Optimizing eciencies in biosecurity surveillance and monitoring: testing colour and multi-lure traps on tomato potato psyllid, other plant-lice (Psylloidea) and stinkbugs (Pentatomoidea)

Cost eciency in biosecurity surveillance is vital, and the ability to survey for multiple pest species using just one trap therefore highly appealing. The Psylloidea, or plantlice, contain signicant horticultural pest species that act as vectors for a number of deleterious plant bacteriums. We examine the ecacy of using two different coloured sticky traps, and two lure types on the general Psylloidea and Pentatomoidea fauna, and a target extant pest psyllid; tomato potato psyllid (TPP) Bactericera cockerelli (Šulc). Specically, we test the effect of lure (no lure, Asian citrus psyllid ACP lure, brown marmorated stink bug BMSB lure, combined lures), sticky trap color (green vs yellow), and sentinel plant (tomato vs citrus) on psyllid and stink bug species in 104 urban backyards across Perth, Australia. We found that tomato sentinel host plants and green traps signicantly increased the capture rate of TPP, but that all lures decreased the capture of TPP. Green traps also increased the capture rate of all other Psylloidea. Although BMSB lures reduced TPP capture, these lures increased abundances of other Psylloidea and the pest stinkbug Plautia anis (Dallas) on traps. Thus, our experiment demonstrates that increased eciencies can be gained with combination traps and lures for particular groups, provided that they have been tested on focal organisms in the rst instance, as reactions to non-target lures are unpredictable and species specic.


Introduction
Biosecurity surveillance is an ever-escalating cost due to increased global travel, free trade agreements, and open markets (Cope et al., 2019); (Meurisse et al., 2019). Accumulation rates of introduced taxa are not slowing (Seebens et al., 2017), suggesting current surveillance and control methods are only partially effective in many regions (Meurisse et al., 2019). Targeted risk analysis and surveillance for invasive plant pests increases the likelihood of early detection and eradication, lessening the potential of establishment and spread, thus reducing the impact of these pests (e.g., (Cope et al., 2019). In recognition of the importance of rigorous surveillance for plant pests, the European Food Safety Authority provided guidelines for designing risk-based surveys of plant pests ( European Food Safety Authority, 2020), and delivered RiBESS, a tool to calculate the sample size needed to substantiate absence of a pest and survey sensitivity (European Food Safety Authority, 2012). This tool provides more impetus to jurisdictions to conduct cost-effective and statistically sound surveillance for plant pests. Given the cost of surveillance, the rigour required and the increasing numbers of invasive pests able to exploit global movement, opportunities to increase the surveillance 'bang for the buck' should be embraced. Examples of such opportunities include optimising efforts in high-risk pathways and value-adding to surveillance efforts by targeting more than one pest species at each trapping site. Combining the monitoring of established nonnative insect pests with targeted biosecurity surveillance of exotic pests is attractive but is not always employed because of the vast array of insect species with different ecological and biological characteristics, taxonomic challenges, difference in trap types and nite resources (Brockerhoff et al., 2013).
The Psylloidea, or plantlice, contain signi cant horticultural pest species including Asian citrus psyllid Diaphorina citri Kuwayama, African citrus psyllid Trioza erytreae Del Guerci, tomato potato psyllid Bactericera cockerelli (Šulc, 1909), and carrot psyllids, particularly Bactericera trigonica Hodkinson and Dyspersa apicalis (Foerster) (Burckhardt, 1994). It is not the insect feeding per se that causes the most damage, but their ability to act as vectors for some devastating plant bacteriums, particularly Candidatus Liberibacter spp (e.g., (Catling, 1970 We aim to determine whether a psyllid species-speci c lure (Asian citrus psyllid), and a non-psyllid lure (brown marmorated stink bug) dissuaded a target organism, tomato potato psyllid (Bactericera cockerelli). Additionally, Asian citrus psyllid speci c yellow-green sticky traps (herein called 'green' traps) were tested to determine the capture rate of TPP compared to standard yellow sticky traps. Finally, we question the e cacy of the lures and coloured sticky traps across other psyllids including many native species (Psylloidea) and the stinkbug fauna (Pentatomoidea) to discern patterns. We conducted these trials in an urban setting as major entry points such as ports and airports in urban areas have been shown to be high risk pathways for incursions of exotic species (Colunga-Garcia et al.  (Greenway & Rondon, 2018). The damage is attributable to the psyllid feeding on host plants, as well as a phloem-limited, unculturable bacterium Candidatus Liberibacter solanacearum (CLso) that TPP is a vector for, and which causes a disease called 'psyllid yellows' or 'zebra chip' (Munyaneza et al., 2007). The psyllid's host range includes many Solanaceous plants such as tomato, capsicum, chilli, eggplant, and tobacco, plus some other families such as Convolvulaceae which includes sweet potato (for a full list see CABI 2021).
In February 2017 TPP was discovered in an inner suburban region of Perth, Western Australia. This was the rst mainland Australian incursion, after TPP established in New Zealand in 2006 (Teulon et al., 2009) and on the Australian territory of Norfolk Island in 2015 (Walker et al., 2015). As TPP was deemed unfeasible to eradicate from Western Australia, surveillance focused on detecting the presence of CLso.
Monitoring occurs in Perth's autumn (March) and spring (October) growing seasons, and began in March 2017. Fortunately, the associated bacterium has not been detected, however, TPP has been detected in Albany in the south of Western Australia, Esperance in the east, and north to Geraldton and Carnarvon, presumably through accidental human-mediated dispersal.

