PACIFIC COOPERATIVE STUDIES UNIT UNIVERSITY OF HAWAI`I AT MĀNOA Dr. David C. Duffy, Unit Leader Department of Botany 3190 Maile Way, St. John #408 Honolulu, Hawai’i 96822 Technical Report 173 The Haleakala Argentine ant project: a synthesis of past research and prospects for the future June 2011 Paul Krushelnycky 1,2 William Haines 1,2 Lloyd Loope 3 Ellen Van Gelder 2,4 1 Department of Plant and Environmental Protection Sciences, University of Hawaii, 3050 Maile Way, Gilmore 310, Honolulu, HI 96822 [email protected] 2 Prior affiliation: Pacific Cooperative Studies Unit, Department of Botany, University of Hawaii, Honolulu, HI 96822 3 U.S. Geological Survey, Pacific Island Ecosystems Research Center, Haleakala Field Station, P.O. Box 369, Makawao, HI 96768 4 U.S. Fish and Wildlife Service, Endangered Species Program, 4401 N. Fairfax Drive, MS 420, Arlington, VA 22203 PCSU is a cooperative program between the University of Hawai`i and U.S. National Park Service, Cooperative Ecological Studies Unit. Organization Contact Information: Department of Plant and Environmental Protection Sciences, University of Hawaii, 3050 Maile Way, Gilmore 310, Honolulu, HI 96822. Office: (808) 356-8261. Recommended Citation: Krushelnycky, P.D., W. Haines, L. Loope and E. Van Gelder. 2011. The Haleakala Argentine ant project: a synthesis of past research and prospects for the future. Pacific Cooperative Studies Unit Technical Report 173. University of Hawai‘i at Mānoa, Department of Botany. Honolulu, HI. 127 pp. Key words: Argentine ant, Linepithema humile, ant control Place key words: Pacific islands, Maui, Haleakala National Park Editor: Clifford W. Morden, PCSU Deputy Director (e-mail: [email protected]) 2 TABLE OF CONTENTS Executive summary 3 I. Introduction 4 II. Background 4 A. Patterns of spread 4 B. Ecological effects 6 III. Research on Argentine ant management at HALE 7 A. Rationale for attempting ant control 7 B. General strategy 8 C. Testing baits for eradication potential 10 1. Initial bait preference test 10 2. 1995 eradication plots 11 3. 1996-7 eradication plots 12 4. 1998 eradication plots 13 5. 2001-2002 eradication plots 15 6. 2002-2003 bait preference trial and eradication plots 16 7. 2004-2005 eradication plots 18 8. 2007 Gourmet Liquid Ant Bait eradication plots 21 9. 2007 0.5 HP Granular Ant Bait eradication plots 22 10. 2008 Advion Insect Granule eradication plots 24 D. Efforts at Argentine ant containment 25 1. 1996 experimental perimeter plots 26 2. 1997-2004 containment efforts 26 IV. Summary of management efforts to date and prospects for the future 29 V. Acknowledgements 33 VI. Literature cited 33 Appendix 1. Argentine ant patterns of spread at Haleakala National Park 37 Appendix 2. Results from the 1998 eradication plots 41 Appendix 3. Results from the 2001-2002 Pharorid and Maxforce plots 44 Appendix 4. Results from the 2002-2003 bait trials and eradication plots 46 Appendix 5. Results from the 2004-2005 eradication plots 49 Appendix 6. Report on evaluation of Gourmet Liquid Ant Bait 52 Appendix 7. Report on evaluation of 0.5 HP Granular Ant Bait 85 Appendix 8. Report on evaluation of Advion Insect Granule bait 101 Appendix 9. Results of 1996 experimental perimeter plots 120 Appendix 10. 1997-2004 experimental containment treatments 123 3 EXECUTIVE SUMMARY 1. The Haleakala Argentine Ant Project is an ongoing effort to study the ecology of the invasive Argentine ant in the park, and if possible to develop a strategy to control this destructive species. 2. Past research has demonstrated that the Argentine ant causes very significant impacts on native arthropods where it invades, threatening a large portion of the park’s biodiversity in subalpine shrubland and alpine aeolian ecosystems. 3. Patterns of spread over the past 30+ years indicate that the invasion process is influenced to a substantial degree by abiotic factors such as elevation, rainfall and temperature, and that the ant has not reached its potential range. Predictions of total range in the park suggest that it has only invaded a small fraction of available suitable habitat, confirming that this species is one of most serious threats to the park’s natural resources. 4. Numerous experiments have been conducted since 1994 in an attempt to develop a method for eradicating the Argentine ant at Haleakala using pesticidal ant baits. Thirty baits have been screened for attractiveness to ants in the park, and ten of these were tested for effectiveness of control in field plots. While some of these baits have been very effective in reducing numbers of ants, none has been able to eliminate all nests in experimental plots. 