header image
Revision of the World species of Xeris Costa (Hymenoptera: Siricidae)
CJAI 28 -- September 25, 2015
doi: 10.3752/cjai.2015.28
Henri Goulet1, Caroline Boudreault1 and Nathen M. Schiff2


Although the preponderance of this work is a worldwide morphological revision of the genus Xeris, DNA barcoding was also used to look for cryptic species and develop a database of sequences that could be used to identify larvae, the life stage most often intercepted in commerce (Schiff et al. 2012).

DNA barcodes, as we use them (i.e. 658 bp of Cytochrome Oxidase 1), were originally introduced as an easy, rapid, inexpensive way for investigators with no specialized taxonomic knowledge to assess biodiversity (Hebert et al. 2003). The methodology proved to be popular and barcodes were used to identify animal species including fish, birds and arthropods, to associate life stages and to uncover cryptic species (Ball and Armstrong 2006, Hajibabaei et al. 2006, Hebert et al. 2004, Hebert et al. 2004A, Hogg and Hebert 2004, Smith et al. 2006, Ward et al. 2005).

However, as more taxa were barcoded a variety of pitfalls and problems were identified including; heteroplasmy, where more than one haplotype is present in a single individual; accidentally sequencing nuclear pseudogenes of mitochondrial origin (NUMT’s); bacterial mediated mitochondrial introgression; misleading results due to hybridization; insufficient variation and taxon discrimination (see discussion in Rubinoff et al. 2006, Blaxter et al. 2005, Linnen and Farrell 2007, 2008, Smith et al., 2012, Whitworth et al. 2007). These limitations made using barcodes more complicated and to clarify when and how to use them. DeSalle (2006) drew a distinction between species discovery and species identification. He argued that barcodes alone were probably not sufficient for species discovery but that if there were a sequence database derived from identified specimens, barcodes could be used to identify unknown specimens with the caveat that some unknowns might not be identifiable. He further proposed that a novel barcode (haplotype) should be considered as a species hypothesis that should only be accepted with verification by a second method. Thus, DNA barcodes should have taxonomic utility but only if there is a database of knowledge with good taxon coverage and appropriate sampling.

DNA barcodes have already proved useful in understanding siricid taxonomy. Based on barcodes, Schiff et al. (2012) synonymized color morphs that had been described as separate species, identified new species that were later supported by morphological characters and hypothesized two new cryptic species that they chose not to describe for lack of morphological characters. Based on these findings it seems likely that DNA barcodes would have utility in a worldwide revision of Xeris.


Cytochrome oxidase 1 DNA barcodes, including 144 that were new for this study, were obtained for 149 specimens of the genus Xeris (see Table 2, under Appendices). 110 sequences (74%) were obtained from adult specimens identified using morphological keys to siricid genera and species and 39 sequences (26%) were obtained from larvae identified as Xeris by their placement in the barcode tree to Siricidae (Schiff et al. 2012). At least one complete sequence (658bp) was obtained for each taxon although only 117 of the barcodes (78%) were full length. Thirteen sequences (9%) were longer than 600bp, nine (6%) were longer than 500bp, eight (5%) were longer than 400bp and two (2%) were between 250 and 300bp in length. The distribution of sub full length sequences was not random. Four of five sequences (80%) of a new species, Xeris degrooti, were less than full length including the two shortest sequences used in the study (289bp and 290bp respectively) and four of six sequences (67%) of Xeris morrisoni were incomplete whereas all other taxa had at least 50% complete sequences. The length of each sequence is reported at the end of each species description in the section listing specimens for molecular studies.

Prior to sequencing, seven Xeris species could be morphologically recognized among the adult specimens. When a Neighbor-Joining phylogenetic tree was constructed from the 149 larval and adult sequences of this study, the resulting tree had 10 branches indicating three potential extra taxa, one from adult and two from larval specimens. Bootstrap analysis showed strong support (above 90) for all major branches except for X. caudatus and X. indecisus with bootstrap values of 40.4 and 62.6 respectively (Fig. D1.1). Figures D1.2a, D1.2b, D1.2c, D1.2d and D1.2e graphically represent the within and between species variation and clearly show that 100% of specimens assort to their respective taxa. Pairwise comparisons show that the divergence between all species pairs (45 comparisons) was greater than 10% except for X. caudatus and X. melancholicus (3.4%), X. morrisoni and X. indecisus (3.0%), X. malaisei and X. spectrum (4.1%) and X. pallicoxae "type 1" and X. pallicoxae "Type 2" (2.2%) (Table 1,. under Appendices).


