Capture-recapture surveys are widely used to study survival rates and population trends in animal populations. They consist of a series of occasions, on which animals are captured and released in the population. Individuals captured for the first time are marked with a unique identification number, while recaptured individuals have their identifier number recorded. Oftentimes, a single population is studied using additional types of surveys, e.g. visual count surveys, carcass recoveries or telemetry surveys. The integration of all the data in a single statistical analysis is a very exciting mechanism to fully exploit the potential of the data. This is typically achieved by approximating the full joint likelihood of the data as a product of likelihoods.

In this talk I will introduce a new Bayesian modeling approach which integrates capture-recapture data with other sources of data in a fully explicit manner. I will focus on the simplest case of integrating capture-recapture data along with count data and compare the performance of the new approach with the usual approach (product of likelihoods) using both a simulation study and a real world application. If time permits I will present a more complex use of the method which integrates capture-recapture data, dead recovery data and snorkel survey data to study the movement of Chinook salmon (Oncorhynchus tshawytscha) from the ocean to spawning grounds on the West Coast of Vancouver Island, Canada.

Wolfgang Wefelmeyer

Discrete-time real-valued Markov chains are sometimes observed at certain time points only. The unobserved time points may be deterministic and periodic, or the gap length may be random and may also depend on the last observed state. This is no problem for estimating the one-dimensional marginal distribution. However, if we want to estimate joint or conditional distributions of the chain, in particular the transition distribution, can we exploit the information in pairs of observations separated by an unobserved gap? We discuss when and how this is possible in nonparametric and in autoregressive models. We point out similarities to mixture models and to regression models with observations missing at random.

In community ecology, building model is often a complex task for a few reasons. First, the multivariate nature of community data is technically challenging to handle, resulting in difficulties in making inferences and predictions. Also, obtaining reliable inferences when constructing species-specific models is a difficult task because most species in a community are rare. Lastly, to better understand the complexity of nature, ecologists are using an increasing diversity of data (e.g. habitat characteristics or species traits); linking these different data types in an ecologically meaningful way require technical developments beyond that of traditional statistics. In this presentation, I will present a flexible and comprehensive statistical framework that can be used to model species association by estimating the positive and negative correlations among species within a specious community and that also accounts for species traits and habitat characteristics (both as fixed and random effects). Through this framework it is also possible to make species- and community-level predictions. This framework relies on a Bayesian hierarchical modelling. I will illustrate the potential of this modelling approach by applying it to fisheries stock data gathered from 1950 to 2010 in the Gulf of Alaska Large Marine Ecosystem, where for each species traits information were gathered with FishBase and SeaLifeBase. As this is a 60 years time series, I will use asymmetric eigenvector maps (an eigenfunction-based method developed to model the effect of directional processes) to account for temporal autocorrelation. With these data, I will show how this statistical framework can be used to approach different ecological questions on marine harvesting in a community ecology context.

Fisheries science has traditionally concerned itself with the interplay of fish population abundance, fishing effort, and fishery catch.  However, fisheries managers must increasingly cope with changes over time in fish productivity (e.g., changing individual growth and juvenile survival rates).  One hypothesis for changing productivity is that interactions among species will be modified by climate change and fishing impacts, and that changes in these interactions cause the parameters of single-species models to be nonstationary.  Estimating community dynamics and species interactions has historically been difficult using time-series data.  However, recent research suggests that spatio-temporal analyses have greater statistical efficiency than previous time-series approaches because they use spatial variation as a form of replication. 
 
In this talk, I discuss ongoing collaborations to estimate community dynamics and interactions using multivariate spatio-temporal point process models.  I start with a global meta-analysis of a classic hypothesis for nonstationary catch rates, i.e., that fish populations collapse to a core habitat during declines in population size.  Using bottom trawl data for 120 populations worldwide, colleagues and I estimate a 0.6% decrease in “effective area” for every 10% decline in abundance, but also show that this relationship varies widely among populations and regions.  I then use “spatial dynamic factor analysis” to summarize community dynamics for the Eastern Bering Sea.  This case study captures the decline and recovery of cod-like species in the mid-2000s, and shows that species with similar evolutionary history have more similar dynamics than unrelated species.  Finally, colleagues and I propose a new procedure for estimating the matrix of pairwise species interactions, where this approach bridges between unregulated (“neutral”) and highly-regulated (“niche”) approaches to community ecology.  Using the marine community in the Gulf of St. Lawrence as case study, we show a mixture of regulated and unregulated dynamics, where the unregulated component is associated with a recovering grey seal population that is negatively impacting productivity for three prey species of fish. 
 
