The genetic region used to identify species in the B. tabaci species complex is the 3′ end of the mtCOI gene. There are two sets of primers commonly used 1) C1-J-2195 and TL2-N-3014 (Simon et al., 1994) and Btab-Uni primers (Shatters et al., 2009). The PCR conditions are: a 30 μl PCR reactions heated at 94°C for 2 min followed by 35 cycles of 30 s at 94°C denaturation, 30 s at 53°C annealing, 1 min at 72°C extension and a final extension of 72°C for 10 min in a MJ Research PTC-200 Peltier thermal cycler. The PCR reactions are composed of 27 μL Platinum PCR SuperMix (Invitrogen, catalog number 11306-016), 1 μL forward primer (10 ρ mol), 1 μL reverse primer (10 ρ mol), and 1 μL DNA template. Prior to sequencing, the amplified products were cleaned using the montage PCR filter units from Millipore (catalog number UFC7PCR50). It is important for the whitefly community to continue using this region of mtCOI to avoid diversion of resources and to build on the informatics power of the historical data that is already publicly available (Boykin et al., 2012a). Publicly available data has been abstracted, curated, and placed in a database that can be found at: http://dx.doi.org/10.4225/08/50EB54B6F1042 (supported through the Commonwealth Scientific and Industrial Research Organisation, Australia).

Prior to conducting phylogenetic analyses, new sequences should be named in a format that will easily integrate into the global dataset. The recommended format is: Species_Country_GenBank Accession number_host. Prior to identification analyses, the species will be unknown, it is recommended to use UNKNOWN_country_GenBank Accession Number_host. After the analyses described below, replace the UNKNOWN with the species identified in the global phylogeny.

All 34 putative species in the B. tabaci complex and closely related sister taxa must be included in the systematic investigation of the B. tabaci species. B. tabaci has a global distribution and removing the other members (to focus on one geographic region) of the complex will mislead the investigation due to incomplete taxon sampling (Boykin et al., 2007). Taxon sampling is crucial for phylogenetic systematics and much of the early confusion in B. tabaci systematics was due to incomplete taxon sampling. In addition, accuracy of inferences about evolutionary processes obtained from phylogenetic analyses improved significantly through taxon sampling (Heath et al., 2008; Boykin et al., 2013). The global dataset is designed strictly as a repository of unique haplotypes to aid in species identification by having a large proportion of the known diversity available in the one place; it is not designed to track global distribution. All sequences in the dataset are originally lodged in GenBank and have accession numbers. Sequences have been checked for gaps, provenance, pseudogenes and ambiguous bases are <0.8%; as such complies with many of the data quality elements of DNA barcoding. Sequence length is 657 bases. The dataset also contains 21 outgroup species Aleurocanthus camelliae, Aleurocanthus spiniferus, Aleurochiton aceris, Aleurodicus disperses, Aleurodicus dugesii, Aleurotrachelus camelliae, Bemisia afer, Bemisia atriplex, Bemisia berbericola, Bemisia emiliae, Bemisia on Rhagodia parabolic, Bemisia subdecipiens, Bemisia tuberculata, Neomaskellia andropogonis, Tetraleurodes acacia, Trialeurodes abutilonea, Trialeurodes lauri, Trialeurodes ricini, Trialeurodes vaporariorum, Trialeurodes ricini, Vasdavidius concursus. The model of molecular evolution for the global data set is GTR + I + G.

Figure 1 shows three options for phylogenetically identifying members of the B. tabaci species complex. We recommend that analysis be done using Geneious (http://www.geneious.com/) because it is user friendly and contains all necessary programs for alignment, model selection and phylogenetic analyses. Geneious is available with a 14 day free trial period. It requires a license. Geneious offers student, non-commercial and commercial licenses and special prices are available for academics based in countries with annual GDP <$USD 2000 per capita. Within Geneious the alignment programs MAFFT (Katoh and Toh, 2008), Muscle (Edgar, 2004) and ClustalX (Thompson et al., 1997) are available. In addition, the phylogenetic programs RaxML (Stamatakis, 2006), GARLI (Brauer et al., 2002), PhyML (Guindon et al., 2010), and MrBayes (Ronquist et al., 2012) are implemented in a user friendly format.

Depending on the specifications of your computer, it might be necessary to conduct the analyses utilizing online high performance computing (HPC) resources. The Cyberinfrastructure for Phylogenetic Research (CIPRES) Science Gateway (Miller et al., 2010) is a portal to conduct all aspects of phylogenetic analyses using high performance computing (HPC) facilities. Users can create a login at: http://www.phylo.org/portal2/ and all necessary programs for phylogenetic analyses are available. In addition, Phylogeny.fr (http://www.phylogeny.fr/version2_cgi/index.cgi) is available for Robust Phylogenetic Analysis For The Non-Specialist (Dereeper et al., 2008), which contains multiple sequence alignment and phylogenetics programs as well as tree visualization software. Finally, a resource for sequence manipulation is available from the European Bioinformatics Institute, which is part of the European Molecular Biology Laboratory (EMBL-EBI): http://www.ebi.ac.uk/services however, it has limited phylogenetic tree reconstruction resources.

All the necessary programs needed for a robust phylogenetic analysis of the B. tabaci species and subsequent species delimitation are available freely from independent sources. The first step is to download the global dataset (see above) and align unknowns using MAFFT (Katoh and Toh, 2008). To verify the quality of the data the sequences are translated to make sure there are no stop codons or inadvertent frame shifts using JalView (Waterhouse et al., 2009). The model of molecular evolution should be determined using jModeltest (Posada, 2009) and the phylogenetic method of choice is a Bayesian approach using MrBayes (Ronquist et al., 2012). To assess the convergence of the Bayesian runs utilize Tracer (Rambaut and Drummond, 2010) and visualize the.con.tre file in FigTree (Rambaut, 2012). Further species delimitation analyses maybe necessary and should be done following the methods of Boykin et al. (2012b) references and therein.

The global dataset is curated to include the names of the species as the first part of the sequence title, this facilitates the identification of unknowns when all sequences in the dataset are included in the phylogenetic analyses. Assignment of unknown sequences is based on phylogenetic placement of the unknown in an already defined species, thus, emphasizing the need to include all sequences in the global data set. There is the possibility that the unknown will not be assignable to an already identified species clade i.e., the sequence divergence exceeds the delimitation bounds, when this occurs, species delimitation tests are needed and have been outlined in Boykin et al. (2012b). Researchers need to name species in accordance with established nomenclature. This consistency will ensure ease of communication and help minimize confusion over identity.

Beyond the phylogenetic programs listed in Figure 1, there are online resources for phylogenetic species delimitation including an online discussion board- phylobabble (http://phylobabble.org) and the evolutionary directory, evoldir, http://evol.mcmaster.ca/evoldir.html. The most comprehensive list of software pertaining to genetic analyses is maintained by Professor Joe Felsenstein and found at: http://evolution.genetics.washington.edu/phylip/software.html

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.