Here we examine the robustness of a recently developed method for studying social transmission of behaviour in groups of animals: network based diffusion analysis (NBDA). We fit an NBDA model to diffusion data derived from observations of foraging bouts in starlings (Sturnus vulgaris) given knowledge of their patterns of association, as well as to simulated data, and derive estimates of key parameters in the model. By deploying known values of key variables in simulated data we assess the robustness of the NBDA method. We employ a reversible jump Markov chain Monte Carlo (RJMCMC) algorithm to discriminate between models, and thereby determine whether the null model (with no social component) or the full model (containing both an asocial and social component) is most likely. We also show proof of principle by accurately reconstructing the original parameter values of the simulated data. Our analysis extends the current use of NBDA models to incorporate random effects and facilitate model discrimination. Furthermore, we show that NBDA models can be used to analyse diffusions and derive association patterns from common behaviours that do not involve learning. This methodology is likely to be particularly useful to deal with datasets that include many covariates and that can be fitted with a correspondingly large number of competing models.

host:

refID: 1098

hide details

I had meant to give this talk over a year ago when I returned to CREEM, so
hopefully better late than never I’ll give it now! During the talk I’ll go into
some depth on the main modelling work I did during my time in Ilkka Hanski’s
Metapopulation Research Group (MRG) in Finland. Owing to the fact that the
Glanville fritillary has been extensively studied since 1993 there exists a
large amount of data and a detailed understanding of its life history.
Consequently we were able to formulate a very detailed individual-based model
(IBM) to describe this system. During the talk I’ll describe the trials I went
through getting the IBM to fit the data and the strategy that worked in the end
including merging some ideas from particle filtering into the MCMC updating. We
also derived from the IBM a population-based stochastic patch occupancy model
(SPOM). The IBM was fit to bi-annual counts of larval groups, whereas the SPOM
was fit to a simplified version of the data (the annual presence or absence of
local populations). I’ll show comparisons of the parameter estimates, model
fits, and predictive abilities of two models of metapopulation dynamics (the
IBM and SPOM). Finally I’ll discuss the biological insights we were able to
gain from this modelling exercise. The work was published in Ecological
Monographs in 2011 with the same title as I have for the talk (but with the
American spelling for modelling).
 

host: Dr Janine Illian

refID: 1102

hide details

The UK is planning an enormous expansion of its offshore renewable energy capacity over the next 10-15 years that is likely to require an investment of more than £100 billion.  The construction and, to a lesser extent, the operation of these facilities could disturb thousands of marine mammals over periods of months and years.  It is, however, a legal offence to cause disturbance that will “impair the ability of animals to survive, breed, reproduce, or rear young” or which “significantly affects local distribution or abundance”.  The government can only issue developers with a licence to disturb if the disturbance does not affect the favourable conservation status of the species concerned.  However, quantitative assessments of the population-level effects of disturbance are currently lacking. We will describe our work on the interim PCoD project, which has been funded by the government departments that must decide whether or not to grant consent for these developments to go ahead. Its modelling framework provides assessments of the effects of disturbance at the population level, using the limited empirical data that are currently available, together with estimates of the levels of uncertainty associated with these predictions. The offshore renewable energy industry is now considering whether or not to fund a multi-million pound research programme that will fill in the gaps in our knowledge and reduce these uncertainties. 

host: Dr Janine Illian

refID: 1100

hide details

Natural and anthropogenic stressors are presumed to impact the health and survival of right whales, yet the effect of these stressors has proven difficult to quantify. To address this issue we built a Hierarchical Bayesian model for survival of individual right whales and fit this model to 30 years worth of sighting data. We assimilate the photographic evidence of visual health parameters for individuals as observations of true, but hidden, health. We use the model to make inference on movement, individual survival, and individual health. In this talk I will present and discuss estimates of health of individuals, population sub-categories, and the entire population.

Estimates of individual health fluctuated across broad ranges, with a mean "healthy" score equaling 84 (on a 0 to 100) scale. We were able to quantitatively link discrete health observations to underlying continuous states, though estimates are less certain for animals with sparse sighting histories. For individuals, discrete observations of poor skin and body condition in particular, appeared to have a strong impact on health. We explored the health trajectories of different population classes, which suggest that females who bring their calves to known habitat regions (i.e. the Bay of Fundy, Canada) fared worse in the 1990's. Since that time, however, this population class appears in better health than females who do not bring their calves to the Bay of Fundy. At the population level, health was stable throughout much of the 1980's. Health values in the 1990's decline from a population average of 85 to a low of 72 in 1999. Population health stabilizes and increases in the 2000's, though the decade scale average (76) is lower than the 1980's.

In the future we hope to use these estimates of movement, survival and health to suggest possible management scenarios that increase survival among individuals and the population as a whole.

host: Dr Janine Illian

refID: 1097

hide details

Physical knowledge about spatial phenomena often require models with non-stationary dependence structures.
Geostatistical models have traditionally been stationary, thus there has been an interest in the literature
to provide flexible and computationally efficient models and methods for non-stationary phenomena. In this
work, we demonstrate that the stochastic partial differential equation (SPDE) approach to spatial modelling
provides a flexible class of non-stationary models where explanatory variables easily can be included in the
dependence structure. We explore the proposed modelling framework using a simulation study, with focus on
distinguishing between a stationary and non-stationary model. Further, we model annual precipitation in southern
Norway using a non-stationary model with dependence structure governed by elevation, and compare this model with
a stationary model. In addition, the SPDE approach enables computationally efficient Bayesian inference with
integrated nested Laplace approximations available through the R-library INLA.

host: Dr Janine Illian

refID: 1099

hide details

Biosecurity risk management is characterised by urgent decisions, large
numbers of poorly understood threats and contentious and substantial
national and international ramifications. Intelligence is insight that can
be used to improve the quality of decisions in this context. It serves a
range of activities including risk assessment, reporting to local and
international regulators and stakeholders, operational and strategic
decision making. While some data sets are extensive, few are complete and
most are inadequate or absent entirely. Gaps are filled by expert
judgments. This presentation outlines some of the most pervasive and
manageable frailties of expert assessments in such contexts. It describes
structured protocols that have been developed and tested over the last 5
years to improve the accuracy and calibration of expert judgments of
facts. It outlines how judgments are supported by data, foresight
activities and dedicated software designed to search efficiently for
relevant open-source information. Lastly, the presentation outlines the
application of methods designed to support biosecurity risk management to
a project in geopolitical risk assessment sponsored by IARPA, the research
arm of the US Intelligence Service.

host: Dr Janine Illian

refID: 1064

hide details

A number of surveys have been conducted in recent years from small vessels in areas of harbour porpoise habitat with the aim of providing data for Environmental Impact Assessment of off-shore wind farm developments. Estimates of the numbers of animals likely to be affected and comparisons between areas require estimates of absolute abundance but these are difficult to obtain from small vessels which cannot accommodate independent observer platforms. Passive acoustics provide an automated and efficient way of detecting harbour porpoises and target motion analysis techniques can now provide estimates of the perpendicular distance to each detection. Combined visual and acoustic data have been used to set up trials to estimate g(0) for each detection method. Although results have been promising, there still remain some issues including duplicate identification that could benefit from further work.

host: Dr Janine Illian

refID: 1043

hide details

One of the purposes of the CREEM seminar series is to promote communication
within the group about research projects we’re involved in.  In March 2010 I
gave a seminar that outlined the various projects I was working on at the time.
  Since then there have been some new folks join the group.  Also, I’ve finished
a few projects (wow!), and started a few new ones.  Hence, I’ll give an update.
  Many projects will be familiar to many of you, but hopefully there will be
something new for everyone.

host: Dr Janine Illian

refID: 1025

hide details

One purpose of the CREEM seminars is to promote communication within the group about our own research. This talk will therefore introduce those umfamiliar with my work on spatial modelling to past projects I have been involved in. However, those who have known me for a while now will hear about more recent challenges and future projects.

host: Dr Janine Illian

refID: 1026

hide details

host: Prof Stephen Gillespie

refID: 1044

hide details

Many statistical ecologists have begun working with data from dtags or other marine-mammal-borne sensors, but not all have had the opportunity to go to sea and see first hand how these data are collected.  In this combination travelogue/research seminar I will try to provide an idea of what field work to tag marine mammals is really like, drawing mostly from my recent experiences as a tag technician on the SoCal Behavioural Response Study (which measures whale behaviour changes in response to simulated Naval sonar sounds).  I promise astonishing audiovisuals, insider stories, (startling?) new insight into where tag data really come from, and at least one equation.

host: Dr Janine Illian

refID: 1023

hide details

Next year is the 80th anniversary of the Loch Ness phenomena. Here I review 79 years of eyewitness testimony from Loch Ness and consider what it tells us about the reporting of anomalous phenomena.

host: Dr Janine Illian

refID: 1032

hide details

With increases in sea surface temperature and ocean acidification associated with climate change Australia's Great Barrier Reef (GBR) is likely to face increased challenges to its diversity and integrity. Globally we need to increase efforts to reduce greenhouse gas emissions but as far as Australia is concerned the most effective action that can be taken locally is to reduce the pressure on the reef from land-based pollutants that flow out of the rivers into the GBR lagoon. The primary pollutants leaving the land are nutrients (N&P), sediments and pest/herbicides. The Federal and State Governments have recently implemented new funding and policy initiatives (e.g. ReefRescue), based on by good science, to encourage the agricultural sector to change management to reduce the amount of pollutants leaving farms and protect the reef in the face of climate change.

 

host: Dr Janine Illian

refID: 1014

hide details

host: Dr Janine Illian

refID: 988

hide details

The multi-analysis engine in Distance is being created to allow issues commonly encountered in large cetacean surveys, such as the Eastern Tropical Pacific (ETP) surveys, to be accounted for within Distance. Such issues include unidentified sightings, model uncertainty and covariate uncertainty. I will also discuss the problems of incorporating unidentified sightings in the analyses in more detail. I will introduce some data that was collected during the ETP surveys to try to establish how difficult different species are to identify and will describe a potentially useful method to improve the allocation of unidentified sighting abundance to specific species.

host: Dr Janine Illian

refID: 987

hide details

host: Dr Janine Illian

refID: 980

hide details

The first part of this talk is based on joint work with Victoria Landsman (Dept of Statistics) and with Yohay Carmel and Ronen Kadmon (Dept of Evolution, Systematics and Ecology) of the Hebrew University.  I will discuss a method we developed (Analyzing Spatially Distributed Binary Data Using Independent-Block Estimating Equations, Biometrics 2007) to analyze a large (6,000 pixels) set of spatially distributed data, in which a binary response was used to describe vegetation changes in a natural forest area.  I will first discuss the computational difficulties encountered when we tried to apply the Composite Likelihood approach of Heagerty and Lele (A Composite Likelihood Approach to Binary Spatial Data, JASA 1998).  I will then describe our alternative approach, which is essentially a GEE estimator in which the working covariance matrix is defined by breaking the region into disjoint blocks and modelling the spatial dependence within the blocks, while assuming independence between the blocks.  

  In the second part of the talk I will discuss some possible additional questions arising from this work: (1) comparing our approach with other methods for handling relatively large sets of spatial data; (2) extending our method from binary to ordinal responses; and (3) in the context of a Generlized Linear Model with spatial dependence defined by latent normal variables, how to graphically examine the form of the spatial autocorrelation on the latent scale.

host: Dr Janine Illian

refID: 951

hide details

host: Dr Janine Illian

refID: 957

hide details

In this talk I present the work I have been implementing over the last few years, in an attempt to maximize interaction with local researchers at CREEM during my two week visit.

