ABSTRACT. Protection of large natural forest landscapes is a highly important task to help fulfill different international strategic initiatives to protect forest biodiversity, to reduce carbon emissions from deforestation and forest degradation, and to stimulate sustainable forest management practices. This paper introduces a new approach for mapping large intact forest landscapes (IFL), defined as an unbroken expanse of natural ecosystems within areas of current forest extent, without signs of significant human activity, and having an area of at least 500 km 2 . We have created a global IFL map using existing fine-scale maps and a global coverage of high spatial resolution satellite imagery. We estimate the global area of IFL within the current extent of forest ecosystems (forest zone) to be 13.1 million km 2 or 23.5% of the forest zone. The vast majority of IFL are found in two biomes: Dense Tropical and Subtropical Forests (45.3%) and Boreal Forests (43.8%). The lowest proportion of IFL is found in Temperate Broadleaf and Mixed Forests. The IFL exist in 66 of the 149 countries that together make up the forest zone. Three of them-Canada, Russia, and Brazil-contain 63.8% of the total IFL area. Of the world's IFL area, 18.9% has some form of protection, but only 9.7% is strictly protected, i.e., belongs to IUCN protected areas categories I-III. The world IFL map presented here is intended to underpin the development of a general strategy for nature conservation at the global and regional scales. It also defines a baseline for monitoring deforestation and forest degradation that is well suited for use with operational and cost-effective satellite data. All project results and IFL maps are available on a dedicated web site (http://www.intactforests.org).

In the recent years, many protocols aimed at reproducibly sequencing reduced--genome subsets in non--model organisms have been published. Among them, RAD--sequencing is one of the most widely used. It relies on digesting DNA with specific restriction enzymes and performing size selection on the resulting fragments. Despite its utility, this method is of a limited use with degraded DNA samples, such as those isolated from museum specimens, as these are either less likely to harbor fragments long enough to comprise two restriction sites making possible ligation of the technical sequences required or performing size selection of the resulting fragments. In addition, RAD--sequencing also reveals a suboptimal technique when applied to an evolutionary scale larger than the intra--specific level, as polymophisms in the restriction sites cause loci dropout. Here, we address both of these limitations by a novel method called hybridization RAD (hyRAD). In this method, biotinylated RAD fragments, covering a random fraction of the genome, are used as baits for capturing homologous fragments from samples processed through a classical genomic shotgun sequencing protocol. This simple and cost-effective approach allows sequencing orthologous sequences even from highly degraded DNA samples, opening new avenues of research in the field of museum genomics. Not relying on the restriction site presence, it improves among--sample loci coverage, and can be applied to broader phylogenetic scales. In a trial study, hyRAD allowed us to obtain a large set of orthologous loci from fresh and museum samples from a non--model butterfly species, with over 10.000 single nucleotide polymorphisms present in all eight analyzed specimens, including 58 years old museum samples.

In the recent years, many protocols aimed at reproducibly sequencing reduced-genome subsets in non-model organisms have been published. Among them, RAD-sequencing is one of the most widely used. It relies on digesting DNA with specific restriction enzymes and performing size selection on the resulting fragments. Despite its acknowledged utility, this method is of limited use with degraded DNA samples, such as those isolated from museum specimens, as these samples are less likely to harbor fragments long enough to comprise two restriction sites making possible ligation of the adapter sequences (in the case of double-digest RAD) or performing size selection of the resulting fragments (in the case of single-digest RAD). Here, we address these limitations by presenting a novel method called hybridization RAD (hyRAD). In this approach, biotinylated RAD fragments, covering a random fraction of the genome, are used as baits for capturing homologous fragments from genomic shotgun sequencing libraries. This simple and cost-effective approach allows sequencing of orthologous loci even from highly degraded DNA samples, opening new avenues of research in the field of museum genomics. Not relying on the restriction site presence, it improves among-sample loci coverage. In a trial study, hyRAD allowed us to obtain a large set of orthologous loci from fresh and museum samples from a non-model butterfly species, with a high proportion of single nucleotide polymorphisms present in all eight analyzed specimens, including 58-year-old museum samples. The utility of the method was further validated using 49 museum and fresh samples of a Palearctic grasshopper species for which the spatial genetic structure was previously assessed using mtDNA amplicons. The application of the method is eventually discussed in a wider context. As it does not rely on the restriction site presence, it is therefore not sensitive to among-sample loci polymorphisms in the restriction sites that usually causes loci dropout. This should enable the application of hyRAD to analyses at broader evolutionary scales.

Summary1. The evolution of continuous traits is the central component of comparative analyses in phylogenetics, and the comparison of alternative models of trait evolution has greatly improved our understanding of the mechanisms driving phenotypic differentiation. Several factors influence the comparison of models, and we explore the effects of random errors in trait measurement on the accuracy of model selection. 2. We simulate trait data under a Brownian motion model (BM) and introduce different magnitudes of random measurement error. We then evaluate the resulting statistical support for this model against two alternative models: Ornstein-Uhlenbeck (OU) and accelerating/decelerating rates (ACDC). 3. Our analyses show that even small measurement errors (10%) consistently bias model selection towards erroneous rejection of BM in favour of more parameter-rich models (most frequently the OU model). Fortunately, methods that explicitly incorporate measurement errors in phylogenetic analyses considerably improve the accuracy of model selection. 4. Our results call for caution in interpreting the results of model selection in comparative analyses, especially when complex models garner only modest additional support. 5. Importantly, as measurement errors occur in most trait data sets, we suggest that estimation of measurement errors should always be performed during comparative analysis to reduce chances of misidentification of evolutionary processes.