When using more than one method to discriminate species one hopes for congruence of results. In this case, we expected that all the morphologically defined taxa would exactly match those identified by DNA sequencing of Cytochrome Oxidase 1. The neighbor joining tree (Figs. D1.2a, D1.2b, D1.2c, D1.2d and D1.2e) shows 149 specimens segregated into ten well differentiated haplotype groups but unfortunately, morphological analysis was not always able to resolve the same taxa. The most complicated problem was the resolution of the new species X. degrooti from the widespread North American species X. indecisus. Although we recognized color variation in X. indecisus, there was nothing to suggest a new species, especially in light of the considerable color variation in other Siricidae (Schiff et al. 2012), until specimens were barcoded. Five specimens formed a distinctive clade approximately 12% divergent from X. indecisus. Initially we were leery of the result, because the samples were obviously degraded (they were not collected into ethanol but another preservative and only later transferred to ethanol), most of the sequences used were incomplete with numerous individuals collected at the same time not producing any readable sequence and the divergence was quite large for North American Xeris species. However, the single complete sequence was a powerful hypothesis. Eventually, we were convinced, because the single complete sequence did not contain any stop-codons suggesting it was not a NUMT (a nuclear pseudogene of mitochondrial origin) one of the possible errors in barcoding (Lopez et al. 1994, Song et al. 2008, Pamilo et al. 2007, Koutroumpa et al. 2009), its closest blastn search match was Xeris morrisoni (89.2% identity, searched 20 March 2015) and its position in the tree was within, not basal to, the other Xeris species. Once we accepted the new species hypothesis, we used the sequence information to make sense of the morphological variation. The results are provided in detail under the species treatments for X. degrooti and X. indecisus but basically X. degrooti females can be separated from X. indecisus females with black abdomens and X. indecisus females with reddish abdomens and clear wings but not from X. indecisus females with reddish abdomens and darkly tinted wings. We further believe that putative male X. degrooti can be separated from male X. indecisus with black abdomens but not from those with reddish brown abdomens. Since none of the five sequenced specimens are males, we cannot be positive that the specimens we posit to be male X. degrooti actually are X. degrooti so we have chosen not to provide a male description (page 47). Although we are convinced of the validity of X. degrooti, we would still like to generate barcodes for more specimens of both genders and all color morphs over more of its putative range.

Perhaps the most surprising result of this study was the independent discovery by both barcoding and morphology of the new species X. pallicoxae sympatric with X. spectrum. The current morphological analysis of X. spectrum of Western Europe revealed two species, X. spectrum and X. pallicoxae and barcode analysis of larval specimens revealed at least two and maybe three taxa that we refer to as X. spectrum, X. pallicoxae “Type 1” and X. pallicoxae “Type 2”. The results are considered to be independent because all the sequences of X. pallicoxae “Type 1” and “Type 2” and most of the sequences of X. spectrum were derived from larval specimens and larvae could not be assigned to a species a priori because there are no keys to larvae of any Siricidae. Fortunately, we were able to obtain sequences of three adults of X. spectrum positively associating the name to the haplotype group but we were unable to obtain sequences of adult X. pallicoxae and therefore had to associate the species to the haplotype group by elimination. As there are two closely related (2.2% divergence see Table 1) X. pallicoxae haplotype groups, we believe one of them is X. pallicoxae and the other is a cryptic species close to X. pallicoxae waiting to be described. Unfortunately, we do not know which haplotype group is associated with the holotype of X. pallicoxae and which is associated with the new species. Consequently, we are forced to call the species X. pallicoxae “Type 1” and X. pallicoxae “Type 2” until adults of at least one species can be sequenced. Initially, we considered that the cryptic species might only be variation within X. pallicoxae, but a fairly large sample size, relatively high bootstrap support (Fig. D1.2e) and a second annual emergence peak (most siricids only have one, see see (Fig. C11.9) support the new cryptic species hypothesis.

The remaining barcode species complement morphological species nicely and support the morphological phylogenetic analysis fairly well (see "Notes on Affinities" under Xeris). The X. indecisus lineage; X. indecisus, X. morrisoni and X. degrooti is supported as is the X. caudatus lineage of X. caudatus and X. melancholicus. Third, Xeris malaisei is recognized as a distinct species from X. spectrum. Finally, we were able to obtain a sequence of the Old World X. himalayensis from genbank. We were surprised to see that it was so divergent from the other Xeris species (16.9%-19.7%) but gratified to see that it clustered with the other Xeris within the Siricidae (Fig. D1.1).

The combination of classical morphological and DNA barcoding methods have allowed us to revise the siricid genus Xeris on a worldwide basis and add to the DNA database that enables identification of siricid larvae. DNA barcodes can unambiguously identify all species for which we were able to obtain sequences (9 of 16) and suggest there is a new cryptic species in Western Europe awaiting morphological description. One new North American species, X. degrooti, can only be positively identified using barcodes at this point but we expect additional sequences of different color morphs over more of the species range will help us clarify its morphological characteristics. This work demonstrates the utility of barcoding for generating species hypotheses and associating color morphs and different life stages.

Table of contents Abstract Introduction Materials and Methods Biology Hosts Parasitoids Morphology Key DNA References Citation Appendices PDFs