Counting large breeding colonies has always been important to understand population dynamics and for wildlife management, but has encountered several methodological and cost-benefit difficulties. Besides, the more endangered is a species the more funds generally it receives, so non-threatened species are given less fieldwork effort, which usually means less accuracy of results.  

The Ebro delta is the most important wetland in Catalonia, Spain, and holds a huge colony of Yellow-legged gull (Larus michahellis). This species breeds next to one of the most endangered gulls of the world, the Audouin’s Gull (Larus audouinii), of European conservation concern SPEC Category 1. Therefore, Yellow-legged gull monitoring is even more important as there is a real opportunity of identifying interespecific competition as well as changes in distribution, abundance and diversity.

From 2005 until 2012 I have used distance sampling methods to attempt to obtain an accurate and feasible sampling method. Changes in spatial distribution have been recorded during this period. The existence of some terrestrial predators in the breeding area has been subject of some debate either in favour or against their presence, as they act as predators of the species and hence regulate population size. On the other hand, presence of predators not only has affected the distribution of Yellow-legged but Audouin’s Gull colony, which has progressively moved to other suboptimal areas.

The results of this analysis will provide critical information for conservation managers and  will determine if a management of Yellow-legged Gull colony is required.

host: Dr Janine Illian

refID: 950

hide details

Many diving mammals undergo substantial changes in their body density and mass, body condition parameters which have been shown to affect vital rates in some species of seals.   Nutritive body condition reflects a complex integration of foraging effort (energy expenditure), foraging success (energy acquisition), and within-body allocation to lipid stores and/or growth (or replacement) of non-lipid body stores. At-sea measurement of body density has been demonstrated with drift-dives of elephant seals and has enabled identification of locations and times in which foragers build fat stores.  The drift dive method, however, does not track changes in body mass and is not applicable to species that do not perform drift dives.  In this seminar, I will demonstrate that drift dives are just one type of ‘gliding’ behaviour, and that the speed of animals during glides can be described analytically with a mathematical model which includes the influences of drag and buoyancy forces on the gliding diver.  Using this model, we have been able to consistently estimate the body density of elephant seals using different types of glides that are routinely performed by most marine divers.   I will present efforts to apply this approach to measure the body density of deep and shallow diving cetaceans, and discuss challenges related to the influence of diving lung volume on shallow divers.

host: Dr Janine Illian

refID: 949

hide details

According to niche theory, resource partitioning by competing plant species contributes to the maintenance of species richness in plant communities. Hyper-diverse tropical lowland rain forest tree communities present a challenge to this theory because ecologists have difficulty defining sufficient distinct niches to accommodate the large number of competing species. Trade-offs among multiple resource axes may help to resolve this problem, especially when biotic interactions are considered. In this seminar I discuss mechanisms of coexistence among competing species of Dipterocarpaceae, which is the dominant family of trees in Southeast Asian rain forests.  Among dipterocarps, per capita seed production appears to be equalised across species despite orders of magnitude variation in flower production among species. This equalization of seed production can be interpreted as contributing to an equalization of fitness among species that, all else being equal, would aid the coexistence of multiple competing species.  Our data suggest that this equalization of per capita seed production among species is determined by interactions of flower size, pollinator body size, pollination success and pollen dispersal distance. Maximum flower production was two orders of magnitude greater in small-flowered than large-flowered species of Dipterocarpaceae, but small-flowered species also had smaller-sized pollinators, lower average pollen dispersal distances and lower mean pollination success than large-flowered species. Paternity analysis revealed that mating between related individuals was more frequent in a smaller-flowered species. We conclude that a trade-off embedded in the relationship between flower size and pollination success contributes to equivalence of fruit production among species. Our findings also demonstrate the potential for differential vulnerability of species to the deleterious ecological and genetic consequences of forest fragmentation. They have clear implications for forest restoration, as seed collection from small-flowered species may be especially vulnerable to cryptic genetic erosion.  Our interpretation highlights the importance of maintaining pollinator communities and gene flow in order to sustain the species richness of tropical tree communities.

host: Dr Janine Illian

refID: 927

hide details

Gene flow is one of the most critical processes determining the genetic structure of plant populations, with seed dispersal primarily affecting spatial structure and pollination having important consequences for the maintenance of genetic diversity. Natural landscapes have been suffering from constant anthropogenic modifications that ultimately affect pollination and seed dispersal processes. However, little is known about how gene dispersal via pollen and seed movement varies across modified tropical landscapes. Here we aimed to investigate the contribution of pollen and seed dispersal to gene flow of the Amazonian plant Heliconia acuminata (Heliconiaceae), a common understory species pollinated and seed-dispersed by birds. The study was conducted in two continuous forest sites and three 1-ha fragments in the Biological Dynamics of Forest Fragments Project, Manaus, Brazil. Data consisted of flowering plants and established seedlings (from 1999 to 2009) genotyped at 10 microsatellite markers. We used a Bayesian approach to jointly estimate parentage and characterize pollen and seed dispersal kernels by making use of genetic data, spatial location and plant reproductive phenology. Contrary to the expectation that forest fragmentation disrupts gene flow, we found that populations in forest fragments presented higher averages of pollination and seed dispersal distances. Also, seed and pollen movement were more distance restricted in the continuous forest with higher density of flowering plants. Pollinators and seed dispersers are likely to be using fragments and moving across the landscape bringing pollen and seeds from outside the fragments. Giving continuous and elevated local spatial availability of flowers and fruits, birds may concentrate foraging within smaller areas generating a more distance-limited dispersal in continuous forest sites. In conclusion, gene flow is enhanced in forest fragments indicating that behavior of pollinators and seed dispersers change across a heterogeneous landscape due to configuration of habitats and spatial and temporal availability of food resources.

host: Dr Janine Illian

refID: 920

hide details

Moose (Alces alces) is not only the most important game animal but also one of the leading causes of economic forest damages in Finland. Also, hundreds of moose-vehicle collisions take place every year. Hence, strict controlling of moose density is required. Knowledge of local moose stocks is essential in order to successfully manage the population. Our main goal is to construct a reliable statistical model for estimating the size and structure of local moose populations.

First, we present a concise binomial model for daily moose observations. We modify and compare the best estimation methods for binomial distribution’s parameters suggested in the literature. Finally, we apply the best method, a hierarchical Bayesian model, to Finnish moose hunting data. The observation data is collected annually by hunters during the hunting season.

Our primary approach is a more comprehensive Bayesian model for moose population dynamics. Several independent sources of data containing information of the population size and structure are combined within a state-space model framework. The data consists of daily observations and harvest records, harvest structure indices, estimated sex ratio and fecundity parameters, catch and observations per effort indices, and hunters’ estimates of post-harvest stock sizes. We apply the model to the hunting data from 1995—2010 resulting in annual estimates of population size and age-sex structure.

host: Dr Janine Illian

refID: 930

hide details

The fine scale foraging behaviour of southern elephant seals was investigated by using a combination of loggers (accelerometer, magnetometers, and time depth recorders) alowing the 3 dimension reconstruction of the dive simultaneously with the recording of a number of oceanographic parameters (temperatures, salinity, fluorescence, dissoleved oxygen and light) and Argos /GPS tracking devices. This biologging approach was completed by the use of stable isotopes to identify the  foraging habitat along a latitudinal gradient (C13) and the trophic levels (N15).  In this presentation I will present you the most recent results obtained at CEBC-CNRS relating the changes in foraging success with the foraging habitat  and oceanographic conditions at sales ranging from the foraging trip to the dive and the output of 10 years of fruitfull collaboration with SMRU reserchers and engienners in developping new satelitte data relayed oceanographic loggers.

host: Prof Michael Fedak

refID: 928

hide details

http://soi.st-andrews.ac.uk/staffProfile.aspx?sunID=steve&mp=&b=

host: Prof Anne Magurran

refID: 910

hide details

Measuring diversity: the importance of species similarity

(Joint work with Christina Cobbold)

There have been decades of debate about how best to measure
biodiversity. Dozens of measures have been proposed. The situation
had seemed forbiddingly complicated until quite recently, when work
of Jost and others cut a straight-line path through the tangle of
proposed measures. I will begin by explaining this new,
conceptually clear, understanding of diversity measurement. (No
previous knowledge of the subject will be assumed.) I will
emphasize the role of diversity profiles, a useful graphical device
for comparing one ecological community with another.

But there is still a problem. Most existing diversity measures are
completely insensitive to the varying similarities between
species, contrary to the common perception of diversity as the
"variety of life". I will show how to repair this defect, and I
will demonstrate with examples that taking inter-species similarity
into account can make a real difference to the assessment of
biodiversity.




 

host: Dr Janine Illian

refID: 872

hide details

Very few long-term studies have been conducted in deep-water systems. As a result we know very little about natural variability or how these systems respond to human impacts. In this talk I will present results from long-term studies from the North Pacific, North Atlantic and South Atlantic. These studies demonstrate the major changes that can occur over time in deep-water fish communities at depths of 1000-4000 m and also the great depths to which anthropogenic impacts reach.

host: Dr Janine Illian

refID: 794

hide details

http://www.bristol.ac.uk/biology/people/njDuCp365xFexnW532D9vM92gtDcWU/index.html

host: Dr Vincent Janik

refID: 879

hide details

Many processes contribute to tree mortality in tropical rainforests, and it
is often hard to distinguish them from one another.  Because of their
potential role in supporting the coexistence of species, however, they are
the subject of much interest and speculation, and it has been suggested
that density-dependent mortality caused by species-specific pests or
pathogens may alone be capable of producing observed levels of species
diversity.  Examples of this effect have been found, but its general
strength and form remains unknown.  In particular, its sensitivity to the
spatial proximity and genetic relatedness of trees is poorly understood.
 We use Bayesian maximum-likelihood modelling to try and identify the
causes of mortality in a 50 hectare rainforest plot in Panama, and to look
for evidence of different kinds of negative density dependence.  We find
ambiguous support for several different models, and discuss issues of model
selection and interpretation.

host: Dr Janine Illian

refID: 868

hide details

For as long as there has been competitive football, spectators have tried to predict match outcomes in advance – both for entertainment and for financial gain. However, only with the development of modern statistical techniques (coupled with advances in computing) has it become possible to build sophisticated forecasting models that can accurately assess the probability of each possible scoreline. The output of such models can be compared against the “market view” expressed by traders on a betting exchange such as Betfair, thereby highlighting promising betting opportunities.

host: Dr Janine Illian

refID: 832

hide details

Traditional log-likelihood based methods for choosing between models, be they nested, non-nested or overlapping, all concentrate on log-likelihoods evaluated at the maximum likelihood estimates of the model parameters. The true model parameters may in fact differ considerably from their maximum likelihood estimates. We propose a method that examines the relative merits of the models over a cross-section of likely parameter values; this method is based upon testing simple hypotheses, and hence avoids pitfalls associated with compound null hypotheses such as biased estimation of p-values. The development of this method from an extension of an approximation to Cox’s test for comparing non-nested models will also be outlined.

host: Dr Janine Illian

refID: 812

hide details

Diagnostic goodness-of-fit tests for capture-recapture data are used prior to most analyses. However, when data include a mixture of live recaptures and dead recoveries, it is still standard practice to remove the dead recoveries from the data set before evaluating the existing goodness-of-fit tests.

