Distinct barcodes for the Cereal leaf beetles Oulema melanopus and Oulema duftschmidi (Coleoptera: Chrysomelidae), two syntopical sibling species

Oulema melanopus (Linnaeus, 1758) and Oulema duftschmidi (Redtenbacher, 1874) (Coleoptera: Chrysomelidae) are two native West Palaearctic species developing on various cultivated and wild grasses. Along with O. obscura they are considered to be secondary pests of cereal crops. However, local outbreaks have been recorded recently and their status as secondary pests may evolve, especially as the use of broad-spectrum insecticides is now greatly reduced. Oulema melanopus and O. duftschmidi are considered to be sibling species. They are morphologically very close and diffi cult to distinguish from each other, which makes it diffi cult to study them. We tested the reliability of the standard barcode fragment (COI) for distinguishing between these species. A total of 92 samples of the two species, covering the majority of their natural range, was sequenced for the barcode fragment and interand intraspecifi c genetic distances were estimated. Our results confi rm those of Bezděk & Baselga (2015, Acta Entomol. Mus. Nat. Prag. 55: 273–304) in that this marker cannot differentiate between all the species of the Oulema melanopus complex, which in the Mediterranean basin contains several described and possibly some undescribed cryptic species. However, this marker may be useful in an agricultural context in areas where only O. melanopus and O. duftschmidi occur (such as in cereal crops in France) where it can be used to reliably and rapidly separate all stages of these two taxa and can therefore help in studying their ecology and dynamics.


INTRODUCTION
Cereal leaf beetles belong to Coleoptera of the subfamily Crio cerinae within the family Chrysomelidae and their larvae feed and develop on various cultivated or wild grasses (Venturi, 1942;Jolivet, 1997 (Bezděk & Schmitt, 2017;Rilet et al., 2003), among which two new species were recently described from Italy (Oulema mauroi Bezděk & Baselga, 2015) and Spain (Oulema verae Bezděk & Baselga, 2015). Only two species of Oulema are frequently cited as cereal pests in the Palearctic region: O. melanopus and O. obscura (Bala-COI gene (Hebert et al., 2003a) to successfully distinguish between and identify several species of Criocerinae in the genus Crioceris. Similarly, Bezděk & Baselga (2015) use this DNA fragment for identifying European species of Oulema, but their results, based on a small number of specimens, indicate that the COI barcode is not appropriate for the molecular identifi cation of these species as there are discrepancies between the species boundaries revealed by morphology and DNA barcodes. In this study the effectiveness of this gene for identifying the species of Oulema feeding on cereals in Europe is re-evaluated using a larger number of specimens.

Sampling and morphological identifi cation
Specimens were collected between 2005 and 2017 in France, Portugal (Madeira), Greece (Crete), Spain and Italy (Table 1 and Fig. 1). At each site sampled, one to three adults or larvae were collected and killed directly in 96.5% ethanol. The adult specimens were identifi ed to species, based on external morphological characters (Warchałowski, 2003;Bezděk & Mlejnek, 2016), except for specimens belonging to the Oulema melanopus/duftschmidi species pair (Fig. 2) for which the identifi cation was based on the structure of dissected male genitalia as only by examination of the fl agellum can the two species in this complex be reliably separated (Chapelin-Viscardi & Maillet-Mezeray, 2015). O. duftschmidi has a thin, elongated fl agellum ( To account for these species in an evolutionary context and validate our means of identifi cation, 25 additional specimens of 8 different species belonging to the subfamily Criocerinae were also sampled. The species Epitrix pubescens (Chrysomelidae: Galerucinae) was used as an outgroup to root the phylogenetic analysis reported below. After validation of the barcoding method, we tested the molecular identifi cation of 17 females and 7 larvae of the O. melanopus/duftschmidi complex. As a result, the sample for molecular analyses included a total of 185 specimens (Table 1).