5. Research into a secondary management goal of ant population containment was initiated in 1996. By treating only expanding margins of the park’s two ant populations with an ant pesticide, rates of outward spread were substantially reduced in some areas. While this strategy was implemented from 1997 to 2004, it was ultimately discontinued after 2004 because of the difficulty and insufficient effectiveness of the technique. 6. In order to achieve the types of results necessary for eradication, the project would probably need to explore the possibility of developing a specialized bait, rather than relying on a commercially produced bait. An alternative would be to pursue approval to use Xstinguish bait, a commercial bait manufactured in New Zealand and not registered for use in the US, which has yielded good results against Argentine ants. Either route would involve significant regulatory hurdles. Because the baits ultimately used would likely be liquid or paste in form, there would also be major logistical challenges in devising methods to successfully apply the baits across the two large ant populations at Haleakala. 4 I. INTRODUCTION The Haleakala Argentine Ant Project has been an effort, spanning the past 30 years, to understand the ecology of the invasive Argentine ant in Haleakala National Park (HALE) and, if possible, to develop management solutions for this important threat to park natural resources. Although the Argentine ant was recognized as a serious cause for concern soon after its detection in the park in 1967, research into management did not begin in earnest until the mid 1990’s. Some of this research on Argentine ant control at HALE has previously been published in journal articles or agency reports, but much of it is reported for the first time here. This technical report provides an overview and summary of all of the Argentine ant control efforts that have taken place at HALE from 1994 through 2010. II. BACKGROUND It is believed that there are no native ants in Hawaii. Nearly 60 ant species have been introduced to the state in the past several hundred years, however, and some of these have caused substantial impacts to native Hawaiian biodiversity (Reimer 1994, Krushelnycky et al. 2005a). Most of the ant species introduced to Hawaii have thrived in the warmer lowlands, and have probably contributed to the extinction of numerous native insects and related arthropods. A handful of species are better adapted to cooler climatic regimes, and several of these become extremely abundant, dominant invaders in Hawaii’s middle to high elevation zones. The Argentine ant, Linepithema humile (Mayr), is among the latter group, and has become one of the most important threats to subalpine shrubland and alpine zone ecosystems in Hawaii. The Argentine ant was first recorded in the park at Hosmer Grove in 1967 (Huddleston and Fluker 1968). Because this ant had such a notorious reputation as a destructive invasive species in California, the southeastern U.S., Australia, Europe, and South Africa, HALE staff and collaborating scientists initiated several research projects in order to learn more about its invasiveness and probable impact in the park. Beginning in 1980, researchers periodically mapped the ant’s distribution to track its spread. Subsequently, studies were conducted in the mid 1980’s and again in the mid-2000’s to estimate the ant’s effects on the resident arthropod community. The results of these efforts confirmed that the Argentine ant was indeed one of the most important threats to the park’s endemic biota; a wide range of native arthropods were significantly reduced in abundance as the ant continued to steadily spread into larger areas of the park, with wider ecological impacts possible. These results are summarized below. A. Patterns of spread The record of Argentine ant invasion at HALE represents one of the most detailed accounts of alien species invasion. The two ant populations have been mapped periodically from 1980 to 2004, with increasing frequency and accuracy since 1993. Synthesis of these data highlights several important trends. First, the patterns of spread clearly indicate that this species is not restricted to areas of human disturbance, as is sometimes the case with invasive ants. The vast majority of the two populations occupy habitat unaffected by roads, trails or structures. Second, 5 abiotic factors, most importantly rainfall and soil temperature, have strong influences on where, and at what rate, the ant populations spread within the park. When the Argentine ant was first mapped in 1980, thirteen years after its detection at Hosmer Grove, it occurred over an area of about 