Diet analysis
The samples used in this study were previously analysed by da Silva et
al. (2019a) to describe the limitations of single markers in
metabarcoding analysis of generalist birds, and to describe a novel
method to integrate metabarcoding dietary data from multiple markers.
Here we use a subset of that data corresponding to the first capture of
110 adult birds, thereby avoiding biases that might result from
including data from a few birds captured more than once
(pseudo-replication), as well as eventual confounding effects of
including a small number of 1st calendar year birds.
Laboratory analysis and bioinformatic processing followed the procedures
described in da Silva et al. (2019a). Shortly, the DNA of the droppings
was extracted in batches of 23 samples plus a negative control, using
the Stool DNA Isolation Kit (Norgen Biotek Corporation) and following
the manufacturer’s protocol. DNA extracts were then subjected to four
independent PCR reactions, each targeting a different gene region: 18S
(Jarman et al., 2013), 16S (da Silva et al., 2019a), COI (Zeale, Butlin,
Barker, Lees, & Jones, 2011) and trn L (Taberlet et al., 2007).
PCR products were diluted 1:4 and amplified again to incorporate
Illumina indexes. Resulting fragments were purified using AmPure Beads,
quantified in Nanodrop, normalized and pooled per primer. Each library
was further quantified using qPCR, normalized to 4nM and pooled. The
final pooled library was sequenced in an Illumina MiSeq using a partial
V2 2x250bp kit with an expected sequence coverage of 12,000
reads/primer/sample. Bioinformatic procedures were done using ObiTools
and consisted in pairwise alignment of reads, removal of primer
sequences, collapsing of reads into exact sequence variants (ESVs), and
removal of non-target and potential spurious sequences using obigrep and
obiclean (more detailed methods in da Silva et al., 2019a). Finally,
reads were assigned to a prey item by blasting each ESV against BOLD and
NCBI online databases and COI sequences from arthropods collected in
Portugal (Ferreira et al., 2018). Each possible taxon was checked for
its occurrence in the Iberian Peninsula and discarded if not known to
occur in either Portugal or Spain. Species level identifications were
usually made at identity levels above 98.5% with a single species,
except for rare cases where no other species of the genus were known to
exist. If the same ESV matched different species, genus, or families,
identifications were made to the lowest taxonomic level possible that
encompassed all the closest hits. Whenever different ESVs matched the
same taxa they were joined into a single molecular unit.
For diet analysis, we only considered molecular operational taxonomic
units (MOTU’s) of prey identified to the order, family, genus or species
levels. We excluded all items that were likely sampling contaminations
(e.g. human, fungi and mealworm DNA), and other items not likely to be
intentionally ingested by wheatears, as bird parasites and plants that
do not have ripe fleshly fruits during the sampling period and are
likely the detection of secondary consumption (da Silva et al., 2019a;
Sheppard et al., 2005). We integrated all the dietary items obtained
from the four molecular markers into a single dataset (Table S2) using
the Python script provided by da Silva et al. (2019a).