DNA sequencing and analysis of sequences
The extraction and amplifi cation protocol was that used by Streito et al. (2018): extraction of the total genomic DNA was carried out in a non-destructive manner, on whole specimens, using the DNeasy 96 Blood & Tissue extraction kit (Qiagen, Hilden, Germany) following the manufacturer's protocol. The standard barcode fragment (Hebert et al., 2003a) was amplifi ed using a mixture of tailed primers (based on Cruaud et al., 2010;Germain et al., 2013 andIvanova et al., 2007) (Table 2). PCRs were carried out in 25 μl of reagents with 2 μl of matrix DNA, 0.07 mM of each primer, 2.5 mM of MgCl 2 , 0.05 mM of dNTPs and 0.025 U/μl Dreamtaq DNA Polymerase (Thermo Scientifi c, Waltham, USA). The PCR conditions used were 94°C for 3 min followed by 5 cycles at 94°C for 30 s, 45°C for 40 s, 72°C for 60 s, then 35 cycles at 94°C for 30 s, 51°C for 40 s, 72°C for 60 s, with a fi nal extension phase at 72°C for 10 min. PCR products were purifi ed, then sequenced directly by Eurofi ns MWG Operon according to their protocol using M13 sequencing primers (M13F and M13R). Forward and reverse overlapping strands were assembled 2015; Chapelin-Viscardi & Maillet-Mezeray, 2015;Leroy & Chapelin-Viscardi, 2018), which makes their study particularly diffi cult for non-specialists, especially as it is not possible to identify females and the immature stages. The existence of both species in France was reported by Berti (1989), who provides reliable morphological criteria for identifying them (Fig. 2) and reports the existence of specimens of O. duftschmidi identifi ed as O. melanopus in the collections of the French National Museum of Natural History (MNHN, Paris). Berti (1989) also states that both species are sympatric and widely distributed in France. More recently, Bezděk & Baselga (2015) revised the Oulema melanopus species complex in Western Europe. They recognize fi ve species, including two new ones and review the taxonomy of the group.
Oulema melanopus and O. duftschmidi can damage crops, especially the larvae that feed on leaves of cereal plants (Philips et al., 2011). These sibling species could potentially harm cereal crops in France (Bonnemaison, 1962;Labeyrie, 1963;Anglade et al., 1976;Chambon et al., 1983;ACTA, 2016 etc.) and other European countries (Labeyrie, 1963): Romania (Knechtel & Monolache, 1936), Hungary (Sajó, 1893), Spain (Urquijo, 1940), Greece (Pélé cassis, 1951) and Italy (Bechini et al., 2013). Most of these publications refer only to O. melanopus whereas O. duftschmidi may be also involved (Chapelin-Viscardi & Maillet-Mezeray, 2015). Recently, high population densities of larvae of cereal leaf beetles were recorded in various parts of mainland France, such as the Ille-et-Vilaine, Loiret and Allier departments (pers. obs.). In addition, extreme climatic events, which are becoming increasingly frequent, are conducive to pest outbreaks, particularly of species that are highly dependent on the climate, such as cereal leaf beetles (Guppy & Harcourt, 1978;Olfert & Weiss, 2006;Bechini et al., 2013). The greatly reduced use of neonicotinoid insecticides, and more generally that of pesticides, along with the increase in organic farming in Europe, could also provide suitable conditions for future outbreaks of these pests. Their status could shift from secondary to major pests, as has occurred in the United States and Asia (Philips et al., 2011).
Due to the impossibility of identifying the larvae of these species little is known about the life history traits and relative abundance of these two species. Preliminary agricultural monitoring indicates that O. duftschmidi is the more common in several French agricultural landscapes (Chapelin-Viscardi & Maillet-Mezeray, 2015). These surveys also reveal that the fl ight activity of the adults of both these species is synchronous, which indicates simultaneous larval development. A more recent biogeographical study provides clear evidence that both species are sympatric and coexist throughout France (Leroy & Chapelin-Viscardi, 2018).
In order to better understand the structure of the O. melanopus/O. duftschmidi species complex it is crucial to have a reliable and routine method for identifying all the development stages of the species. Kubisz et al. (2012) used the standard barcode fragment of the mitochondrial   Table 1. Pairwise nucleotide sequence divergences were calculated using the Kimura 2-parameter model of substitution (Kimura, 1980) in MEGA 7 software (Kumar et al., 2016) and the "pairwise-deletion" option. A preliminary phylogenetic tree of the genus Oulema, including all the specimens sequenced, was reconstructed based on the COI sequences. For a clearer view of the inferred tree, only the sequences of one or two specimens per locality are presented in Fig. 3 (specimens not highlighted in Table 1). We reconstructed a tree (Fig. 3) including the standard barcode COI solely for specimens reliably identifi ed to species (only dissected males for the melanopus/duftschmidi complex) and the sequences published by Bezděk & Baselga (2015). Phylogenetic analyses were conducted using the Maximum Likelihood The aLRT approximation (approximate likelihood ratio test) was used to calculate the bootstrapping values for each node (Anisimova, 2006) (5000 replicates). The resulting phylogenetic trees were edited using iTOL software (Letunic & Bork, 2016).

