Oligonychus perseae Tuttle, Baker & Abbatiello (Acari: Tetranychidae) is an economically important foliar pest of avocados from Mexico. Invasive O. perseae populations became established throughout the commercial avocado system in California (USA) during the early 1990s, but the putative geographic origin(s) of the California O. perseae populations has not been investigated. To address this shortcoming, a series of population genetic analyses for O. perseae populations representative of a broad geographic sample range was conducted. This study identified a single mitochondrial cytochrome c oxidase subunit I (COI) haplotype match between O. perseae populations collected in California with those from 4 municipalities in Michoacán, Mexico, part of the presumptive native range of this pest. Interestingly, this haplotype also was collected from avocados at 2 locations in Baja California, Mexico, and it was identified from a representative sample from Israel where O. perseae is an invasive avocado pest. Molecular data confirm the likely Mexican origin of invasive O. perseae populations in California, and may help delimit the geographic area to be searched for coevolved natural enemies of O. perseae that could be introduced into California as part of a future classical biological control program targeting this pest. Moreover, molecular results uncovered significant and concordant genetic divergence in both mitochondrial (COI) and ribosomal DNA markers, i.e.,internal transcribed spacer 2 (ITS2) and a section of the 28S gene region, pointing to the potential occurrence of a cryptic species complex within O. perseae. The implications of these findings on future taxonomic and molecular work for O. perseae are discussed.
The persea mite, Oligonychus perseae Tuttle, Baker & Abbatiello (Acari: Tetranychidae), is a foliar pest of avocados, Persea americana Miller (Lauraceae). Both pest and host are native to Mexico, which historically has been the world's largest commercial producer of avocado fruit (Tuttle at al. 1976; Chen et al. 2009; USDA—FAS 2015). Foliar feeding by O. perseae populations induces premature defoliation, which in turn increases the risk of sunburnt fruit and reduced fruit yield (Aponte & McMurtry 1997; Maoz et al. 2011). Crop damage resulting from O. perseae infestations is also a concern in other avocado growing countries where adventive populations of this spider mite have successfully established, including Costa Rica (1978), the United States (California) (1990), Israel (2001), Spain (2004), Portugal (2005), and Italy (2014) (Lara & Hoddle 2015a; Zappalà et al. 2015).
In California, O. perseae was first recorded in 1990 on residential avocado trees in San Diego County. Thereafter, O. perseae spread rapidly throughout commercial avocado growing areas, covering an area extending more than 400 km from San Diego County to San Luis Obispo County (Bender 1993). Oligonychus perseae is 1 of several exotic species that have successively established in the California avocado system since the 1980s. This complex of exotics includes the red banded whitefly, Tetraleurodes perseae Nakahara (Hemiptera: Aleyrodidae) (detected in 1982), avocado thrips, Scirtothrips perseae Nakahara (Thysanoptera: Thripidae) (detected in 1996), the avocado lace bug, Pseudacysta perseae (Heidemann) (Hemiptera: Tingidae) (detected in 2004), and the thrips Neohydatothrips burungae (Hood) (Thysanoptera: Thripidae) (detected in 2004) (Hoddle 2005). All of these species require avocado foliage on which to feed to complete their life cycle, but O. perseae is the primary foliar pest of avocados in California, and tends to be most problematic over the spring and summer months.
California produces 90% of all fresh avocados in the USA with an annual value exceeding $300 million (US), and more than 90% of avocados produced in California are cultivar ‘Hass’ (CAC 2015). This high market-value cultivar, which was developed in California, has proven to be highly susceptible to foliar feeding injury from O. perseae (Kerguelen & Hoddle 2000). The severity of O. perseae infestations in California can readily exceed the recommended action threshold of 50 to 100 mites per leaf (Maoz et al. 2011; Lara & Hoddle 2015b). Thus, effective sampling and control options are needed to prevent O. perseae damage during the growing season. Although pesticide applications are cost-effective, pesticide resistance has been detected among O. perseae populations sampled from California (Humeres & Morse 2005). To mitigate resistance development, new pesticide chemistries have been registered, and complementary sampling plans have been developed to promote judicious pesticide use in commercial avocado orchards (Hoddle & Morse 2012; Lara & Hoddle 2015b). Nevertheless, heavy reliance on chemical control remains an unsustainable management strategy for O. perseae.
