centers (ACs), are defined as the locations around which the individuals move. These ACs are typically assumed to be independent, and their spatial pattern is modeled using homogeneous Poisson processes. However, this assumption is often unrealistic, since individuals can interact with each other, either within a species or between different species. In this article, we consider a vector of point processes from the general class of interaction point processes and develop a model for CR data that can account for interactions, in particular repulsions, between and within multiple species. Interaction point processes present a challenge from an inferential perspective because of the intractability of the normalizing constant of the likelihood function, and hence standard Markov chain Monte Carlo procedures to perform Bayesian inference cannot be applied. Therefore, we adopt an inference procedure based on the Monte Carlo Metropolis Hastings algorithm, which scales well when modeling more than one species. Finally, we adopt an inference method for jointly sampling the latent ACs and the population size based on birth and death processes. This approach also allows us to adaptively tune the proposal distribution of new points, which leads to better mixing especially in the case of non-uniformly distributed traps. We apply the model to a CR data-set on leopards and tigers collected at the Corbett Tiger Reserve in India. Our findings suggest that between species repulsion is stronger than within species, while tiger population density is higher than leopard population density at the park.
The paper, titled “Capture-recapture models with heterogeneous temporary emigration”, first published online on the 14th of September 2022, develops a novel modelling framework for capture-recapture data without relying on the assumption of permanent emigration. The model is built within a Bayesian non-parametric & changepoint process framework, and is demonstrated on data on salmon anglers in Norway.
The paper, A Generic Method for Estimating and Smoothing Multispecies Biodiversity Indicators Using Intermittent Data, by Stephen Freeman, Nicholas Isaac, Panagiotis Besbeas, Emily Dennis and Byron Morgan has just been published in the Journal of Agricultural, Biological and Environmental Statistics
https://doi.org/10.1007/s13253-020-00410-6
Abstract
Biodiversity indicators summarise extensive, complex ecological data sets and are important in influencing government policy. Component data consist of time-varying indices for each of a number of different species. However, current biodiversity indicators suffer from multiple statistical shortcomings. We describe a state-space formulation for new multispecies biodiversity indicators, based on rates of change in the abundance or occupancy probability of the contributing individual species. The formulation is flexible and applicable to different taxa. It possesses several advantages, including the ability to accommodate the sporadic unavailability of data, incorporate variation in the estimation precision of the individual species’ indices when appropriate, and allow the direct incorporation of smoothing over time. Furthermore, model fitting is straightforward in Bayesian and classical implementations, the latter adopting either efficient Hidden Markov modelling or the Kalman filter. Conveniently, the same algorithms can be adopted for cases based on abundance or occupancy data—only the subsequent interpretation differs. The procedure removes the need for bootstrapping which can be prohibitive. We recommend which of two alternatives to use when taxa are fully or partially sampled. The performance of the new approach is demonstrated on simulated data, and through application to three diverse national UK data sets on butterflies, bats and dragonflies. We see that uncritical incorporation of index standard errors should be avoided.
The paper Size‐ and stage‐dependence in cause‐specific mortality of migratory brown trout by Chloé R. Nater, Yngvild Vindenes, Per Aass, Diana Cole, Øystein Langangen, S. Jannicke Moe, Atle Rustadbakken, Daniel Turek, Leif Asbjørn Vøllestad and Torbjørn Ergon was published in Journal of Animal Ecology.
The paper, by Alex Diana, Eleni Matechou, Jim Griffin, and Alison Johnston, defines the Hierarchical Dependent Dirichlet process for modelling phenological patterns using multi-season multi-site capture-recapture data.
The paper, titled “A Polya Tree based model for unmarked individuals in an open wildlife population”, has been accepted for the conference “Bayesian Statistics: New Challenges and New Generations – BAYSM 2018”.
This is Alex’s first paper, well done Alex!
The work presents novel dynamic mixture models for the monitoring of bumblebee populations on an unprecedented geographical scale, motivated by the UK citizen science BeeWalk.
The models allow us for the first time to estimate bumblebee phenology and within-season productivity, defined as the number of individuals in each caste per colony in the population in that year, from citizen science data.
All of these parameters are estimated separately for each caste, giving a means of considerable ecological detail in examining temporal changes in the complex life cycle of a social insect in the wild. Due to the dynamic nature of the models, we are able to produce population trends for a number of UK bumblebee species using the available time-series. Via an additional simulation exercise, we show the extent to which useful information will increase if the survey continues, and expands in scale, as expected.
Bumblebees are extraordinarily important components of the ecosystem, providing pollination services of vast economic impact and functioning as indicator species for changes in climate or land use. Our results demonstrate the changes in both phenology and productivity between years and provide an invaluable tool for monitoring bumblebee populations, many of which are in decline, in the UK and around the world.
The review paper, by Dormann, C.F., Calabrese, J.M., Gurutzeta, G., Matechou, E., Bahn, V., Bartoń, K., et al. to appear in Ecological Monographs, explores different model averaging techniques in terms of ways to calculate the model weights and to combine predictions from different models. The advantages and disadvantages of model averaging are discussed and code for methods falling under three categories (Bayesian, information theoretical and tactical) is provided.
Read a blog post written by the main contributors of the paper here.
Anita and Rachel have had a paper accepted in Journal of Agricultural Biological and Environmental Statistics.
A Test of Positive Association for Detecting Heterogeneity in Capture for Capture–Recapture Data
Anita Jeyam, Rachel S. McCrea, Thomas Bregnballe, Morten Frederiksen and Roger Pradel
The Cormack–Jolly–Seber (CJS) model assumes that all marked animals have equal recapture probabilities at each sampling occasion, but heterogeneity in capture often occurs and should be taken into account to avoid biases in parameter estimates. Although diagnostic tests are generally used to detect trap-dependence or transience and assess the overall fit of the model, heterogeneity in capture is not routinely tested for. In order to detect and identify this phenomenon in a CJS framework, we propose a test of positive association between previous and future encounters using Goodman–Kruskal’s gamma. This test is based solely on the raw capture histories and makes no assumption on model structure. The development of the test is motivated by a dataset of Sandwich terns (Thalasseus sandvicensis), and we use the test to formally show that they exhibit heterogeneity in capture. We use simulation to assess the performance of the test in the detection of heterogeneity in capture, compared to existing and corrected diagnostic goodness-of-fit tests, Leslie’s test of equal catchability and Carothers’ extension of the Leslie test. The test of positive association is easy to use and produces good results, demonstrating high power to detect heterogeneity in capture. We recommend using this new test prior to model fitting as the outcome will guide the model-building process and help draw more accurate biological conclusions.
The full article is available as open access and can be found here: https://link.springer.com/content/pdf/10.1007%2Fs13253-017-0315-4.pdf