Analysis of newly obtained sequences
The barcode fragment was successfully amplifi ed for all specimens, regardless of sex or stage. In total, 185 specimens belonging to 12 species were sequenced ( Table 1). The observed minimum interspecifi c genetic distance between species of Chrysomelidae in the neighbouring genera of Oulema based on COI ranged from 10.2 to 19.3% (Table  3). In the genus Oulema, O. obscura showed a minimum divergence of 18.5% with Oulema melanopus and Oulema duftschmidi. O. melanopus showed a minimum divergence of 3.1% with O. duftschmidi, its closest relative. Between genera of the Criocerinae, the minimum divergence varied from 16.5% between Lema and Oulema to 21.8% between Crioceris and Oulema. Epitrix had a minimum distance from the Criocerinae species tested ranging from 22.2 to 26.2%.

Phylogenetic reconstruction
The substitution models selected by PhyML 3.0 as the most appropriate were the GTR + G + I model for the tree in Fig. 3 (AIC = 10403,93488). That tree is based on both our sequences and those published by Bezděk & Baselga (2015). The genus Oulema is monophyletic (support 83.26%) with two sister clades. One (support 91.40%) contains O. obscura and O. hoffmannseggii from Spain (as in the phylogenetic tree of Bezděk & Baselga, 2015, in which, however, the latter species is paraphyletic); O. obscura contains two sister groups, one comprising specimens from the Iberian Peninsula and the other from France and the Czech Republic. The other clade containing the O. melanopus complex (including the 5 species recognized by Bezděk & Baselga, 2015) is highly supported (99.42%), but its internal relationships are problematic. O. duftschmidi forms a well-supported clade * (91.36%) which, however, * Among the dissected males, only one mismatched specimen (JSTR02905_0101 from Alentejo, Portugal, morphologically clearly belonging to O. duftschmidi) was nested with specimens of O. melanopus. This could be a labelling error or contamination (see Discussion); therefore the specimen was removed from the fi nal analyses, from the trees in Figs 3 and S1 and consequently from sequences deposited in Bold (Table 1).  When we included the unidentifi ed specimens of the O. melanopus complex from France and Italy (see Table 1) in the above analysis, they were clearly placed in one or the other species (Fig. S1).

DISCUSSION
During this study, we sequenced 184 specimens of 11 species of Crio cerinae in four genera (Crioceris, Lema, Lilioceris and Oulema) with a view to testing the possibility of routinely using the standard barcode COI for highthroughput and reliable identifi cation of the three species of Oulema of agronomic interest, O. obscura, O. melanopus and O. duftschmidi. We included European specimens (O. duftschmidi was also available from Madeira) of 9 out of 28 species and subspecies of Criocerinae known to occur in Europe, and added specimens of 2 more Lema species from the Mascarenes. Combining our sequences with   (2015), giving a total of 122 sequences. Each colour corresponds to a morphologically identifi ed species. The sequence code, identity and geographical origin are given for each specimen. The bootstrap values located at branch nodes are percentages and only those greater than 64% are presented (over 5,000 replicates). The country name is coded according to ISO 3166-1 alpha-2 (ISO 3166, 2016). those of Bezděk & Baselga (2015), 7 European species of Oulema (from the 11 known in Europe) were documented. Our results confi rm those of Kubisz et al. (2012) and show that the standard DNA barcode can reliably differentiate between most European species of Criocerinae, but confi rm also the results of Bezděk & Baselga (2015) that this marker cannot differentiate between all species of the genus Oulema. A complex of species occurs in the Mediterranean basin including at least O. melanopus, O. verae, O. duftschmidi, O. mauroi and possibly some undescribed cryptic species, which cannot be reliably distinguished by the standard barcode. In this study, sampling and marker selection were used to address agronomic questions. The lack of material coming from non-cultivated Mediterranean ecosystems prevented us addressing the problem of the species occurring around the Mediterranean basin where much more extensive sampling and the use of other markers and other methods would be needed to clarify the taxonomy of Oulema.