As an alternative, augmentative biological control through releases of a commercially-available phytoseiid mite, Neoseiulus californicus (McGregor), have been shown to reduce O. perseae populations to non-economic levels when deployed at a rate of 2,000 predators per tree (Hoddle et al. 2000). Unfortunately, releases of N. californicus are required annually to ensure O. perseae control and the associated expense renders this approach impractical for management of this pest in commercial orchards. Additionally, native phytoseiids, Euseius species, may also feed on O. perseae in commercial orchards, but their primary feeding ecology as pollen specialists limits their peak abundance to spring months when avocado trees are blooming (McMurtry & Johnson 1966; McMurtry et al. 2013). Oligonychus perseae densities are typically low during the spring, and consequently, this temporal asynchrony of Euseius species with O. perseae phenology results in these predators being unable to provide reliable control of O. perseae populations during summer when outbreaks of this pest are likely to occur (Yee et al. 2001).
The limited efficacy of currently-available biological control options has generated interest in foreign exploration efforts to search for potentially effective natural enemies of O. perseae from its native range in Mexico for possible introduction into the California avocado system. Classical biological control attempts to re-associate host-specific coevolved natural enemies from the native range of a target pest with the invasive population in the introduced range (Hoddle 2016). Thus, determining the geographic origin of invasive O. perseae populations in California is a fundamental step in the development of a classical biological control program targeting O. perseae in California. Population genetics can provide insight into the probable geographic areas of origin for invasive pests (Rugman-Jones et al. 2007, 2012) but limited molecular work has been conducted on O. perseae populations in Mexico with samples from the Mexican states of México and Michoacán only (Guzmán-Valencia et al. 2014). A broader geographic sample is required before meaningful inferences about the origin of Californian populations can be drawn. Here, we present results from the first attempt to characterize genetic variation among native O. perseae populations across Mexico (with a focus on important avocado growing regions) and exotic populations in California, Israel, and Spain.
Materials and Methods
Adult O. perseae were collected by JRL and MSH from infested avocado foliage in California (USA), 6 states in Mexico, and cultivar ‘Hass’ production areas in Costa Rica (Fig. 1). Oligonychus perseae specimens from Spain were collected by E. Hernandez-Suarez (Canarian Institute of Agrarian Research, Tenerife, Spain) and M. Montserrat (Institute for Mediterranean and Subtropical Horticulture, Málaga, Spain) (Table 1). In Mexico, sampling was primarily conducted in the state of Michoacán because it is the primary commercial producer of cultivar ‘Hass’ avocados in Mexico (Sánchez-Colín et al. 2001; USDA—FAS 2015), and prior to 27 Jun 2016, was the only Mexican state that could legally export avocado fruit to the USA (Boriss et al. 2006; USDA-APHIS 2016), thus making Michoacán a recognized potential source of O. perseae populations in southern California. Specimens were collected from multiple leaves on individual trees directly into labeled vials containing 95% ethanol, and subsequently stored at −20 °C. Identity of the specimens was initially assessed using morphological and behavioral diagnostic characteristics of O. perseae (Ochoa et al. 1994; Aponte & McMurtry 1997), and subsequently confirmed by sequencing the internal transcribed spacer 2 of (ITS2) ribosomal RNA (see below), and matching the sequence with deposited sequences of O. perseae from Mexico (accessions KC568365-KC568386), Spain (accession GU565305), and Israel (accessions DQ656456-DQ656458), through BLAST searches using GenBank.
