DNA metabarcoding of macrobenthic communities in the North Sea
is most comparable with morphology based analyses when using the Leray
primer set
There is considerable support for the use of the mitochondrial COI gene
to assess metazoan diversity from bulk samples (Andujar et al., 2018;
Leray & Knowlton, 2015; Lobo et al., 2017), yet finding the best primer
set for a specific target group and study requires empirical data rather
than relying on general rule of thumbs (Alberdi et al., 2018). In a
study comparing the performance of five primer sets on mock communities
of marine macrobenthos using the 454 sequencing platform, primer set E
was amongst the best primer sets in terms of detecting many species
(Hollatz et al., 2017). The amplification of the 3’ region of the Folmer
region has resulted in better detection of marine macroinvertebrates
than the full length Folmer fragment (Aylagas, Borja, et al., 2016;
Hollatz et al., 2017; Lobo et al., 2017). In silico testing showed that
the primer sets in the 3’ region generally detected almost twice the
amount of species compared to the primer set in the 5’ region (Table 1),
which was confirmed by our wet lab results: primer sets amplifying the
3’ region outperformed primer set D in recovering macrobenthic species.
Our data clearly show that primer sets used by Leray et al. (2013)
(Primer set A) and Lobo et al. (2013) (Primer set E) outcompete the
other three primer sets in the number of high quality and non-chimeric
reads (ESM Fig 1). Both primer sets amplify exactly the same fragment of
the COI gene, but the reverse primer of primer set A has a higher
degeneracy than the reverse primer of primer set E (Table 1). A higher
degeneracy leads to a better performance of COI primers for insect
communities (Marquina, Andersson, & Ronquist, 2019). Based on the
wetlab results, primer set A outperformed primer set E at various
levels: 1/ twice the amount of sequences received taxonomic assignment
(22.6 vs 10.9 %); 2/ the best correspondence with morphological species
numbers in the different locations was obtained (ESM Fig 4) and 3/ the
highest number of morphologically identified species was detected with
primer set A (Fig 3). Finally, the Leray primer set was able to
differentiate the four macrobenthic communities. Consequently, for
monitoring studies targeting marine macrobenthic communities the use of
the Leray primer set is recommended.
Metabarcode studies typically detect a higher number of species compared
to morphology based analyses (Elbrecht, Vamos, et al., 2017; Lobo et
al., 2017), which has been explained by the fact that specimens are not
lost during the sorting process and also small pieces of animals –
which are morphologically not identified- are included in the bulk DNA
and eDNA. In addition, metabarcode studies have reported the detection
of species that were lacking in the morphological analyses from the same
samples (Aylagas, Borja, et al., 2016). This was also the case for our
study, but in contrast to what we expected, less species were detected
using DNA metabarcoding (Fig 3): for the four samples that were
processed both morphologically and with DNA metabarcoding, 57 species
were morphologically identified, while 52 and 45 species were identified
with the bulk DNA and eDNA, respectively. This may be partly explained
by the reference database and algorithm used to assign taxonomy: 51 and
76 species for the bulk and eDNA, respectively, were identified when
using the MIDORI reference database, while 60 and 103 species for the
bulk and eDNA, respectively, were identified when using Blast and the
NCBI database. The latter procedure was also used by Lobo et al. (2017)
and Elbrecht, Vamos, et al. (2017).