Distinguishing Oulema obscura from the O. melanopus complex
While adult specimens of O. obscura can be easily distinguished from those of the melanopus group on the basis of their general coloration (body entirely blue versus red pronotum and legs, respectively) ( Fig. 2a versus 2b, c), the use of DNA barcodes for species identifi cation can also be used to reliably identify the immature stages of these species. The sequence of the COI gene tested makes it possible to distinguish this species from the entire melanopus complex which includes the other two Oulema cereal pests (O. melanopus and O. duftschmidi). The minimum divergence of 18.5% between O. obscura and O. melanopus / O. duftschmidi is rather high and comparable to distances recorded between species in different genera in the same subfamily, such as Crioceris and Lema (Table 3).

Distinguishing between O. melanopus and O. duftschmidi
The minimum interspecifi c divergence of 3.1% between specimens of O. melanopus and O. duftschmidi (Table 3) means that these sibling species are more closely related than all the other species studied, which is consistent with their similar morphology and biology. In addition, such a value is congruent with what is reported for other sibling species in the family Chrysomelidae (Cognato, 2006). However, we recorded lower genetic distances between specimens of O. duftschmidi from very distant populations (North of France and the Italian province of Puglia) or isolated populations (such as those in Madeira and Crete), than between specimens of O. melanopus and O.   (Tables 3 and 4). Lastly, the interspecifi c percentage divergences (Table 3) were well above the maximum percentages of intraspecifi c divergence recorded (Table 4) and there was no overlap between the intra-and interspecifi c distances of O. melanopus and O. duftschmidi. Consequently, this argues in favour of a clear genetic differentiation of the sibling species O. duftschmidi and O. melanopus, which is supported by the phylogenetic trees (Figs 3 and S1). Of the 92 males studied, only one (JSTR02905_0101, a male specimen of O. duftschmidi from Alentejo, Portugal), was placed in a cluster that does not correspond with the species identifi cation based on the morphology of its genitalia. A posteriori examination of the preserved adult and its dissected genitalia defi nitively excluded misidentifi cation. However, we cannot exclude an error in tube labelling during handling or contamination. Wherever possible, we deliberately selected specimens of the two species for our dataset that were collected on the same day at the same location (see Table 1), to maximize the chances of recording potential hybridisation. Apart from this specimen, for which it was not possible to rule out a handling error, no other individual was incorrectly assigned in our data set. A second case of a mismatch between molecular and morphological identification was that of the Moroccan specimen (PK406715.1), which was genetically assigned to O. duftschmidi, whereas it was identifi ed as O. melanopus by Bezděk & Baselga (2015). These authors (pers. com.) suspected that there was an undescribed cryptic species in Morocco to which this specimen belonged. Indeed, they noted differences between the genitalia of this specimen and typical specimens of O. melanopus with which it was tentatively identifi ed. The method we used (extraction, amplifi cation and sequencing of a gene) enabled correct assignment of males previously identifi ed on the basis of dissected genitalia. The unidentifi ed females and larvae that we tested were also unambiguously assigned to one of the two taxonomic groups. It would be interesting to test the method on a larger number of specimens in order to check whether introgression has occurred and assess its percentage of occurrence. Breeding tests would also be required to test this hypothesis.

