A simple PCR-based method for the rapid and accurate identification of spider mites (Tetranychidae) on cassava

The morphological identification of mites entails great challenges. Characteristics such as dorsal setae and aedeagus are widely used, but they show variations between populations, and the technique is time consuming and demands specialized taxonomic expertise that is difficult to access. A successful alternative has been to exploit a region of the mitochondrial cytochrome oxidase I (COI) gene to classify specimens to the species level. We analyzed the COI sequences of four mite species associated with cassava and classified them definitively by detailed morphological examinations. We then developed an identification kit based on the restriction fragment length polymorphism–polymerase chain reaction of subunit I of the COI gene focused on the three restriction enzymes AseI, MboII, and ApoI. This set of enzymes permitted the simple, accurate identification of Mononychellus caribbeanae, M. tanajoa, M. mcgregori, and Tetranychus urticae, rapidly and with few resources. This kit could be a vital tool for the surveillance and monitoring of mite pests in cassava crop protection programs in Africa, Asia, and Latin America.


Scientific Reports
| (2020) 10:19496 | https://doi.org/10.1038/s41598-020-75743-w www.nature.com/scientificreports/ identification kit based on restriction fragment length polymorphism-polymerase chain reaction (RFLP-PCR) using subunit I of COI to rapidly and definitively identify four species of cassava mites, and to democratize its access to field practitioners to ameliorate the current labor-intensive and time-consuming taxonomic identification methodology.

Materials and methods
Establishment of mite colonies. Cassava leaves with mites and mite damage were collected from four locations at Valle del Cauca, Colombia: a producer field located at the locality of Potrerillo (municipality of Palmira); an experimental field; and two greenhouses at the International Center for Tropical Agriculture, Palmira (CIAT, from its Spanish acronym) ( Table 1). At each location, at least 30 males and 30 females of the same morphotype were collected under a stereomicroscope and placed on one-month-old cassava plants for rearing. Each colony was checked every three days until teliochrysalids, the last developmental stage before adult emergence, were detected. From each colony, one to three couples of newly emerged females (virgin) and non-virgin males were placed on a new cassava plant in a rearing room (25 °C and 70% relative humidity) at the CIAT (Table 1) to establish a new colony from these founders. The new colonies were maintained until at least a hundred adult mites were detected; at this point, the leaves with mites were collected and placed into a freezer at − 20 °C for several minutes. Dead mites were collected and placed into an empty vial (Eppendorf Tubes™, Germany). For each colony, 13-28 adults of both sexes were mounted for morphological classification; the remaining individuals were preserved at − 80 °C and used for molecular studies.
Morphological identification. Seventy-seven adults from the four colonies indicated in Table 1 were cleared in lactophenol solution and mounted on slides in Hoyer's solution, following the guidelines of Krantz and Walter 14 . Each specimen was examined, and the most relevant morphological structures for identification were photographed with scales using a Canon Eos 60D camera attached to an Olympus light microscope BX43, and with a Nikon Digital Sight DS-Ri1 camera attached to a Nikon Eclipse Ni-U 90 microscope. The measurements (micrometers μm) were developed in Adobe Illustrator (Adobe Systems Incorporated, USA). The body length represented by the idiosoma, was measured with and without gnathosoma, and the setae were measured from the base to the tip; the description of characters follows Lindquist 15 . Identification to the genus level was performed using the taxonomic key of Bolland et al. 16 , and the species were identified with different taxonomic keys 17 , morphological descriptions, illustrations [18][19][20]  The morphological identifications were made by Aymer Andrés Vásquez-Ordóñez (the CIAT, see the "Morphological identification" section). The number of females and males, respectively, examined for identification is indicated in parentheses. c The location, geographic coordinates, and collection date correspond to data for the collection of mites used to establish the first colonies. The mites collected in the greenhouses were wild mites, not cultured mites. . COI sequence contigs were assembled from three clones per species to identify unique haplotypes with a discrimination requirement of 100% identity. All mite sequence identities and similarities were calculated using BLAST. Genetic distances were calculated with the alignment sequences for the four species. Additionally, we included COI sequences reported in GenBank from other species belonging to the same tribe (Tetranychini) ( Table 2). The alignment process was carried out using both Mezquite and MEGA v7.0 software 23 . The phylogenetic reconstruction analyses were performed using the parameters from the Neighbor-Joining model 24 and Kimura-2 model 25 for nucleotides (nt) with 1000 bootstrap replications.
Restriction mapping and COI-RFLP assay. COI sequences belonging to each of the four mite species morphologically identified in this study, and the 15 Tetranychini and 1 Bryobia COI sequences obtained from GenBank for the species identified as closely related to them (Table 2), were digested in silico through the web portal NEBcutter V2.0 26 . In total, 36 restriction enzymes were evaluated to estimate whether the COI region's restriction pattern could be used to identify the four mite pest species associated with cassava and to select the most informative restriction pattern and its enzyme(s). Of the 36 commercially available restriction enzymes, three allowed a clear differentiation of the four mite species in silico. Two units of the restriction enzymes AseI, MboII, and ApoI were used to digest 2 µg of the PCR-amplified COI region from T. urticae, M. caribbeanae, M. mcgregori, and M. tanajoa. The final reaction volume was 50 µl and included 10% of 1X supplied buffer and ultrapure water. Reactions were incubated for 1 h at 37 °C or 50 °C according to the manufacturer's specifications. Restriction digestion products were run in 2% agarose gel in BS buffer (10% boric acid, 2% sodium hydroxide) at 80 V for 3 h. Gel staining was carried out with ready-made SyBR Safe (Invitrogen, USA). Restriction pattern images were captured with a GelDoc™ BioRad documentation system, and size estimation of the digestion patterns was performed using PyElph V.1.4 software 27 .

