Performance of indicators and the effect of grain size in the discrimination of plant communities for restoration purposes

There is strong pressure to embrace indicators for practical goals such as nature conservation and management and to evaluate the restoration success, but the selection of appropriate indicators is not straightforward. In addition, the grain and the type of data collected and data transformation adopted can influence restoration monitoring results. In this paper, we assessed the effect of changing indicator, grain size (i.e., plot dimension) and data transformation in discriminating different mapped plant communities, relying on vascular plant composition data. We considered flora entities at different taxonomic scales of resolution as indicators and used biological forms such as life forms, growth forms and a combination of the two types, i.e., life and growth forms, as rough plant traits. We also analysed the contribution of species as indicators of the different land cover classes by performing an Indicator Species Analysis. We evaluated the effect of changing indicator (taxonomic resolution, life and growth forms and indicator species), grain size and data transformation using permutational multivariate analysis of variance on cover data expressed in percentages and as simple presence/absence. Our results demonstrated that indicators such as taxonomic resolution and biological forms have partial success in discriminating between plant communities, only for the analysis performed on presence/absence data, and that the effects of changing indicator varied depending on the data transformation used. On the contrary, indicator species are coherently effective and changing the grain size has a moderate influence on their ability to discriminate among the habitat types investigated. Hence, indicator species emerged as a promising tool in restoration monitoring. Although indicators are not supposed to substitute comprehensive surveys of vegetation, their use can help redirect considerable time, resources and expertise to more replication and better sampling design.