Cooperative National Park Resources Studies Unit University of at Manoa Department of Botany 3 190 Maile Way Honolulu, Hawaii 96822 (808) 956-8218

Technical Report 86 (Vesp- mvlvanh) Biology and Abatement in the National Parks of Hmii

Parker Gambino and Lloyd Loope

Haleakala National Park P.O.Box 369 Makawto HI 96768

Cooperative Agreement CA8018-2-0001

February 1992 TABLE OF CONTENTS

ABSTRACT ... LIST OF TABLES ...... u ... LIST OF HGURES ...... u PREFACE ...... 1 SECTION 1. YELLOWJACKET NATURAL HISTORY ...... 2 INTRODUCTION ...... 2 MATERIALS & METHODS ...... -3 Population Monitoring ...... 3 Nest Sites ...... 5 Colony Analysis ...... 5 Queen Physiology & Behavior ...... 6 RESULTS ...... ,...... 6 Population Monitoring ...... 6 HALE (Haleakala National Park) ...... 6 HAVO (Hawaii Volcanoes National Park) ...... 7 Nest Sites ...... 8 HALE (Haleakala National Park) ...... 8 HAVO (Hawaii Volcanoes National Park) ...... 9 Colony Analysis ...... 9 HALE (Haleakala National Park) ...... 9 HAVO (Hawaii Volcanoes National Park) ...... 11 Queen Physiology & Behavior ...... 12 HALE (Haleakala National Park) ...... 12 HAVO (Hawaii Volcanoes National Park) ...... 12 DISCUSSION ...... 13 SECTION I1 . YELLOWJACKET PREDATION ...... 15 INTRODUCTION ...... 15 MATERIAL & METHODS ...... 16 Qualitative Studies ...... 16 Quantitative Studies...... 16 RESULTS & DISCUSSION ...... 17 Qualitative Studies...... 17 Quantitative Studies...... 19 SECTION n~. YELLOWJACKET ABATEMENT ...... 22 INTRODUCTION ...... 22 NEST TREATMENTS ...... 22 Materials & Methods ...... 22 Results & Discussion...... 23 FACTORS AFFECTING BAITING ...... 23 Materials & Methods ...... 23 Bait Attraction...... 23 Bait Acceptance ...... 23 Enhancement of Bait Acceptance ...... Repellency of Aged Bait ...... 24 Foraging Behavior in Ola'a Tract ...... -25 Results & Discussion...... 25 Bait Attraction...... 25 Bait Acceptance ...... 25 Enhancement of Bait Acceptance ...... 25 Repellency of Formulated Bait ...... 25 Attraction of heptyl butyrate in Ola'a Tract ...... 26 YELLOWJACKET ABATEMENT TRIALS ...... 27 Materials & Methods ...... 27 Results & Discussion...... -28 1988: HAVO: Kipuka PuaulUlKipuka Ki ...... 28 1988: HAVO: Kipuka Kulalio/Kipuka Maunaiu ...... 28 1988: HALE: HALEl/HALE2...... 29 1989: HAVO: Footprints/Route 11-Ka'u Desert ...... 29 1989: HALE: HALE1/HALE2 ...... 30 1990: HAVO: Narnakani Paio/Kipuka Puaul ...... 30 1990: HALE: ...... 30 ACKNOWLEDGEMENTS ...... 30 LITERATURE CITED ...... 31 Appendix 1 ...... 34 Appendix 2 ...... 35 LIST OF TABLES Table 1. Yellowjacket population monitoring programs on Maui and Hawaii...... 4 Table 2. Seasonal activity of workers at Haleakala National Park...... 7 Table 3. Elevation effects on Y.pemybmh forager populations at Haleakala National Park...... 7 Table 4. Nest Sites of Y. on mainland North America and Hawaii...... 8 Table 5. Nest sites of Y.pen&a&a at Hawaii Volcanoes National Park and vicinity...... 9 Table 6. Characteristics of colonies from Haleakala National Park ..... 10 Table 7. Characteristics of colonies from Hawaii...... 11 Table 8. Summary of prey taken by Y. pensvlvaruca in Hawaii...... 19 Table 9. Prey Intake at Thtee colonies at Puhirnau, HAVO, 24 November 1989...... -20 Table 10. Effects of yellowjacket predation on the standing crop of ohia foliage atthropods (potential prey) ...... -21 Table 11. Tests of Ficarn D treatments on nests of Y. -...... 24 Table 12. Tests of Whitmire PT5 15 Waspfreeze treatments on nests of Y. pensvlvahica ...... 24 Table 13. Numbers of Y. workers attracted to ten baits ...... 26 Table 14. Amounts of baits collected by Y. workers...... 26

"2 Table 15. Yellowjacket abatement trials, 1988- 1990...... 29

LIST OF FIGURES Figure 1. Worker seasonality at HALE. Figure 2. Worker seasonality at HAVO. Figure 3. Queen capture data. Figure 4. Prey trap. Figure 5. Foraging rates at three Puhimau colonies. Figure 6. Foraging success at three Puhimau colonies. Figure 7. Ohia abundance at Puhimau - weights. Figure 8. Ohia arthropod abundance at Puhimau - small and large size . Figure 9. Yellowjacket abatement: 1988: HAVO: Kipuka Puaulu/Kipuka Ki. Figure 10. Yellowjacket Abatement: 1988: HAVO: Kipuka Kulalio/Kipuka Maunaiu. Figure 11. Yellowjacket Abatement: 1988: HALE: HALEl/HALE2. Figure 12. Yellowjacket Abatement: 1989: HAVO: Footprints/Route 11-Ka'u Desert. Figure 13. Yellowjacket Abatement: 1989: HALE: HALEl/HALE2. Figure 14. Yellowjacket Abatement: 1990: HAVO: Namakani Paio/Kipuka Puaulu. Figure 15. Yellowjacket Abatement: 1990: HALE. Figure 16. Yellowjacket baiting program: distance effects.