Within this talk I will present adapted diagnostic tests for joint recapture and recovery data, and through simulation evaluate their power to detect specific departures from basic model assumptions. The importance of including all data available is demonstrated through simulation. I will include tests for both single- and multi-site joint recapture and recovery data and apply the new tests to a data set of great cormorants.

This is collaborative work with Byron Morgan (University of Kent) and Roger Pradel (CEFE, CNRS).

host: Dr Janine Illian

refID: 776

hide details

I got my PhD in "Mathematical Statistics" in 1973, solving a problem in abstract multivariate analysis that no one has ever had. This is a travesty, because statistics is the ultimate service discipline: "Statisticians are to numbers what librarians are to words." This presentation will deal with some of the data challenges that have come my way from biology. Decent statistical analyses of the primary moult of birds were not feasible until the "Underhill-Zucchini moult model" was introduced in 1987. At one stage, the "chain index" used by the BTO to generate trends in wader populations was performing so badly that results for some species were being repressed. The bird atlas projects in South Africa have produced new opportunities to quantify characteristics in bird distributions, the latest being a measure of fragmentation of distribution. GPS tracking of animal movements at intervals of seconds provides opportunities to analyse behaviour in a way never attempted before. Statisticians need to be alert to opportunities to solve problems that researchers in general (and biologists in particular) really have.

host: Dr Janine Illian

refID: 788

hide details

Natural and anthropogenic environmental changes affect the behavior and physiology of individual organisms, but their ecological significance depends on their consequences at the population level. Understanding how individual condition can be affected by behavior-altering disturbance, and how in turn changes at the individual level can cascade through to population level effects in vital rates is a key goal of marine ecology. Despite the advances in biotelemetry, we still lack the ability to monitor changes in body condition at fine temporal scales while marine predators are at sea. To address this discrepancy, we examine changes in body condition in three different species of marine predator: northern and southern elephant seals, and north atlantic right whales. To estimate condition in these species we used two different modeling approaches. The first case study, elephant seals, relied on inferring daily changes in lipid content in pregnant adult females using observations on drift rates through the water column. The second case study, right whales, relied on estimating daily changes in health as a function of sporadic observations of body condition using photo mark-recapture. For elephant seals we have successfully fit a hierarchical Bayesian state-space model to 29 northern elephant seals, and 30 southern elephant seals, providing daily estimates of lipid status with uncertainty. For right whales, though we are in the early stages of modeling, we have been able to assimilate many types of data from a wide range of sources, and provide initial estimates of underlying health at a monthly time-step for each individual in the population. The modeling approach can also be used to infer condition in a variety of species - both marine and terrestrial. Results from this effort can be used to link disturbance to individual and population health, and inform both where and when management efforts will be most effective.

host: Dr Janine Illian

refID: 784

hide details

The talk focuses on top-down versus bottom-up drivers on the regulation of savanna herbivore populations.  Specifically, I will investigate how these patterns switch based on body size, and how this affects the spatial distribution of these herbivores and possibly their movement patterns.

host: Dr Janine Illian

refID: 777

hide details

The adaptation of distance sampling to estimate bear abundance in Alaska has many unique challenges.  These surveys are conducted with small fixed-wing aircraft, the resulting distance data has maximum detection occurring away from the line transect line.  A gamma-shaped detection model (Becker and Quang 2009) that allows covariates is used to model the distance data.  The resulting covariate models will have multiple distance locations where detection is maximized, one location for each level of the covariate variables.  Mark-recapture data is used to model the probability of maximum detection.  These 2 models are combined using a point-independence model (Laake 1999, Borchers et al. 2006) to produce bear population estimates.  Progress on implementing a limiting independence model (Buckland et al. 2010), which has no independence assumptions, will also be presented. 

host: Dr Janine Illian

refID: 775

hide details

Modelling species presence/absence models using local environmental factors has been a growing problem in ecology in recent years. I'll briefly discuss an approach to address this issue that I have developed this year. I propose a Bayesian hierarchical model to predict the occurrence of species incorporating environmental and spatial variables using data from commercial fishery. Maps of predicted probabilities of presence have been generated using Bayesian kriging. I applied this approach to the Mediterranean mackerel (Trachurus mediterraneus) in the western Mediterranean, but the methodology could also be used for any other species. However, this predictive model identifies the distribution of species based only on ecological relationships between the environment and habitat selection of species. My interest now is to analyze how the biotic relationship affects the spatial distribution of species. I'll propose a particular case of study of "predator-prey", and I'll explain my guess to address this issues.

host: Dr Janine Illian

refID: 774

hide details

Having (successfully?) avoided giving a (proper) CREEM seminar over my
six years with the department, my last day seemed an opportune time to present
my research. Starting with the PhD on birds in Scotland, to a postdoc looking a
sea mammals in the Southern Hemisphere, and a future with the Marine Mammal
Commission in which I'll consider species in the Bahamas, state-space models
have been the connecting theme. I'll present finished research, work in
progress and ideas for the future in an attempt to make up for my previous
years of dodging the presentation bullet.

host: Dr Janine Illian

refID: 759

hide details

We report on the use of network sampling of tracks in the snow to estimate animal population size over large geographic areas in Alaska.  Network sampling, a special form of probability sampling, is typically used to estimate the prevalence of rare diseases in human populations.  Treating animal tracks in the snow as a network allows us to obtain population estimates of various wildlife populations, especially low density populations (3-15 animals/1000 km²).  The sample design incorporates available information on the likelihood of encountering fresh tracks in different regions of the study area.  The design is modified to incorporate important biological facts of the target species.  We illustrate this approach by estimating wolverine (Gulgo gulgo) and wolf (Canis lupus) population size.  The presentation will also involve an introduction to tracking wolverines and wolves in the snow.

host: Dr Janine Illian

refID: 768

hide details

Catriona Harris, Becci Jewell and Nicola Quick will give an overview of three manuscripts that have resulted from a database of line-transect surveys and associated abundance estimates, compiled by CREEM and SMRU Ltd over the last 5 years.  We will describe the taxonomic, spatial, temporal and methodological distribution of line transect surveys between 1980 and 2005, focussing on ~50 species of interest.  We will then describe analyses that were carried out to determine whether trends in abundance could be detected for any species when we accounted for all of the factors mentioned above.  Finally, we will briefly discuss the overlap between available data and marine development activities.

host: Dr Janine Illian

refID: 747

hide details

Bitterling are small fish that lay their eggs in the gills of
living freshwater mussels. Over the past 16 years I have undertaken research on
these fascinating animals to understand their relationship with mussels and
their mating system. Female bitterling make sophisticated decisions about which
mussel to lay their eggs in and choose mates based on genetic compatibility.
Males are sensitive to the risk of sperm competition from rivals and modulate
the way in which they distribute sperm in space and time. The bitterling mating
system lends itself to experimental studies, but also to a mathematical approach
to develop new hypotheses and predictions. In this talk I will outline the
mating system of bitterling and suggest questions that might benefit from a
modelling approach

host: Dr Janine Illian

refID: 740

hide details

Local assemblages, the set of species present, differ from those of neighbouring areas and are not constant in time either. This turnover in composition is a constant in nature and the reasons why there exists this change presents some interesting ecological questions. First, are some areas inherently prone to have greater turnover than others, and second, what is it about these areas that might cause this? Here we analyse mid to macro scale patterns in turnover of birds concentrating on two contrasting mechanistic hypotheses: (i) mean climatic conditions generate differences in turnover from place to place – because of their average climatic conditions, some areas are inherently more subject to local extinction/recolonisation, (ii) variance in climate causes turnover. The latter hypothesis seems a priori to have the stronger intuitive appeal: variance in climate begets variance in species composition. Using simple mixed modelling we assess the support for these two mechanisms to see if this intuition is correct.

In the second part of the talk we discuss the notion of optimal conservation planning in the context of British bird species associations with climate and habitats and the contrasting requirements/limitations of species. We show how these contrasting requirements with respect to habitat creation/management might be reconciled across many species both in the current environment and in the future under environmental change.

Jack J. Lennon (The James Hutton Institute), Colin M. Beale (University of York), James W. Pearce-Higgins (British Trust for Ornithology), Roger Catchpole (Natural England), Kevin J. Gaston (Sheffield University).

host: Dr Janine Illian

refID: 720

hide details

As a quantitative palaeobiologist, the challenge I am faced with is extracting
the signal of genuine biodiversity change from a noisy fossil record. Although
palaeobiologists have made considerable progress in adapting ecological tools
and methods to tackle sampling biases, the rock record is a unique challenge.
My talk will give a broad overview of the problems of reconstructing the
evolutionary trajectories of fossil taxa and communities and explain some of
the spatial explicit analytical approaches I am attempting to adapt from my
experience in ornithological fieldsurveys to tackle these problems.I will also
discuss the challenges of data intensive research in biodiversity and my new collaboration with ScotGrid that is aimed at improving our ability to search and process large, complex data sets.

host: Dr Janine Illian

refID: 738

hide details

Population viability analysis (PVA) is a method to evaluate the probability of an unexpected state (e.g. extinction, decline population) as a function of several biological and environmental variables. In fisheries science there is not consensus in the way to implement the PVA, which has covered from qualitative model-free approach to sophisticated mathematical frameworks including spatially explicit and stochastic models. In this work I will expose some ideas about the application of PVA in fisheries science using a probabilistic and mathematical approach. PVA method such shown here accounts and compares the multi-criteria decision problem underlying fisheries management.

host: Dr Janine Illian

refID: 743

hide details

Adaptation is important for reducing the vulnerability of local communities
to global change. Many responses, however, can worsen existing problems or
reduce capacities to respond to future events. This paper reports on the
trajectories of vulnerability in Kahua communities in Solomon Islands and
asks whether current responses to environmental change are maladaptive or
adaptive. Over the past five years, interdisciplinary work has been
investigating a range of subjects from perceptions of change, valuation of
ecosystem services, education, and cross-community governance. Findings
suggest that changes are driven by a relatively small number of key drivers
(population growth, desire for monetary prosperity) which are acting
synergistically in combination with increasing opportunities for income
generation to reinforce and increase vulnerability to future change. The
presentation will outline key findings, then open up the discussion to
consider how ecological modelling and related approaches might be able to
add further insights to understanding the Kahua social-ecological syst

http://www.fazey.webeden.co.uk/

host: Dr Janine Illian

refID: 726

hide details

Population viability analysis (PVA) is a method to evaluate the probability of an unexpected state (e.g. extinction, decline population) as a function of several biological and environmental variables. In fisheries science there is not consensus in the way to implement the PVA, which has covered from qualitative model-free approach to sophisticated mathematical frameworks including spatially explicit and stochastic models. In this work I will expose some ideas about the application of PVA in fisheries science using a probabilistic and mathematical approach. PVA method such shown here accounts and compares the multi-criteria decision problem underlying fisheries management.

host: Dr Janine Illian

refID: 742

hide details

Extreme value theory (EVT) is concerned with drawing inferences about the statistical properties of rare events, and the univariate version of EVT is widely used as a basis for risk assessment within disciplines such as hydrology, climatology and finance. Multivariate EVT describes the relationship between the extremes of different processes - or between the extremes of the same process at multiple locations or timepoints - and is substantially more challenging from both a theoretical and practical perspective.