165 ha that encompassed the park service area and headquarters (Fellers and Fellers 1982). The Argentine ant, like several other invasive ant species, does not have mating flights, and its populations therefore spread through budding: newly mated queens disperse with a retinue of worker ants a short distance on foot, creating a general pattern of outward population expansion. In 1982, a second ant population was discovered higher up the mountain, perched on the crater rim around Kalahaku Overlook (Fellers and Fellers 1982). This second population was recent in origin and still fairly limited in distribution, and was most likely initiated through inadvertent human movement of nest material from the first population. Over the fifteen years from 1982 to 1997, both populations increased dramatically in size, reaching a total area of about 555 ha. This spread is illustrated in Figures 1- 3 in Appendix 1. (We have designated the two ant populations as the lower population and upper population, referring to the original and secondary populations, respectively). The pattern of spread in the lower population indicates that the Argentine ant appears to have an upper rainfall tolerance of about 1600 mm/yr at this location. Hawaii’s northeast tradewinds create the predominant weather patterns on the mountain: the north and northeast slopes of Haleakala receive high levels of annual rainfall, while the leeward south and southwest slopes sit in a rain shadow and are considerably drier. In between these areas, such as at Hosmer Grove on Haleakala’s northwest slope, steep rainfall gradients are formed (Appendix 1, Fig. 1). Over the course of 30 years, the lower ant population has not spread to the northeast beyond Hosmer Grove. Instead, 94% of the eventual colonization occurred southwest of Hosmer Grove, or towards the lee of Haleakala volcano (Appendix 1, Fig. 2). Spread towards the northeast is most likely limited by cold soil temperatures, which result largely from the increased vegetative cover that prevails as rainfall increases along the windward gradient. Unlike the lower population, the upper population has not yet encountered environmental conditions that prevent its outward expansion. However the rate of spread within the upper population has not been equal in all directions. Namely, expansion from Kalahaku overlook westward outside the crater rim has been relatively slow (averaging 23 meters per year), while eastward spread down the crater walls and across the crater floor has been much more rapid (increasing to 91 meters per year) (Appendix 1, Fig. 3). Rates of spread in both populations are strongly tied to soil temperatures, and can be predicted with fairly high accuracy using a simple degree-day model based on soil temperature. As the upper population has reached warmer temperatures down the crater walls and on the crater floor, rates of population growth and opportunity for above-ground foraging have likely increased, resulting in the faster rates of spread. Similarly, the lower population has expanded more rapidly towards the lee of the volcano, where sparser vegetative ground cover results in warmer soil temperatures. Based on these past patterns of spread, we made a rough estimate in 2004 of the total potential range of the Argentine ant in the park (Appendix 1, Fig. 4), as well as the rate at which it would likely fill this range. As indicated in Figure 4, this potential range included most of the remainder of the west slope shrubland and aeolian zone, including the summit, and much of the crater and 6 Kaupo Gap. At 5360 ha, the predicted potential range encompassed approximately 48% of the total park area (based on the park land holdings at the time), and 75% of the park’s subalpine shrubland and alpine aeolian zone habitats. More recent modeling of potential range based on additional soil temperature measurements and degree-day estimates (Hartley et al. 2010) suggests that this earlier estimate was probably conservative, and that even more of the park is suitable for Argentine ant invasion. Moreover, suitable habitat will increase as temperatures increase at the top of Haleakala, and this warming is currently occurring at a rapid rate (Giambelluca et al. 2008). We also estimated that without management intervention, the two populations would merge on the west slope in less than 20 years, would invade half of the predicted total range in the park in 30 years, and would cover roughly 90% of this area in about 70 years. We estimate that the establishment of the second population at Kalahaku around 1977 has accelerated the invasion process into the park by about 50 to 60 years, underscoring