Eradicating little � re ants ( Wasmannia auropunctata ) from the island of Kaua i , Hawai i : process , challenges , and results to date

Michelle Montgomery (  michelle.montgomery@little reants.com ) Institute for Applied Ecology, University of Canberra Casper Vanderwoude Hawai i A n t L a b , P a c if i c C ∞ p e r a t i v e S t u d i e s U n i t , U n i v e r s i t y o f H a w a i i A. J. Jasmyn Lynch Conservation and Water Planning Policy Branch, Environmental, Planning, and Sustainable Development Directorate, ACT Government Tiffani Keanini Kaua i I n v a s i v e S p e c i e s C o m m i e e , U n i v e r s i t y o f H a w a ii Craig Kaneshige Plant Pest Control Branch, Hawai`i Department of Agriculture


Introduction
Invasive ants are especially successful invaders because they possess the ability to adapt and invade a wide range of habitats.Special biological and behavioral traits, such as polygyny, unicoloniality, high interspeci c aggression, symbiotic relationships with honeydew producing insects, and use of human mediated long-distance dispersal, contribute to the success of invasive ants as invaders (Holldobler and Wilson 1990;O'Dowd et al. 2003;Passera 1994;Wetterer and Porter 2003).Of the 147 ant species which have been transported beyond their native range, only a small subset are known as IAS (McGlynn 1999) with well-documented negative impacts.Five ant species are the most common targets of ant eradication programs worldwide: Anoplolepis gracilipes Smith, Linepithema humile Mayr, Pheidole megacephala Fabr, Solenopsis invicta Buren and Wasmannia auropuncata Roger.Despite the frequency at which ant eradications take place, the outcomes of ant eradication efforts are still poorly documented and frequently unsuccessful (Hoffmann et al. 2016).
Hawai`i has no known native ant species, yet over 60 species have been introduced and established to date.Many of Hawai`i's introduced ant species are considered invasive "tramp" species owing to their propensity to be distributed long distances via human-mediated transport and a unique set of biological characteristics such as polygyny and unicoloniality (Loope and Krushelnycky 2007;Passera 1994).Tramp ants are also known for their negative impacts on residential, agricultural and natural ecosystems ( As is common among invasive tramp ants, LFA primarily spread by people unwittingly moving infested materials (e.g.nursery stock, agricultural materials and commodities, aggregate, vehicles, machinery, etc…) to uninfested locations.Economic costs of LFA infestations in Hawaii are projected to be $6.1 billion over the next 35 years provided that management efforts are maintained at the current "status quo" level (Lee et al. 2015).Because of this, the state of Hawai`i has a vested interest in eradicating LFA infestations when and where feasible.

