International Biometric Society
British and Irish Region

Statistical Modelling in the Environment with Special Reference to Biodiversity and Spatio-temporal Approaches

Paris
Wednesday May 24^th 2006

Together with the Environmental Statistics Sections of the French Statistical Society and the Royal Statistical Society and the French Region of the International Biometric Society we are organising a joint meeting to be held in Paris at The National French Institute for Rural Engineering, Water and Forest Management (ENGREF), 19 av du Maine, 75 732 Paris Cedex 15, France, tel +33 (0)1 45 49 89 30, near Gare Montparnasse (for location details, please see map at http://www.engref.fr/planfco.pdf).

There is no registration cost for attending the meeting (though please note that lunch is not provided – plenty of local establishments where lunch can be bought!), but participants are required to register their intention to attend the meeting by emailing pingeon@engref.fr.

It is hoped that the timing of the meeting will allow many delegates from the British and Irish Region to attend the meeting without needing to stay in Paris overnight (though of course you are welcome to do so!). A list of local hotels will be provided if requested when emailing to indicate your intention to attend. It is suggested that travel time from Paris airport to the meeting location should be around 1 hour, though you might wish to allow a little longer to include time for getting lost!

The proceedings of the meeting will be published in a special issue of the Journal of French Statistical Society.

Programme (and some abstracts)

(other abstracts will be posted to the IBS-British Region website as available)

10h25 – 10h30 Words of welcome
Cyrille Van Effenterre (Head of ENGREF)

10h30 – 11h00 State-space modelling of climate extremes
Philippe Naveau (LSCE)

Loss of life and economic damage from extreme weather and climate events has been recurrent in human history. Although our understanding of the mean behaviour of most climatic processes is well understood, the statistical modelling of extreme events in time and space remains a difficult mathematical challenge. This is mainly due to the intrinsic rarity of extreme events, their non-Gaussian amplitudes and the different spatio-temporal scales involved.

In this presentation, we will discuss new statistical models for one important geophysical research topic: data assimilation of extreme events (state-space modelling). The fundamental problem of data assimilation may be simply stated as follows: given the state of the atmosphere at one time, what is the state of the atmosphere at a later time if one knows the observational data with the underlying dynamical principles governing the system under observation? Mathematically, this corresponds to a state-space formulation in which the state equation drives the dynamics of the system and the observational equation integrates measurements with the state variables.

We aim at taking advantage of recent developments in the field of Extreme Value Theory (EVT) and to offer mathematically sound models.

This is a joint-work with Paul Poncet.

11h00 – 11h15 Coffee break

11h15 – 12h00 Contribution of spatial point process modelling to biodiversity theory
Janine Illian(Univ. of Aberdeen)

Ultimately, the earth as a system is dependent on the functioning of natural and managed ecosystems since it is regulated by the biogeochemical processes derived from them. Recent decades have seen an increasing decline in species' biodiversity as a result of human interference. The potential ecological consequences of biodiversity loss have led to a growing concern about the future survival of ecosystems and their functioning. Consequently, the relationship between biodiversity and ecosystem functioning constitutes a major scientific issue today. However, understanding the impact of biodiversity loss requires an understanding of the processes that organise ecosystem communities and the mechanisms that sustain biodiversity. Substantial research focuses on modeling structures and processes in plant ecosystems and communities. Plants are primary producers and hence represent the basal component of most ecosystems with many terrestrial ecosystems' survival and diversity depending on the resources plants provide as well as on their structures and diversity.

Key research in community ecology thus aims at revealing the mechanisms that allow a large number of species to coexist. Coexistence primarily concerns the inter- and intra-specific interactions in a community. Since individual plants interact mainly with their neighbours interactions between plants in plant communities typically take place in a spatial context and hence current mathematical modelling approaches consider individuals in spatially explicit models.

Spatial point process models are statistical models that describe the exact locations of objects in space. They model the pattern formed by these objects based on interaction between them and on their properties, taking potential spatial heterogeneity into account. The spatial pattern formed by the individuals in a plant community may be the result of species interaction and environmental heterogeneity. Spatial point processes may hence be used as models of plant communities allowing inference on interaction structures and dependence on local growing conditions.