In this talk I will introduce the basic concepts of EVT and outline some of the key statistical ideas that are involved in studying dependence between variables at extreme levels. I will present some recent research in this area on quantifying extremal dependence within Markov chains, and will give an overview of current research on the application of Reversible Jump MCMC to multivariate extreme value models.

host: Dr Janine Illian

refID: 727

hide details

Conserving biodiversity in the face of deforestation, climatic change
and other threats is one of the foremost challenges facing ecologists.
Effective conservation policy relies on understanding both the
mechanisms that maintain biodiversity and the ways in which human
activity threatens its persistence.

There are many theories that seek to explain the coexistence of
species in diverse natural communities, such as tropical rainforests.

host: Dr Janine Illian

refID: 733

hide details

Is there really a chance undiscovered giant
marine life might exist? Ignore all the clichés
and what you see on TV and join Dr Charles Paxton
from the University of St Andrews as he delves
into the deep and attempts to discover the truth about sea monsters.

host: Dr Janine Illian

refID: 735

hide details

Last year in the fall, the Southwest Fisheries Science Center/NOAA, La Jolla conducted a
survey to estimate the abundance and distribution of cetaceans around the Hawaiian Islands
including the entire island chain and surrounding waters out to 200nm (the Hawaiian
Economic Exclusive Zone). Two NOAA research vessels were used for this survey, the
McArthur II (for about 4.5 months) and the Oscar Elton Sette (for about 2 months). This time I
was part of the acoustics team on the McArthur II conducting line-transect surveys using
passive acoustics methods. During my presentation I will talk about survey design, survey
methods, equipment we used and what we found.

host: Dr Janine Illian

refID: 728

hide details

host: Dr Janine Illian

refID: 721

hide details

Bayesian Model Selection and Averaging are common tools to deal with
model uncertainty, not just in the ecological community but in
statistics at large. It is usual in such circumstances to choose an
"uninformative'' prior over the model space, but we demonstrate how it
is important to exercise care in choosing the model parameter priors as
well, so as not to bias the prior towards certain models. We present an
example in modeling density dependence in the American Wigeon and show
how applying some simple rules and a little thought can lead to improved
posterior inference and noticeably better predictive power.

host: Dr Janine Illian

refID: 695

hide details

I will discuss some of the work of I have been doing during my visit to
CREEM on integrating genetic information into occupancy models. I will
present some very preliminary results and leave plenty of time for
discussion."

host: Dr Janine Illian

refID: 718

hide details

Design -based and Model -based approaches are used for abundance
estimation of cetaceans. To compare performance between both approaches,
many simulation studies are reported from with synthetic data. It is
useful to know the precision and robustness of both approaches so that
an informed choice can be made under plausible survey and distribution
conditions. Under a gradient distribution assumption (which is standard
in this area), a line transect survey is carried out, but crucially
including estimates of robustness and precision... Parameters and
covariates that are needed for generating a virtual space and whales
were based on Japanese whale research programs under a special permit in
the Antarctic. A virtual survey was carried out under a framework of
stratified systematic sampling of a line transect survey. It was assumed
that g(0)=1, and all animals within a fixed width are detected.
Abundance estimation from GLM, GAM and line transect methods were
carried out and their precision (CV: Coefficient of Variation) and
robustness (RMSE: Root Mean Square Error) were compared under 144
scenarios. GLM showed better performance than for CV in almost all
scenarios, whereas line transect method performed better in terms of
RMSE. The CVs and RMSEs were analyzed using a random effects model ANOVA.
 The result of the ANOVA indicated an interaction term between the true
distribution of whales and the pattern of stratification. This term
always represented more than 10% of the total variance for each CV and
RMSE from both the GLM and the line transect method.

host: Dr Janine Illian

refID: 694

hide details

The 2010/2020 biodiversity target, set by the Convention of Biological Diversity and adopted by the EU’s Environmental Council, seeks to “halt the loss of biodiversity and the degradation of ecosystem services” at a global, national and regional level.  To assess progress or failure in achieving this goal, long-term trends have to be monitored and analysed.
Due to their wide range of habitat and position in the food chain, birds are widely used as indicator species for the state of ecosystems.   Here, we look at data from the UK Breeding Bird Survey (BBS). Collected yearly on a national level since 1994, these data offer a sufficiently long time series to investigate trends in bird diversity across different habitat groups. Traditional and more recent diversity indices are examined. By evaluating 2^nd derivatives of these trends we are able to identify points where the rate of change in diversity has been significantly accelerated or slowed down.

host: Dr Janine Illian

refID: 688

hide details

In this seminar I will give an overview of some of the work I do and what I intend to do over the next three weeks while here at CREEM. Much of my research focuses on the development of tools for the conservation of wildlife in landscapes that have been highly modified by humans. Wildlife conservation in these landscapes presents unique challenges because of complex interactions between multiple threats that drive species persistence. Despite this complexity, easy to use principles and rules-of-thumb upon which to base management and monitoring decisions are still desirable. I will illustrate a range of approaches we have been using to inform conservation decision-making in these landscapes using the koala as an example. I will then outline some novel approaches for integrating species’ distribution data and genetic data to improve species’ distribution models that I am interested in working on while here at CREEM.

host: Dr Janine Illian

refID: 697

hide details

Point pattern analyses such as the estimation of Ripley’s K-function or the pair-correlation function g are commonly used in ecology to characterise ecological patterns in space. However, a major disadvantage of these methods is their missing ability to deal with spatial heterogeneity. A heterogeneous intensity of points causes a systematic bias in estimates of the K- and g-functions, a phenomenon termed ‘‘virtual aggregation’’ in the recent literature. To address this problem, we derive a new index, called K2-index, as an extension of existing point pattern  characteristics. The K2-index has a heuristic interpretation as an approximation to the first derivative of the g-function.
We estimate the K-, g- and K2-functions for six different types of simulated point patterns and show that the K2-index may provide important information on point patterns that the other methods fail to detect. The results indicate that particularly the small-scale distributions of points are better represented by the K2-index. This might be important for
testing hypotheses on ecological processes, because most of these processes, such as direct neighbour interactions, occur very locally.

When applied to empirical patterns of molehill distribution, the results of the K2-analysis show regularity up to distances between 0.1 and 0.4 m in most of the study areas, and aggregation of molehills up to distances between 0.2 and 1.1 m. The type and scale of these deviations from randomness agree witha priori expectations on the hill-building behaviour of moles. In contrast, the estimated g-functions merely indicate aggregation at the full range from 0 to 7 m (or even above). Considering the advantages and disadvantages of the different methods, we suggest that the K2-index should be used as a complement to existing approaches, particularly for point patterns generated by processes that act on more than one scale.

host: Dr Janine Illian

refID: 693

hide details

host: Dr Debbie Russell

refID: 681

hide details

 Stochastic filtering is a fundamental concept in the theory of estimation of dynamic systems. In its discrete-time formulation, the stochastic filter can be described by the Chapman-Kolmogorov equation, which models the evolution of a posterior distribution over time, and Bayes rule, to update the posterior based on new observations. The optimal solution to the multi-object filtering can be found through a direct generalisation of the Bayes filter to multi-object systems using point process theory. This solution, derived by Ron Mahler has led to mathematically principled and tractable multi-target tracking algorithms that now have been deployed in a number of industrial applications for autonomous robotics navigation and radar surveillance. This talk will present the theoretical foundation for these results and discuss new developments in this field.

host: Dr Janine Illian

refID: 687

hide details

David Borchers
I'll be giving an overview of my current and recent research. This includes mark-recapture distance sampling methods, spatially explicit capture-recapture methods, using hidden Markov models (HMMs) to model Scottish fishing vessel fishing behavour from "vessel monitoring system" data (i.e. GPS tags on vessels),  using HMMs to model animal availability, dealing with distance measurement errors on distance surveys, some methods for accommodating non-uniform animal distribution on distance sampling surveys, Markov modulated Poisson process models for animal availability on distance sampling surveys, and a few other things.

Carl Donovan
I'll similarly be giving an overview of current/recent research and the projects I am involved in. Specific examples:
i) (more on) simulation methods for Environmental Impact Assessments of Sonar/Seismic surveying and Marine installations.
ii) modelling weather impacts on power distribution networks.
iii) adaptive smoothing methods, in particular over "complex" 2D spaces.
iv) boosting methods.

host: Dr Debbie Russell

refID: 679

hide details

host: Dr Debbie Russell

refID: 678

hide details

Designing efficient algorithms is considered to be one of the most challenging
tasks in Scientific Computing.
The aim of the talk is to give a gentle introduction to some techniques for
designing and implementing an efficient algorithm.
The first part of the talk will focus on some key issues such as memory
management, definition and use of data structures, iterations, profiling.
For this purpose, source codes written in R will be used as examples.
In the second part, several possible approaches for parallel computing in R will
be presented.

host: Dr Debbie Russell

refID: 677

hide details

Realised Levy flights are not super-diffusive and are only a special case of a simple random walk.

host: Dr Brett McClintock

refID: 654

hide details

If only there was no missing data...

Imagine that we knew certain biological features of our population of interest. For example, suppose that the times of birth and death were known without error, as well as associated covariates. In this ideal world, our inference would be straightforward and our focus would be on specifying models and relationships of scientific interest that describe the dynamics of the population. In reality, however, values such as the times of birth and death, as well as covariates are unknown. Therefore, we construct elaborate sampling schemes (e.g. capture-recapture experiments) in order to get partial information about these quantities of interest. This means that the capture-recapture experiment can be thought of as a missing data problem. Classical approaches to overcoming this problem focus on accounting for the complex sampling process that generated the missing data. This wrestles our focus away from the scientific questions of interest, evidenced by the proliferation of literature that describe how to account for subtle (as well as not so subtle) sampling differences.

Here we will show how the missing data inherent in capture-recapture data can be accounted for using data augmentation where `we model the data we wish we had'. This allows our focus to return to modeling relationships that describe the dynamics of the population. We will proceed step-by-step through this modeling framework using several examples.

host: Dr Debbie Russell

refID: 643

hide details

These days many, even rather complex statistical models can be fitted routinely
as a result of widely accessible high-level statistical software, such as R and
winbugs. For instance, even the non-specialist user can estimate parameters in
generalised linear mixed models or run a Gibbs sampler to fit a model in a
Bayesian setting both of which formerly required expert programming skills. In
addition, assessing a model's fit as well as the comparison of different models
has become straight forward. Researchers from many different disciplines are
now able to analyse their data with sufficiently complex methods rather than
resorting to simpler yet non-appropriate methods.

For spatial point process models the
situation is very different. Even though a rich theory has been developed
fitting a point process model to data is complex and requires expert knowledge.
There have been basically no attempts at providing methods for model
comparison for Cox
processes.

However, due to rapidly improving technology and a growing awareness of the
importance and relevance of small scale spatial information, spatially explicit
data sets have become increasingly available in many areas of science,
including plant ecology, animal ecology, genetics, geology and medicine.
Currently, these data sets are often analysed with methods that do not make
full use of the available spatially explicit information. Hence there is a need
for making the existing point process methodology available to applied
scientists.