Experimental Design
Adopt-a-trappers from within the greater Perth metropolitan region were recruited to host traps over a 4week period in October 2020. Trappers were selected from previous surveillance for CLso, which demonstrated that they had TPP in their gardens in the previous two years. Trappers were given a tomato plant and a native citrus plant (Murraya sp.) as sentinel plants for TPP and ACP, respectively. Seven trappers used their own established citrus trees (lemon or orange) instead of the offered Murraya plant. Trappers were randomly assigned traps with one of the below designs (Table 1). *Lures were only attached to the yellow sticky trap near tomato Participants were each allocated a minimum of eight yellow sticky traps (four with a white dot, four with a red dot), two trap cages, 16 barcodes, eight trap protectors (clear plastic sleeves), four return post envelopes, four trap return sheets and a lure. Lured traps were yellow sticky traps with red dots, with the lure attached to the cage of the trap, and placed near the tomato plant on trapper's properties (Fig. 1). The control lure was a square of felt glued on to the cage. Yellow traps with the white dot were placed in their cage near the Murraya plant without any lure, and approximately 2m from the tomato plant ( Fig. 1). A subset of trappers were given a third trap and cage, with two 'green' sticky traps, to be placed near their tomato plant. Volunteers were asked to submit a report and photos using MyPestGuide™ Reporter (see https://www.agric.wa.gov.au/pests-weeds-diseases/mypestguide ) in the rst week of trapping to report their sticky trap locations. Each week the sticky traps were collected and replaced. Used sticky traps were placed in protective plastic sleeves with a barcode identi er. Traps were returned in a prepaid envelope along with a trap return sheet that also included a duplicate barcode to ensure that traps could be linked to location and date.
Once returned, the traps were triaged, assigned a lab identi cation number and entered in the Sample Manager (Version 12.2; Thermo Fisher Scienti c, 2019) database. We used Leica M80 dissecting microscopes to identify and count insects on the traps. Pentatomoidea were identi ed to species level, while Psylloidea were classi ed as follows "TPP", "Other Triozidae" and "Other Psylloidea". In addition, we scanned the traps for Pentatomoidea and Psylloidea exotic to Western Australia. Results were recorded on the individual lab sheet for that trap, which included its unique identi cation number, and data was subsequently entered into the database. When TPP were found, specimens were cut from the trap using a scalpel and cleared of adhesive material using household-grade orange oil and Decon 90, then sent for PCR testing to monitor for any presence of the exotic bacterium CLso. Weekly emails were sent to adopta-trappers informing them of their TPP count and the presence of any concerning species, as well as their MyPestGuide reports replied to with any interesting nds. The traps were stored in a refrigerator and associated paperwork led.