the critical importance of preventing additional founder populations. A more detailed account of these trends and predictions is provided in Krushelnycky et al. (2005c) and Hartley et al. (2010). B. Ecological effects In the mid-1980’s park scientists and colleagues used pitfall sampling and under-rock surveys to assess the effects that the Argentine ant invasion was having on the arthropods in the shrubland and alpine aeolian ecosystems (Medeiros et al. 1986, Cole et al. 1992). They found that a wide range of endemic arthropods is significantly reduced in abundance within ant-invaded areas. The affected groups include, among others, herbivores (Agrotis spp. [Lepidoptera: Noctuidae]), predators and scavengers (Mecyclothorax and Blackburnia spp. [Coleoptera: Carabidae], Lycosa hawaiiensis Simon [Araneae: Lycosidae]), and pollinators (Hylaeus spp. [Hymenoptera: Colletidae]). These results were largely substantiated, and expanded upon, in studies conducted in the mid-2000’s (Krushelnycky and Gillespie 2008, 2010). These latter investigations found that native arthropod species richness declines by over 50% in ant-invaded areas of the park. In addition, the overall abundance of native arthropods drops dramatically, and communities become numerically dominated by introduced arthropods. Affected native species represent all trophic groups, and include species active on the ground, inhabiting the leaf litter, and even those resident on shrubs. Rare endemic predators are most vulnerable, but up to 65% of native species overall are drastically reduced in abundance in invaded areas. Many Hawaiian arthropod species have very limited distributions (Howarth 1990), and the expansive populations of Argentine ants could easily spread to occupy the entire natural ranges of some of HALE’s most localized and rare endemic species, placing them at high risk of extinction. Recently, three species of native carabid beetles have been rediscovered adjacent to the Argentine ant populations at HALE after going undetected for over 100 years (Krushelnycky et al. 2005b). None of these rediscovered carabid species have been found within ant-invaded areas, and most appear to have extremely localized distributions. If the ongoing spread of the two ant populations cannot be halted or reversed, we may be placed in the unfortunate position of watching these beetle species go extinct. Moreover, the invasion by Argentine ants of the majority of the subalpine shrubland and aeolian zone habitats at the top of Haleakala, as predicted, has the potential to eliminate other spectacular examples of Hawaiian evolution. For instance, a flightless moth that hops around the cinders (Thyrocopa apatela [Walsingham]), a 7 flightless lacewing with spiked, beetle-like forewings (Micromus cookeorum [Zimmerman]), and a silversword-feeding long-horned beetle (Plagithmysus terryi [Perkins]) are only found there. Argentine ant invasion has also led to community-level impacts at HALE. Arthropod trophic structure is strongly altered in invaded areas, as is overall arthropod biomass (Krushelnycky and Gillespie 2008). These alterations signal changes in energy flow and, potentially, changes in ecosystem function in invaded areas. One hypothesized functional impact is that Argentine ants may negatively affect reproduction of obligate outcrossing plant species, such as silverswords, by reducing numbers of native pollinators (Cole et al. 1992). For instance, seemingly moderate reductions in rates of silversword seed set (e.g. from 30% to 20%) are predicted to result in dramatic population declines over the long term (Forsyth 2002). More recent work indicates that Argentine ants are not currently depressing silversword seed set rates (P. Krushelnycky unpub. data). However, ants have only invaded a small portion of silversword habitat so far, and future expansion into much larger areas of silversword habitat could very well have a much stronger impact on the silversword’s pollinators and reproduction. In the early 1990’s park staff became concerned about the spread of the upper ant population into endangered Uau (or Hawaiian petrel, Pterodroma sandwichensis) nesting habitat. Invasive ants have been found to prey upon the nestlings of a number of ground-nesting birds (Holway et al. 2002), and the Argentine ant could therefore pose a significant new threat to this imperiled bird that constructs burrows in Haleakala’s cliffs. Monitoring of nesting colonies both within and outside the ant population by USGS and NPS, however, revealed that the Uau currently appears to be unaffected by ants, with similar reproductive success rates in both areas (Krushelnycky et