Question How reliable is the process of delimiting plant species by morphotyping sterile specimens from a highly diverse Amazonian forest plot? Location Biological Dynamics of Forest Fragments Project (BDFFP), Central Amazon, Manaus, Brazil. Methods A taxonomic exercise was conducted during a Center for Tropical Forest Science (CTFS) Taxonomy Workshop held in Manaus in April 2011, using specimens collected in a 25‐ha forest plot. The plant species from this plot had been previously delimited by morphotyping of ca. 80 000 sterile specimens, a process that resulted in the recognition of 115 cases (accounting for 38% of all trees) in which species delimitation was problematic. For the workshop, we selected a subsample of specimens for eight of these difficult cases (taxonomic groups/complexes) and asked 14 participants with different levels of botanical training to independently sort these specimens into morphospecies. We then compared the classifications made by all participants and explored correlations between botanical training and plant classification. Results The classification of specimens into morphospecies was highly variable among participants, except for one taxonomic group/complex, for which the median pair‐wise similarity was 95%. For the other seven taxonomic groups/complexes, median pair‐wise similarity values ranged from 52% to 67%. Training did not increase the similarity in the definition of morphospecies except for two taxonomic groups/complexes, for which there was higher congruence between the classifications made by participants with a high level of botanical training than in comparisons that included less‐experienced participants. The total number of morphospecies defined by participants was highly variable for all taxonomic groups/complexes, with the total number varying from 12 to 46 (a 383% difference). Conclusions Local plant species delimitation by morphotyping sterile specimens is prone to large uncertainties, and botanical training may not reduce them. We argue that uncertainty in species delimitation should be explicitly considered in plant biodiversity inventories as diversity estimates may be strongly affected by such uncertainties. We recommend that species delimitation and identification be treated as separate processes and that difficulties be explicitly recorded, so as to permit error estimates and the refinement of taxonomic data.

Litterfall and litter decomposition are vital processes in tropical forests because they regulate nutrient cycling. Nutrient cycling can be altered by forest fragmentation. The Atlantic Forest is one of the most threatened biomes in the world due to human occupation over the last 500 years. This scenario has resulted in fragments of different size, age and regeneration phase. To investigate differences in litterfall and leaf decomposition between forest successional phases, we compared six forest fragments at three different successional phases and an area of mature forest on the Atlantic Plateau of Sao Paulo, Brazil. We sampled litter monthly from November 2008 to October 2009. We used litterbags to calculate leaf decomposition rate of an exotic species, Tipuana tipu (Fabaceae), over the same period litter sampling was performed. Litterfall was higher in the earliest successional area. This pattern may be related to the structural properties of the forest fragments, especially the higher abundance of pioneer species, which have higher productivity and are typical of early successional areas. However, we have not found significant differences in the decomposition rates between the studied areas, which may be caused by rapid stabilization of the decomposition environment (combined effect of microclimatic conditions and the decomposers activities). This result indicates that the leaf decomposition process have already been restored to levels observed in mature forests after a few decades of regeneration, although litterfall has not been entirely restored. This study emphasizes the importance of secondary forests for restoration of ecosystem processes on a regional scale.

No abstract