In an attempt to provide evidence for or against the random drift theory, Stephen Hubbell and his colleagues established a 50 ha plot on Barro Colorado Island (BCI) in Panama in the early 1980s recording the locations and sizes (diameter at breast height) of 235,349 individuals of 304 (rainforest) tree species in 1982. In addition, a large number of soil variables has been collected. Since then, these have been repeatedly collected at regular times. Similar plots have been established in a network of 16 forest plots in several countries coordinated through Center for Tropical Forest Science (CTFS). This presentation describes how spatial point process modelling may contribute to the discussion and outlines current results in the context of the rain fort data sets and other plant communities.

12h00 – 12h30 Spatial models for marine resource assessment and management
Nicolas Bez (IFREMER)

12h30 – 14h00 Break for lunch in nearby restaurants

14h00 – 14h30 Panel discussion: Avner Bar-Hen, Liliane Bel, Marian Scott, Ruth King

14h30 – 15h15 Measuring biodiversity: using birds as indicators of environmental change in Europe
Richard Gregory, Petr Vorisek & Arco van Strien (RSPB)

In 2002 world leaders pledged 'a significant reduction in the rate of biodiversity loss by 2010' and these commitments have been echoed at regional and national levels. Meeting such commitments requires a way of measuring progress towards the targets and concerted conservation action to improve the status of species and their habitats. Measurement alone poses a considerable challenge because biodiversity trend information is often sparse and patchy in its coverage of species and ecosystems, and synthesis is rare. Delivering conservation action on a sufficient scale and with sufficient intensity to bring about change is a massive challenge. Here we use European birds as an example to show how workable environmental indicators can be constructed at national and continental scales, and show how they can then be interpreted. We have developed statistical methods to create national and supranational multi-species indices and indicators based on data from national annual land bird surveys in Europe. The resulting indicators show, for example, that common farmland birds have declined steeply across Europe over the last three decades. Evidence from elsewhere suggests that agricultural intensification is the prime driver of biodiversity decline on farmed land. We argue that, with some care, wild bird indicators of this kind can often provide a useful proxy for ecosystem health and trends in other elements of biodiversity. The purpose of such indices is to communicate with and aid decision makers in reviewing and formulating environmental policies. The indicators we describe have been adopted by national governments in Europe and by the European Union. Our work provides a template for other taxa, habitats and regions, and provides a step towards fully representative biodiversity indicators.

15h15 – 15h45 Stochastic models and statistical inference for pollen dispersal in homogeneous and heterogeneous environments
Catherine Laredo and Sylvie Huet (INRA)

Studying pollen dispersal has important implications in various biological fields: population genetics and ecology (for gene flows and populations structure), pollination biology (for reproductive systems and hybridization), agronomy (for genetic purity of crops and seeds), and recently escape of transgenes by pollen (in relation with Genetically Modified Organisms).

We investigate various stochastic models for wind dispersed pollens for homogeneous environments (Grimaud and Laredo, 2006 for corn pollen dispersal). Then we present an approach coupling these models with semi-parametric models for isotropic dispersions and heterogeneous environments (Milhem et al., 2005 for oilseed rape dispersion).

Using data from field experiments, a global statistical framework is developed taking into account competition in pollen clouds.

15h45 – 16h00 Coffee break

16h00 – 16h30 Bayesian Clustering using Hidden Markov Random Fields in Spatial Population Genetics
Olivier François, Sophie Ancelet, Gilles Guillot (IMAG)

We present a new Bayesian clustering algorithm for studying population structure using individually geo-referenced multilocus data sets. The algorithm is based on the concept of Hidden Markov Random Field which models the spatial dependencies at the cluster membership level. We argue that (i) a Markov Chain Monte-Carlo procedure can implement the algorithm efficiently, (ii) it can detect significant geographical discontinuities in allele frequencies, and can regulate the number of clusters, (iii) it can check whether the clusters obtained without the use of spatial priors are robust to the hypothesis of discontinuous geographical variation in allele frequencies, (iv) it can reduce the number of loci required to obtain accurate assignments. We illustrate and discuss the implementation issues with two datasets: the Scandinavian brown bear and the human CEPH data set.

16h30 – 17h00 Hierarchical Bayesian models accounting for spatial dependence and zero-inflation in sapling patterns: examples from a French Guianan rainforest
Fred Mortier (CIRAD)

17h00 – 17h45 Monitoring regional biodiversity: some statistical issues
Steve Buckland and E Rexstad (Univ. of St Andrews)

At the 2002 World Summit on Sustainable Development in Johannesburg, political leaders agreed to strive for 'a significant reduction in the current rate of loss of biological diversity', by the year 2010. This goal assumes that we can measure the rate of loss of biological diversity over large regions; otherwise, we have no means of assessing whether the goal has been achieved. We consider shortcomings of classical measures of diversity such as species richness, Simpson's index and the Shannon index for measuring trends. We also look at the role of composite indices, formed from combining species-specific estimates of trend in relative abundance. Issues considered include population definition, detectability, design of monitoring surveys, and alpha and beta diversity. We use Alaskan data from the North American Breeding Bird Survey to illustrate these issues.

