Seasonal animal migration is a widespread phenomenon. At the species level, it has been shown that many migratory animal species track similar climatic conditions throughout the year. However, it remains unclear whether such niche tracking pattern is a direct consequence of individual behaviour or emerges at the population or species level through behavioural variability. Here, we estimated seasonal niche overlap and seasonal niche tracking at the individual and population level of Central European White Storks (Ciconia ciconia). We quantified niche tracking for both weather and climate conditions to control for the different spatio-temporal scales over which ecological processes may operate. Our results indicate that niche tracking is a bottom-up process. Individuals mainly track weather conditions while climatic niche tracking mainly emerges at the population level. This result may be partially explained by a high degree of intra- and inter-individual variation in niche overlap between seasons. Understanding how migratory individuals, populations and species respond to seasonal environments is key for anticipating the impacts of global environmental changes.

We trapped 62 adult white storks in the state of Saxony-Anhalt, Germany, and equipped them with solar GPS-ACC transmitters (e-obs GmbH; Munich, Germany) that weighed 55 g including harness, ca. 2% of the average stork’s weight (see  [25]). The transmitters recorded GPS fixes every 5 min when solar conditions were good (95% of the time) or every 20 min otherwise. This dataset include a set of maximum 100 GPS locations randomly selected per day and individual to estimate the seasonal niche and to avoid over-fitting the data to some locations.

In addition, each datapoint were associated to three type of environmental variables at two scales, weather and climate. For weather variables, the datapoints were annotated with environmental data of temperature (Land Surface Temperature & Emissivity 1-km Daily Terra; MOD11A1 V6), precipitation (ECMWF Interim Full Daily SFC-FC Total Precipitation; 0.75 deg.; 3 hourly)  and Normalized Difference Vegetation Index (NDVI; MODIS Land Vegetation Indices 1km 16 days Terra) using the Env-DATA track annotation tool of MoveBank. For the climate data, we used long-term averaged monthly temperature and precipitation patterns for the time period 1979-2013 at 1 km resolution (CHELSA), and monthly NDVI for the time period 1982-2000 (GIMMS AVHHR Global NDVI), and extracted the values of each variable for all selected points using the “raster” package.

 

There is two csv files for each environmental scale (weather and climate).

Both data files have the GPS information for the 100 randomly points selected for each individual and day. Longitude ("location.long"), Latitude ("location.lat"), date (day_format"), and id for the individual ("tag.local.identifier").

Besides, three columns for each climate variable:

1- For climate data,  tmean_value (temperature mean, from the CHELSA dataset), prec_value (precipitation mean, from the CHELSA dataset), and ndvi_value (monthly NDVI for the time period 1982-2000).

2- For weather data,  tmean_value (temperature mean, Land Surface Temperature & Emissivity 1-km Daily Terra; MOD11A1 V6), prec_value (precipitation from ECMWF Interim Full Daily SFC-FC Total Precipitation; 0.75 deg.; 3 hourly), and ndvi_value (MODIS Land Vegetation Indices 1km 16 days Terra).