Cryptic diversity of the jewel beetles Agrilus viridis (Coleoptera: Buprestidae) hosted on hazelnut

Abstract The genus Agrilus (Coleoptera: Buprestidae) represents a taxonomic puzzle, since the boundaries between species, subspecies and morphotypes tied to different host plants are sometimes difficult to establish on morphological characteristics alone. Some Agrilus species can cause severe agricultural damage; this makes correct distinctions of the taxon and knowing whether the insects switch from one host plant to another important. This study of mtDNA examined the genetic characteristics of lineages of A. viridis, a jewel beetle recently found causing damage to the hazelnut Corylus avellana in NW Italy. Three mitochondrial markers (a portion of the 12S rDNA and a DNA-fragment including partial NADH dehydrogenase subunit I gene, the tRNA Leucine gene and partial 16S rDNA, and partial Cytochrome c oxidase) were compared between individuals collected on birch Betula sp., beech Fagus sp., willow Salix sp., alder Alnus sp. and hazelnut. We found a high genetic distance between A. viridis sampled on different host plants, while individuals sampled on the same host plant were similar despite a considerable geographic gap between sampled areas. Our study supports the general pattern for strong ecological separation between populations living on different host plants.


Introduction
The jewel beetles, family Buprestidae, represent a large group comprising about 15,000 species (Evans et al. 2015). Classification within the Buprestidae family is controversial (Bellamy 2003). The group includes four subfamilies: Schizopodinae, Julodinae, Buprestinae and Agrilinae (Lawrence & Newton 1995), with the latter comprising the genus Agrilus (Lawrence & Newton 1995). The genus Agrilus includes about 2880 species (Bellamy 2008), making it probably the biggest living animal genus (Curletti 2001(Curletti , 2010, and represents a taxonomic puzzle with the boundaries between species, subspecies and morphotypes tied to different host plants making them difficult to be characterised on morphological traits alone. The classification of Agrilus is complex (Curletti 1994), and classification errors have been multiplied by the previous description of numerous races, varieties and forms within the same species.
One of the best-known examples of taxonomic difficulties in this genus is represented by the Palearctic species A. viridis (Linnaeus, 1758) (Jendek 2016). This jewel beetle is polyphagous, a feeding habit rarely encountered in this genus, which could enhance our ability to discover sister species with little morphological differentiation. According to the late C. L. Bellamy (pers. comm. to GC), A. viridis includes 24 synonyms, 15 unavailable names and only one subspecies, the poppiusi Obenberger 1924 from east Siberia (see also Jendek & Grebennikov 2011). Recent studies have shown the presence of different varieties or ecotypes strictly linked to geographical area, altitude or arboreal host species (Brechtel & Kostenbader 2002;Bernhard et al. 2005). These studies considered the birch Betula and the willow Salix and indicated correlation between genetic variants of A. viridis and tree species (Bernhard et al. 2005;Pentinsaari et al. 2014b). The hazelnut tree can host A. viridis as the primary xylophagous species (Corte et al. 2009), as well as A. angustulus (Illiger, 1803), A. graminis Laporte & Gory, 1837, and A. olivicolor Kiesenwetter, 1857 as secondary xylophagous hosts within the jewel beetles (Ciampolini & Ugolini 1975). Jendek and Poláková (2014) reported A. viridis association with 23 genera of host plants through their entire distribution range.
In recent years, the Piedmont hazelnut agriculture has had to deal with an increase in A. viridis attacks. This woodboring jewel beetle digs long and winding tunnels in the cambium of the hazelnut, thus causing a slow decline until desiccation. Probably a combination of factors favoured the intensification of attacks, with the main cause attributed to climatic and agricultural factors, such as the repeated superposition of dry years with production overhead. The phenomenon causes stress in hazelnuts, and the weakened trees easily become victims to the parasite. The symptoms are foliage yellowing during summertime, drying of branches and, in severe cases, the death of the plant due to lymphatic cycle breakage (Ciampolini & Ugolini 1975;Moraglio et al. 2013). During inspections, only undulating or linear cortical reliefs can be observed, without the ability to find holes with larval penetration, while it is possible to discover the emergence holes of adults. The larval dig includes a first involvement of sub-cortical tissue and then proceeds into the xylem (Pollini 2006).
In this study, we investigated the links between A. viridis genetic lineages and their host trees. We utilised three markers (a portion of the 12S rDNA and a DNA-fragment including partial NADH dehydrogenase subunit I gene, the tRNA Leucine gene and partial 16S rDNA and partial Cytochrome c oxidase) of the mitochondrial DNA in jewel beetles sampled on different angiosperms: birch Betula sp., beech Fagus sp., willow Salix sp., poplar Populus sp. and hazelnut Corylus avellana. The aim of our study was to test whether A. viridis hosted on different trees show a clear genetic separation. This could indicate that different lineages are not likely to switch from one host plant to another, which would help in planning future actions targeted at containing this harmful species. Interventions can be limited to plants of agricultural interest without needing to operate on wild plants present in the same area (Maloy 2005).

