Discovery and Identification of Meloidogyne Species Using COI DNA Barcoding

Figure 1 displays a ML tree of 322 Meloidogyne sequences including 117 sequences from GenBank and 205 sequences from the University of Nebraska–Lincoln Nematology Laboratory. ML partitions these sequences into 19 groups with bootstrap support values from 93 to 100 (Tables 1, 2, Fig. 1). Three unique GenBank sequences represent Meloidogyne haplanaria Eisenback et al., 2004, Meloidogyne duytsi Karssen et al., 1998, and Meloidogyne artiellia Franklin, 1961 as distinct from other sequences in the dataset, but without additional supporting sequences.

Groups 1 to 3 form a clade characterized by the loss of a single amino acid (3 bp) resulting in a 721 bp sequenced region. This shared deletion unites M. haplanaria, and M. enterolobii Yang & Eisenback, 1983 with the so-called “major” tropical apomictic species of Meloidogyne (Elling, 2013). Included in this group are sequences representing M. arenaria (Neal, 1889) Chitwood, 1949, M. incognita (Kofoid & White, 1919) Chitwood, 1949, M. javanica (Treub, 1885) Chitwood, 1949, as well as M. hispanica Hirschmann, 1986, M. floridensis Handoo et al., 2004, M. konaensis Eisenback, Bernard & Schnitt, 1995, M. luci Carneiro et al., 2014, and M. inornata Lordello, 1956 (Table 2). The same amino acid deletion is also found in unnamed group 12. Within group 1, the COI sequences are nearly identical with a few notable exceptions. Four substitutions are shared by three specimens identified as M. konaensis, including GenBank accession KU372176, identified as Meloidogyne sp. 2 TJ-2016 T316 on Beta vulgaris from Spain in Janssen et al. (2016). Two substitutions are shared by specimens identified as M. incognita grahami, originally described as M. grahami Golden & Slana, 1978, and considered distinct from M. incognita based on reproduction on NC-95 tobacco, a cultivar with resistance to M. incognita, plus a greater juvenile length and a distinctive perineal pattern (Golden and Slana, 1978).

Outside of clades 1-3 there are 11 other described species represented by a minimum of a single COI sequence. Meloidogyne hapla Chitwood, 1949 is represented by specimens from 10 U.S. states and two specimens from Nepal. There are multiple haplotypes within M. hapla and possibly some population substructure within the species. Group 17 identified as M. chitwoodi Golden, O’Bannon, Santo & Finley, 1980 and M. fallax Karssen, 1996 is virtually homogeneous except for a 5-bp difference between the two species. Within group 6, identified as M. partityla Kleynhans, 1986 one specimen collected from Big Thicket National Preserve, Texas comes from a native lowland plant community, compared with other specimens from New Mexico collected from commercial pecan (Carya illinoinensis (Wangenh.) K. Koch) production.

There are seven groups labeled as unnamed, all with sequence derived from j2 stage specimens except for N4431 and N4496 which were males collected from native chestnut (Castanea dentata (Marshall) Borkh.) in Great Smoky Mountains National Park (GRSM), North Carolina. All specimens in the unnamed groups 4, 5, 9 to 13 were isolated from soil samples within Gulf Coast or Eastern North American forests. Groups 9 and 12 were associated with American beech, (Fagus grandifolia Ehrh.) and chestnut or oak, respectively. Measurements of the unidentified juveniles are presented in Table 3, and Fig. 2 illustrates juveniles from three of the unnamed groups.

Fig. 1

Fig. 2

The COI gene region used as a diagnostic marker in this study appears to discriminate many of the described species of Meloidogyne. It does not separate the apomictic “major species” and their close relatives, except possibly M. konaensis and M. incognita grahami. Other mitochondrial genes such as NAD 5 may help resolve some of those species boundaries (Janssen et al., 2016). Aside from an inability to discriminate among the tropical clade 1 species, there are advantages to using COI as a DNA barcode. As a protein coding gene, nucleotide alignment is easier compared with non-protein coding genes. Taxonomic resolution is at the population and species level, although for many genera, mutational saturation, lineage extinctions, or inadequate sampling may obscure deeper relationships that aid in the recognition of species groupings. Nonetheless, COI barcodes in combination with an adequately curated sequence database, provide a powerful tool for identification and discovery. The limitation of DNA barcoding without a corresponding database is illustrated by the unnamed groups in the Meloidogyne dataset. For example, there was an expectation that focal samples from soil around individual chestnut and oak trees in GRSM might yield Meloidogyne querciana Golden, 1979 which was described from northern red oak (Quercus rubra L.) and chestnut hosts within the same ecoregion. Indeed Meloidogyne specimens were found in these samples, however, the barcode data demonstrate that multiple COI lineages were associated with chestnut and oaks in the park. Similarly, unnamed lineages were also discovered associated with American beech and baygall plant communities in Big Thicket National Preserve, Texas (www.nps.gov/bith/plant-communities.htm). These results indicate that considerable Meloidogyne diversity exists in the primary and secondary forests of eastern and southern United States. Characterization of this diversity by COI barcoding allows us to rule out described species with representation in the COI database, yet neither COI barcode nor morphometrics of juvenile specimens permits unequivocal assignment of a species name to these specimens. For these unknown specimens a more complete taxonomic analysis that includes obtaining adult stages will be required before a barcode sequence can be linked to a formal Latin binomial.