17h45 – 18h00 Close of meeting
Ron Smith, Eric Parent

10h25 – 10h30	Words of welcome Cyrille Van Effenterre (Head of ENGREF)
10h30 – 11h00	State-space modelling of climate extremes Philippe Naveau (LSCE)
	Loss of life and economic damage from extreme weather and climate events has been recurrent in human history. Although our understanding of the mean behaviour of most climatic processes is well understood, the statistical modelling of extreme events in time and space remains a difficult mathematical challenge. This is mainly due to the intrinsic rarity of extreme events, their non-Gaussian amplitudes and the different spatio-temporal scales involved. In this presentation, we will discuss new statistical models for one important geophysical research topic: data assimilation of extreme events (state-space modelling). The fundamental problem of data assimilation may be simply stated as follows: given the state of the atmosphere at one time, what is the state of the atmosphere at a later time if one knows the observational data with the underlying dynamical principles governing the system under observation? Mathematically, this corresponds to a state-space formulation in which the state equation drives the dynamics of the system and the observational equation integrates measurements with the state variables. We aim at taking advantage of recent developments in the field of Extreme Value Theory (EVT) and to offer mathematically sound models. This is a joint-work with Paul Poncet.
11h00 – 11h15	Coffee break
11h15 – 12h00	Contribution of spatial point process modelling to biodiversity theory Janine Illian(Univ. of Aberdeen)
	Ultimately, the earth as a system is dependent on the functioning of natural and managed ecosystems since it is regulated by the biogeochemical processes derived from them. Recent decades have seen an increasing decline in species' biodiversity as a result of human interference. The potential ecological consequences of biodiversity loss have led to a growing concern about the future survival of ecosystems and their functioning. Consequently, the relationship between biodiversity and ecosystem functioning constitutes a major scientific issue today. However, understanding the impact of biodiversity loss requires an understanding of the processes that organise ecosystem communities and the mechanisms that sustain biodiversity. Substantial research focuses on modeling structures and processes in plant ecosystems and communities. Plants are primary producers and hence represent the basal component of most ecosystems with many terrestrial ecosystems' survival and diversity depending on the resources plants provide as well as on their structures and diversity. Key research in community ecology thus aims at revealing the mechanisms that allow a large number of species to coexist. Coexistence primarily concerns the inter- and intra-specific interactions in a community. Since individual plants interact mainly with their neighbours interactions between plants in plant communities typically take place in a spatial context and hence current mathematical modelling approaches consider individuals in spatially explicit models. Spatial point process models are statistical models that describe the exact locations of objects in space. They model the pattern formed by these objects based on interaction between them and on their properties, taking potential spatial heterogeneity into account. The spatial pattern formed by the individuals in a plant community may be the result of species interaction and environmental heterogeneity. Spatial point processes may hence be used as models of plant communities allowing inference on interaction structures and dependence on local growing conditions. In an attempt to provide evidence for or against the random drift theory, Stephen Hubbell and his colleagues established a 50 ha plot on Barro Colorado Island (BCI) in Panama in the early 1980s recording the locations and sizes (diameter at breast height) of 235,349 individuals of 304 (rainforest) tree species in 1982. In addition, a large number of soil variables has been collected. Since then, these have been repeatedly collected at regular times. Similar plots have been established in a network of 16 forest plots in several countries coordinated through Center for Tropical Forest Science (CTFS). This presentation describes how spatial point process modelling may contribute to the discussion and outlines current results in the context of the rain fort data sets and other plant communities.
12h00 – 12h30	Spatial models for marine resource assessment and management Nicolas Bez (IFREMER)
12h30 – 14h00	Break for lunch in nearby restaurants
14h00 – 14h30	Panel discussion: Avner Bar-Hen, Liliane Bel, Marian Scott, Ruth King
14h30 – 15h15	Measuring biodiversity: using birds as indicators of environmental change in Europe Richard Gregory, Petr Vorisek & Arco van Strien (RSPB)