DNA EXTRACTION AND AMPLIFICATION
Whole genomic DNA was extracted from individual female specimens using the EDNA HiSpEx tissue kit (Saturn Biotech, Perth, Australia), following the manufacturer protocol for extraction from 1 mm3 of tissue, but reducing the reagent volumes by 75% (total extraction volume 25 μL). The mitochondrial (mtDNA) cytochrome c oxidase subunit I (COI) and ribosomal (rRNA) internal transcribed spacer 2 (ITS2) region were amplified using the primer pairs, 5′-ATATGCTTAAATTCAGCGGG-3′ and 5′-GGGTCGATGAAGAACGCAGC-3′; and, 5′-TGATTTTTTGGTCACCCAGAAG-3′ and 5′-TACAGCTCCTATAGATAAAAC-3′, respectively (Navajas et al. 1998). PCR was performed in 25 μL reactions containing 2 μL of DNA template (concentration not determined), 1× ThermoPol PCR buffer (New England Biolabs, Ipswich, Massachusetts), 200 μM each dNTP, 1 μL of bovine serum albumin (BSA) (New England Biolabs, Ipswich, Massachusetts), 0.2 μM of each primer, 2 mM MgCl2, and either 1 (ITS2) or 2 (COI) units of Taq polymerase (New England Biolabs, Ipswich, Massachusetts). Amplification was performed in a Mastercycler 5331 (Eppendorf, Hamburg, Germany) programmed for: an initial denaturing step of 4 min at 95 °C, followed by 35 cycles of 1 min at 92 °C, 1 min at 45 °C, 1.5 min at 72 °C; and a final extension of 3 min at 72 °C.
Based on unforeseen but corroborating levels of variation in COI and ITS2 sequences (see Results), a second, more conserved region of rRNA, 28S, also was amplified for representatives of each mitochondrial haplotype, using the 28sF3633 and 28sR4076 primers and protocol detailed in Rugman-Jones et al. (2010).
AMPLIFIED DNA CLEANING AND SEQUENCING
Successful amplification of each marker was confirmed by visualizing PCR products after electrophoresis on 1% agarose gels stained with ethidium bromide. PCR products were purified using the Wizard® PCR Preps DNA Purification System (Promega, Madison, Wisconsin) and sequenced in both directions at the Institute for Integrative Genome Biology, University of California (Riverside, California). Forward and reverse sequence reads were aligned manually using BioEdit version 7.1.11 (Hall 1999).
COI sequences were collapsed into haplotypes, and the number and nature of polymorphic sites were characterized using DnaSp v. 5.10.01 (Librado & Rozas 2009). The haplotypes generated in our study were then combined with a further 29 congeneric haplotypes, and those of 4 outgroup taxa (Acari: Tetranychidae), Tetranychus pueraricola Ehara & Gotoh, Tetranychus kanzawai Kishida, Tetranychus truncates Ehara, and Eotetranychus celtis Ehara, retrieved from GenBank (see Fig. 2 for accession numbers). Sequences were aligned using MUSCLE (Edgar 2004), and then trimmed to 305 base pairs using Gblocks (Castresana 2000) to remove overhanging sequence tails, before genealogical relationships among the haplotypes were estimated using maximum likelihood (ML) analyses conducted with PhyML (v3.0 aLRT) (Guindon et al. 2010); all via the online Phylogeny.fr platform (Dereeper et al. 2008). The HKY85 substitution model was selected assuming an estimated proportion of invariant sites (of 0.376) and 4 gamma-distributed rate categories to account for rate of heterogeneity across sites. The gamma shape parameter was estimated directly from the data (gamma = 0.431). Branch support was assessed using the approximate likelihoodratio test (SH-Like). ML analyses revealed the existence of 3 divergent clades (see Results), and so, divergence within and between these clades was estimated as average Kimura 2-parameter (K2P) distances calculated using MEGA version 6.06 (Tamura et al. 2013).
Sequences of the ITS2 rRNA region were aligned with MAFFT version 7 using the Q-INS-i strategy (Katoh & Standley 2013) and manually collapsed into genotypes. These genotypes were combined with additional ITS2 sequences retrieved from GenBank, encompassing O. perseae samples from Mexico (Guzmán-Valencia et al. 2014; accessions KC568365–KC568386), Israel (DQ656457–DQ656458), and Spain (GU565305), and realigned. Five sequences, KC568367 and KC568378–381, from Guzmán-Valencia et al. (2014) were subsequently discarded because they were particularly short, or because their inclusion induced multiple single gaps at the 5′ end of the alignment matrix, a trait typical of low quality reads. The sequences were aligned for a third time and non-overlapping regions were removed, resulting in an alignment incorporating 25 sequences in a 366-base-pair matrix (including gaps). Relationships among the ITS2 genotypes were investigated using MEGA version 6.06, by constructing a neighbor joining (NJ) tree based on pairwise uncorrected p-distance. Gapped positions were removed for each sequence pair, and branch support was assessed by bootstrapping with 1,000 replicates.
