a b s t r a c tBesides species Red Lists, recently, a variety of frameworks have been proposed for assessing higher levels of biological organisation, i.e. ecosystems, habitats, plant communities. Most of these protocols refer to 'plant species assemblages' or 'vegetation types' as proxies for ecosystems or habitats. Indeed, the habitat concept based on plant communities has acquired a central role as a key approach for biodiversity conservation above the species level. Plant communities, like every complex biological system, hold scaledependent 'emergent' properties which vary as a function of the scale of observation. With reference to red-listing, these scale-dependent properties have far-reaching consequences for both identification and classification, as well as for representation and evaluation, and become particularly challenging when dealing with criteria regarding decline in distribution or restricted distribution. The recent discussion on the red-listing protocols has evidenced several aspects that claim special efforts for a suitable use. In the present paper, starting with the analysis of some recently proposed protocols for the red-listing of habitats and ecosystems, we discuss and test some 'emergent' properties of species assemblages, providing cues for reflection. Based on a variety of theoretical models and scientific outcomes in literature from the last decades, we theorise that plant communities own some intrinsic, ecologically based and scale-dependent spatial features, which give rise to different types of pattern of spatial occupancy. We discuss a model where, in natural conditions, the possible patterns of spatial occupancy are referred to 3 basic types: areal, linear and point. This approach is here proposed as a tool to discriminate among different broad categories of plant community-based habitat types and optimise their assessment in the red-listing process. Starting from a homogeneous data set, the proposed case studies prove that the choice of the scale affects the comprehension of the habitats' occurrence, with a substantial relapse on the estimates of their distribution size. In particular, habitats with linear and point distribution, often naturally small in size and dispersed, are more susceptible to biased evaluation of their actual distribution and consequently of their threat status. The intrinsic spatial attributes of plant communities should not be neglected in a red-listing process and claim for a 'habitat-tailored' approach. The use of different gridcell sizes and thresholds for the three main patterns of spatial occupancy here proposed, might certainly avoid inaccurate statements.