Geometric Morphometric Analysis and Molecular Identification of Coconut Mite, Aceria guerreronis Keifer (Acari: Eriophyidae) Collected from Thailand

Simple Summary The coconut mite is one of the most well-known and serious pests of coconut fruits worldwide, and it has spread to most regions where coconuts are produced; in Thailand, Aceria guerreronis Keifer (Acari: Eriophyidae) is a quarantine pest. We conducted a geometric morphometric analysis and molecular identification on coconut mites collected from Thailand to obtain their origin and history. Our findings will provide a genetic resource for future functional studies on the relative phylogenetic relationship of coconut mites, which will be performed to understand how coconut mite species interact with their host plant. These findings will be helpful in designing pest management strategies against quarantine pests in Thailand. Abstract One of the most impactful pests in several coconut production regions across the world is the coconut mite, Aceria guerreronis Keifer. Scholars can obtain some necessary biogeographic information about coconut mites from studies that explore the geographic patterns of morphological variations and molecular properties among coconut mite populations from various locales. To investigate the geographical origin, ancestral host associations, and colonization history of the mite in Thailand, we obtained DNA sequence data from two mitochondrial (16s and COI) and one nuclear region (ITS) from coconut mite samples originating from 25 populations; additionally, we analyzed the morphological variations in the prodorsal shield and the coxigenital and ventral regions of the mite idiosoma. From the results of experiments using both identification methods, we identified the mite as the coconut mite, A. guerreronis (Acari: Eriophyidae). According to the phylogenetic analysis results of the 25 mite samples, we classified the mites as being closely related to mites found by the authors of a previous report in India. We are the first to report the results of a geometric morphometric analysis and molecular identification of A. guerreronis in Thailand, and our findings support the idea that the mites’ origin and invasion history are not well documented, which makes it difficult to apply quarantine procedures and search for biological pest control agents.


Introduction
The invasive coconut mite Aceria guerreronis Keifer (Acari: Eriophyidae) has spread and become well established in the main coconut (Cocos nucifera L. (Arecaceae))-growing  The mites were observed under a phase contrast light optical microscope (Leica DM100 LED) (100× objective). The morphology and nomenclature follow Lindquist [22] and the systematic classification is based on Amrine et al. [23]. The morphological characteristics essential for the determination of species were compared with the original description of this species [4]. The phase contrast optical microscope (Leica DM100 LED, Leica Microsystems Ltd., Heerbrugg, Switzerland) was linked to a digital camera (Leica MC170 HD), which was then linked to a computer to capture the images of the body regions of the chosen specimens. Images of the prodorsal shield and coxigenital region were obtained using a 100× magnification objective, and images of the ventral region were obtained using a 40× magnification objective. To conduct the landmark digitization, the prodorsal shield, coxigenital, and ventral sections of the A. guerreronis body were each individually assessed (including the coxigenital region and opisthosoma). These regions were selected because of their taxonomic importance and because a high number of landmarks could be defined. Ten landmarks in the prodorsal shield (Figure 2A), twelve in the coxigenital ( Figure 2B), and nineteen in the ventral section ( Figure 3) were selected. The classification of landmarks was based on [24,25]. Landmark data were produced with a series of programs called TpsUtil64 ver. 1.81 and Tpsdig264 ver. 2.32 software [26,27] and plotted. Deformation grids were obtained as thin-plate spline warps using MorphoJ software version 1.07a [28,29] and were plotted and used to explain deviations in the shape of each species from the average landmark configuration (consensus). Using a PCA of the covariance matrix of the population-averaged Procrustes coordinates, shape differences among the studied populations were further investigated. PCA was carried out using the MorphoJ program [28,30].
Leica Microsystems Ltd., Heerbrugg city, Switzerland) was linked to a digital cam (Leica MC170 HD), which was then linked to a computer to capture the images of body regions of the chosen specimens. Images of the prodorsal shield and coxigenital gion were obtained using a 100× magnification objective, and images of the ventral reg were obtained using a 40× magnification objective. To conduct the landmark digitizat the prodorsal shield, coxigenital, and ventral sections of the A. guerreronis body were e individually assessed (including the coxigenital region and opisthosoma). These regi were selected because of their taxonomic importance and because a high number of la marks could be defined. Ten landmarks in the prodorsal shield (Figure 2A), twelve in coxigenital ( Figure 2B), and nineteen in the ventral section ( Figure 3) were selected. classification of landmarks was based on [24,25]. Landmark data were produced wi series of programs called TpsUtil64 ver. 1.81 and Tpsdig264 ver. 2.32 software [26,27] plotted. Deformation grids were obtained as thin-plate spline warps using MorphoJ s ware version 1.07a [28,29] and were plotted and used to explain deviations in the sh of each species from the average landmark configuration (consensus). Using a PCA of covariance matrix of the population-averaged Procrustes coordinates, shape differen among the studied populations were further investigated. PCA was carried out using MorphoJ program [28,30].