Sampling
Jewel beetles were collected in the field during autumn through selective cutting of branches infested by A. viridis, after the typical damage caused by the larvae was recognised by one of the authors (GC). The tree branches with larvae were stored during the autumn and winter seasons in a fresh cage, at ambient temperature, until the emergence of adults in spring. Individuals that emerged were identified as A. viridis and were subsequently transferred into test tubes containing pure ethanol and stored in a freezer at −20°C until the DNA extraction procedure. Samples and localities are listed in Table I.

DNA extraction, PCR amplification and sequencing
Each adult individual was sectioned along the median sagittal plane and one half was used entirely for DNA extraction. The genomic DNA was isolated using a commercial kit for the extraction of nucleic acids to silica columns (NucleoSpin® Tissue Machery Nagel). First, thermal shocks in liquid nitrogen were carried out to facilitate the rupturing of the cell. In addition, a lysis buffer (Buffer FLB Macherey Nagel) designed for small amounts of DNA or degraded DNA was used. The DNA extracted after the final centrifugation and resuspension in standard buffer solution (Tris-EDTA pH 8-9) was stored at −20°C for subsequent molecular analysis.
The reactions were set up in a final volume of 25 μL with 10 mM polymerase chain reaction (PCR) buffer, 25 mM MgCl 2 , 2.5 mM for each dNTP, 25 μM for each primer, 0.5 unity/μL of Taq DNA polymerase, 10% BSA, and 0.3-0.5 ng/μL of extracted DNA. The amount of DNA and TAQ polymerase was varied according to the state of degradation of the biological sample. The amplification reactions were obtained in a Bio-Rad C1000 thermal cycler using the following protocol (94°C × 5ʹ), 29-32 cycles at (94°C × 30") (Ta°C × 30") (72°C × 1ʹ), final extension 7 min at 72°C. The DNA concentration was determined on 1.8% agarose gel (TBE 1%) using GelRedTM nucleic Acid stain gel (Biotium Inc., Hayward, California, USA) and quantified by UV-transilluminator Gel Doc XR (Bio-Rad Laboratories Inc., Hercules, California, USA) with Molecular Imager ChemiDoc XRS Systemand Quantity One (Bio-Rad) software. The amplification products intended for sequencing were purified by treatment with ExoSAP-IT ® using the following protocol -(37°C × 45ʹ), (85°C × 15ʹ) -and were sequenced on an ABI 3730XL (Applied Biosystems, Foster City, CA, USA) automated DNA sequencer at the BMR genomics lab (Padova, Italy).
The genetic distances were calculated using the MEGA 6.0 software applying the Kimura 2-parameter model (Kimura 1980;Tamura & Nei 1993). Phylogenetic trees were determined through the MEGA 6.0 software applying the neighbour-joining (NJ) algorithm (Saitou & Nei 1987). The confidence levels of the tree internodes were estimated using the bootstrap procedure (Felsenstein 1985), which reported the indexes as a percentage after 1000 replications and rejected values less than 50.