Collection information for representative Oligonychus perseae populations studied (see Fig. 1).
Seventeen representative sequences of 28S rRNA also were aligned separately using the Q-INS-i strategy in MAFFT version 7 and manually collapsed into genotypes. The resulting 28S alignment contained no gaps and was 641 base pairs long. Relationships among the 28S genotypes were investigated again by constructing a NJ tree based on pairwise uncorrected p-distance using MEGA version 6.06.
Among COI sequences collected from 75 individuals, 35 polymorphic nucleotide sites were identified comprising a total of 11 haplotypes, H1 to H11 (Table 1). California specimens were fixed for a single COI haplotype (H8). The same haplotype also was recovered from specimens collected in: 2 municipalities (counties), Tijuana and Ensenada (Table 1, sites D and E), in the Mexican state of Baja California; 4 municipalities, Uruapan, Charo, Morelia, and Tarímbaro (Table 1, sites F, G, I, J, and K), in the Mexican state of Michoacán; and, also was previously recorded from an Israeli specimen (DQ656485). Surprisingly, the ML analysis provided strong evidence of 3 divergent COI haplotype clusters (branch support > 0.9; Fig. 2). Cluster 1 included haplotypes H6 to H11 and represented samples from California (USA) (Table 1, sites A, B, and C), Mexico (Table 1, sites D–G, I–K, and M–V), and Israel (accession DQ656485). Cluster 2 included haplotypes H1 to H3, corresponding to samples only from Mexico (Table 1, sites H and L). Cluster 3 included haplotypes H4 and H5, corresponding to samples from Costa Rica (Table 1, site W). Variation within each cluster ranged from 0.2 to 0.4 % (mean pairwise K2P distance ± SE (standard error); cluster 1 = 0.004 ± 0.002, cluster 2 = 0.004 ± 0.003, and cluster 3 = 0.002 ± 0.002). Divergence between clusters was around 20-fold higher (mean pairwise K2P distance; between clusters 1 and 2 = 0.083 ± 0.016, between 1 and 3 = 0.079 ± 0.016, and between 2 and 3 = 0.073 ± 0.016).
Four genotypes were identified in the ITS2 sequences (Fig. 3; Table 1). These genotypes grouped our specimens into the same 3 genetic clusters as the COI data (Fig. 4). All Californian specimens (n = 11) shared a single ITS2 genotype (G1) containing 2 heterozygous nucleotides at positions 284 and 317; in both cases A to T transversions (Fig. 3). This genotype also was present in the majority of our specimens from Mexico (n = 61). Heterozygosity was absent at those 2 positions (A and T, respectively) in a second, less abundant genotype (G2) that was otherwise identical (Fig. 3). This homozygous genotype was detected in all 5 specimens collected from Atlixco, Puebla, Mexico (Table 1) and a single specimen from Zitácuaro, Michoacán, Mexico (KY474287). All specimens with a COI cluster 1 haplotype (H6–H11), had one of these 2 ITS2 genotypes. All of our Mexican specimens with a COI cluster 2 haplotype (H1–H3), shared a single ITS2 genotype (G3) that differed from that of cluster 1 at multiple positions (Fig. 3). Finally, Costa Rican samples (COI cluster 3, haplotypes H4 and H5) shared a further ITS2 genotype (G4), that contained a heterozygous nucleotide at position 224 (a C to T transition), and that also differed from the others at multiple positions (Fig. 3). Combination of our 4 ITS2 genotypes with existing O. perseae sequences retrieved from GenBank revealed the following patterns: ITS2 sequences from Israel (DQ656456–458; Ben-David et al. 2007) and Spain (GU565305; Perez-Sayas et al., unpublished) grouped with COI cluster 1, along with 4 sequences from Mexico (KC568382–385; Guzmán-Valencia et al. 2014) (Fig. 4). With the exception of KC568382 and KC568385, all variation within this cluster was restricted to the heterozygous positions 284 and 317. The remaining 14 Mexican sequences that were included from Guzmán-Valencia et al. (2014) grouped with cluster 2 (Fig. 4). Nine were an exact match for our cluster 2 ITS2 genotype, and 3 sequences (KC568366, KC568368, and KC568374) differed by only a single nucleotide substitution, each at a different position. The remaining genotype represented by KC568377 differed at multiple positions.