Sample Collection
A total of 25 coconut mite samples were collected and stored at −20 • C. Thus, all subsequent extractions were performed with approximately 200 pooled adult mites. These samples were obtained from one breed distributed across 5 provinces in Thailand (Table 2). Animal handling and experimentation followed the animal experimental procedures and guidelines approved by the Ethics Committee of Kasetsart University (ID Code ACKU65-AGK-039).

Sample Collection
A total of 25 coconut mite samples were collected and stored at −20 °C. Thus, all subsequent extractions were performed with approximately 200 pooled adult mites. These samples were obtained from one breed distributed across 5 provinces in Thailand ( Table  2). Animal handling and experimentation followed the animal experimental procedures and guidelines approved by the Ethics Committee of Kasetsart University (ID Code ACKU65-AGK-039).

Total DNA Isolation
The total DNA was isolated from the coconut mites using the CTAB buffer method. As a first step, 0.6 mL of the CTAB buffer was added, and the buffer was homogenized with the sample. Then, the homogenized sample was incubated overnight at 65 • C, frozen for 15 min at 20 • C, and incubated for a further 15 min at 65 • C. Then, 0.5 mL of chloroform/isopropanol (24:1) was added, thoroughly mixed by shaking, and then incubated for 2 min at 25 • C. Samples were centrifuged at 12,000× g for 5 min. The DNA exclusively remained in the upper aqueous phase, and was transferred to a fresh tube. The DNA in the aqueous phase was precipitated by adding 0.5 mL of isopropyl alcohol. Then, the sample was incubated overnight at 4 • C. Afterwards, the supernatant was removed, the DNA pellet was air-dried, and then dissolved in 20 µL of RNase-free water before being stored at −20 • C. Finally, the DNA concentration and purity were assessed via gel electrophoresis using a NanoDrop Spectrophotometer 2000 (ThermoFisher Scientific, Waltham, MA, USA).

PCR Amplification and Sequencing
The amount of DNA was multiplied using a PCR with a random hexamer primer. The PCR was amplified using the tree primer set of rDNA-ITS (F-rDNA-ITS: AGAGGAAG-TAAAAGTCGTAACAAG and R-rDNA-ITS: ATATGCTTAAATTCAGGGGG [31]), 16S mtDNA (F-mtDNA-16S: CCGGTCTGAACTCAGATCACG and R-mtDNA-16S: CGCCT-GTTTAACAAAAACAT) [32], and COI mtDNA (F-mtDNA-COI: GGATCACCTGATATAG-CATTCCC and R-mtDNA-COI: CCCGGTAAAATT AAAATATAAACTTC [32]). PCR amplifications were performed using a Taq ® 2X Master Mix (New ENGLAND BioLabs ® inc., Ipswich, MA, USA) in a final volume of 25 µL containing 12.5 µL of 2X Master Mix, 1 µL of each primer (10 pmol/µL), 8.5 µL of dH2O, and 2 µL of DNA, which was multiplied with a random hexamer primer (100 ng/µL). The cycling profile included a 5 min preliminary denaturation cycle at 95 • C, followed by 40 denaturation cycles at 95 • C for 30 s, annealing at 50 • C for 30 s, and extension at 72 • C for 30 s, with a final extension at 72 • C for 5 min. The PCR products were separated via electrophoresis on a 2% agarose gel and visualized under ultraviolet light. Sequencing of the PCR product samples was carried out at ATGC Co., Ltd. (Ward Medic IDT, Bangkok, Thailand).