Little Fire Ants in Hawai`i
Originally detected in 1999 on Hawaii island (Conant and Hirayama 2000), LFA have persisted and spread to four of the main Hawaiian Islands despite early attempts at eradication.The size and number of infestations identi ed in 1999 suggested that LFA had been introduced several years prior.
Although it is uncertain exactly when and how they were introduced, Hawaii's LFA are genetically distinct and identical to those in Florida, suggesting Florida as the likely source of the Hawai`i population (Foucaud et al. 2010).
Results from response surveys and publicity relating to these detections indicated that the number of LFA-infested private properties and commercial nurseries on Hawaii island at the time of original detection was such that eradication was not feasible with the resources available at that time.
Additionally, in 1999, an isolated infestation was identi ed at a private property on the north shore of Kauai island.The origin of that infestation was likely from a shipment of infested palms from a nursery on the island of Hawai`i (Conant and Hirayama 2000; Null and Gundersen 2007).
An island-wide survey in 1999 did not detect any other LFA infestations on Kauai (Null and Gundersen 2007), and there appeared to be an opportunity for an island-wide eradication.In October 1999, the Hawaii Department of Agriculture (HDOA) along with the Hawaii Department of Health Vector Control Branch launched the rst LFA eradication effort for the state of Hawaii focused on the Kaua`i infestation.The infestation was treated using ready-to-use granular insecticidal ant baits (Amdro®, hydramethylnon 0.739% a.i.), however no documentation could be found regarding methods used to treat the original infestation such as application rates, frequency, and number of applications.At the time, the eradication was considered successful with LFA declared "eradicated" in HDOA's Annual Report for FY 2000 (Hawaii Department of Agriculture 2001).Unfortunately, follow-up surveys of the area in 2003 by the Kauai Invasive Species Committee (KISC) revealed that the infestation had rebounded and spread to two adjacent privately-owned properties.A monitoring and containment strategy was initiated until effective eradication methods and procedures could be developed for this species (Null and Gundersen 2007).
Since the original eradication attempt in 1999, considerable research has been conducted and dedicated to understanding LFA biology and behaviors and improving management strategies and bait e cacy (Montgomery et al. 2015;Montgomery et al. 2020).This new knowledge was used to help formulate appropriate species-speci c management strategies.A lipid-based gel bait and new eradication procedure was developed by the Hawaii Ant Lab (HAL) (hereafter referred to as the HAL gel bait) and this showed great promise against LFA infestations (Vanderwoude and Nadeau 2009;Vanderwoude et al. 2010).In 2011, the HAL partnered with the KISC and the HDOA to develop and implement an LFA eradication plan for the second time on Kauai.Here, we detail the process and current status of the second LFA eradication attempt.
Landscape features included open mown lawn, palm arboretum, dense tropical landscaping and non-native coastal forest, sheer cliffs, and rocky outcrops which were surrounded by ocean at high tide ( gure 2).Approximately one-third of the infested area consisted of steep terrain requiring the use of specialty equipment and training to navigate.The steep cliffs were wholly infested and when the eradication began there was no viable means of accessing this area nor were there treatment methods developed for such site features.As a result, the eradication plan was divided into two phases ( gure 3).Phase I consisted of treating all areas accessible without the use of specialty equipment between September 2012 and July 2013.Phase II consisted of treating the remainder of the infested area between September 2014 and June 2015 and once specialty equipment, training and appropriate treatment methods became available.A 20 m buffer surrounding the remaining infestation during Phase II resulted in overlap of the Phase I and Phase II treatment areas.

Survey methods
Little re ant surveys consisted of placing 5-dram plastic vials (20.6 x 52mm, BioQuip® Products, California, USA) laced with peanut butter (hereafter referred to as sample vials) which were collected after 60 min exposure time.The vials then were capped and labelled, with GPS coordinates recorded for each sample site using a GARMIN Foretrex 401 GPS (Garmin International, Missouri, USA).Ants captured in the vials were identi ed under a dissecting microscope and attributed with the geospatial data.All LFA captured were counted and numbers recorded in the survey database.Treatments were focused on the use of insecticidal baits applied to the ground and all vegetation throughout the treatment area.Baits were applied at six-week intervals for 12 months during Phase I and Phase II treatment regimens.The HAL gel bait containing 0.25% s-methoprene (Tango™, EPA reg. 2724-420, Wellmark International, Illinois, USA), an insect growth regulator (IGR), was applied to the ground and all vegetation for the rst half of treatments and the HAL gel bait with 0.18% indoxacarb (Provaunt®, EPA reg.100-1487, Syngenta Crop Protection LLC., North Carolina, USA), an oxadiazine insecticide, was applied to vegetation only for the nal half of treatments during each phase of the treatment regimen.Broadcast applications of Probait® (0.73 hydramethylnon, EPA reg.73342-1-2724, Wellmark International, Illinois, USA), a ready-to-use granular bait, were made one week following each HAL gel bait application during Phase I; however, Phase II treatments consisted solely of the HAL gel bait with s-methoprene and indoxacarb as described above.The omission of granule baits during Phase II was due to the terrain making it di cult to apply granular products throughout the area.
Spot treatments were made as needed throughout the eradication effort and consisted of insecticidal bait applications or use of residual insecticides, such as Talstar® P (

