Is the observer effect significant in vegetation assessment of restored metallicolous grassland?

Projects that prospects habitat restoration require protocols of vegetation assessment to monitor the evolution of the restored ecosystems over time. In most cases, multiple assessors are needed and some studies have shown that the variability among observer can be significant. In the context of tropical grassland, few studies quantified the impact of the observer on the results, reducing the capacity to replicate conclusions, and, therefore to plan restoration strategies at large scale. This study aims to quantify the variation among observers using two phytosociological methods on restored metallicolous grassland and to test whether variation in the assessment could involve the conclusions of the vegetation assessment. Four observers were selected to perform monitoring of quadrats in two distinct habitats identified in four restored ecosystems. All observers sampled seven metrics based on phytosociological methods and two of them also performed a complete vegetation sampling of the quadrat, including species identification and cover in quadrat. For metrics based on cover, variation of measure exists among the observers. For metrics based on counting, no variation is highlighted and the observers are able to distinguish and identify species. These results emphasize that a same method could lead to distinct results according to observer and, in restoration projects, it is crucial to ensure a correct transfer of protocols of vegetation assessment in order to reproduce equally the method.