Hi,
This is my interpretation. I did not find any reference source describing this, but this notation is also used by other taxonomic databases out there.
So, currently many studies rely on metabarcoding, the use of a taxonomic marker, such as ITS (Fungi), 16S (Prokaryotes) and 18S (Eukaryotes) rRNA genes, to classify taxonomically communities in a given environment/habit/system. This is done by extracting the environmental DNA (DNA extraction from soil/water etc without previous enrichment or selection for any particularly organism) and, then, amplifying the taxonomic marker gene of interest by PCR and sequencing by next-generation sequencing, such as Illumina. In the past, these sequences were amplified and isolated by cloning the environmental DNA into E. coli and subsequent Sanger DNA sequencing of the clones. In each case, this process results in many thousand-to-millions of taxonomic marker sequences that are analyzed bioinformatically. The aim is to classify taxonomically each unique or cluster of identical sequences. Many of these sequences or clones are novel, meaning that they do not share high similarity with any sequence available in the database (this happens because the majority of microbes cannot be cultured or their conditions are not known or difficult to reproduce in vitro). Still these sequences or clones may share some degree of similarity with a particular order or family, therefore, they are classified until that level and then they remain as environmental sample, being called environmental sample - clone Family/Order (where Family or Order is the lower taxonomic rank at which these environmental sequences/clones were confidently classified).
So, environmental sample refers to any environmental sequence or clone obtained from an environmental sample, i.e., a community sample (it can be an enriched sample - but still is a mixture of DNA and the organism was not isolated), that is novel in relation to the database.
António