In the otherwise excellent special issue of Trends in Ecology and Evolution on long-term ecological research (TREE 25(10), 2010), none of the contributors mentioned the importance of natural history collections (NHCs) as sources of data that can strongly complement past and ongoing survey data. Whereas very few field surveys have operated for more than a few decades, NHCs, conserved in museums and other institutions, comprise samples of the Earth's biota typically extending back well into the nineteenth century and, in some cases, before this time. They therefore span the period of accelerated anthropogenic habitat destruction, climate warming and ocean acidification, in many cases reflecting baseline conditions before the major impact of these factors.Natural history collections (NHCs) provide a rich source of data at the taxic and community levels, and can contribute to a wide range of studies [1]. These include biogeographic range changes (spatial and/or altitudinal) [2]; phenological shifts (e.g. in flowering time [3]); and evolutionary change (genetic or morphological). They can also (if unsorted bulk samples are available) document changes in community composition in the recent past (historical samples) and through deeper geological time (fossil samples). In addition, museum specimens provide source material for a range of genetic, biochemical, isotopic and trace-element studies into organismal responses to environmental change (e.g. use of preserved feathers to trace changes in the diet and migration of birds [4]).NHCs comprise not only the products of opportunistic collecting but are also (particularly in the major national museums and institutions) repositories of major surveys. The Natural History Museum (NHM) in London, UK, for example, houses thousands of jars from the Discovery and Challenger marine expeditions that were collected at hundreds of stations in the late nineteenth and early twentieth centuries. These samples provide an outstanding (and largely untapped) resource for comparison with modern survey data. Collections made with a coherent and systematic sampling strategy are inevitably more likely to provide research-quality material than ad-hoc or point samples. Unknown or inconsistent sampling strategy can be a problem, but this is not unique to museum collections; methodological changes or gaps also occur in long-term field datasets and need to be accommodated in the analysis [5].Of particular value, a proportion of historical and modern collections comprise time-series (i.e. the same locality and/or taxon has been regularly collected over many years). For exploration of long-term ecological responses, NHCderived datasets can be integrated with local or regional climatic and other records, such as the Central England Temperature Record, which is continuous back to 1659 (http://hadobs.metoffice.com/hadcet/). Time-series through much longer intervals of the Earth's history are available in palaeontological collections.Central to all such endeavours is the availability of accurate provenance...