Many latitudinal insect migrants, including agricultural pests, disease vectors and beneficial species, show huge fluctuations in the year-to-year abundance of spring immigrants reaching temperate zones. It is widely believed that this variation is driven by climatic conditions in the winter-breeding regions, but evidence is lacking. We identified the environmental drivers of the annual population dynamics of a cosmopolitan migrant butterfly (the painted lady Vanessa cardui), using a combination of long-term monitoring, climate and atmospheric data within the western part of its Afro-Palearctic migratory range. Our population models show that a combination of high winter NDVI (normalised difference vegetation index) in the Savannah/Sahel of sub-Saharan Africa, high spring NDVI in the Maghreb of North Africa, and frequent favourably-directed tailwinds during migration periods, are the three most important drivers of the size of the immigration to Western Europe, while our atmospheric trajectory simulations demonstrate regular opportunities for windborne trans-Saharan movements. The effects of sub-Saharan vegetative productivity and wind conditions confirm that painted lady populations either side of the Sahara are linked by regular mass migrations, making this the longest annual insect migration circuit so far known. Our results provide the first quantification of the environmental drivers of large annual population fluctuations of an insect migrant, and hold much promise for predicting invasions of migrant insect pests, disease vectors and beneficial species.

The dataset was derived from a 22 year's data (1994-2015) of the Butterfly Monitoring Scheme [BMS] records from the Mediterranean (northeastern Spain) and NW Europe (the United Kingdom and the Netherlands).  BMS data follow a standardized data collection (see more detail in: E. Pollard, T. J. Yates, Monitoring Butterflies for Ecology and Conservation (Chapman & Hall, 1993)). In this dataset, the population density of the painted lady butterfly were calculated for each record.  

The dataset contains six columns: transect_id, Year, Month, Day, Region and Population density.  Information on how the measurements were done can be found in the associated manuscript referenced above.