Here we suggest an approach to fitting complex log Gaussian Cox processes.
We demonstrate that this allows us to fit
many realistically complex data sets to spatial point pattern data and
that it enables
the routine fitting of complex point process models to spatially explicit data
sets.

host: Dr Debbie Russell

refID: 642

hide details

Brook trout were abundant in many streams in the eastern United States
in the 1700s.  Human perturbation has lead to the extirpation of brook
trout in many streams.  The Eastern Brook Trout Joint Venture (EBTJV)
was formed in 2005 to form partnerships in preserving and restoring
brook trout habitat.  To aid in the management process, a large
database was compiled by EBTJV covering roughly 5000 subwatersheds
with information on brook trout status and land use and other metrics.
 The data have been analyzed using several approaches using CART and
spatial clustering.  Although a CART model provides adequate
classification, a spatial approach improved classification.  Measures
of restoration and preservation are developed based on the
probabilities of extirpation using model-based estimates and also
estimates that incorporate information from neighbors.

host: Dr Debbie Russell

refID: 641

hide details

My current work aims at modelling the distributions and spatial behaviours of vertebrate terrestrial species with principal goal the control of possible impacts of anthropogenic disturbances on ecosystem dynamics and human health of local populations. My thesis focused on the transmission ecology of Echinococcus multilocularis, a parasite responsible for fatal zoonosis when ingested by humans. I analysed the distributions of host species for this parasite in Chinese remote areas (Sichuan province), at different scales and levels of biological organisations: i) the spatial distributions of small mammal assemblages on the basis of presence/absence data sets collected on the field and, ii) the nocturnal space occupancy for some individual domestic dogs based on data collected using GPS collars.

In order to summarise the great diversity observed in sampled habitats and trapped species, we first defined small mammal assemblages by reducing the redundancy that was found in the habitat classes chosen as explanatory variables for the multinomial model fitted to the trapping success observations. The spatial distributions of assemblage presences/absences were then modelled along environmental gradients extracted from satellite images such as elevation, slope and vegetation indices. We obtained that Multiple Adaptive Regression Splines provide lower bootstrapped classification error rates than GLM's or linear or non-linear discriminant analysis. Finally, while predictions from locally trained models were not transferable on study sites distant from one hundred km, a regional classification of assemblages, trained on the whole regional data set, provided low predictive errors. Finally, analysis of dog nocturnal trajectories versus spatial distributions and contaminations counts of canid faeces allowed to assess and quantify the predominant contribution of dog over fox in the environmental contamination and to localise areas of high transmission risk, close to human habitations. Dogs spend the majority of the day near their owners' houses but also travel excursive paths outside the mean activity area of the village dog populations where small mammal presence indices were the most frequent.

Based on this first experience in statistical and spatial modelling of eco/epidemiological data and the encouraging results obtained using recent statistical methodology, I seek to go on working on the interface between field data collection alongside with ecologists and the development of pertinent statistical approaches in close collaboration with statisticians. Furthermore, the developing availability of large data sets of multiple species presences/absences, consisting of a collection of several field surveys or of atlas data, along with satellite images actually enhances model building and evaluation on large spatial ranges at which effects of global environmental changes (climate warming or landscape disturbances) on population distributions and biodiversity can be observed. Concerning the analysis of individual trajectories, I am currently working on the modelling of dog movements using dynamic models based on Brownian motion and Levy flights. One interesting expected result would be to identify model switching behaviours in dog nocturnal trajectories.

host: Dr Debbie Russell

refID: 616

hide details

host: dm60

refID: 631

hide details

Recent developments in animal tracking technology have permitted the collection of detailed movement paths from individuals of many species. Despite this rapidly increasing wealth of information, model development for the analysis of complex movement data has not kept pace with these technological advancements. To better understand complicated animal movements in heterogeneous landscapes, we propose that complex movement paths can be dissected into a few general movement strategies among which animals transition as they are affected by changes in the internal and external environment. We develop a suite of discrete-time individual animal movement models based on biased and correlated random walks that include different behavioural states for migration, exploratory, and resident movements. Models may then be “custom-built” for a wide variety of species applications, thereby allowing the simultaneous estimation of movement behaviour states, state transition probabilities, locations of migration or resident centres of attraction, and the strength of attraction to specific locations. The inclusion of memory or covariate information in the modeling of state transition probabilities permits further investigation of specific factors related to different types of movement. Using Markov chain Monte Carlo methods to facilitate Bayesian inference, we apply the proposed methodology to grey seal movements among haul-out and foraging locations in eastern Scotland.

host: Dr Debbie Russell

refID: 615

hide details

The early Earth was a “water world”—one ocean enveloped the entire planet. The atmosphere then consisted mostly of the greenhouse gas, CO2. Warm alkaline springs continually exhaled into this carbonated ocean carrying that most effective of fuels, hydrogen. The H2 and CO2 reacted together, and with hydrothermal ammonia, to produce organic molecules in one of the many mounds precipitated at these springs on the seafloor. The organic molecules so produced then reacted together to form ever more complex organic molecules, eventually reproducing as the first organic cells within this hydrothermal hatchery. The acetate and methane wastes generated during this early biosynthesis were discharged into the ocean. Bacteria migrated from the mound into the deep biosphere where some were carried by ocean floor spreading into the photic zone. Here a few managed to exploit the solar flux, even gaining H2 for biosynthesis from the photolysis of water.

host: Dr Debbie Russell

refID: 614

hide details

One of the purposes of the CREEM seminar series is to promote communication within the group about research projects we’re involved in.  With that in mind, I will give a brief overview of the various projects I’m working on at present.  Many will be familiar to most of you, but hopefully there will be something new for everyone.

host: Dr Debbie Russell

refID: 613

hide details

Extensive volunteer-based programmes for monitoring avian abundance and demography allow us to investigate some of the factors that determine bird population changes at very large spatial scales. Intensive studies of small areas often provide good insights into factors determining local population size, but scaling up from these studies to regional and continental scales is not straightforward. The BTO/JNCC/RSPB Breeding Bird Survey (BBS) and Constant Effort ringing Schemes (CES) co-ordinated through the European Union for Bird Ringing (EURING) illustrate the value of well designed monitoring projects. The BBS has been central to measuring declines amongst farmland birds and is now being used to evaluate the effectiveness of agri-environment schemes. A recent study of the possible effects of predation on bird populations in England illustrates both the power of such large-scale analyses and the limitations of a purely correlative approach. European scale modelling of the demography of a suite of Palaearctic-African migrants is providing insights that could not be obtained using data from individual countries. There is an urgent policy need for better predictions of the impact of climate and environmental change on biodiversity which we are starting to address through spatial modelling of BBS data. In many of these areas close collaboration between ecologists and statisticians can help us to answer challenging ecological questions and to provide more effective advice to policy makers.

host: Dr Debbie Russell

refID: 612

hide details

The BUGS project is very ancient. The software aims to make the

host: Dr Debbie Russell

refID: 611

hide details

Members of RUWPA have been involved with analysing aerial surveys of marine mammals around the coast of Greenland for several years. In May 2009, I was invited to participate in a survey along the northwest coast. The main species of interest were beluga, narwhal and walrus and in order to take account of imperfect detection a double-observer search protocol was implemented. For the first few days bad weather hindered our progress to the study region and we hop-scotched our way up the coast, sightseeing as we went. Fortunately, the weather improved and so we were able to start surveying. I will describe (briefly) double-observer methods and what happened in practice, show some photographs of the trip north and finish with some results.

host: Dr Debbie Russell

refID: 610

hide details

Originally thought to be restricted to humans or animals kept in captivity, there is currently compelling evidence that the process of ageing, or senescence, is common in nature. Investigating the ecological conditions that alter senescence in wild animals as well as between-individual heterogeneity regarding ageing, is likely to shed light on the evolution of the decline in fitness contribution with age. We studied the pattern of ageing at the individual level in tawny owls, a relatively long-lived nocturnal raptor subject to important environmental variation driven by cyclic prey fluctuations. We further investigated whether conditions encountered early in life have long-lasting effect and alter the extent of senescence  in this species.

host: Mr Phil Le Feuvre

refID: 588

hide details

Once upon a time information about a journal article was kept on a note card,
and some filing system was used to dredge that information back at a later
time.  Now, we download pdfs or have a mass of bookmarks in our browser.  The
more sophisticated use software such as Endnote or Jabref.  I will briefly
describe a browser plugin (Zotero) that provides for a) organization of
bibliographic information for individual researchers, b) opportunity to share
bibliographic collections among colleagues and c) retreive stored bibliographic
material while using your normal writing tools in preparing manuscripts and
reports.

host: Dr Debbie Russell

refID: 577

hide details

Most methods used to estimate population parameters in fisheries with catch-at-age data assume that all individuals follow the same growth pattern.  Models that involve reconstruction of age structure from length frequencies, fitting to length composition data as a surrogate for age, and/or predict selectivity schedules based on size distributions at age, typically assume instantaneous renewal of the size structure at age at each time step.  This assumption is seriously violated in size-selective fisheries where the faster growing individuals of an age-class suffer cumulatively greater fishing mortality. The “growth type group” (GTG) model accounts for individuals of the same age growing at different rates, thus explicitly considering distortions to the population size structure.  Accounting for this distortion results in different estimates of fishing mortality from size-structure data. Using retrospective analysis on data for the shrimp fishery in the Gulf of Mexico (US EEZ), as well as simulated data, we compared VPA reconstructions using age compositions estimated from length data to GTG estimates based on fitting to total catch and effort data.  We show here that different fishing mortality rates can be estimated and the GTG can therefore be a useful tool in the management and rebuilding of fish stocks.

host: Mr Phil Le Feuvre

refID: 573

hide details

Relative fit is routinely assessed when performing model selection to choose between a set of candidate models. Examples of relative assessment approaches are the use of information criteria, or hypothesis tests such as likelihood ratio or score tests (McCrea and Morgan, 2009). This talk begins with a brief discussion about the assessment of relative fit of models; however it then focuses on the need to assess the absolute fit of a selected model.

Diagnostic goodness-of-fit tests exist for both single and multi-site mark-recapture models. However, at present tests determining the absolute fit of models are lacking for joint recapture and recovery data. Because of this, current approaches ignore the recoveries of dead animals when assessing goodness-of-fit. We demonstrate the problems associated with this approach.

The sufficient statistic formulation of joint recapture and recovery models provides the solution to absolute goodness-of-fit assessment of these models. We explain the differences between the King and Brooks (2003) and Catchpole et al. (1998) sufficient statistic formulations and demonstrate why the King and Brooks (2003) formulation is the necessary approach for the assessment of absolute fit.

The methodology is applied to data on Great Cormorants (McCrea et al., 2009), as well as demonstrated using simulations.

References Catchpole, Morgan, Freeman, Harris (1998) Integrated recapture/recovery data analysis. Biometrics. 54, 33-46. King and Brooks (2003) Closed-form likelihoods for Arnason-Schwarz models. Biometrika. 90, 435-444. McCrea and Morgan (2009) Multi-state mark-recapture model selection using score tests. Biometrics. Revision submitted. McCrea, Morgan and Bregnballe (2009) One size fits all? EURING proceedings, Journal of Ornithology. Submitted.

host: Mr Phil Le Feuvre

refID: 572

hide details

ST Buckland, SR Baillie, JMcP Dick, DA Elston, AE Magurran, EM Scott, RI Smith, PJ Somerfield and A Watt

The 2010 Biodiversity Target of the Convention on Biological Diversity, set in 2002, is that there should be ‘a significant reduction of the current rate of biodiversity loss’ by 2010. We consider whether the approaches used to measure biodiversity trends are effective for assessing whether targets of this nature are achieved. We conclude that few monitoring programmes have sufficient elements of survey design to allow valid inference at a regional level, as distinct from quantifying trends at the sampled sites only. We also note the failure of most schemes to quantify detectability and how it varies by species, and discuss how this might lead to biased conclusions. We further conclude that precision is often not estimated or not adequately quantified. We also note that ‘gamma’ diversity is not dependent on the scale (size) of the sample plots, whereas the classical partitioning of this into ‘alpha’ and ‘beta’ diversity is scale-dependent, reducing their relevance and usefulness for regional monitoring. We consider the types of survey typically conducted, and identify which types are appropriate for different objectives. We present this in the form of a guide which might be used as an aid for the design of new surveys.

host: Dr Debbie Russell

refID: 564

hide details

 Spatially Explicit capture-recapture (SECR) models are varieties of capture-recapture models which use spatially referenced capture-recapture data to draw inferences about the dependence of capture probability on spatial location. One of their distinguishing features is that they allow rigorous estimates of density to be obtained from the (spatially referenced) capture-recapture data itself.