al. 2001). In this case, cold temperatures within the petrel burrows likely discourage heavy ant foraging to the nest chambers. This situation may need to be re-examined as temperatures continue to increase on upper Haleakala, because petrel burrows may begin to become thermally accessible to ants. III. RESEARCH ON ARGENTINE ANT MANAGEMENT AT HALE A. Rationale for attempting ant control As the implications of the Argentine ant invasion became clear, research focus shifted towards developing a technique to control, and ideally eradicate, the ant in the park. Because of their budding modes of dispersal, populations of some invasive ant species like the Argentine ant can occur as discrete entities even when their overall distributions are wider. The lack of mating flights means that eradication of incipient populations could result in the permanent removal of these species from particular areas of concern, as long as re-introduction by humans can be prevented or quickly detected. In the 1980’s and 1990’s, for example, a 3 ha population of the invasive little fire ant (Wasmannia auropunctata) was eradicated from the Galapagos island of Santa Fe using an ant pesticide (Abedrabbo 1994). This success was an important impetus for the Argentine ant project at HALE. The two populations in the park are well isolated from other Argentine ant invaded sites on the island, and a similar technique could possibly result in substantial control, or even eradication, of Argentine ants at HALE. Numerous additional ant eradication campaigns, many with explicit conservation goals, have been conducted since the 8 early success on Santa Fe Island (Hoffmann et al. 2010). Most of these, however, have been much smaller in scale compared to the Argentine ant infestation at HALE. The use of pesticides may seem to contradict the conservation mandate of a national park. However, the control of invasive species often represents the most critical management actions required in Hawaii’s protected areas. Moreover, pesticides developed for ant control have several important improvements over previous wide-spectrum contact insecticides. These pesticides are developed as baits that are attractive to ants, and are typically formulated with only a small amount of active ingredient (toxicant). Ants locate the baits while foraging and bring them back to the nest to share among the colony, meaning that relatively small amounts of the baits need to be used to reach all of the target nests. And because the baits are often highly attractive to the ants and invasive ant populations often achieve very high densities, the target ants are much more effective at retrieving and consuming the baits than non-target arthropods. These attributes have the effect of making ant baits both more effective and less likely to result in severe non- target impacts; ant baits have represented by far the most commonly used technique when attempting ant eradication in natural areas (Hoffmann et al. 2010). Despite the improved safety of today’s ant baits, it is virtually impossible to avoid all non-target impacts when using pesticides. Even if the target ant species retrieves all of the bait applied, secondary poisoning may occur among organisms that consume dead or moribund ants. In addition, if multiple applications of the baits are needed, the likelihood that some non-target arthropods will directly ingest the baits increases greatly. At HALE, these potential non-target impacts have been deemed to be tolerable because the ecological impacts from Argentine ants are so high. The same species that may be impacted by ant control efforts are likely already severely impacted by the ants themselves. Moreover, as summarized above, Argentine ants currently only occupy a small fraction of the total area that they will almost certainly eventually invade in the park. Some temporary non-target impacts from pesticide use is preferable to the long term impacts resulting from unrestricted Argentine ant invasion into the park. B. General strategy Successful control of invasive ants involves the seemingly simple objective of delivering an effective bait to the infested area. There are therefore two components that are inter-related and impinge on each other: finding or developing an effective bait, and devising an application system that can deliver the bait to the target area. The difficulties in actually achieving control or eradication lie not only in finding successful solutions to these two challenges, but also stem from the reality that the best solutions to the two challenges can be in conflict. The situation at HALE exemplifies this issue. Argentine ants are highly