Results
Morphological identification. In the Valle del Cauca region of Colombia, cassava is affected by different mite species. A survey conducted in 2012, in which cassava leaves infested by mites were collected and inspected under a stereoscope, identified at least four morphotypes on a single leaf. Four mite colonies, one from each morphotype, were then established on fresh cassava leaves. From the resulting offspring, mating couples were selected and allowed to breed on new cassava leaves to enable a full taxonomic and molecular identification ( Table 1). A complete morphological characterization was undertaken on each mite colony, allowing the identification of four species: Mononychellus caribbeanae 28 , M. mcgregori 29 , M. tanajoa 30 , and Tetranychus urticae 31 . A description of the diagnostic morphological characteristics used to identify these species is provided in detail.
Tetranychus urticae Koch, 1836 31 . Diagnosis. Female: Length of body with gnathosona of 500 38 , with elliptical body shape (Fig. 3). Empodia with three pairs of proximoventral hairs and without dorsomedian spur. Peritremes with a hook longer than 15. Dorsohysterosomal striae longitudinal between members of setae e1 and members of setae f1, forming a diamond shaped pattern. Tarsus I with the socket of proximal duplex setae distal (> 10) to the socket of four proximal tactile setae. Preginital striae broken medially and solid laterally. Male: Empodium of leg I-II with mediodorsal spurs longer than 2 µm. Terminal knob of aedeagus less than twice (about 1.5x) as wide as neck and parallel or forming a small angle (0-20°) with axis of shaft; knob rounded or angulated dorsally and with anterior and posterior acute projection (Fig. 3) (Fig. 1B,C). Male: Knob of aedeagus with anterior margin convex and with margin dorsal and ventral on acute angles approximately equal (Fig. 1E)…Mononychellus caribbeanae.
Sequencing and phylogenetic analysis. Standard PCR was performed using the COI-barcode primers on DNA extracted from the four mite species. A positive amplification of the COI region was resolved in 1.5% agarose ( Fig. 2A). The COI region of each sample was then cloned, and positive clones containing an insert of the relevant size were confirmed by fragment amplification using T7 and SP6 primers (Fig. 2B). A total of 12 samples, three individual clones per species, were sent for sequencing. The length of the DNA sequences obtained was 709 bp, with a unique sequence haplotype for each mite species. Sequence analysis confirmed the presence of the 219 amino acids of the COX1 gene present in the mitochondrial genome and commonly referred to as the COI sequence or the Barcode of Life 42 .
A BLAST search of the four mite haplotype sequences against the NCBI's GenBank revealed six mite genera closely related to these samples. Thus, we proposed a global alignment that included 16 COI DNA sequences plus the four haplotype sequences from this study; this alignment was used in a phylogenetic reconstruction.
A phylogenetic tree was constructed based on the nucleotide sequence alignment, with bootstrap values above 70% based on 1000 replicates; these values are indicated at each node (Fig. 3). Most of the nodes in the tree were highly supported. An example species from the genus Bryobia was used as an outgroup.
This phylogenetic tree clearly separates the genus Tetranychus (T. urticae) from the genus Mononychellus (M. tanajoa, M. mcgregori, and M. caribbeanae). The T. urticae example from this study clusters together with the GenBank example of the same species collected in Korea. Likewise, the three species of Mononychellus collected in the Valle del Cauca clustered together; they are the first to be sequenced and reported to GenBank.
COI-RFLP analysis. COI sequences belonging to 20 mite species were digested in silico with 36 restriction enzymes with the aim of identifying a set of enzymes that could be used for easy identification. We found that the digestion patterns obtained with AseI, MboII, and ApoI allowed for the full identification of the four mite species described here. Restriction digestion of the PCR-amplified COI regions of the four mite species confirmed the in silico results. The banding patterns obtained from the agarose gel electrophoresis showed that AseI and MboII resolved eight and five distinctive bands, respectively, across all mite species with sizes above 100 bp, demonstrating the ease and accuracy of this method. ApoI showed only three distinctive bands, and did not distinguish between M. tanajoa and M. caribbeanae, limiting its potential use for quick diagnosis (Fig. 4).
The banding patterns obtained after restriction enzyme digestion of the COI region could be replicated in two independent molecular laboratories of the cassava program at the CIAT (the Genetics and the Virology laboratories), confirming the reproducibility and accuracy of RFLP-PCR. AseI showed the best profile for differentiating among the four mite species, followed by MboII. Even though ApoI only allowed the definitive identification of two mite species, its inclusion alongside the other two enzymes would make an identification kit for the four species more robust.