Finally, the more conserved 28S rRNA sequences representative of the 11 COI haplotypes detected in this study, also contained 4 genotypes (Fig. 5; Table 1) and recovered the same genetic clustering pattern as both COI and ITS2 (Fig. 6). Within COI cluster 1, all specimens had one of two 28S haplotypes, which differed from each other by only a single nucleotide substitution at position 579 (e.g., KY474115 and KY474121). All specimens with a COI haplotype from cluster 2 shared a 28S haplotype that differed from those of cluster 1 at 11 nucleotide positions (Fig. 5). Similarly, specimens with a COI haplotype from cluster 3 shared a 28S haplotype that differed from those of cluster 1 at 10 nucleotide positions (Fig. 5). Cluster 2 and 3 haplotypes differed from each other at 3 nucleotide positions (Fig. 5).
The primary goal of this study was to circumscribe the potential geographic origin of O. perseae populations in California from within the assumed native range of this pest in Mexico. Clarification of the origin of exotic O. perseae populations found in California would help document its invasion history as has been done for other exotic avocado pests established in California that originated from Mexico (Rugman-Jones et al. 2007, 2012), and also guide the development of future classical biological control programs for O. perseae in the California avocado system.
DNA extraction and amplification of COI, 28S and ITS2 genes from single adult O. perseae were successful. All ITS2 sequences from California clustered with O. perseae ITS2 sequence data (Fig. 4) representative of Israel, Spain and various regions in Mexico (Ben-David et al. 2007; Guzmán-Valencia et al. 2014). Examination of COI haplotypes revealed that one, H8, is fixed in California. Haplotype H8 has also been recorded in an invasive population in Israel (DQ656485). Within Mexico, we recovered H8 from 2 locations in Baja California, and 3 locations in Michoacán, but not from states neighboring Michoacán (i.e., Puebla, Guanajuato, México, and Morelos). Like California (USA), Baja California (Mexico) is not part of the native home range of avocado (Bost et al. 2013), and so, O. perseae is likely also invasive there, thus implicating the state of Michoacán as the probable source of O. perseae populations found in California, as well as in Baja California and Israel. Given that 37% of our samples from Mexico (i.e., 23 out of 61 Mexican specimens) shared the H8 haplotype, it is possible that there may have been multiple introductions of O. perseae from Mexico into other avocado growing regions. However, it is also plausible that invasive populations may have subsequently acted as invasion bridgeheads from which infestations spread. However, we cannot unambiguously discern this based on our data. For now, the present detection of a fixed COI haplotype, H8, from O. perseae populations in California (i.e., San Luis Obispo, Ventura, and San Diego counties) indicates that there has been at least one successful introduction event of O. perseae into California, and the ultimate phylogeographic source of that invasive lineage was most likely Michoacán, Mexico.
A series of introduction pathways for exotic avocado pests, including O. perseae, from Mexico into the USA have been discussed previously by Hoddle (2005). The key limiting factor for any successful O. perseae introduction pathway is that this pest is strictly a foliage feeder and like many other small leaf feeding plant pests introduced into California, it may have been introduced on live plant material (Dowell et al. 2016). The requirement of O. perseae for fresh foliage to survive indicates there is minimal risk of O. perseae introduction associated with the movement of washed fruit from Mexico to other countries, including California (but see Morse et al. 2009, 2016). Instead, the biology of O. perseae and the molecular evidence presented here implies that its introduction from Mexico, presumably Michoacán, into California was most likely mediated through the movement of infested plant material possibly intended for avocado propagation (i.e., budwood or whole plants). Interestingly, Tuttle et al. (1976) first described O. perseae from avocado plant material interdicted at a quarantine facility in Texas (USA), which originated from Mexico.