I conclude by outlining opportunities for future research in statistical ecology.  Throughout, I stress that continuing progress will likely combine methodological improvements (e.g., Riemann MCMC) with increased biological realism in models (e.g., advective-diffusive movement in community models).  Given the increasing role of statistics in ecological theory (e.g., neutral and maximum entropy theories), I hypothesize that this two-pronged approach will yield improvements in both the theory and practice of fisheries science.
 

Predicting the impacts of environmental change on species requires a mechanistic understanding of biological processes such as foraging, migration, and reproduction. However, the continuous behavioural data needed to assess how these processes change through time is often impossible to gather, particularly for Arctic and marine species. Thus, ecologists increasingly rely on animal telemetry to monitor activity patterns. In this talk, I will demonstrate how emerging statistical methods and movement data can be used to model the behaviour of a range of species (e.g. polar bear, rhinoceros auklet), and discuss how the information provided by movement models can help us answer fundamental ecological questions and solve conservation problems.

The occupancy model has become increasingly popular in ecology as a means to account for imperfect detection of a species when predicting where it is likely to occur. The dynamic, multi-season occupancy model extends the framework to account for open populations with occupancies that change over time through local colonizations and extinctions. However, few versions of the model relate these probabilities to the occupancies of neighboring sites or patches. I will present a version that does incorporate this information, where a site is more likely to be colonized if more of its neighbors were previously occupied and if it provides more appealing environmental characteristics than its neighboring sites. Additionally, a site without occupied neighbors may become colonized through long-distance dispersal. In my presentation, I will describe the concept and mathematics of the occupancy model, how we incorporate the spatial and temporal processes, and use the model to obtain inference for the ongoing Common Myna (Acridotheres tristis) invasion in South Africa. The results suggest that the Common Myna continues to enlarge its distribution and its spread via short-distance movement, rather than long-distance dispersal. Overall, the new modeling framework provides a powerful tool for managers examining the drivers of colonization, including short- vs. long-distance dispersal, habitat quality, and distance from source populations.

In a three part review of my latest research I describe new developments and findings in (1) landscape-scale social ecological systems modelling for a BC recreational fishery, (2) statistically rigorous approaches for accounting for uncertainty from data processing in population analyses and (3) using simulation to establish robust management for Atlantic bluefin tuna and data-limited global fisheries.

The Metropolis-Hastings algorithm is often used to obtain Markov Chian Monte Carlo samples from highly complex posterior distributions, such as those involved in full inference of hyper-parameters in Gaussian Process models. Here we compare one new and three existing implementations of the Metropolis-Hastings algorithm in the context of sampling from the posterior distributions of hyper-parameters in a Gaussian Process. The implementations are compared in terms of their initialization biases and convergence rates, as well as in terms of their performance on higher dimensional data. Our experiments involve sampling from GP posterior distributions using the four different implementations and comparing the quality of these samples. A discrepancy measure is devised based on the Kolmogorov-Smirnov test to measure the convergence rate of each algorithm. Issues in generating MCMC samples for high dimensional data are discussed and an optimal approach to sampling is proposed called the Laplace approach. All of the algorithms in this paper are implemented in an R package called ‘gpMCMC’.

This presentation will describe advances in identifying anomalies in datasets of internet of things (IoT) devices found in commercial buildings. Leveraging properties specific to machine­-to­-machine communications, data models and algorithms can be constructed from the time series datasets to identify security threats (i.e. cyber attacks) and failing devices (predictive maintenance). Devices modeled include security cameras, lighting and heating control, and various sensors and actuators. 

Big data techniques can also be applied to the communication datasets, including various forms of clustering and fuzzy logic/fuzzy inferencing to identify deviations from real­-world normal operations. In such situations, modularity is key to proper implementation in order to allow for proper data processing, algorithm training and algorithm testing. 

Results of current research and future roadmap will be discussed.

About Optigo Networks

Optigo Networks is shaping the future of the commercial Internet of Things (IoT) by redefining how smart buildings are connected and operated. By applying visualization and anomaly detection to the building system, Optigo allows the IoT to scale, driving down the cost to maintain and operate the technologies that make buildings comfortable and efficient.

With its award-­winning software, Optigo Networks allows building operators to quickly identify faults and security threats in the building system, cutting troubleshooting time down from hours to minutes. Built­in analytics rein in the building IoT, reducing OpEx and maintenance costs with tools and reports to visualize the health and security of the building network.