Tracking potential biosecurity incursions using publicly available images: A case of coconut rhinoceros beetle

The coconut rhinoceros beetle, Oryctes rhinoceros (L.), (CRB) is a major pest of coconut and oil palms that causes significant damage in the Asia/Pacific region. Traditional methods for surveillance and detection of CRB involve in‐person inspections, but these are expensive to carry out and were made impossible by travel restrictions during the COVID‐19 pandemic. In this paper, a method to remotely detect possible CRB incursions and assess damage levels using publicly available images is described. The method involves searching for pictures of palms and identifying those with the distinctive V‐shaped cuts in fronds characteristic of CRB feeding. The method was validated by examining palm images from known outbreak areas and those free from the pest. We found evidence of CRB damage in the previously unreported site of Dili, Timor Leste and possible evidence of CRB in the Solomon Islands even before it was first reported in 2015. Potential limitations and broader applications of this approach are also discussed.

the larvae is provided from palms damaged during natural disasters or redevelopment.
Before the COVID pandemic, surveillance relied on local reporting and visits by experts to the coconut and oil palm growing areas to confirm CRB incursions in new areas and to assess damage levels. Once social distancing and severe travel restrictions were put in place across the globe during the COVID epidemic, alternatives were sought to conduct CRB surveillance. Here, we explain a method using images from the public domain and demonstrate its potential to detect biosecurity incursions and damage levels of CRB.
Palm images from the public domain provide a resource to check remotely for indications of the presence of the beetle as tourists and amenity managers have lodged dated and GPS fixed images from various locations at different time points. Fortunately, attack by CRB leaves evidence through distinctive cutting of the palm fronds that could be easily recognized from images ( Figure 1).
Based on invasive pest databases (CABI, 2022;EPPO, 2022) and through published literature, we selected four sites each from where CRB is (i) recorded and (ii) has not yet been recorded. Palm images uploaded to the internet from these sites were examined and, if palms were present with the crowns clearly visible, were downloaded for assessment. Then, we demonstrate the potential of our protocol in pest surveillance based on damage symptoms. Broader application of this approach along with its limitations are discussed and summarized. To our knowledge, this is the first case of using publicly available images to detect biosecurity incursion from a pest and to assess damage levels.

| Description of CRB damage
The distinctive notches produced by feeding adults are a clear marker of CRB and will often be the first indicator of their presence (Bedford, 1980). Adult rhinoceros beetle attack palms by entering the crown of the palm and moving down within the palm axils until they are located in a gap of 1-2 cm between the leaf axil and the stem (Young, 1972). From this position, they will bore horizontally into the growing frond bases, first encountering a petiole and then the developing spear. Once in the soft meristem tissue, they will bore downwards and macerate tissue, feeding on the plant juices.
The initial horizontal hole will cut through developing fronds producing the characteristic damage notches and V-shapes after the fronds have emerged (Figure 1). Feeding on the meristem will produce leaf distortion or death of the frond. Damage to upper fronds indicates recent attack (in the last 6 months), middle fronds (6-18 months) and lower fronds (about 2 years; Jackson et al., 2020).

| Data collection
Distribution maps from CABI and EPPO indicated that CRB is well spread through the islands of SE Asia and the western Pacific

| Framework for damage assessment
Each screenshot was copied to a Microsoft publisher file (Microsoft-Corporation, 2021) and the images of palms were examined for CRB-like damage symptoms. The number of damaged palms was recorded. Damage on each palm was further assigned a ranking based on the severity of the defoliation (Jackson et al., 2020). Palm trees from each picture were marked with a white circle and the grades were added subjectively with reference to a standards chart ( Figure 1; Data S1). Palms with no CRB damage were graded as '1'.
Palms with notching or tip damage and a foliar loss of 20% or less were graded as '2'. Palms with multiple damaged fronds and a foliar loss of 20-50% were graded as '3'. Palms with more than 50% foliar loss were assigned grade '4'. Palms with growing point destroyed and no chance of recovery were assigned grade '5'. Pictures were magnified, by zooming-in, as required to ensure uniform grading.

| Data analyses
The binary data (palms damaged vs. undamaged) were compared between sites using a generalized linear model with binomial distributions through a logit link function with 'palm damage' as a single factor. Data were analysed using 'R Studio' software with data package 'gee' (Højsgaard et al., 2006).
The proportion of palm damage from CRB from different sites was summarized using descriptive statistics (frequencies and percentages). The location of pictures was mapped using 'R Studio' software with data package 'leaflet' (Cheng et al., 2019).