Distinguishing other species in the melanopus complex and their intraspecifi c differences
The intraspecifi c diversity of specimens from the Mediterranean basin, especially those from the Iberian Peninsula, was greater than that of the French and Czech specimens (Fig. 3). This increase in genetic diversity with increase in the geographical coverage is documented (Bergsten et al., 2012) and due to the presence of Mediterranean glacial refugia and their associated biological diversity (Hewitt, 2001). Currently the data from the Mediterranean areas is limited and more extensive sampling could provide additional insights into the biological or biogeographical processes that resulted in the present diversity.
The other issue is distinguishing between the cereal pests and the rarer or more localized species that are described in the genus Oulema. Phylogenetic relationships between the fi ve species of the melanopus group have been studied by Bezděk & Baselga (2015) based on the COI gene. They conclude that the relationships between the different species in this group were not well resolved on the basis of this gene. The results we obtained by combining their work with ours are slightly more optimistic at least in the possibility of distinguishing O. duftschmidi from any European specimens currently being morphologically identifi ed as O. melanopus and support the idea that these two groups are genetically well separated. In Fig. 3 (Bezděk & Baselga, 2015) is included.
The remaining 12 BINs are represented by a limited number of sequences (one to ten). Several species names are associated with several BINs: O. erichsonii (four different BINs among which is AAK5928); O. duftschmidi (AAO0694 and ADK1309 for one Indian sequence); O. obscura under the name O. gallaeciana (AAN1559 and ABW1444 for the 7 sequences from Spain, Galicia, see Fig. 3); O. hoffmannseggii (ABV0207 and ADU7791 for two sequences from Spain); O. melanopus (AAK5928 and four other BINs for sequences from different European and non-European countries, of which ABW1460 contains among other the 5 sequences from Spain, Galicia, sister to O. rufocyanea in our Fig. 3). That sequence of O. rufocyanea (KP406722.1) is associated with a unique BIN (ACJ0414). 44 sequences were not identifi ed to species and 46 were not associated to a BIN due to their poor quality, insuffi cient length, etc.
BOLD provides the state-of-the-art barcoding information of the genus Oulema and highlights the need of clarifi cation of the taxonomy in this group. The association of one species with several BINs and conversely several species within the same BIN may not be only due to misidentifi cation. We cannot exclude some cryptic species such as the 7 specimens identifi ed as O. gallaeciana from Spain that form a separate BIN. At the present state of knowledge, the use of the database for routine identifi cation of Oulema can only be considered in a limited geographic context, keeping in mind possible misidentifi cations and the partly unresolved taxonomy. For that reason we chose to compare our results only to sequences resulting from a taxonomic study (Bezděk & Baselga, 2015). Bergsten et al. (2012) highlighted that limited sampling, and thus a restricted set of sequences refl ecting local biodiversity, improves the identifi cation by barcoding. This is supported by our results. Depending on the geographical context and the agronomic versus natural context, the identifi cation of Oulema species by barcoding may be more or less effi cient.

CONCLUSION
This study showed that the standard COI barcode can be used to distinguish between some Oulema species, including O. melanopus and O. duftschmidi, but cannot distinguish some other species in the melanopus complex, suggesting that further analyses might be needed to validate their taxonomic status.
Very extensive sampling has been carried out recently in agricultural regions in France (Chapelin-Viscardi & Maillet-Mezeray, 2015;Leroy & Chapelin-Viscardi, 2018). Several thousand specimens were identifi ed based on the shape of the male fl agellum. Given the relatively clear morphological criteria that characterise O. mauroi and O. verae (Bezděk & Baselga, 2015), it is unlikely that they would have been confused with O. melanopus and O. duftschmidi in those studies. Identifi cation based on male genitalia, and especially fl agella, tested by specialists, was validated by our study, which also confi rmed the quality of the morphological identifi cations. These surveys provide evidence that only three species of Oulema are present in cereal crops in France: O. obscura, O. melanopus and O. duftschmidi. In the absence of the other species of the melanopus complex, the results obtained show that DNA barcoding is a good method for differentiating between species of the genus Oulema in cereal crops in France, regardless of the developmental stage or sex of the specimens. In order to meet the needs of plant protection professionals, the method must provide both unambiguous and reliable results. To achieve this, it will be necessary fi rst of all to associate the reference sequences present in the database (on which the identifi cation of sequences will be carried out) to a given geographical area and context, in our case cereal fi elds in mainland France.  S1. Phylogenetic tree constructed using the Maximum Likelihood method (ML) and the COI gene sequences of identifi ed specimens from Table 1 plus unidentifi ed specimens (in bold, 11 females and 3 larvae), belonging to the complex Oulema melanopus/duftschmidi. Epitrix pubescens (in bold) was used to root the tree (106 sequences in total). Each colour corresponds to a morphologically identifi ed species. The sequence code, identity and geographical origin are given for each specimen. The bootstrap values at the branch nodes are percentages and only those greater than 64% are presented (over 5,000 replicates). The country name is coded according to ISO 3166-1 alpha-2 (ISO 3166, 2016).