 In recent years a variety of SECR estimation models and estimation methods have appeared in the literature.  In this talk I focus on the models underlying the estimation methods. It turns out that there is a neat connection between SECR models and some distance sampling models. You can construct a kind of continuum of increasingly complex models, starting with circular plot sampling, moving through varieties of distance sampling model and ending (for the moment) with SECR models. The talk includes a brief overview of SECR models in the literature. Areas of current and possible future research in the area are also briefly discussed.

host: Dr Debbie Russell

refID: 563

hide details

Smoothing over complex 2-D regions is difficult. Several approaches have been proposed in past years including finite element analysis (Ramsay, 2002), within-area distance (Wang & Ranalli, 2007) and recently soap film smoothing (Wood, Bravington & Hedley, 2008.) Here I investigate two alternative methods.

The first (following from Eilers (2006)) is based on the Schwarz-Christoffel transform from complex analysis. This takes the region and 'morphs' it to a rectangle or disk in a prescribed way using a conformal mapping.

The second uses a "new" algorithm to create a set of within-area distances. These distances are then mapped to points in the new domain using multidimensional scaling.

In both cases the aim is to find a new configuration of points that takes into account the intrinsic structure of the domain. It is possible to then smooth over the transformed area using penalized regression splines and transform this smooth back to the original domain in order to perform analysis.

host: Dr Charles Paxton

refID: 553

hide details

Clifford Bay Marine Farms Ltd is seeking to establish a new marine farm – for green-lipped mussels - in Clifford Bay, on the east coast of New Zealand. During the consenting phase of this project, concerns were raised about potential impacts on a local population of the endangered Hector’s dolphins (Cephalorhynchus hectori hectori), residing in Clifford Bay and neighbouring Cloudy Bay. In order to establish baseline distribution and abundance (for subsequent post-installation monitoring), three years of aerial surveys were carried out, using distance sampling methods (MRDS). Four surveys per year were carried out per year, corresponding approximately with spring, summer, autumn and winter. This talk will describe the survey work, and discuss the results.

host: cgp2@st-and.ac.uk

refID: 543

hide details

 Since 2005, stratified random trawl surveys of the monkfish stock which occupies the northern European shelf have been conducted annually to obtain age-structured estimates of absolute abundance for this stock. Survey data are unusual in that almost all fish are aged as well as lengthed. We describe an estimation method which incorporates statistical models for herding, length-based selectivity and missing age data and incorporates uncertainty from all of these sources in variance estimation. Design-based estimates are presented and model-based methods are investigated as an alternative.

host: Dr Charles Paxton

refID: 507

hide details

Time-varying, individual covariates are problematic in experiments with marked animals since the covariate is typically only observed when the animal is captured. We consider three methods to incorporate time-varying, individual covariates in capture-recapture or mark-recapture-recovery experiments: simple imputation, a Bayesian approach based on modelling the joint distribution of the covariate and the capture-history, and a conditional approach considering only events dependent on the observed data. We will describe each method in turn and use a simulation study to compare the three approaches under different scenarios. We discuss the pros and cons of each method before finally applying the methods to real data (Soay sheep).

host: Dr Charles Paxton

refID: 491

hide details

Vaquita /Phocoena sinus/ of the upper northern Gulf of California, Mexico are now considered the most endangered cetacean species in the world after the baiji /Lipotes vexillifer/ of China’s Yangtze River were recently declared functionally extinct. As part of an effort to protect the vaquita, Mexico’s Instituto Nacional de Ecología (INE) requested collaborative support from the international scientific community to develop new autonomous acoustic methods to monitor this rare species effectively. In response to this invitation, US and Mexican scientists together with expert acousticians from the UK, the US and Japan conducted a multi-platform research cruise in fall of 2008.
To produce a current abundance estimate that would tie in with the newly developed acoustic methods, scientists aboard the NOAA research vessel /David Starr Jordan/ conducted line-transect surveys in the chore area of vaquita habitat. I was part of this team and will present research objectives, field methods and preliminary results with main emphasis on the visual survey.

host: Dr Charles Paxton

refID: 490

hide details

abstract to follow

host: Dr Charles Paxton

refID: 464

hide details

In 2006 and 2008, an international collaborative team of scientists
from St Andrews, FFI in Norway, and TNO in Holland conducted a series
of dose-response experiments to describe the influence of sonar
frequency on behavioural responsiveness of cetaceans.  The
experiments consist of attaching a sound and movement-recording Dtag
to a subject whale, and visually observing the tagged animal and
group members before, during, and after controlled exposures by 1-2
kHz or 6-7 kHz sonar signal transmitted by a boat approaching the
whales at 8knots.  Control stimuli include no-sonar silent approaches
and playback of killer whale sounds.  To date, we have conducted 3
experiments with killer whales, 4 with pilot whales and 1 with a
sperm whale.  Another trial is scheduled for May-June 2009.  In our
dataset, we have noted striking behavioural changes coincident with
the sonar presentations, but these changes seem quite variable both
within and across species, likely reflecting the diverse behavioural
context in which the sonar signals were played to the
animals.  Standard frequentist statistical approaches are likely to
be challenging with our data, due to a small sample size and
variability in responses observed, so I hope to consider other
statistical approaches by which we might describe what we can infer
from our dataset about response thresholds, response type, and
biological significance of responses.

host: Dr Charles Paxton

refID: 478

hide details

Abstract:

This talk will outline three current research areas in the dynamics of marine ecosystems, and potential directions for such research in the future. 

1) Dynamics of marine size spectra.  These describe the biomass transfer from phytoplankton to large marine animals.  Knowledge of these dynamics is needed in a period of changing physical environments and changing patterns of exploitation. 

(2) Minimalistic dynamics of exploited open-access resources.  This is an attempt to achieve a dynamical coupling of ecology, economics and social interactions.  The work is motivated by the need to think across traditional subject boundaries to tackle the problems raised by human activities in natural ecosystems. 

(3) Selection pressures and genetic change caused by exploitation.  If we are to pass on healthy marine ecosystems for the future, the longer-term effects fisheries-induced evolution need to be understood and properly managed.

host: Dr Charles Paxton

refID: 476

hide details

host: Dr Theoni Photopoulou

refID: 456

hide details

Portfolio selection and hedging problems in finance require the quantification of the risk arising from extreme events involving a collection of variables. I will describe statistical methods which aim to combine financial structure, empirical findings and asymptotically motivated semi-parametric extreme value models. The methods will be applied to closing price stock market data from different countries.

host: Dr Charles Paxton

refID: 458

hide details

host:

refID: 453

hide details

Temporal turnover is a fundamental characteristic of ecological communities and underpins key concepts, such as MacArthur and Wilson's theory of island biogeography. However, in contrast to spatial patterns of biodiversity - which have received a great deal of attention in recent years - we know rather little about changes in diversity through time. This is an important omission since we need to understand the background rate of change in ecological assemblages if we are to evaluate the consequences of extrinsic events, such as climate change. In this seminar I will explore temporal patterns in plankton and fish data sets and examine the relationship between changes in overall structure, and variation in the abundances of the individual species.

host: Dr Charles Paxton

refID: 451

hide details

Controlled laboratory studies have shown that social learning, resulting in social transmission, is common in a wide range of animal taxa. This has lead to claims of animal cultures in some species, such as chimpanzees. However, skeptics claim that we cannot be sure that social transmission is responsible for the observed behavioural variation. To resolve this debate, we need methods to detect social transmission in populations of animals. In this talk I describe two statistical methods we have been working on to resolve this issue: detecting "option bias" and inferring social transmission from the in which individuals acquire a behavioural trait.

host: cgp2@st-and.ac.uk

refID: 448

hide details

The main body of theory concerning population dynamics and community structure is based on population models that only account for reproductive processes and for mortality, either from predation or from other causes. A major process that is not considered by these classical approaches is the fact that most if not all individual organisms develop during their life history and often grow significantly in body size during this development. In addition, this growth in body size is often food-dependent. If both reproduction as well as development depends on food-density, one of these two processes may turn out to limit population growth more than the other. As a consequence, a population in equilibrium can be regulated in two distinct ways: either through total population reproduction as limiting process or through total population maturation. We will show that in case of reproduction regulation the population is dominated by adult individuals and that any increase in mortality may lead to an increase in juvenile biomass. Similarly, in case of maturation regulation the population is dominated by juvenile individuals and increases in mortality may increase adult biomass. This overcompensation in biomass occurs with increases in both random and stage-specific mortality, even when the latter targets the stage exhibiting overcompensation. Such biomass overcompensation in prey species may have important community consequences, as it may give rise to emergent Allee effects for stage-specific predator species or to facilitation between two stage-specific predator species if these predators exploit different stages of prey.

host: Dr Charles Paxton

refID: 428

hide details

Environmental history looks both at what a natural resource has meant to human society and what human exploitation has done to the resource. The Bass gannets have a documentary record that goes back to before 1500, involving a degree of scientific observation almost from the start. In the last 500 years they have been variously exploited for their fat, feathers and flesh, for the fun of killing them and for the fun of observing them.Mostly this has been extremely bad for the colony, and an attempt is made to estimate its size back to the sixteenth century. This is a paper that will provide a rest from strict modelling!

host: Dr Charles Paxton

refID: 343

hide details

Abstract to follow

host: cgp2@st-and.ac.uk

refID: 333

hide details

Abstract:

 Program MARK, a Windows Vista or XP program, provides  parameter estimates from marked animals when they are re-encountered at a later time. Re-encounters can be from dead recoveries (e.g., the animal is harvested), live recaptures (e.g. the animal is re-trapped or re-sighted), radio tracking, or from some combination of these sources of re-encounters. The time intervals between re-encounters do not have to be equal, but are assumed to be 1 time unit if not specified. More than one attribute group of animals can be modeled, e.g., treatment and control animals, and covariates specific to the group or the individual animal can be used. The basic input to MARK is the encounter history for each animal.