| RE SULTS
Altogether, 467 palms were assessed for damage from the eight sites.
The sites known to be infested by CRB (Honiara, Upolu, Guam and Kerala) all showed evidence of CRB-damaged palms ( Figure 2 and Table 1). Surprisingly, high levels of damage were observed in Dili.
The highest level of CRB damage symptoms in palms was observed F I G U R E 2 Sites monitored for CRB, Oryctes rhinoceros (L.)damage symptoms from images uploaded to the googlemaps website (googl emaps.com). Red indicates damaged palms while palms without visible CRB damage are represented in blue. The 'purple' coloured sites are those where both damaged and undamaged palms were observed. [Colour figure can be viewed at wileyonlinelibrary.com] from Honiara (71%) followed by Dili (51%), Upolu (38%), Kerala (34%) and Guam (28%) ( Table 1) Table 1).

| D ISCUSS I ON AND LIMITATI ON S
Publicly available images uploaded to the internet provided a view of remote locations which could be used for surveillance and detection of CRB. CRB damage was clearly identified in the known outbreak areas with levels of damage corresponding to published data and common knowledge among the CRB expert community. For example, Honiara, Solomon Islands, is in the grip of a severe outbreak of CRB (Marshall et al., 2023), while Upolu, Samoa, has had CRB present for >100 years, with the population largely under control . No evidence of CRB was observed from Tahiti, the Cook Islands or Cairns, but, interestingly, we observed signs of CRB incursion from Dili, Timor Leste, which has not been previously reported. Further, we provide evidence of CRB damage in Honiara, Solomon Islands well before 2015 when it was first reported.

| Evidence of CRB damage in Timor Leste
Distribution maps from CABI and EPPO indicate that CRB is well spread through the islands of Southeast Asia and the western Pacific. However, no data are presented from the country of Timor Leste, which indicates either that CRB is not yet present in the country or is not yet recognized. Our study indicates that CRB is present in the country and is widespread with more than 50% of palms showing symptoms of CRB damage. As Timor Leste seeks to improve its cash crop economy, including coconut, which is grown by 40% of the households (FAO, 2022), it is important to recognize the threat of CRB and start to implement methods for its control. This is es-

| CRB damage in Honiara, Solomon Islands from 2013
CRB was officially reported from Honiara, Solomon Islands in 2015 where it has caused severe damage to coconut palms (Paudel et al., 2021;Tsatsia et al., 2018). However, based on the level of damage seen in 2015, it is believed that CRB may have entered the Solomon Islands two or more insect generations earlier, possibly during the '11th Festival of Pacific Arts' in 2012, which was held in Honiara, Solomon (Marshall et al., 2023). In the present study, CRB damage symptoms are seen in two out of the 20 pictures from Honiara that were taken during 2013. This further supports the assumption that CRB was present in Honiara well before 2015.

| Effectiveness and limitations
To ensure its effectiveness as a monitoring tool against CRB incursions, this approach of using publicly available images was examined against three major questions (Welvaert et al., 2017). First, are people likely to post these kinds of pictures? Second, is this information reliable and accessible? Third, is there a way to validate the data? Palm trees are the major host of CRB. Tourists are attracted towards large, lush palms and spectacular waterfront views and take photos and upload them to several online sites. Therefore, there is a high probability that people will continue posting palm pictures. TA B L E 1 Number (#) and percentage (%) of palms with CRB, Oryctes rhinoceros (L.) damage symptoms in pictures selected from eight sites.  .

Site
These potential incursion reports should be verified using pheromone traps (Bedford, 1980), alongside surveys of palm damage (Jackson et al., 2020). Photographic pictures from a site overtime will also help determine the level of infestation at different time points and support decisions about control measures. For example, a decline in CRB damage level provides evidence that control measures are working.

| Conclusion and future perspective
In conclusion, the study has shown that the use of publicly available palm images as a monitoring tool for CRB damage in palm trees is a cost-effective and easy-to-use approach. Overall, the use of publicly available images as a monitoring tool for CRB damage in palm trees has the potential to be a valuable tool for biosecurity management and should be further explored and validated. Conceptualization; methodology; data curation; investigation; funding acquisition; writing -original draft.

ACK N O WLE D G E M ENTS
This work has been funded by the New Zealand Ministry of Foreign Affairs and Trade (MFAT). Laura Villamizar (AgResearch, NZ) and anonymous referees are also thanked for reviewing this paper and providing comments/feedbacks. Open access publishing facilitated by AgResearch Ltd, as part of the Wiley -AgResearch Ltd agreement via the Council of Australian University Librarians.

CO N FLI C T O F I NTE R E S T
The authors declare that the research was conducted without any commercial or financial relationships construed as a potential conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available