	In 2002 world leaders pledged 'a significant reduction in the rate of biodiversity loss by 2010' and these commitments have been echoed at regional and national levels. Meeting such commitments requires a way of measuring progress towards the targets and concerted conservation action to improve the status of species and their habitats. Measurement alone poses a considerable challenge because biodiversity trend information is often sparse and patchy in its coverage of species and ecosystems, and synthesis is rare. Delivering conservation action on a sufficient scale and with sufficient intensity to bring about change is a massive challenge. Here we use European birds as an example to show how workable environmental indicators can be constructed at national and continental scales, and show how they can then be interpreted. We have developed statistical methods to create national and supranational multi-species indices and indicators based on data from national annual land bird surveys in Europe. The resulting indicators show, for example, that common farmland birds have declined steeply across Europe over the last three decades. Evidence from elsewhere suggests that agricultural intensification is the prime driver of biodiversity decline on farmed land. We argue that, with some care, wild bird indicators of this kind can often provide a useful proxy for ecosystem health and trends in other elements of biodiversity. The purpose of such indices is to communicate with and aid decision makers in reviewing and formulating environmental policies. The indicators we describe have been adopted by national governments in Europe and by the European Union. Our work provides a template for other taxa, habitats and regions, and provides a step towards fully representative biodiversity indicators.
15h15 – 15h45	Stochastic models and statistical inference for pollen dispersal in homogeneous and heterogeneous environments Catherine Laredo and Sylvie Huet (INRA)
	Studying pollen dispersal has important implications in various biological fields: population genetics and ecology (for gene flows and populations structure), pollination biology (for reproductive systems and hybridization), agronomy (for genetic purity of crops and seeds), and recently escape of transgenes by pollen (in relation with Genetically Modified Organisms). We investigate various stochastic models for wind dispersed pollens for homogeneous environments (Grimaud and Laredo, 2006 for corn pollen dispersal). Then we present an approach coupling these models with semi-parametric models for isotropic dispersions and heterogeneous environments (Milhem et al., 2005 for oilseed rape dispersion). Using data from field experiments, a global statistical framework is developed taking into account competition in pollen clouds.
15h45 – 16h00	Coffee break
16h00 – 16h30	Bayesian Clustering using Hidden Markov Random Fields in Spatial Population Genetics Olivier François, Sophie Ancelet, Gilles Guillot (IMAG)
	We present a new Bayesian clustering algorithm for studying population structure using individually geo-referenced multilocus data sets. The algorithm is based on the concept of Hidden Markov Random Field which models the spatial dependencies at the cluster membership level. We argue that (i) a Markov Chain Monte-Carlo procedure can implement the algorithm efficiently, (ii) it can detect significant geographical discontinuities in allele frequencies, and can regulate the number of clusters, (iii) it can check whether the clusters obtained without the use of spatial priors are robust to the hypothesis of discontinuous geographical variation in allele frequencies, (iv) it can reduce the number of loci required to obtain accurate assignments. We illustrate and discuss the implementation issues with two datasets: the Scandinavian brown bear and the human CEPH data set.
16h30 – 17h00	Hierarchical Bayesian models accounting for spatial dependence and zero-inflation in sapling patterns: examples from a French Guianan rainforest Fred Mortier (CIRAD)
17h00 – 17h45	Monitoring regional biodiversity: some statistical issues Steve Buckland and E Rexstad (Univ. of St Andrews)
	At the 2002 World Summit on Sustainable Development in Johannesburg, political leaders agreed to strive for 'a significant reduction in the current rate of loss of biological diversity', by the year 2010. This goal assumes that we can measure the rate of loss of biological diversity over large regions; otherwise, we have no means of assessing whether the goal has been achieved. We consider shortcomings of classical measures of diversity such as species richness, Simpson's index and the Shannon index for measuring trends. We also look at the role of composite indices, formed from combining species-specific estimates of trend in relative abundance. Issues considered include population definition, detectability, design of monitoring surveys, and alpha and beta diversity. We use Alaskan data from the North American Breeding Bird Survey to illustrate these issues.
17h45 – 18h00	Close of meeting Ron Smith, Eric Parent

International Biometric Society British and Irish Region

Statistical Modelling in the Environment with Special Reference to Biodiversity and Spatio-temporal Approaches

Paris Wednesday May 24th 2006

Programme (and some abstracts)

International Biometric Society
British and Irish Region

Paris
Wednesday May 24^th 2006