Data Analysis
DNA sequences were manually checked using BioEdit [33] and then aligned using the ClustalW algorithm in MEGA 11.0 [34]. The evolutionary history of the genomic DNA and mtDNA was inferred by using the maximum likelihood method, the Tamura 3-parameter model [35], and the Kimura 2-parameter model [36]. Initial trees for the heuristic search were obtained automatically by applying the Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances, which were estimated using the Tamura 3-parameter model. The topology was then selected using the superior log likelihood value. Evolutionary analyses were conducted in MEGA11 [34].

Coconut Mite Sampling and Identification
Based on comparing the morphological characteristics of the mite from our research with the original description of this species [4], we concluded that it was Aceria guerreronis. The symptoms of coconut fruits with A. guerreronis attack have yellowish-to-brownish triangular scars that start at the margin of the bracts and increase as the fruit grows ( Figure 4).

Coconut Mite Sampling and Identification
Based on comparing the morphological characteristics of the mite from our research with the original description of this species [4], we concluded that it was Aceria guerreronis. The symptoms of coconut fruits with A. guerreronis attack have yellowish-to-brownish triangular scars that start at the margin of the bracts and increase as the fruit grows ( Figure  4).

Landmark-Based Morphometric Methods
Analyses were carried out in order to learn more about the morphological differences across A. guerreronis populations. The consensus shapes, based on the prodorsal shield of the 9 landmarks, the coxigenital area of the 12 landmarks, and the ventral regions of the 19 landmarks belonging to 75 A. guerreronis individuals from five localities, are shown in Figures 5A, 6A and 7A, respectively. Moreover, in Figures 5B-F, 6B-F and 7B-F, thin-plate spline deformation grids are depicted as variations in the A. guerreronis morphology from five locations in Thailand.
By applying a multivariate analysis (PCA), we discovered that the A. guerreronis populations from the sampled localities morphologically varied. The PCA performed on the shape coordinates of 19 landmarks in the ventral region of 75 specimens from the five studied populations resulted in 34 principal components, with the first two components explaining 51.75% of the total variation (PC1 29.29%, PC2 22.46%). Then, the PCA performed on the shape coordinates of nine landmarks in the prodorsal shields of 75 specimens from the five studied populations resulted in 16 principal components, with the first two components explaining 70.89% of the total variation (PC1 60.75%, PC2 10.14%). Moreover, the PCA performed on the shape coordinates of 12 landmarks in the coxigenital areas of 75 specimens from the five studied populations resulted in 20 principal components, with the first two components explaining 47.83% of the total variation (PC1 35.99%, PC2 Insects 2022, 13, 1022 8 of 17 11.84%). When we plotted the populations against their respective values for PRIN1 and PRIN2 ( Figures 5-8), we found that several mite populations, namely Ratchaburi (d) and Samut Sakhon (e), were dispersed along both axes, showing considerable morphometric diversity within each of them. On the other hand, mites from other populations, including Chachoengsao (a), Nakhon Pathom (b), and Pathum Thani (c), were concentrated in a very small area of the graphic, indicating a higher degree of morphometric similarity within each of those populations. explaining 51.75% of the total variation (PC1 29.29%, PC2 22.46%). Then, the PCA performed on the shape coordinates of nine landmarks in the prodorsal shields of 75 specimens from the five studied populations resulted in 16 principal components, with the first two components explaining 70.89% of the total variation (PC1 60.75%, PC2 10.14%). Moreover, the PCA performed on the shape coordinates of 12 landmarks in the coxigenital areas of 75 specimens from the five studied populations resulted in 20 principal components, with the first two components explaining 47.83% of the total variation (PC1 35.99%, PC2 11.84%). When we plotted the populations against their respective values for PRIN1 and PRIN2 ( Figure 5-8), we found that several mite populations, namely Ratchaburi (d) and Samut Sakhon (e), were dispersed along both axes, showing considerable morphometric diversity within each of them. On the other hand, mites from other populations, including Chachoengsao (a), Nakhon Pathom (b), and Pathum Thani (c), were concentrated in a very small area of the graphic, indicating a higher degree of morphometric similarity within each of those populations.  (E) (F)   Table 1: (A) ventral region, (B) coxigenital, and (C) prodorsal shield.