An alternative explanation for the invasion of O. perseae into California centers on the recovery of the H8 haplotype from Tijuana in Baja California, Mexico. Tijuana is connected to San Diego California via a pedestrian walkway, and avocados are relatively common in residential gardens in Tijuana and San Diego (Hoddle 2011). The invasion of O. perseae into San Diego may have occurred from avocado trees infested with O. perseae that were planted in urban areas in Tijuana. Oligonychus perseae produces silk strands, which catch the wind and allows mites to disperse aerially (Fig. 7A–C) with the potential to reach neighboring avocado trees (Aponte & McMurtry 1997; Bell et al. 2005). Consequently, O. perseae may have dispersed aerially or phoretically (e.g., on humans that may have come in contact with infested plant material) into California and established on trees growing on the USA side of the border (Hoddle 2011). Once established in San Diego, O. perseae likely spread to major avocado production areas in California via accidental human-assisted transport of infested plant material and contaminated equipment to uninfested orchards, nurseries, and retail stores.
Molecular data presented here suggests that future explorations for natural enemies of O. perseae, for potential introduction into California as part of a classical biological control program, should focus on key areas in Michoacán, i.e., Uruapan, Charo, Morelia, and Tarímbaro. These 4 sites are less than 100 km from each other (Fig. 1) and natural enemies in these areas are likely best adapted to the invasive O. perseae haplotype that established in California. Uruapan, in particular, is a major producer of cultivar ‘Hass’ avocados in Michoacán (Salazar-García et al. 2005). Ideally, this search would consist of finding candidate natural enemy species that can subsist year round in avocado orchards, exhibit a high propensity to feed on O. perseae when it is available, and are capable of accomplishing density-dependent regulation of O. perseae populations. An inventory of natural enemies, primarily predator mites, on avocados in Michoacán already exists (Estrada-Venegas et al. 2002), but to our knowledge, the efficacy of these predators as biological control agents for O. perseae within the cultivar ‘Hass’ avocado system has not been evaluated. Field evaluation of phytoseiid mites associated with O. perseae in Mexico in combination with the inclusion of additional molecular markers such as microsatellites (Rugman-Jones et al. 2007, 2012), could optimize the geographic search and candidate selection for effective O. perseae natural enemies for potential use in a classical biological control program in California.
In addition to providing insight on the putative geographic origin of O. perseae populations in California, this study found strong molecular evidence for the occurrence of 3 potential cryptic species within O. perseae. Sequences of 2 regions of the nuclear ribosomal cistron (ITS2 and 28S) and a single mitochondrial gene (COI) independently grouped our O. perseae specimens into 3 congruent divergent genetic clusters (Figs. 2, 4, and 6). Such correlative clustering across independently evolving nuclear and mitochondrial loci can be seen as strong evidence for the existence of cryptic species (Mallet 1995; Rugman-Jones et al. 2010). Furthermore, in an earlier study of several other Oligonychus species, Matsuda et al. (2012) reported non-overlapping intra- and inter-specific thresholds for genetic distances for COI (intra: <2.9%, inter: 7.3–18.3%) and 28S (intra: <0.1%, inter: 0.4–10.7%). The genetic distances reported herein for the 3 COI and 28S genotype clusters in this study fall within the respective intra- and inter-specific thresholds established by Matsuda et al. (2012). Specifically, within-cluster variation in COI was less than 0.5%, and variation between clusters ranged from 7.3 to 8.3%. For 28S, within-cluster variation was almost absent, and variation between clusters ranged from 0.5 to 1.8%. Evidence of cryptic species has previously been reported in 24 superfamilies of Acari, including Tetranychoidea (Skoracka et al. 2015). In the context of agricultural systems, the occurrence of cryptic species has been documented in the economically important genus Tetranychus (Acari: Tetranychidae) (Matsuda et al. 2013), to which the genus Oligonychus is confamilial.