Results
The number of LFA positive detections and the mean number of ants per sample vial within the Phase I monitoring plots rapidly decreased upon commencement of the treatment regimen ( gure 4).Results from the Phase I midpoint monitoring survey indicated the LFA infestation had greatly reduced in size with only 10.5% of deployed sample vials detecting LFA ( gure 5).Based on the number of sample vials containing LFA, the estimated Phase I infested area had been reduced from 3.22 to 0.39 ha.By January 2014, 4.6% of deployed sample vials were containing LFA, the majority of which were along the boundary dividing the Phase I and II treatment areas.Discounting the positive detections along the Phase I and II boundary, the Phase I infested area was estimated in 2014 to be 0.05 ha.A slower decline was observed during the Phase II treatment regimen.Results from the Phase II midpoint monitoring survey indicated that the infested area had been reduced from 1.35 ha to 0.76 ha with 37.9% of the sample tubes collected containing LFA ( gure 5).By January 2016, no LFA were detected in either Phase I or Phase II treatment areas.
A total of 53,757 sample tubes were deployed and collected during post treatment monitoring surveys between 2016 and 2021.Isolated LFA detections (hotspots) were detected on several occasions within the Phase I treatment area; the last being in September 2019.No hotspots have been detected within the Phase II treatment area since area-wide post-treatment monitoring began.

Discussion
Ant eradications are notoriously di cult and many factors can in uence the success or failure of an eradication project.The development of the HAL gel bait has provided practitioners with the ability to effectively treat tall vegetation as well as the ground, and has increased the chances of successful eradication and management programs.Since site features and vegetation varied between the Phase I and II treatment areas, it is di cult to tell with certainty whether the different treatment methods or site features and vegetation had a greater impact on the results of monitoring efforts for each phase.However, the general trends observed at the midpoint and post treatment monitoring surveys showed rapid population knockdown during Phase I and a gradual population decline over time during Phase II ( gure 5).The difference was likely due to the incorporation of toxic granular baits throughout the Phase I treatment regimen which did not occur during Phase II.At the same time, the early reduction in population due to the use of toxic granular baits concurrently with the HAL gel bait containing s-methoprene likely reduced the amount of IGR shared throughout the super colony; thus, the residual effects of the IGR may not have been as pronounced during Phase I as in Phase II.It is also possible that the vegetation composition throughout the Phase I treatment area was such that more areas of refuge were available for hot spots to persist compared to the Phase II treatment area.
Treatment methodology has greatly improved since the original eradication attempt in 2000 due to dedicated research on species speci c treatment methods for LFA.However, the ability to detect incipient remnant colonies is as important as nding suitable and effective treatment methods.Our post treatment monitoring procedure included closer spacing of sample tubes, palm tree crown surveys, and expanded surveys beyond the treatment area in order to increase the chances of detecting lone, remnant colonies within the landscape.Little re ant hotspots continued to be detected occasionally until September 2019.
It is not uncommon for incipient colonies to go undetected for multiple years using the ground survey methods described here.We have identi ed two main factors that likely contributed to the persistence of LFA hotspots within the treatment area and detections outside of the treatment area, despite repeated thorough surveys: 1) under-treated areas or features such as tall trees, dense vegetation, large mulch piles, steep terrain, or a combination of these act as reservoirs and safe havens for LFA to persist; and 2) movement of infested yard waste containing colony fragments to mulch piles beyond the treatment area allowed ants to be transported after treatments had ceased.
Although the application equipment allowed for baits to be applied up to 9 m horizontally and vertically, and mapping of treatment tracks indicated su cient coverage, treatments were likely insu cient for palm trees ≥ 15 m or for penetration of dense mulch piles.Additionally, although we surveyed all palm trees throughout the treatment area multiple times, sample tube placement is critical for accurate results.Non-ideal sample tube placement may have resulted in occasional false negative survey results.
Although no LFA have been detected within the Phase II treatment area since area-wide post-treatment monitoring began, post-treatment monitoring should be maintained long term.Not detecting ants during a survey does not necessarily indicate absence, but that the size of a population is below detectable levels.It is possible that remnant LFA colonies remain within small areas that were completely inaccessible despite the use of specialty rappelling equipment.
Treatment options and methodology for LFA control have improved substantially over the past decade.However, the accuracy of monitoring programs continues to be a weak point in the process.The use of baited tubes or other lure-based monitoring methods can be labor intensive and project managers often nd themselves in the position of sacri cing accuracy due to funding and personnel constraints.More effective detection methods, that don't sacri ce accuracy, are needed to seek out and destroy incipient remnant colonies within as soon as possible after cessation of treatments.
Surveys were conducted throughout the eradication effort (2011 -present) for infestation delimitation prior to treatment, population monitoring, and detection of nascent remnant colonies after cessation of treatment regimens.For delimiting surveys, the sample vials were deployed at approximately 10 m intervals along transects radiating outward from known infested areas.The outer boundary of the infestation was determined when sample results revealed zero LFA detections for a minimum of 50 m.Midpoint monitoring surveys were conducted throughout the immediate treatment area (i.e.Phase I vs. Phase II treatment area) and immediately prior to the 5 th treatment during both Phase I and Phase II treatment regimens.Sample vials were spaced in a 10 x 10 m grid-like pattern throughout the treatment areas for Phase I and Phase II midpoint surveys.The purpose of the midpoint monitoring survey was to assess treatment e cacy and identify areas requiring further attention.This allowed for adjustments to the treatment procedure to be made in a timely manner if and when needed.Immediately prior to each Phase I treatment (September 2012 -July 2013), sample vials were deployed in eight permanent monitoring plots placed randomly throughout the treatment area (table1).The monitoring plots were used to track population dynamics in various habitat types during the Phase I treatment regimen.No monitoring plots were established during Phase II.Post treatment monitoring throughout the Phase I treatment area only occurred between January 2014 and June 2015 with area-wide (Phase I and II treatment areas) post treatment monitoring beginning in January 2016 and continuing to date.vegetable garden approximately 3 x 5 m Ground sample only 1 1

