The open data debate: a need for accessible and shared data in forest science

Abstract

During the last decade, large data sets have been increasingly used to address key questions in the field of forest science, including: (1) the impact of climate change on productivity and species distribution; (2) the long-term course of carbon, water, and nutrient cycles; (3) the spread and virulence of pathogens; (4) the genetic basis of local adaptation; and (5) sustainable socio-economic strategies (Rehfeldt et al. 2001, 2002; Diaz-Balteiro and Romero 2008; Cappa et al. 2012; Benito-Garzon et al. 2013; Porth et al. 2013; Stephenson et al. 2014). Sound data are difficult to produce in forest science because trees are long lived, are elements of complex ecosystems, and are not easily amenable to simple experiments. Yet, foresters have observed, monitored, and measured trees and forest ecosystems for a very long time, producing impressive data sets. International provenance tests are carefully monitored since the early twentieth centrury (Rehfeldt et al. 2001, 2002); they compare in common gardens trees from seeds collected in different localities (provenances) in order to record the genetic diversity of traits of importance for forestry and adaptation. Similarly, long-term records are now available for the carbon budget, water use, and nutrient cycling of a large number of forest ecosystems in temperate, boreal, as well as tropical forests (Luyssaert et al. 2007). However, like in other fields of research, the fate of these data remains, in many cases, uncertain, which has certainly detrimental effects for the advancement of forest science and for the improvement of forest ecosystem management. In the best of cases, they were published (usually not in the form of raw data) along with companion articles discussing the results. In most cases, they are stored under heterogeneous formats in the personal files of researchers and risk disappearing when these researchers change interests or retire. Recent European and international projects have taken this concern very seriously and have initiated the construction of large metadata and databases (e.g., TreeBreedex and Evoltree for genetic data, European Fluxes Database Cluster like ICOS, CarboExtreme, GHG-Europe, InGOS, for long-term ecological monitoring). International networks that monitor functional and morphological changes in forests are pushing in this direction. Institutes and research departments simultaneously consolidate available data into standardized and interconnected databases. Such a wealth of data requires very specific database management and data sharing procedures (Michener and Jones 2012). In addition to producing increasingly large data sets from monitoring and automated machine collections, forest science also relies on a large number of short-term experiments whose main results are published in scientific journals, usually without providing the corresponding data that are sometimes lost after publication. There is a general feeling that such data are often under-analyzed by their authors and that they should be made available for re-use through synthesis and analysis to generate novel ideas and test theories at unprecedented scales. Handling Editor: Erwin Dreyer B. Fady : C. Pichot INRA, UR 629, Ecologie des Forets Mediterraneennes, Domaine St Paul, Site Agroparc, 84914 Avignon, France