Although we have strong molecular evidence for O. perseae being a complex of 3 cryptic species, biological (e.g., interpopulation mating studies) and ecological data (these putative cryptic species occupy near identical niches and cause very similar feeding damage) to support their status is currently lacking, and we strongly recommend these as future lines of research. Interestingly, Guzmán-Valencia et al. (2017) recently assessed genetic variation among sympatric O. perseae and O. punicae populations on 6 perennial host trees in the state of México, which included the hardwood species Salix bonplandiana Kunth (Salicaceae), Alnus jorullensis Kunth (Betulaceae), and Alnus acuminata Kunth (Betulaceae), alongside 3 avocado cultivars (‘Hass’, ‘Fuerte’, and ‘Criollo’). As it relates to our study on avocado, Guzmán-Valencia et al. (2017) found no genetic variation among the COI sequences of morphologically-identified O. perseae populations sampled from the 3 avocado cultivars. However, examination of their sequences (GenBank accessions KX072889-921; Guzmán-Valencia et al. 2017) reveals levels of divergence between the COI of their avocado populations and those they sampled from the 2 Alnus species, similar to those reported herein (~8%), suggesting that they may also have encountered 2 cryptic species (in their case, on 2 different types of host plant). Unfortunately, the regions of COI used in our study and that of Guzmán-Valencia et al. (2017) do not overlap, and additional genetic marker sequences are not available from Guzmán-Valencia et al. (2017). Thus, it is uncertain whether or not there is overlap in the potential cryptic species reported in the 2 studies. Previously, Guzmán-Valencia et al. (2014) detected similar levels of divergence in COI among morphologically identified O. perseae populations from the states of México and Michoacán, but again these specimens were not recognized by the authors as potential cryptic species, nor can we be certain how they relate to ours. Combined, these key studies provide very strong evidence for the existence of a cryptic species complex in O. perseae, and highlight the importance of considering multiple genetic loci in the assessment of such complexes (i.e., not just COI).
Both (Acari: Tetranychidae) Oligonychus yothersi (McGregor) and Oligonychus punicae (Hirst) have been reported to co-occur with O. perseae in some avocado growing regions (Ochoa et al. 1994; Aponte & McMurtry 1997; Lara & Hoddle 2012), but it seems unlikely that the findings of the current study are the result of misidentification. Oligonychus punicae is easily separable from O. perseae both by gross morphology (O. punicae tends to be brown and O. perseae is yellowgreen) and preferred feeding sites (O. punicae generally feeds on the adaxial surface of leaves and O. perseae inhabits the abaxial surface of leaves) (Lara & Hoddle 2012). Furthermore, COI sequences can be used to clearly separate O. punicae from O. perseae (Fig. 2; Guzmán-Valencia et al. 2014, 2017). Similar, relatively easy, morphological discrimination also can be made between O. yothersi and O. perseae (Ochoa et al. 1994) although DNA sequences are not available for the former. Importantly, the ITS2 sequences of all our Californian and Mexican O. perseae specimens (i.e. clusters 1 and 2) clustered with those of specimens from Guzmán-Valencia et al. (2014), the identity of which were all confirmed by morphological taxonomy. With regards to our samples from Costa Rica, our specimens originated from a field site where O. perseae was the only species reported based on morphological taxonomy (Solano-Guevara 2011). However, a possibility remains that specimens from Costa Rica that were used in this study (and by extension those from Guzmán-Valencia et al. 2014) could correspond to another species, Oligonychus peruvianus (McGregor) (Acari: Tetranychidae), which has previously generated taxonomic confusion due to its very close morphological similarity to O. perseae (McMurtry 1993; Sandoval et al. 2011). Unfortunately, no molecular data is currently available for O. peruvianus and we were unable to secure authoritatively identified specimens for inclusion in the current study. Ultimately, the insight from our study and Guzmán-Valencia et al. (2014, 2017) highlight the need to develop and validate unambiguous species identification methods for pest mites infesting avocados (Skoracka et al. 2015).
This work was supported in part by the University of California Thelma Hansen Trust and the California Avocado Commission. Additional financial support to JRL was provided by the Robert van den Bosch Scholarship. We thank the numerous growers and homeowners who provided access to their avocado trees for the collection of mite specimens. We are indebted to Dr. Hugo Aguilar Piedra and his research staff for providing transport and generous access to laboratory space in Costa Rica.