Discussion
In the present study, we undertook a classical morphological identification of four mite species associated with the cassava crop in Valle del Cauca and then used those results to confirm the feasibility of a proposed molecular kit for easier identification.
The morphological identification of the four mite species assessed here was challenging because it required extensive technical expertise to prepare and mount the samples as well as a deep knowledge of the taxonomy of mites. Rogo et al. 43 and Guerrero et al. 37 reported problems in the identification of Mononychellus specimens, whereby apparent differences in morphological characteristics could have been the result of bad mounting or deterioration of the microscope slide. The other limiting factor affecting the morphological characterization of Mononychellus spp. is the limited number of morphological characteristics available for accurate characterization. For instance, in M. tanajoa, the traditional characteristic is the longitude of dorsal setae that presents a continuous variation among the populations, a factor that does not facilitate distinguishing them from nearby species 37,43 . Another important characteristic is the shape of the aedeagus; although its descriptions have presented problems, comprehensive research indicates the same form as documented in this paper 37,43 , demonstrating its usefulness.  www.nature.com/scientificreports/ Not only is the identification process for mites using morphological traits time consuming and demanding of specialized taxonomic expertise, but the task is also amplified by the fact that the Tetranychidae family is composed of > 1200 species 16 . The organism's small size and significant phenotypic plasticity also increase the complexity associated with its morphological identification 44 .
DNA barcoding could circumvent these limitations 45 . The COI gene sequence has been used extensively for insect identification; for instance, it has been used successfully in the identification of whitefly 12,46 , rhopalids 47 , and mirids 48 . On this basis, we sought a method to reduce the time and increase the accuracy of mite species identification that would be usable by other scientists or agriculture research practitioners responsible for pest surveillance and monitoring, choosing to characterize the genetic information contained in the COI region of the COX-1 gene of four mite species taxonomically identified and associated with the cassava crop in the Valle del Cauca, Colombia.
The COI region has been used successfully to characterize the mite species complex [49][50][51] ; thus, we believed this type of data would aid the species identification of mites associated with cassava. Among the most important species, only a partial COI sequence for M. progressivus is available at NCBI 52 ; thus, the four haplotype COI sequences reported here constitute the first to be available through GenBank for the genera Mononychellus and Tetranychus. Comparing molecular markers with diagnostic morphological traits allows the building of a species-specific sequence library to aid the accurate identification in the absence of a well-trained entomologist. COI sequences from mites associated with other plants, including T. urticae 44,[53][54][55] , were used to conduct a phylogenetic relationship analysis, permitting the construction of a phylogenetic tree from 20 COI sequences, including the outgroup Bryobia sp. 44 . Two well-defined clades were apparent: one from a monophyletic group that corresponded to the Tetranychus genus (and in which our T. urticae sample was grouped in close relationship with a sample reported by the Korea Research Institute of Bioscience and Biotechnology); and the second from a polyphyletic group composed of four different genera but including a well-defined Mononychellus clade.
Most mite identification studies agree on the challenge of identifying the species by morphology alone, because that is time consuming and defeats any attempt to rapidly provide information on pest distribution. Large-scale surveys involving DNA sequencing may be possible, but are probably too costly. Hence, DNA-based methods not involving sequencing become attractive. Restriction enzyme-mediated genotyping is a more costeffective approach for identification with large sample sizes. It requires that a restriction digestion of a PCR product (RFLP-PCR) be obtained from a variable well-known gene, such as the COI sequence. We have shown that a standard PCR amplification of the COI region, followed by enzymatic digestion with AseI or MboII allows the unequivocal identification of all four mite species associated with the cassava crop, with no sequencing of the COI region necessary. Our method does not compare sequences to classify one sample into a species; therefore, our method does not take into account the number of intra-or interspecific variations, which have been problematic, given that in some species of mite, intraspecific variation may exceed interspecific variation 44 . Our kit is based on the target sequences of the restriction enzymes and the patterns they generate when digesting the sequence of the COI region. This approach to a rapid and accurate mite species identification is cost effective and easy to implement in a very basic molecular biology laboratory. Thus, incorporating RFLP-PCR for the COI gene into a routine mite surveillance and monitoring program should quickly and simply identify any potential outbreak of an exotic type not reported previously. The restriction digestion patterns from AseI and MboII clearly showed how the COI-RFLP is as effective a method for mite identification as that proposed for whiteflies and the oriental fruit fly 12,56 .
In conclusion, our COI-based RFLP-PCR kit, developed using a multi-disciplinary approach that included gene data, morphological traits, and bioinformatics pipelines, has been shown to unequivocally identify T. urticae, M. caribbeanae, M. mcgregori, and M. tanajoa. This kit can identify mite species accurately, cheaply, and rapidly, and could become a useful tool in crop protection programs monitoring and surveying mite pests.