Data Analysis
Logistic regression models (family "Poisson") were used to analyse the data in

Results
In total, 9,180 insects were identi ed from 820 traps, set at 104 adopt-a-trapper's households. The number of traps assessed were less than the maximum (882) because some trappers were only recruited for two weeks, other trappers did not return the full complement of traps, and quality control removed data when a trapper's lure set up was determined to be less than optimal (e.  (Fig. 2A). Unsurprisingly, yellow sticky traps placed with a citrus sentinel plant attracted far fewer TPP at 2.89 per trap ( Fig. 2A). When lures were added to traps, these reduced the number of TPP captured signi cantly, regardless of whether the lure was BMSB (to 4.32 TPP individuals), ACP (6.17 TPP), or a combination (6.58 TPP), however, BMSB lures were the most dissuasive (Fig. 2D).
There appeared to be an interaction with time, in that the impact of ACP lures on TPP slightly decreased after week 1 (Fig. 2G), although this interaction wasn't signi cant. Control traps indicated a peak in TPP numbers in week 2. In contrast, both lures combined showed no difference in TPP capture over time, maintaining low levels, while BMSB lures continued to dissuade TPP into week 4 (Fig. 2G). However, there was no signi cant interaction with time overall, likely due to the large weekly uctuations of the control, particularly in week 2.

Other Triozidae
Despite being in the same family as TPP, there were far fewer other Triozidae represented on the traps overall ( Fig. 2B) and 25% of properties didn't record any (26 households). Although there were no clear patterns of preference with regards to lures (Fig. 2E), there was a general reduction of individuals captured over time, by approximately 50% for the control, ACP and BMSB lures (Fig. 2E). A peak in triozid abundance was noted in week 2 for ACP and both lures combined, while it occurred in week 3 for the control traps (Fig. 2H).

Other Psylloidea
Other Psylloidea were found at every property except two (98% had other Psylloidea). They were also attracted to the green sticky traps in signi cantly higher abundance at a mean of 9.98 per trap than the yellow sticky traps (3.61 per trap) (Fig. 2C). In contrast to TPP, yellow sticky traps placed with a citrus sentinel plant attracted similar abundance of psyllids as the tomato traps at 3.71 per trap. When lures were added to traps, ACP had no impact (3.01 per trap), but the BMSB lure had a positive effect on the Psylloidea, increasing capture rates signi cantly to 5.94 per trap (Fig. 2F). However, when ACP lures were added in combination, this interfered with BMSB lures, and reduced the e cacy back to control levels of 3.82 individuals per trap (Fig. 2F). The attractiveness of BMSB lures on other psyllids signi cantly decreased over time (P < 0.05) until reaching the levels of all other traps by week 4 (Fig. 2I). Control traps indicated a peak in psyllid numbers in week 3. In contrast, ACP and both lures combined showed little difference in psyllid capture over time, maintaining lower and not displaying the peak in week 3 of the control traps levels (Fig. 2I).

Stinkbugs
Outside of the lure experiment, and for surveillance of the bacterium CLso vectored by TPP, 447 yellow traps were placed near tomato but no stinkbugs were captured on any of these. Only 18% of lure experiment trappers captured stinkbugs (19 households). Of these households, very few stinkbugs were captured near citrus (Fig. 3A). Stinkbugs were predominantly captured on lured yellow traps near tomato (Fig. 3B) 3B). Interestingly though, no stink bugs were captured with the ACP lure alone. Without any lure, only two P. a nis were recorded (Fig. 3B). Capture rates also decreased with time, with 15, 10, 5, 1 stink bugs caught in weeks 1 to 4, respectively (Fig. 3C).

