PDF of the announce
PhD project in spatial statistics and stochastic geometry
Spatio-temporal modeling and simulation of agricultural landslides, with an
application to pest regulation
Thomas Opitz, Julien Papaı̈x (BioSP, INRA Avignon, France); Édith Gabriel (Université
Motivation : agro-ecology
In agroecosystems, species interactions between pests and auxiliaries are often strongly influenced
by the proportion and spatial organization of semi-natural elements of agricultural landscapes.
By conducting sensitivity analyses for biological control strategies of pests with respect to the
composition and configuration of agricultural landscape elements, it will be possible to assess the
effect of such interactions and to improve existing strategies of biological control.
Research questions : stochastic geometry and sensitivity analysis
This PhD project proposes to utilize methods from the field of stochastic geometry to model and
simulate realistic agricultural landsapes through the development of models based on a small
number of interpretable parameters characterizing the landscape structure. Mathematically, we will
represent a study region as a tessellation of space, whose edges and cells are the spatial support for
other elements such as hedgerows (i.e., linear segments) and crop types. A first goal of the thesis
will be to propose realistic stochastic generator models for linear segments (hedges, ditches,
grassy field margins...), and for the occupation of agricultural fields by various crops, while
taking into account crop rotation. Our approaches will be based on censored Gaussian random
fields and on marked Gibbs point processes.
In the following step, in order to model the spatial distribution of species, we will couple partial
differential equations (PDEs) in 2D for modeling the population dynamics within the fields with
PDEs in 1D for the dynamics on linear elements. This approach allows studying the size and the
spatial distribution of the species populations with respect to geometric characteristics of the land-
scape such as the density and the connectivity of the hedgerow network. By simulating a multitude
of scenarios, the sensitivity of population dynamics with respect to landscape structure
will be studied. The framework of stochastic simulation makes it possible to quantitatively assess
the part of uncertainty related to landscape.
In the applied part of the thesis, the methodological results will be applied to the study region
“Basse Vallée de la Durance” with the pest “codling moth” and its auxiliary species in apple
Candidate profile : Statistician with interest in spatial modeling and ecology.
Knowledge of fundamentals of spatial statistics and/or stochastic geometry and of the R program-
ming language is appreciated.
The successful candidate will become part of the Biostatistics and Spatial Processes lab (BioSP)
of INRA Avignon. The starting date is October 2018. This PhD project will be supervised by
Julien Papaı̈x (julien.papaix [a] inra.fr), Thomas Opitz (thomas.opitz [a] inra.fr) and Edith Gabriel
(edith.gabriel [a] univ-avignon.fr). To apply, please contact the supervisors by providing a detailed
CV, a motivation letter and a recommendation letter from at least one former supervisor / professor.