We apply modern regression methods (adapted to spatial, temporal, or spatio-temporal data) such as GLMs, GLMMs, GAMs, or GAMMs to estimate linear and nonlinear trends, population sizes or to visualize distribution patterns. These models can be adapted to various types of data, such as presence-absence data, count data, catch-recatch data or presence-only data.
In the last decade, new wildlife telemetry and tracking technologies have become available. This has led to an impressive growth in the volume of GPS or telemetry tracking data. However, the statistical analysis of these data is challenging, e.g. since they belong to the class of ''presence-only-data'' with high spatio-temporal autocorrelation.
Our research is concerned with the statistical analysis of such tracking data, comprising the following bricks:
Does the animal significantly prefer certain habitats? Does the population significantly avoid certain anthropogenic structures (such as wind turbines)? These questions can be answered with appropriate regression techniques specialized for tracking data. We further developed and adapted a new emerging end increasingly used class of models called ''point process models'' (PPM's) to animal tracking data. These models overcome the statistical disadvantages of the traditionally used ''pseudo-absence models'' and can be combined with mixed or additive modeling.
Which are the regions most frequently used by an animal or a population? There are several methods for the estimation and visualization of home ranges. Each method, however, requires certain assumptions regarding the data. We help to choose and apply the most appropriate method for your data and research question.
Animals usually show different behavioral states (such as ''foraging'', ''resting'', or ''migration''). The automatic categorization into these states based on GPS / telemetry tracking data can be of great interest, e.g. since each of these states is related to specific covariates. Also here, several methods exist and ach method requires certain assumptions regarding the data. We help to choose and apply the most appropriate method for your data and research question.
Sometimes, tracking data are either sparse or show strong uncertainties (e.g. if ARGOS-based data are used). We use different methods (such as state-space-models) to interpolate the most probably true track of the animal. This can be combined with subsequent (PPM) regression analyses.
Does an anthropogenic structure significantly reduce animal densities? To answer such questions, the true impact of the structure has to be distinguished from natural density fluctuations in space and time.