Figures Figure 1
Figures

Figure 2 Photos
Figure 2

Figure 3 Map
Figure 3

Figure 4 Results
Figure 4

Table 1 :
Descriptions of the 2012-2013 population monitoring plots and sample placements.colonies present after cessation of the treatment phases.Expanded surveys, extending beyond the treatment area, occurred at least once per year between 2016 and 2021.Additional sample vials were placed in the crowns of all palm trees higher than 3 m.Sample tubes were placed in the tree crowns via a weighed line.These were left in position for 24 h before collection rather than 60 min due to the time in which it took to deploy arboreal samples.The purpose of the canopy survey was to determine if any arboreal colonies survived.
Area-wide post-treatment monitoring began after all treatment phases were completed.The number of area-wide post-treatment surveys per year varied due to property access, crew availability, and weather (table 2).Vial density was increased in order detect any nascent colonies remaining and the sample grid spacing was reduced to an average of 2.5 x 2.5 m.High sample vial densities for post-treatment surveys increase the likelihood of detecting nascent remnant

Table 2 :
Summary of area-wide post-treatment surveys for the years 2016-present.No surveys were conducted in the year 2020 due to the COVID-19 pandemic and associated restrictions.Total area surveyed was calculated by buffering survey points to 3 m and then calculating the area of the resulting polygon.
Treatment Strategy Approximately 3.2 ha were treated during Phase I and approximately 1.4 ha were treated during Phase II treatment regimens ( gure 3).The overlap in Phase I and II treatment areas ensured su cient treatment coverage.

Table 3 :
.22.0-Białowieża mapping software, buffering each point to 6 m, and calculating the total area of the buffered points.
7.9% bifenthrin, EPA reg.279-3206,FMCCorporation,Pennsylvania,USA) or Talstar® PL (0.2% bifenthrin, 279-3168, FMC Corporation, Pennsylvania, USA).Other ready-to-use granular bait products used during spot treatments were Amdro Fire Ant Bait (0.73% hydramethylnon, EPA reg.73342-1,AMBRANDS,Georgia,USA) and Siesta Fire Ant Bait (0.063% meta umizone, EPA reg.7969-232,BASFCorporation,NorthCarolina USA).Total annual amounts of all pesticides used throughout the project are listed in table3.Total amounts of baits and residual barrier treatments applied annually between 2012 and 2020Data AnalysisDescriptive statistics, such as mean number of LFA per sample vial and number of sample vials with positive detections was used as a surrogate for LFA population and to track treatment e cacy over time.Estimates for infested area were calculated by plotting GIS data for positive detections in QGIS version 3