RSPB Centre for Conservation Science, Edinburgh, United Kingdom
Every autumn, millions of Afro-palearctic migratory birds make the long journey from their breeding grounds in Europe to wintering grounds in sub-Saharan Africa, returning north again in the spring. Over the last 50 years, the European breeding populations of many of these species have been declining, but the reason(s) for this remain unknown. The annual cycle of migratory species means that they could be vulnerable to factors acting on their breeding, wintering and/or staging grounds.
To date, most research into the population declines has focused on breeding grounds, largely due to the greater availability of data and resources here than elsewhere along the flyway. This has improved our understanding of species ecology and the factors affecting breeding, but hasn’t yet yielded any explanation for such widespread declines. We wanted, therefore, to look at the other part of their range – the wintering grounds in sub-Saharan Africa – to see whether there were any changes here that could help explain why so many species are in decline.
Migration is energetically expensive and requires sufficient food availability prior to departure, to fuel the journey, and on arrival in order to recover. Changes in the timing of food availability, with changing land management practices, habitat degradation or climate, could result in birds departing for migration in poorer condition, or arriving too early or too late to replenish reserves at the end of their migratory journeys.
In our paper published in Remote Sensing in Ecology and Conservation, we made use of a newly available dataset on vegetation phenology, based on NDVI, which allowed us to look at the timing of growing seasons in Africa and how they had changed over the last 34 years. This enabled us to map, for the first time, trends in key phenology variables such as the start and end dates of growing seasons, and rates and lengths of green-up and senescence.
However, relating these phenology changes to bird population changes across large areas of sub-Saharan Africa presented several challenges. The greatest limitation in our analysis undoubtedly arose from the paucity of data on the distribution and population trends of migrants in Africa. Instead, we used breeding population data collected in Europe and range maps from BirdLife International to plot European population trends onto African wintering grounds. Even within Europe, data on population trends of migrants were incomplete, meaning that we had to use an imputation method to fill in the gaps. Although we used the best available maps of species’ non-breeding distributions, these varied greatly in accuracy between species, as the wintering distributions of many migrants are still poorly known.
Despite these caveats and constraints, our study represents an important first step in bringing together long-term population trend data of migrating birds in Europe with remotely sensed data on changes in vegetation and seasonality across species’ wintering ranges in Africa. Importantly, and in contrast to most past studies, we did not assume that uniform relationships applied across the whole study area. Instead we used geographically weighted regression to identify how relationships between bird populations and environmental conditions varied across the flyway. Our results reveal a very complex picture of spatially varying relationships. For example, senescence variables were particularly significant across the Sahel and sub-Sahelian region, whilst greening variables appeared to have most influence in southern Africa. Perhaps the timing and the rate at which vegetation dies off after rain are important in the Sahelian region because they influence food availability in spring in areas where birds fuel up before crossing the Sahara. In contrast, ‘green’ conditions in southern Africa may be important because they coincide with the mid wintering period when birds are often further south. Furthermore, groups of species with different habitat associations showed different responses to changes in conditions. While perhaps not unexpected, this does highlight the simplicity of some previous studies that have treated the continent as a whole, or used only cumulative NDVI as a measure of vegetation, without considering phenology.
With advances in satellite remote sensing and the tools available to process data at high spatial and temporal resolutions, the limitation is no longer in the availability of environmental data, but in our knowledge of the non-breeding ecology of migratory birds. Better data on the non-wintering populations and distributions of migrants, including improved knowledge of their migratory connectivity and within-season movements, are needed to fully determine the extent to which they may be impacted by changes in conditions in Africa. This gap in knowledge can only be addressed through fieldwork in non-breeding areas, and the use of new and emerging tracking technologies to identify migratory pathways can help target this fieldwork.
A recent rise in the use of tracking technologies has already begun to greatly improve our understanding of species’ migratory routes and use of stopover and wintering sites. If we are to take advantage of these opportunities and make best possible use of these data as they emerge, cooperation and collaboration will be required. The seabird tracking community are leading the way, with collation of data in a centralized tracking database (seabirdtracking.org) enabling identification of global priorities. An equivalent database for migratory land birds would be a big step forward.
Combining this information with satellite-derived vegetation and phenology data, such as that synthesized in this study, could be a powerful tool in understanding the influence of phenology throughout the entire lifecycle of migratory birds. It could also reveal the extent to which species are flexible in their choice of wintering grounds, and whether distributions change between years and/or within seasons, in relation to environmental conditions. Until then, we suggest that it is no longer assumed that there is a straightforward relationship between migrant populations and NDVI across species’ ranges, and that studies should, at the very least, consider the multiple attributes of phenology and the potential for spatial variation in relationships.