Results
The COI sequences showed little variability, with only two haplotypes found in the beetles collected on hazelnut and two others found in those collected on willow plants (Table II). Beetles collected on poplar showed two different haplotypes.
The NJ tree shows a genetic distinction between individuals hosted on Corylus avellana and those hosted on Salix (Figure 1). The UPGMA (Unweighted Pair Group Method with Arithmetic Mean) tree (not shown) confirmed the division into major and minor clades which was identified from the NJ tree. The haplotype network (Figure 1) shows that the haplogroup which includes the sequences of individuals on C. avellana is separated by a minimum distance of 40 mutations from that which comprises the sequence on Salix.
The genetic distances within haplogroups found on the same host plant are smaller than those between haplogroups found on different host plants, which confirms the structure of the tree phylogenetic clades (Table III). The largest genetic distances are observed between the haplotypes of A. viridis hosted on hazelnut and willows (0.068-0.072) and between those of alder and willows (0.066-0.068).
The 12S sequences identified four different haplotypes with four polymorphic sites and three informative sites. The ND1-16S sequences identified seven different haplotypes with 42 polymorphic sites and 32 informative sites.
The NJ tree computed on 12S and ND1-16S concatenated sequences and rooted with one sequence of Trachys troglodytes shows that the sequences of hazel and willow found in the same Piedmont area cluster into two distinct groups ( Figure 2). Willow samples from Piedmont were also separated from a willow sample from Germany. The UPGMA tree (not shown) confirmed the division into major and minor clades which was identified from the NJ tree. The haplotype network shows a phylogenetic structure consisting of two haplogroups comprising, respectively, A. viridis samples from hazelnut plants and samples taken from willow, separated by a minimum distance of 35 mutations (Figure 2).
Haplotype diversity (Hd) and nucleotidic diversity (π) values are reported in Table IV. The genetic distance between the haplotypes found on hazelnut and those found on willows in the same Piedmont area was high, assuming values between 0.044 and 0.049 (Table V).

Discussion
The results obtained in this study demonstrated the existence of a significant genetic differentiation between A. viridis hosted on different arboreal trees. This result is in line with previous observations that, in the absence of the host tree, these beetles are not able to switch to other plants (Heering & Biologie 1956), which is a circumstance that is likely to generate genetic diversification within the same geographical area. Our results highlight the high genetic differences between the A. viridis living on different arboreal types and could suggest an incipient case of sympatric speciation, a speciation model in which the ecological separation is more important than geographical isolation to influence population genetics (Filchak et al. 2000;Lechner et al. 2015).
A cut-off value higher than a 2-3% difference in the COI sequences was used in similar studies in order to separate two genetic clusters as different species or as different OTU (operational taxonomic units ;Blaxter 2004;Smith et al. 2005). In the Piedmont area, the genetic distance between A. viridis hosted on hazelnut vs those hosted on willow was approximately 7%. This value is quite high, even in comparison with the distances found between different species of Agrilus or among other Buprestidae in Germany (Hendrich et al. 2015), and may suggest a future treatment of the species of these jewel beetles as distinct taxonomic entities. However, it should be noted that the distances between species of Coleoptera can be higher than that observed in other animals (Pentinsaari et al. 2014a), and that for the other mitochondrial markers (12S and ND1-16S), the genetic distance is much less. As pointed out in a previous molecular study (Bernhard et al. 2005) and in our data, the genetic distances found utilising the 12S and ND1-16S mitochondrial markers are very low between A. viridis hosted on different host plants. Taken together, all the data suggest that the viridis is probably a species complex (Bernhard et al. 2005), even when the presence of cryptic species within the complex is considered (Hendrich et al. 2015).
Besides molecular investigations, it would be worthwhile to extend the search for new biometric distinguishing features in parallel, which would allow a better discrimination of lineages hosted by the C. avellana trees. Indeed, morphometric differences in the male genitalia have been already reported in A. viridis hosted on Betula and Salix (Pentinsaari et al. 2014b). Moreover, it would be interesting to deepen the knowledge about this species to clarify whether A. viridis using the hazelnut as larvae have reached a complete reproductive isolation from those hosted on other trees, or if they can still interbreed.
Recognising similar but distinct taxonomic units is important for disease and pest control. Our data are of interest for applied agronomy, since the genetic separation between the lineages of viridis growing on different host plants suggests that any action targeted at containing this harmful species can be limited to the plants of agricultural interest without the need to operate on other wild plants present in the same area.
Future studies could examine the genetic differentiation across the whole geographical range of the species, from West Europe to East Asia, embracing samples from different altitudes, climates and host plants.