Molecular Identification
Based on the morphological characteristics, we initially identified the coconut mite strains ( Table 2) as belonging to A. guerreronis. We recovered a total of 42 sequences for

Molecular Identification
Based on the morphological characteristics, we initially identified the coconut mite strains (Table 2) as belonging to A. guerreronis. We recovered a total of 42 sequences for three genomic regions, including 10 for the rDNA ITS, 22 for the mtDNA 16S, and 10 for the mtDNA COI from the 25 A. guerreronis samples that we analyzed. Due to difficulties in obtaining the PCR amplification, which were likely brought on by the degraded state of some of the materials, we did not sequence every DNA template for the three DNA areas.
The tree with the highest log likelihood (−1445.42) is shown (Figure 9), and the percentage of trees in which the associated taxa clustered together is shown next to the branches. This analysis involved 18 nucleotide sequences. We included the 1st + 2nd + 3rd + Noncoding codon positions, which resulted in a total of 833 positions in the final dataset. The percentage of trees in which the associated taxa clustered together is shown next to the branches. We used a discrete Gamma distribution to model the evolutionary rate differences among the sites (five categories (+G, parameter = 6.5149)). This analysis involved 30 nucleotide sequences. We included the 1st + 2nd + 3rd + Noncoding codon positions, which resulted in a total of 636 positions in the final dataset. The tree with the highest log likelihood (−4493.79) is shown (Figure 10). The percentage of trees in which the associated taxa clustered together is shown next to the branches. We used a discrete Gamma distribution to model the evolutionary rate differences among the sites (five categories (+G, parameter = 6.5149)). This analysis involved 30 nucleotide sequences. We included the 1st + 2nd + 3rd + Noncoding codon positions, which resulted in a total of 636 positions in the final dataset. The tree with the highest log likelihood (−4493.79) is shown (Figure 10).
The trees with the highest log likelihood (−2254.25) of the 600 bp PCR product and (−1278.80) of the 400 bp PCR product are shown ( Figure 11). The percentage of trees in which the associated taxa clustered together is shown next to the branches. We automatically obtained an initial tree(s) for the heuristic search by applying the Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances, which we estimated using the Tamura 3-parameter model and then selecting the topology with a superior log likelihood value. This analysis involved 19 nucleotide sequences. We included the 1st + 2nd + 3rd + Noncoding codon positions, and totals of 433 positions (600 bp PCR product) and 405 positions (400 bp PCR product) were present in the final dataset. The trees with the highest log likelihood (−2254.25) of the 600 bp PCR product and (−1278.80) of the 400 bp PCR product are shown ( Figure 11). The percentage of trees in which the associated taxa clustered together is shown next to the branches. We automatically obtained an initial tree(s) for the heuristic search by applying the Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances, which we estimated using the Tamura 3-parameter model and then selecting the topology with a superior log likelihood value. This analysis involved 19 nucleotide sequences. We included the 1st + 2nd + 3rd + Noncoding codon positions, and totals of 433 positions (600 bp PCR product) and 405 positions (400 bp PCR product) were present in the final dataset. The tree with the highest log likelihood (−1353.95) is shown (Figure 12), and the pe centage of trees in which the associated taxa clustered together is shown next to th branches. We used a discrete Gamma distribution to model evolutionary rate differenc among the sites (five categories (+G, parameter = 1.7145)). The rate variation model a  The tree with the highest log likelihood (−1353.95) is shown (Figure 12), and the percentage of trees in which the associated taxa clustered together is shown next to the branches. We used a discrete Gamma distribution to model evolutionary rate differences among the sites (five categories (+G, parameter = 1.7145)). The rate variation model allowed for some of the sites to be evolutionarily invariable ([+I], 8.39% sites). This analysis involved 13 nucleotide sequences. We included the 1st + 2nd + 3rd + Noncoding codon positions, which resulted in a total of 392 positions in the final dataset.

