Animal migration is an intriguing phenomenon which is studied by many ecologists. I am especially interested in the mechanisms that determine the costs and benefits of bird migration for an individual bird during its lifetime as well as a population over several generations. Currently I am investigating the way by which birds use (knowledge about) wind patterns and thermals to minimize the cost of their migratory journey.

Collaborators

## Long distance migration

Recently Judy Shamoun, Jutta Leymer and others (including me) completed a study where the variation in abundance of red knots at a site in France between their departure site in Western Africa and their destination in the German Wadden Sea is explained by the amount of wind support (see the paper).

Simulated spring knot migration for 1985 and 1986 at different pressure levels along the great circle route from Mauritania (circled 1), through a potential staging site in France (circled 2) to their main stopover in the German Wadden Sea (circled 3). Each segment in the trajectories marks a 6 hour time interval. At the right the probability density functions of cumulative flight time up to different latitudes are shown over all years that were studied (1979-2007).

I developed the model for this study together with Judy. It calculates the expected migration route, depending on some predefined air-speed and waypoints via which the species is supposed to travel. The model is coded in matlab and I called it 'LoDiMi'. LoDiMi uses gridded wind data at a resolution of 2.5 degree lat/lon and 6 hours (or more detailed). We often use NCEP/NCAR Reanalysis 1 data when studying long distance bird migration with LoDiMi. The trajectories between waypoints are calculated along a great circle route or a leading line. A theoretical or empirical wind compensation function describes how birds compensate for drift. This function allows to describe anything between complete, partial or no compensation (as well as e.g. asymmetrical responses). The strengths of LoDiMi are that it:

• is simple and contains few parameters that may be fitted to observational data (because of its simplicity the derivation of the wind compensation function on the basis of suitable observations and e.g. a bayesian technique is acutally feasible);
• interpolates the wind field to the point where the value is needed, regardless of the chosen parameters or the resolution of the available wind data.

LoDiMi does not explicitly model stopover decisions and navigation (these are input to the model in the form of waypoints and either a great circle route or leading line). By using ocean current data and appropriate parameters, LoDiMi can describe migration in marine systems similar to bird migration. If you like to use the model for your questions and data I am happy to provide you with the code and collaborate - just send me an e-mail. Currently we are trying to use LoDiMi to beter interpret tracks of long-distance migrants. We think that they can adhere to simple rules in their response to the wind conditions to reach their destination. By integrating observational data via both state and parameter adjustments in LoDiMi, we hope to unravel some of these behavioural rules.

## Migration by soaring birds

Birds that use soaring flight for migration require thermal convection or orographic lift to climb in altitude and then glide in the migration direction; after a gliding phase they need a new thermal or source of uplift to climb again. Although a lot of observations have been collected on soaring migratory flight by sighting from the ground, from motorized gliders, via GPS-transmitters and tracking radar, a synthetic view of the mechanisms leading to the observed behaviours and migration patterns is lacking. We do for instance not know how and to what distance soaring migrants can detect thermals, whether flocking has a function in enhancing migration efficiency, or how cross-country migration speed is affected by thermal strength and density. To make such a synthesis possible I have integrated our knowledge about soaring bird migration in a dynamic individual based model, jointly with others at the CGE-group. Simsoar, as the model is called, is described in this paper.

The next step I would like to undertake is to use Simsoar for planning our observations with UvA GPS-tags on migration of the white stork and honey buzzard. Subsequenly I will try to estimate parameters in Simsoar and identify decision rules on thermal choice through the data collected over the past years. I am eager to calibrate and test Simsoar on fine resolution (order of minutes) GPS-tracking data of other species as well. So if you have this type of data for a migrating soaring bird, preferably in combination with good quality meteo-data, I would like to collaborate and see how far we get.

Simsoar is coded in matlab and you can download the model-code here. If you need help with the model, just send me an e-mail, and I will see what I can do.

## possibilities for MSc thesis research

Currently I am offering an MSc thesis-research opportunity during 2010 and 2011 in relation to soaring bird migration, starting anywhere after May 2010. The thesis research deals with the migration-tracks that we are collecting for white storks and honey buzzards with our GPS tags. It comprises data analysis, comparison of results with theories on soaring bird migration and, if time permits, improving the Simsoar model. If you are an MSc student (regardless from UvA or anywhere else) and would like to work on this, just send me an e-mail to express your interest. I will supply you with a detailed project description. This may be a great project for you especially if you like to process and understand quantitative ecological data, have an analytical mind, and are willing to learn novel computational techniques. You will be supervised intensively and acquire a lot of research skills, but be prepared that I am demanding when it comes to the quality and timely completion of your work.