MARK can also provide estimates of population size for closed populations, including robust design versions of multi-state models. Parameters can be constrained to be the same across re-encounter occasions, or by age, or by group, using the parameter index matrix (PIM). Besides the logit function to link the design matrix to the parameters of the model, other link functions include the log-log, complimentary log-log, sine, log, and identity. Program MARK computes the estimates of model parameters via numerical maximum likelihood techniques. The FORTRAN program that does this computation also determines numerically the number of parameters that are estimable in the model, and reports its guess of one parameter that is not estimable if one or more parameters are not estimable. The number of estimable parameters is used to compute the quasi-likelihood AIC value (QAICc) for the model. Outputs for various models that the user has built (fit) are stored in a database, known as the Results Database. The input data are also stored in this database, making it a complete description of the model building process. The database is viewed and manipulated in a Results Browser window. Summaries available from the Results Browser window include viewing and printing model output (estimates, standard errors, and goodness-of-fit tests), deviance residuals from the model (including graphics and point and click capability to view the encounter history responsible for a particular residual), likelihood ratio and analysis of deviance (ANODEV) between models, and adjustments for over dispersion. Models can be retrieved and modified to create additional models. Context-sensitive help screens are available with Help click buttons and the F1 key. The Shift-F1 key can also be used to investigate the function of a particular control or menu item. Help screens include hypertext links to other help screens, with the intent to provide all the necessary program documentation on-line with the Help System. Recent additions to MARK include Bayesian inference with a Markov Chain Monte Carlo algorithm, new data types including mark-resight models and multi-state models with uncertain identification of states, and the capability to run all possible subsets of a set of variables in the design matrix.

host: Dr Charles Paxton

refID: 410

hide details

abstract to follow

host: Dr Charles Paxton

refID: 341

hide details

Increasingly, distance sampling is being used in designed experiments, to assess the impact of a treatment on animal densities.  We consider the issues that arise, and possible solutions.  We illustrate the approach using a before-after control-impact (BACI) study, in which the effect of burns in ponderosa pine forest on bird densities was assessed.

host: Dr Charles Paxton

refID: 340

hide details

abstracts to follow

host: Dr Charles Paxton

refID: 337

hide details

This talk presents an update on ongoing work, aimed at using particle filtering algorithms to fit Bayesian state-space models for the population dynamics of British and Canadian grey seals to data on the number of pups born each year. The model for British seals now fits the data reasonably well; that for Canadian seals is rather less impressive, but I'm the man with the plan (well, we have some ideas, anyway). There have also been lots of recent developments in the sphere of fitting high-dimensional models to data using computer-intensive methods, and I'll provide some pointers to these.

host: Dr Charles Paxton

refID: 336

hide details

Household surveys have typically been considered inappropriate to estimate the prevalence of problem drug use, such as the use of opiates such as heroin. The capture-recapture method, which was originally developed to estimate the size of animal populations, has been used in various settings across Europe to estimate the size of drug using populations. The capture-recapture method combines data from three or more sources, such as drug treatment agencies and the police and fits a log-linear model to the aggregated overlap data. To obtain the overlap pattern, the method requires individual-based data and therefore can involve substantial data collection. Another method, known as the multiple indicator model, can also be used to estimate the prevalence of drug use where it is not possible to use the capture-recapture method.

In a Home Office funded study in England, estimates of the prevalence of opiate and / or crack cocaine use in all 149 local areas were derived using either the capture-recapture method or a multiple indicator model. In this presentation I will discuss this study, in particular the approach taken in model selection and deriving confidence intervals.

host:

refID: 328

hide details

In forest ecology, many open questions are related to per species
behaviour and 'local' scale behaviour of the trees. To answer these
questions one needs mathematical tools which account for the spatial
structures present in the forest population.

In this talk I will show an example of a rainforest dataset and
discuss how the framework of spatial point processes can help in
answering the spatially challenging ecological questions.

host:

refID: 317

hide details

The Zackenberg Ecological Research Operations (ZERO) and its sister project Nuuk Basic are large scale monitoring programmes set up with the dual aim of investigating patterns and processes in undisturbed high and low arctic ecosystems, and to monitor their response to climate change. Regional climate systems, such as the North Atlantic Oscillation influence local weather conditions, with profound effects for ecosystem functioning in the region. The population dynamics of species in the Arctic is strongly affected by the seasonality of the region, and considerable inter-annual variation in snow cover, snow melt and onset of spring pose challenges for resident and migrating species which need to time their reproduction to a short Arctic summer. Extended snow cover may severely delay growth and flowering of plant species, which in turn affects the spatial distribution of herbivores. Likewise, variations in snow cover affect the availability of suitable nest sites for waders and the spatial distribution of predators and prey.

host: Dr Charles Paxton

refID: 290

hide details

TBA

host: Dr Charles Paxton

refID: 291

hide details

From a flat, floating disk to the spinning sphere we call home, the image of the earth has changed dramatically over time. In the days of the ancient Greeks, the questions of the earth's shape, its position in the cosmos, and the picture of the inhabited world were debated by philosophers and astronomers alike. The aim of this talk is to provide a brief chronology of the Greeks' view of their world, from the earliest known cosmographies, to the astronomical and geographical theories that would endure for over a thousand years after the decline of the Greek civilization.

http://www-maths.mcs.st-andrews.ac.uk/homepages/dm433.html

host: Dr Charles Paxton

refID: 271

hide details

I shall present a simple log-linear model for series of regular bird counts at a number of survey sites. The model is parameterized in terms of growth rates rather than abundance; this permits easy estimation of these parameters and allows us to model the effects upon population growth of covariates. The latter might be local presence of a competitor or predator, or some management regime, which may themselves vary in space and over time. A recursive relationship permits the expected count at a site to be functionally dependent upon the expected count at the previous visit, rather than replacing the latter with their observed counterparts, as has been used previously.

host:

refID: 280

hide details

Density Estimation for Cetaceans using passive Acoustic Fixed sensors (DECAF)

host: Dr Tiago Marques

refID: 249

hide details

The aim of this work is to estimate what proportion of birds breed at various ages. Existing methods such as Pradel cannot be used, as various assumptions are violated. Therefore, I'm trying to estimate the number of pairs which nest in a given breeding season, but are not detected, using closed population mark-recapture methods within the breeding season. I will compare the initial results, and hopefully spark discussion on other possible approaches to this problem.

host: Dr Charles Paxton

refID: 268

hide details

This seminar will explore if statistics can be used to analyse topics on the fringes of science/in the historical record? Can the plural of "anecdote" actually be "data"? I will debate statistical approaches to three different problems: species discovery curves, reports of seemingly unknown species of large marine animals since the 18th century  and the assassination of Abraham Lincoln.This is very much a talk about work in progress and work about to begin. Ifs, buts and caveats will all be debated. Audience participation is encouraged.

host: Dr Charles Paxton

refID: 145

hide details

In this talk we address various issues which arise in the classical parental care conflict, which have, should or could be modelled with game theoretic models. After a brief summary of fundamental concepts in game theory, aspects such as cooperation, benefit and welfare consequences for the young will be put into a game theoretic context and various models will be discussed. These models have been developped in joint work with John McNamara and Alasdair Houston.

host: Charles Paxton

refID: 144

hide details

In the last ten years, the dramatic increase in size of deer and wild boar populations all-over Italy and their impact on native woodland, agricultural land and other wild species have made efficient monitoring essential for an effective management strategy. However, these animals are elusive, nocturnal and prefer densely vegetated habitats, reducing the detection probability and the efficiency of standard survey techniques.

host: Dr Charles Paxton

refID: 146

hide details

Rhododendron ponticum is a serious invasive alien plant in the British Isles and is of considerable conservation and economic concern. While optimal control strategies for single individuals and small stands of R. ponticum are well described, effective regional control of the plant demands an improved understanding of its spatial dynamics, in particular its dispersal ecology. I will describe the results of two field experiments designed to quantify the dispersal pattern of R. ponticum seeds: 1. controlled release over a few seconds at known windspeeds and 2. natural release over the peak dispersal period. We then use these results to assess the potential use of two different mechanistic wind dispersal models (WINDISPER and WALD) as descriptors of seed dispersal ecology for this species.

host: Dr Charles Paxton

refID: 147

hide details

Formulating mark-recapture models as state-space models allows an embedded population dynamics model to be specified. This aims to ensure that estimated changes in the population are consistent with what is assumed to be known about the population biology of the species being modelled.When formulating these state-space models in order to model capture histories we can adopt one of two approaches. Firstly, we consider a 'conditional' approach in which the state of an individual animal is determined not only by characteristics such as age and gender but also by its capture history. Therefore, parameters relating to the probability of capture will appear in the vector of state parameters. Secondly, we consider an 'unconditional' approach in which the capture histories are regarded as observations. Consequently, the capture histories do not influence the state of an animal. Therefore, the parameters relating to capture probability appear in the vector of observation parameters.The conditional approach is so named because for a specific time period, we condition on the known numbers of animals possessing capture histories which include capture in that time period. Under this approach there is no observation error in the model and stochasticity enters only through the uncertainty in the numbers of animals in state elements corresponding to capture histories that do not include capture in that time period. Under the unconditional approach, capture is treated as a stochastic observation process, and the capture histories are considered to be a single random realization of this process. This latter approach is more consistent with traditional mark- recapture methods.We describe how suitable state-space models can be specified, under both conditional and unconditional approaches, in order to simulate state vectors that satisfy the constraints imposed by the observed data and our knowledge of the biology of the population under study. We also discuss the implementation of particle filtering techniques for fitting these models under each approach.

host: creem

refID: 141

hide details

This talk is rooted in the field of mathematical population genetics and considers effects arising in the context of a selectice sweep: A genetic locus carries a strongly beneficial allele which has recently become fixed in a large population. With the quick fixation of the strongly beneficial allele, sequence diversity at partially linked neutral loci is reduced (Maynard-Smith & Haigh, 1974). As it seems not only sequence diversity at single neutral loci is affected but also the joint allelic distribution of several partially linked neutral loci. We studied the latter by means of a genealogical approach: Joint genealogies of selected and neutral loci can be described by the structured ancestral recombination graph, a certain continuous time stochastic process. In the regime of large selection coefficients , recent work by Durrett and Schweinsberg (2005) and Etheridge, Pfaffelhuber, Wakolbinger (2006) introduced a marked Yule tree as an approximation to the exact genealogy of a single neutral locus. We extended the Yule approximation to the case of two neutral loci. This led to the full description of the genealogy with accuracy of O ((log alpha^-2 ) in probability. As an application the expectation of Lewontin’s D as a measure for non-random association of alleles was calculated.

host: Dr Charles Paxton

refID: 148

hide details

Double-observer distance sampling methods are becoming increasingly widespread, especially for the estimation of marine mammal abundance from aerial and shipboard line transect surveys when detection of animals on the line is uncertain. Models that assume full independence between observers typically underestimate abundance in the presence of heterogeneity in detection probabilities. The assumption of point independence (in which independence is only assumed on the line or at the point) allows less biased estimation. We consider extending this idea to the case that independence is only assumed in the limit, as the probability of detection for an observer tends to unity. We show how this method can be used to reduce bias in abundance estimates, or to assess the assumptions of full independence or point independence. We test the methods by simulation, and illustrate them using aerial survey data for narwhals off West Greenland, for which the effect of heterogeneity is large, and for shipboard survey data of minke whales in the North Sea and adjacent waters, for which survey methods minimized any dependence between the observers.

host: creem

refID: 142

hide details

host: creem

refID: 140

hide details

Invasive species are environmentally and economically damaging and management of many species is currently inadequate. Biocontrol is one strategy whereby invasive plant populations can be managed. Using models we can explore complex interactions within and between plant and herbivore populations and incorporate environmental effects inherent in ecological systems, leading to management solutions that were perhaps not intuitively obvious at the beginning of the process. Disturbance is implicated as an important driver of plant invasions and I explore this concept using a very simple model, making the prediction that some times, even for disturbance adapted invaders, it doesn't matter. In other scenarios we can manage disturbance rather than weed removal itself to get control. Underlying all management decisions is the availability of cold, hard cash, I describe an innovative decision making framework which predicts optimal control strategies from the impact-density curve of the invaders and a number of ecological and economic parameters.