attracted to liquid foods, especially plant nectar and the sugary exudates of plant sap-sucking insects (Hemipteran honeydew). Not coincidentally, much Argentine ant control research has focused on developing sugar water- based baits (formulated with a toxicant), because this food source is usually avidly consumed by workers, is easily carried back to the nest internally in the ant’s crop, and is readily shared with nestmates through trophallaxis (regurgitation). Liquid and gel/paste baits are therefore often more effective than drier granular baits against Argentine ants, simply because the baits are more attractive to the ants. However, liquid baits need to be delivered within bait stations, and gel or paste baits usually need to be delivered by hand. The size, topographic complexity and complete 9 inaccessibility of parts of the two Argentine ant populations at HALE raise serious questions about whether any liquid, gel or paste baits could realistically be applied to the entire infested area. In contrast, aerial broadcast systems exist for granular baits, and granular baits could in theory be delivered to all infested areas at HALE. More effort has therefore been directed at testing granular baits at HALE, even though granular baits are expected to usually be less attractive and hence less effective than liquid or gel baits. An effective ant bait must consist of a highly attractive matrix (consisting mostly of a food attractant) combined with a toxicant that is undetectable to foraging ants and which has delayed action, allowing foragers to bring the bait back to the nest and share it with nestmates, including queens and brood (immature stages), before they are killed. The ease of meeting these criteria varies among invasive ant species: some species are much easier to control and indeed eradicate than others. Unfortunately, Argentine ants have been found to be among the hardest of invasive ant species to control, with few cases of successful eradication. The reasons for this greater difficulty are unclear but could arise from a number of factors, possibly including a greater number of queens in each nest (some of which may receive food in reduced quantities or quality), a greater sensitivity and ability to respond to incipient mortality in the nest, and higher selectivity or variability in food gathering and consumption. The challenge at HALE has therefore been to find a bait that is effective enough in experimental test plots to justify attempting eradication on a larger scale, and that at the same time could be applied to large areas in a feasible manner. This will usually mean that a prospective bait can kill all ant nests within test plots with one to several applications (if ants cannot be eradicated from relatively small plots, attempting eradication from hundreds of infested hectares in the park would be pointless), and that it could be applied aerially. A secondary, lesser goal, has been to find a bait that causes enough mortality to prevent outward spread of the two populations when applied to their perimeters. This containment strategy could greatly increase the chances of success of the first goal in the future, even if an eradication strategy is not available at present. Most of the management research summarized below has been directed at these two goals. From 1994 to the present, various baits have been assessed for effectiveness in experimental plots. These have met with varying degrees of success, but none has been able to completely eradicate all ant nests within the plots. Additional baits have been screened less extensively to assess basic attractiveness. Beginning in 1997, a moderately effective granular bait (Maxforce Granular Insect Bait) was broadcast once annually within a 120m wide perimeter zone in an attempt to slow the outward expansion of the two ant populations, while new baits continued to be tested for their potential to eradicate the populations. This containment strategy also met with varying degrees of success, and was discontinued after the summer of 2004 when results were especially poor. As no easily broadcast (i.e. granular) baits appeared to be sufficiently effective, we began conducting research on liquid baits. The rationale for testing these baits was that it would be useful to know if liquid baits were more effective than granular baits at HALE, even if their application on a larger scale seemed unrealistic at present. In addition, effective liquid baits could be used to control ants in smaller sites that may be acting as high-risk source pools for initiating new populations in distant areas of the park. These include park buildings, greenhouses