© 2006 Nature Publishing Group © 2006 Nature Publishing Group Climate change and population declines in a long-distance migratory bird Christiaan Both 1,2 , Sandra Bouwhuis 1 †, C. M. Lessells 1 & Marcel E. Visser 1 Phenological responses to climate change differ across trophic levels 1–3 , which may lead to birds failing to breed at the time of maximal food abundance. Here we investigate the population consequences of such mistiming in the migratory pied flycatcher, Ficedula hypoleuca 4 . In a comparison of nine Dutch populations, we find that populations have declined by about 90% over the past two decades in areas where the food for provisioning nestlings peaks early in the season and the birds are currently mistimed. In areas with a late food peak, early-breeding birds still breed at the right time, and there is, at most, a weak population decline. If food phenology advances further, we also predict population declines in areas with a late food peak, as in these areas adjustment to an advanced food peak is insufficient 4 . Mistiming as a result of climate change is probably a widespread phenomenon 1 , and here we provide evidence that it can lead to population declines. Ongoing climate change leaves a clear global fingerprint on ecosystems. Many organisms bring forward the timing of their seasonal activities, whether it be flowering in plants, budding of trees, emergence of insects or breeding in birds 5–7 . Despite this general advancement, some species may not cope with climate change because their response differs from the response of organisms at lower levels of the food chain 1–4,8,9 , leading to a mismatch between the timing of reproduction and the main food supply 10 . This mistiming can have a clear effect on species population dynamics and ecosystem functioning 2,11 . In general, we expect the populations that are most mistimed to decline most in number. Here we show how populations of a small passerine bird have declined as a consequence of climate change, because the phenology of their main food supply during breeding has advanced more than the birds’ breeding date. We studied the population ecology of the long-distance migratory passerine, the pied flycatcher Ficedula hypoleuca, and its caterpillar food supply. We have previously shown in this long-term study in the Netherlands that the flycatchers have advanced their laying date but not the timing of their spring arrival in the Netherlands, and that the advancement in laying date was not sufficient to track the advance- ment of spring, leading to increased selection for early breeding 4 . The temperate forest habitat of our study area is characterized by a clear peak in caterpillar abundance in spring, and caterpillars are an important food source for nestling flycatchers 12,13 . The timing of this caterpillar peak differs between areas (see Supplementary Infor- mation) and years, with a clear shift forward over the past 20 years in our main study population 14 . We predicted that areas with increased mismatch between the timing of the birds and the peak availability of their prey would show a strong population decline. To test this prediction, we collated annual population counts between 1987 and 2003 from ten nest box populations in the Netherlands that differed strongly in population trends. If increased mistiming is the cause of population declines, we predict that populations in the areas with the earliest food peak will have declined most strongly. This is because these long-distance migrants have a relatively fixed spring migration programme 15 , and in early food phenology areas these birds have a shorter period between their arrival and the time of the food peak. A short time interval between arrival and breeding may act as a constraint, because the birds can not shorten this much further. We expect that populations in these areas of early food peak might also react less flexibly to increases in temperature, and consequently decline in number. We found strong support for our hypothesis: pied flycatchers have declined by about 90% in areas with the earliest food peaks, but have only declined by about 10% in areas with the latest food peaks (Fig. 1a). Because we measured the caterpillar peak date in only one year (2003, at the end of the study period), we used the percentage of great tits producing a second brood over a six-year period (1985–1990) as a second measure of the phenological state of the area, because great tits produce second broods only if caterpillar peaks are late 16 . Flycatcher populations declined most in areas with LETTERS Figure 1 | Population trends of Dutch pied flycatcher populations. a, b, Trends in response to the local date of the caterpillar peak (in days since 31 March) (Spearman rank correlation: r s ¼ 0.80, n ¼ 9, P ¼ 0.013) (a), and the slope of annual median laying date on spring (16 April–15 May) temperature (r s ¼ 20.86, n ¼ 7, P ¼ 0.03) (b). Populations of pied flycatchers with an early food peak and a weak response declined most strongly. Population trend is the slope of the regression of the log number of breeding pairs against year. In b, the x axis shows the slope of a linear regression of median laying date against mean temperature from 16 April–15 May. Error bars represent the standard errors of the slopes of the regression lines. All points in b are also in a, except for one point, for which we had no data regarding the caterpillar peak. 1 Netherlands Institute of Ecology (NIOO-KNAW), PO Box 40, 6666ZG Heteren, The Netherlands. 2 Animal Ecology Group, Centre for Ecological and Evolutionary Studies, University of Groningen, PO Box 14, 9750AA Haren, The Netherlands. †Present addresses: Behavioural Biology Group, University of Groningen, PO Box 14, 9750AA Haren, The Netherlands, and The Edward Grey Institute, Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UK. Vol 441|4 May 2006|doi:10.1038/nature04539 81