host: creem

refID: 139

hide details

Generalist pathogens, such as canine distemper virus (CDV), can have devastating effects on wild felid populations; in 1994 over 1000 Serengeti lions died from CDV. In order to understand the CDV transmission dynamics among lions in this ecosystem, we built a spatially-explicit mathematical model of CDV transmission based on pride location, demographic, and contact data from the Serengeti Lion Project. We use this model to ask whether lion populations can sustain CDV epidemics without repeated introduction from other sympatric carnivore species, such as hyenas and jackals. We find that that other carnivore species were likely involved in the 1994 epidemic. Inspired by these results, we built a multi-species model involving three theoretical carnivore species with differing social structure. We use this model to ask whether interspecific disease transmission could produce the observed patchy spread of CDV in the lion population. Our model suggests that when species with varying degrees of within- and between-group disease transmission are coupled, a patchy spatial spread is produced in the species that could not have sustained the epidemic on its own. The results from both models suggest that lions are a non-maintenance population for CDV and the reservoir for CDV is either domestic dogs alone or combined with other wild carnivore species.

host: creem

refID: 138

hide details

This seminar describes how to invest finite conservation resources to save endangered species. Using a case study of threatened Australian birds, the optimal investment strategy is insensitive to the timeframe of the analysis, but is very sensitive to the chosen objective and the available budget. Money is only allocated to critically endangered species if the budget is sufficiently large. The optimal decision is difficult to determine intuitively, so decisions about investing resources in conservation of species would benefit from using formal decision theory methods rather than, or in addition to, intuitive rules.

host: creem

refID: 137

hide details

Toe clipping is commonly used in population ecology to identify individual amphibians, particularly anurans (frogs and toads). My co-author (Michael McCarthy) and I have demonstrated a consistent, negative effect of toe clipping on the return rate of anurans, sparking controversy in the herpetological world. In this seminar, I will discuss the toe clipping story in the context of the modern scientific method, including issues of logic, statistical inference, the scientific review process and the ethics of field research

host: creem

refID: 136

hide details

The problem of estimating population total of tigers from pugmark measurements poses a number of interesting statistical questions that are not addressed through conventional methods of estimating animal abundance. Unlike elephants producing dung or birds producing nest, tigers do not produce pugmarks at a uniform rate. On the other hand, pugmarks do not carry as accurate and reliable information as fingerprints do. Is it still possible to estimate the population total precisely and accurately? If so, what should be the appropriate number of pugmarks to be collected? Which features of a pugmark carry useful information for discrimination? How can one cluster the pugmarks so that each group of pugmarks belongs to a distinct tiger? Is good clustering a prerequisite for good estimation of population total? Is it possible to account for tigers missed in the sample? How can one utilize positional information for tiger counting? How wide are confidence intervals for the population total? How can the estimation be improved? In this talk, I shall present the currently available answers to these questions, and identify areas where further investigation is needed.

host: creem

refID: 135

hide details

An extensive body of research exists on mathematical models for the optimal control of biological populations. To apply these models to real world situations, population monitoring is required to track population size. However, the cost of monitoring is rarely included when assessing management regimes. We consider finding a combined optimal management strategy for an invasive fox population which specifies when to monitor as well as when to control. In our model, the monitoring data are binomial observations of an index associated with the population size. We use bayesian updating to track the population index and embed this in a Markov decision process to investigate the value of monitoring.

host: creem

refID: 134

hide details

host: creem

refID: 133

hide details

host: creem

refID: 132

hide details

host: creem

refID: 131

hide details

It is sometimes said of controlled fusion that it has been promising unlimited cheap power for the last forty years, but always forty years in the future. I will outline the basic physics and the benefits this process might offer if it can be made to work. I will then discuss where the difficulties lie and what is the current state of progress, paying particular attention to ITER (International Tokamak Experimental Reactor), a 10bn Euro international project whose construction is just beginning in France. This is intended to come into operation in about ten years and demonstrate that reactor conditions can be achieved.

host: creem

refID: 129

hide details

The Key Largo woodrat is an endangered species, restricted to the hardwood hummock on northern Key Largo, Florida. It is typical of many animal populations for which abundance cannot reliably be estimated, but which need to be managed effectively if they are to avoid extinction. Previous methods of estimating abundance have focused on capture-recapture methods using a grid-based, robust design. Instead, we propose trapping point transects as a cost-efficient alternative monitoring method. The methodology and results of the first fieldwork season will be presented, along with many pictures and more snake stories.

host: creem

refID: 130

hide details

In this talk we'll take a wander through the history of Sudoku puzzles, looking at their mathematical origin as Latin squares. The further development of Sudoku type grids was prompted by the design of agricultural experiments, where one needs to separate the effects of the treatment from localised effects in the field. We'll see how a theorem on marriage can be used to solve Sudoku puzzles, and finish with a discussion of how to generate a random puzzle.

host: creem

refID: 128

hide details

Cognitive mapping skills are crucial to many species of animal, but the difficulties of studying cognitive maps under field conditions have hampered scientific understanding of environmental cognition. Data on natural travel patterns are abundant and easy to record, but hard to interpret when travel goals can be inferred only retrospectively. We present a method of finding change points in an animal's travel path, i.e. of deciding the point(s) at which the path became directed at a subsequently-reached location, for instance a feeding site. We illustrate how it can be tailored to particular problems, and how it can aid in determining the cognitive basis of route choice in animals.

host: creem

refID: 127

hide details

Modeling clustering in spatial point processes dates back as early as Bartlett (1964) and Ripley (1976) but only in 2000 did Badderley et al. develop cluster detection methods for processes with heterogeneous intensities. Recent literature has tested spatial point patterns for clustering whilst taking heterogeneous environments into account by using a control group, known to follow an inhomogeneous poisson process, to model the underlying intensity. Often in practice, however, both the cases and the control group are clustered, and the question of scientific interest is the amount of clustering of the cases relative to the controls. Therefore, we propose a method to test for significant differences in the levels of clustering between two processes whilst taking into account the first order heterogeneity that may exist. Existing methods are developed to allow for the control group itself to follow a clustered process whilst individual level covariates, observed at only event locations, can be taken into account to see the affect they have on clustering. Inference and diagnostics are based around the inhomogeneous K-function and methods developed are demonstrated using data from a large study of tropical trees in Sri Lanka.

host: creem

refID: 126

hide details

Classical solutions to inverse and ill-posed problems require regularization, i.e., an attempt to solve a stable variant of the original unstable problem. Theoretically, a successful regularization procedure results in finding the true solution. The aim of this talk is to discuss the classical approach to solving inverse and ill-posed problems in light of data and model uncertainty. A simple example will be presented to exemplify key points.

host: creem

refID: 125

hide details

The triennial EURING Technical Meetings are the closest thing we currently have to an international conference on statistical ecology. The focus is on the analysis of data from marked individuals, particularly birds, but data from non-marked animal populations are also considered. The latest meeting was held in January in Dunedin, New Zealand. In this very informal seminar, I will try to give a flavour of what went on, including highlights from each session. If you want me to cover any particular talks in more detail, please email me in advance - the programme and abstracts are available at http://www.phidot.org/euring/main.html. Note that some of these speakers will, I hope, be giving NCSE seminars over the next year, so any feedback on which you'd like to hear would be useful. I will also show a few holiday snaps.

host: creem

refID: 124

hide details

In recent years there has been increasing concern regarding interactions between recreational activities and the threatened southern sea otter (Enhydra lutris nereis) in California. Kayaking is of particular concern as this activity occurs at high levels along the northern Monterey Peninsula. The current study employed multidisciplinary methods over a four year period to investigate this issue. Interviews with stakeholders determined attitudes toward otters and identified whether relationships between stakeholders contributed to disturbance. A survey of kayaking participants assessed motivation, tourist types, attitudes toward management and satisfaction with the kayaking experience. These social research methods were used as previous studies have shown that involving stakeholders can improve management. Observations were made of sea otter interactions with recreational vessels over a three year period. Disturbance to sea otters from recreational vessels was manifested as short-term changes in otter behaviour including increased vigilance or avoidance of disturbance sources. These changes had impacts on otter time-budgeting which may lead to increased energy expenditure. Kayaks caused most disturbance events, although less than 25% of kayaks provoked a response from otters. Direct and very close approaches were identified as particularly disturbing. There was some evidence of tolerance in otters exposed to high levels of disturbance. Potential causes of disturbance to sea otters were identified as: lack of awareness by some management agencies and tour operators regarding occurrence of disturbance events, insufficient information on the potential impacts of disturbance on sea otters, poor communication between agencies and inadequate public education. The data collected have been used to improve collaboration, communication and information dissemination between stakeholders and to develop wildlife watching guidelines for kayakers. It has also contributed to marine wildlife watching guidelines being developed for the wider state of California. This is the first study to use multidisciplinary research methods to identify underlying social and biological causes of wildlife disturbance and to use this information to improve educational programmes and wildlife watching guidelines. These methods may also be of benefit in identifying and addressing causes of wildlife disturbance in other locations.

host: creem

refID: 123

hide details

During the last five years it has become increasingly apparent that there has been underreporting of landings of NE Atlantic Mackerel. The stock which occupies an area along the shelf break from Portugal to Norway is currently fished by 14 European nations, in national and international waters. In 2005 regulatory authorities in UK and Ireland have investigated vessel landings and processing factory records. The results from UK indicate that landings in to UK have exceeded reported landings and the additional quantities found in this single country amount to between 8-10% of the total quota allocated to all countries. NEAFC, (North Eastern Atlantic Fisheries Commission) the management body for this stock has asked ICES to evaluate the extent of estimates of unaccounted mortality in fisheries for NEA mackerel to provide estimates of historic Spawning Stock Biomass and Fishing Mortality that would be compatible with these estimates of unaccounted mortality. Annual age structured data from the fishery, five triennial egg surveys and 19 years of tag and recovery data on total mortality at age have been used in a WINBUGS based population model, to try to estimate total mortality and fishing mortality for this stock. The model is a development of the population model currently used for the annual assessments. The estimates of unknown removals from the stock are presented and the sensitivity of these to the different models are explored. These results are presented in the context of Bayesian P values and DIC criteria. The influence of different sources of data are discussed to see if further work might be justified. The form of final conclusions are discussed.

host: creem

refID: 122

hide details

SCANS-II took place in July 2006 with the aim of estimating absolute abundance for harbour porpoise, several dolphin species and minke whales. This talk will describe the theoretical and practical aspects of the shipboard survey and whether the statistical theory held up in practice.

host: creem

refID: 121

hide details

Species exist in a web of interactions with other species. Despite this, population and evolutionary ecology has typically studied pairs of interacting consumer-resource organisms (e.g. predator-prey, parasitoid/parasite/pathogen-host). While the study of pairs of interacting species has led to many modern theories and applications of consumer-resource ecology, recent reports have highlighted the importance of studying the effects of multiple interacting species as they reveal novel and unanticipated dynamics. Here I report on single, 2- and 3-species communities and show effects of an insect predator and an insect pathogen on population size, dynamics, cycle period and variance of a shared prey population. Furthermore I demonstrate that in line with a recently published model of insect outbreaks, the period of time within and between peaks in prey abundance are determined by switches in the relative importance of interactions of the generalist predator, the specialist pathogen and the environment of their shared prey population.

host: creem

refID: 120

hide details