Discussion
We first reported and identified coconut mites in Thailand as A. guerreronis based on their morphological and molecular properties, which we obtained with multiple DNA sequences. We found that these methods were useful in supporting or augmenting the conventional morphology and that they enhanced the characterization and validation of genetic barcoding. Morphometric analyses can help researchers resolve the issue that conventional taxonomic approaches are insufficient to delimit morphologically identical eriophyoid mites that inhabit various plants. Because of this, many mites have been regarded as either different entities or as a single oligophagous species. In all existing studies, researchers have performed analyses similar to [2,[37][38][39][40][41][42]. It should be emphasized that several morphological characteristics may help researchers distinguish between congeneric species or species that belong to various eriophyoid genera. In addition, morphological/morphometric and genetic variations in eriophyoid mites such as Colomerus vitis (Pagenstecher), the erineum strain (Eriophyidae) from grape vines, can differ according to geographical areas, sampling seasons, host plant physiology, and environmental factors [43,44].
After analyzing the morphological variability of the mites using geometric morphometric techniques on three body regions of A. guerreronis populations inhabiting five localities of Thailand, our general conclusions are as follows: The results of a principal component analysis (PCA) show generally comparable patterns for all population combinations from the various geographic regions. They clearly demonstrate that A. guerreronis had variations across its geographic distribution range. We anticipate that any of those factors could account for the morphological differences across A. guerreronis populations that resulted from the mites inhabiting coconut hosts in various geographical locations. In some cases, the results of the morphometric investigations show that environmental in-