Discussion
Lures signi cantly increase trap e cacy for targeted insect species. The commercial lures we utilized increase capture of Asian citrus psyllid by 3-4 times (Czokajlo et al. 2015), and brown marmorated stink bug by 120 + times (Weber et al., 2019). Lures can, however, have consequences on the capture rates of non-target taxa, and these impacts are not always predicted by taxonomic boundaries. For example, lures of the Noctuidae moth Helicoverpa armigera (Hübner) attract signi cantly more non-target moths, bees and ladybird beetles to traps (Spears et al., 2016). We found that tomato potato psyllid (TPP) was repelled by both ACP and BMSB lures, despite TPP and ACP both belonging to the superfamily Psylloidea. Psyllids representing the families Aphalaridae and Psyllidae were also repelled by ACP lures. Asian citrus psyllid belongs to the family Liviidae and is not represented by any known species in Western Australia, so the intrafamilial e cacy of the ACP lure remains unknown. Attraction or avoidance to lures corresponds to the chemicals used. The chemicals in ACP lures are a mixture of organic volatiles present in new ushes of citrus host plants, namely aphla-phellandrene, beta-phellandrene, beta-caryophyllene, gamma-terpinene, ocimene and terpineol (Setamou, 2018 Instead of plant volatiles, lures may be based on insect pheromones, which will have an even more unpredictable effect on bycatch. For example, the H. armigera lures mentioned above are based on pheromones, but can signi cantly increase ladybird beetle, bee and nontarget moth captures by 23%, 110% and > 2,000%, respectively (Spears et al. 2016). BMSB lures are pheromone lures and, while TPP was signi cantly repelled by the BMSB lure, psyllids other than TPP were attracted to BMSB lures, despite BMSB and psyllids belonging to different taxonomic suborders of true bugs (Heteroptera versus Sternorrhyncha). Delving into our psyllid data, high capture rates of a diverse composition of different species, mainly native Acizzia species, often dictated the pattern. In many cases, however, Ctenarytaina longicauda Taylor appeared to be driving the association on BMSB lured traps at tomato, and on sticky traps at citrus. Ctenarytaina longicauda is another recent arrival in Western Australia from eastern Australia (Francesco Martoni, Submitted 2021) and although Lophostemon confertus (R.Br.) is the known host plant, in Perth C. longicauda can be found in high abundances feeding on citrus (M.Moir pers. obs.). Why C. longicauda was not attracted to ACP lures, which are derived from the plant volatiles of fresh citrus growth (Setamou, 2018), remains unknown.
In some cases, species-speci c lures will attract an unintended species within the same family. We found that BMSB lures attracted the pest stinkbug Plautia a nis (Dallas), the only known species of Plautia to occur in southwest Western Australia. Plautia a nis is a likely introduction from eastern Australia as records in Western Australia prior to (Cassis & Gross, 2002) are scarce, with (Froggatt, 1907) restricting species distribution to the state of New South Wales. The earliest specimens from Western Australia were collected by L.J. Newman in Geraldton (ICDB 2018), no date is listed, but would have been after his appointment as Government entomologist in 1918 and prior to his death in 1938. BMSB lures contain, in part as a pheromone synergist for BMSB, the pheromone of Plautia stali Scott (methyl (2E,4E,6Z)-2,4,6decatrienoate), which is a sympatric species with BMSB in Asia (SUGIE et al., 1996). We found that combining lures will not counteract one another, nor will they have a cumulative impact on species examined here, supporting similar results for lures on BMSB and pest wood-boring beetles (Cerambycidae and Scolytinae: Curculionidae) (Chase et al., 2018), as well as Lepidoptera lures (Brockerhoff et al., 2013). For example, P. a nis was attracted to BMSB lures whether they were in combination with ACP lures or not, and although TPP was repelled by both lure types, combinations of these lures did not increase repellancy signi cantly. Unsurprisingly, we found that over time lure effectiveness declined as the chemical degrades, which has been an identi ed issue for lure longevity (Suckling, 2000); (Cottrell et al., 2020). The impact of the lures were most prominent for all taxa in the rst two weeks, after which TPP was the only insect still detecting and avoiding BMSB lures.

Conclusion
Surveillance for plant pests is conducted to de ne the pest status of an area, improve the likelihood of early detection and monitor pest prevalence. The data collected from surveillance can help establish pestfree areas and streamline the trade of plants and plant products. While biosecurity agencies attempt to survey for many taxa with fewer resources, experimentally determining trap e cacy prior to initiating surveillance avoids squandering resources in the long-term. While attempting to incorporate additional exotic species into the surveillance for the bacterium CLso (vectored by TPP), we found that ACP and BMSB lures could not be used on traps intended to target TPP, but that 'green' ACP sticky traps were more effective at capturing TPP and other Psylloidea than yellow traps, and could confer a more effective multi-species approach. Our study demonstrates that, when planning multi-taxa surveillance or monitoring programs, lures must rst be tested on both the target and non-target insect taxa to determine trap e cacy as the consequences for species catch can be unpredictable.