Sequencing results
We retrieved DNA barcodes of 368 out of the 376 species represented as adults in the sampling plots. After the cleaning, filtering, and chimaera removal process, the sequencing of juvenile pool samples generated a total of 15,805,993 sequence reads and 3,839,513 unique sequences. After manually adding the sequences of the adults, the Swarm algorithm grouped all sequences in 4,668 non-singleton MOTUs, of which 140 contained exclusively adult artificial sequences. After the curation process (retaining only Araneae, removing non-iberian taxa, and removing artificially added adult sequences), the final dataset for juveniles consisted of 9,956,432 reads distributed in 1,343 MOTUs. The process of removing pseudogenes eliminated 455 MOTUs and left 888 final juvenile MOTUs (final dataset in Table S2, Supporting information). The sequencing of three of the four replicates of plot PM1 was unsuccessful so we omitted these samples from subsequent analyses.
The steps involved in refining the taxonomic assignment of these MOTUs (merging the species that were found to be indistinguishable by the LERAY COI fragment, checking online genetic databases and the COI tree to improve taxonomic assignments, and assigning the remaining unassigned MOTUs to their best match in the reference library) assigned the 888 MOTUs to 524 different species. Filtering out MOTUs with less than 5 reads left 411 MOTUS that, when collapsed by their taxonomic assignment, corresponded to 350 different species. Finally, considering only MOTUs with at least 5 replicates in a plot and presence in at least 2 of its replicates as present in that plot left 331 different species.
Rarefaction curves (Fig. S1, Supporting information) showed that almost every plot reached a plateau in all or some of its replicates. The exceptions were the only remaining replicate of PM1 and two of the replicates of PS2, all of which yielded very low numbers of reads.