Discussion
We first reported and identified coconut mites in Thailand as A. guerreronis based on their morphological and molecular properties, which we obtained with multiple DNA sequences. We found that these methods were useful in supporting or augmenting the conventional morphology and that they enhanced the characterization and validation of genetic barcoding. Morphometric analyses can help researchers resolve the issue that conventional taxonomic approaches are insufficient to delimit morphologically identical eriophyoid mites that inhabit various plants. Because of this, many mites have been regarded as either different entities or as a single oligophagous species. In all existing studies, researchers have performed analyses similar to [2,[37][38][39][40][41][42]. It should be emphasized that several morphological characteristics may help researchers distinguish between congeneric species or species that belong to various eriophyoid genera. In addition, morphological/morphometric and genetic variations in eriophyoid mites such as Colomerus vitis (Pagenstecher), the erineum strain (Eriophyidae) from grape vines, can differ according to geographical areas, sampling seasons, host plant physiology, and environmental factors [43,44].
After analyzing the morphological variability of the mites using geometric morphometric techniques on three body regions of A. guerreronis populations inhabiting five localities of Thailand, our general conclusions are as follows: The results of a principal component analysis (PCA) show generally comparable patterns for all population combinations from the various geographic regions. They clearly demonstrate that A. guerreronis had variations across its geographic distribution range. We anticipate that any of those factors could account for the morphological differences across A. guerreronis populations that resulted from the mites inhabiting coconut hosts in various geographical locations. In some cases, the results of the morphometric investigations show that environmental influences were not the primary variability determinants. Sometimes, according to the results of morphometric assessments, environmental factors were not the primary variability determinants. Following the findings of Navia et al. [25], some populations in neighboring Brazil from climatically comparable regions, as well as those from the northeastern coastal region, did exhibit very noticeable differences. This clearly implies that the observed physical variation in such groups is tightly correlated with the genetic background of the species.
The fact that the morphology of Eriophyoidea mites may be related to their habitat structure is well known [25,37,45]. We found that the A. guerreronis population collected from different localities in Thailand was morphologically similar to the geographically varied ventral regions' closest populations (Chachoengsao, Nakhon Pathom, and Pathum Thani), but that it was different from Samut Sakhon and Ratchaburi.
The shape of Eriophyoidea mites may be related to their habitat structure. Hence, Navia et al. [25] investigated morphological variations in the ventral regions, prodorsal shield, and coxigenital area among A. guerreronis populations in America, Africa, and Asia. Their findings regarding the coxigenital and ventral areas confirmed the origin and invasion history of this species, which agreed with those obtained using molecular markers [46]. Furthermore, their finding that substantial morphological variations existed between American populations supported earlier claims that A. guerreronis originated in the United States, whereas the similar morphology between the African and Asian populations suggested a shared origin and rapid separation of those populations.
Researchers have recently reported that Aceria guerreronis is present in Asia, specifically in India and Sri Lanka [25,46,47]. The connection between the African and Asian populations reveals their related genetic characteristics, suggesting that the Asian populations resulted from an introduction from Africa. In Thailand, A. guerreronis is classified as a quarantine pest that is prohibited under the Plant Quarantine Act B.E. 2507 (1964) (No. 3) B.E. 2550 (2007), and there have never been reports of A. guerreronis in Thailand before. The results of our phylogenetic analyses on the 25 mite samples, which identify the mites that were closely related to mites from an earlier publication in India, suggest that both populations had a common origin.
The results of rDNA ITS and mtDNA COI sequences reveal that the A. guerreronis collected from Chachoengsao and Ratchaburi were closely related to those from an earlier publication in India [48]. Moreover, the results of the mtDNA 16s sequences not only show that the A. guerreronis collected from Samut Sakhon were closely related to those from earlier publications in Brazil, India, Sri Lanka, Tanzania, Benin, Venezuela, America, and Mexico, but they also show that Nakhon Pathom and Pathum Thani were closely related to those from an earlier publication on Samut Sakhon [48].
Recently, the Department of Agriculture (DOA) Thailand reported that the coconut mite, A. guerreronis, has been discovered in 18 provinces of Thailand, including Amnat Charoen, Chainat, Suphanburi, Kamphaeng Phet, Lop Buri, Nakhon Pathom, Nakhon Sawan, Nakhon Ratchasima, Pathum Thani, Phetchabun, Phichit, Phitsanulok, Ratchaburi, Saraburi, Sing Buri, Suphanburi, and Amnat Charoen; additionally, 4.2% of all the trees that were surveyed had fruit damage. The important coconut-growing regions of Thailand's upper north, northeast, and south did not have this species. The results of this study reveal that A. guerreronis is also present in two provinces, Chachoengsao and Samut Sakhon, in addition to Nakhon Pathom, Pathum Thani, and Ratchaburi.
Other Asian and Pacific locations where coconut mites have not yet been detected may be at risk for experiencing coconut mite infestations. This is because the primary way coconut mites are introduced to remote locations entails the movement of any propagation tissue from palm trees, as well as the transit or exchange of propagation host plant material, especially from Thailand, which represents a quarantine risk.
In conclusion, the coconut mite A. guerreronis is not only a serious threat to coconut plantations, but it is also a quarantine pest in Thailand. We isolated coconut mites from infested plants and identified them based on their morphological characteristics and molecular properties, which we obtained using multiple DNA sequences. We are the first to detail a geometric morphometric analysis and molecular identification of A. guerreronis in Thailand. To discover efficient biological control measures, researchers must determine the historical distribution of the mite. Knowing the population spread patterns is also economically important because the coconut mite still poses a threat to Asian nations where it has not yet been found. Understanding its spread could help scholars predict the likelihood of future incursions and could help them direct quarantine measures to stop the spread of the pest.