iii ABSTRACT

The western yellowjacket Jkspula pa&anh of prey at ar near plant and soil surfaces. Populations (Saussure) has become widely established in the of highly precinctive endemic arthropods may in Hawaiian Islands, where it exhibits a high degree of some instances be unable to recover from such in- reproductive plasticity. Although most colonies in tense predation pressure. Although even local eradi- Hawaii Volcanoes and Haleakala National Park ad- cation of Y.pensvlvanica is not feasible, toxic baiting here to the basic annual cycle dominant throughout trials using Knoxout 2FM in canned chicken bait, this 's native range in western North America, with enhancement of bait acceptance with heptyl overwintered colonies were detected in some years. butyrate, were in many instances effective in drasti- Polygyny, achieved by adding queens to an estab- cally reducing yellowjacket forager populations. lished colony, is a likely prerequisite for successful Toxic baiting was most consistently effective in high- overwintering. The large size of overwintered colo- elevation shrubland and least effective in forest situ- nies and some annual colonies (with more than 300 ations. A protocol is presented for use by resource worker sorties per minute from the nest) results in managers in dealing with yellowjacket threats to very heavy predation on local arthropod biota. native biota and to visitors. Workers of lkquh pmi&ah take a wide variety The second category of yellowjacket problems PREFACE concerns their interactions with other organisms, mostly arthropods, and applies only in areas where The invasion of the Hawaiian Islands by non-na- are established as alien . Yel- tive species has been to the overall detriment of lowjackets are predators, feeding on a wide range of native Hawaiian ecosystems. The damage caused by arthropod taxa, with great potential for negative im- large mammals such as pigs and goats is on a scale pact on the native fauna. This is especially signifi- readily apparent to the human eye; perhaps for this cant in the Hawaiian Islands, which have a high reason, the effects of feral mammals, and ecosystem degree of endemism, and where arthropods that have recoveries after reductions in their populations, are evolved in the absence of social may lack the fairly well documented. In contrast, the effects of anti-predator defensive mechanisms selected for alien insects occur on a much finer scale, and are elsewhere. Native insects play important roles in the more easily overlooked. While our understanding of functioning of Hawaiian ecosystems, be it as polli- the roles of invertebrates in native Hawaiian ecosys- nators, predators keeping herbivores in check, detri- tems lags behind that for plants and larger , tivores, or food for vertebrates. Depletion of local native and alien insects are no less worthy of inves- populations of certain species through yellowjacket tigation predation may disrupt the overall system in unpre- The western yellowjacket dictable ways. Thus, yellowjacket abatement in the (Saussure) was first recorded from the Hawaiian Hawaiian National Parks is an important component Islands in 19 19. This wasp species is troublesome of the National Park Service's overall strategy of for several reasons. Fitst, it interferes with human enhancing biodiversity through the preservation of + activites. The sting, which has evolved as an anti- natural ecosystems. vertebrate defense mechanism, is painful. For the This report has several objectives: 1) to docu- small percentage of humans allergic to yellowjacket ment the invasion of Maui and Hawaii by VP, and venom, a sting can cause severe medical problems identify factors influencing its successful estab- such as anaphylactic shock Estimates of the number lishment; 2) to describe VP's impact on the native of sting-related fatalities in the United States range Hawaiian arthropod fauna, to provide justification from several dozen to several hundred per year, but for a VP pest management program; and, 3) to furnish only one, an O'ahu construction worker in the rnid- information on practical and logistical successes and 1980s (0.Komatsu, pers. cornrn.), has been recorded problems encountered in VP abatement research, to from Hawai'i. Most people are familiar with the be taken into consideration by park resource manag- yellowjacket's stinging ability, and dense popula- ers in designing and implementing overall pest man- tions can interfere with people's outdoor activities agement activities for the parks. The report is through intimidation. Foraging yellowjackets are divided into three sections, each summarizing inves- attracted to many of the foods that people eat, and by tigations of different aspects of Y. pensvlvanica bi- perfumes and other fragrances. The higher the yel- ology in the Hawaiian Islands: I. Yellowjacket lowjacket population, the more likely they are to natural history, including signifcant adaptations to sting or otherwise disturb people. However, even at the Hawaiian environment; 11. Yellowjacket preda- low population levels, there can be unfavorable en- tion on arthropods; and 111. Yellowjacket abatement. counters with if a nest is accidentally dis- turbed. Yellowjackets are sensitive to vibrations, and subterranean nests are not easy to detect and avoid. Recreational, research, or resource management ac- tivities may disturb nests sufficiently to induce sting- ing behavior; unfortunately, many nests are discovered this way. nesting activities and colony cycle. This section SECTION I presents a generalized account of these, based on YELLOWJACKET NATURAL HISTORY typical behavior in the native range as outlined in the references listed in the previous paragraph and from INTRODUCTION unpublished observations from California, and pro- Yellowjackets (kiqula v&ids b.],\lesDula vides a basis for analysis and comparison of Y. [Saussure]) are among the non-native -'s biology in Hawai'i. insects introduced into the Hawaiian Islands in his- 'Qpical colonies undergo an annual cycle char- torical times. Both species are native to western acteristic of the subfamily Vespinae. Inseminated North America, the likely source of founders of the queens hibernate in protected places during the win- Hawaiian populations (Miller 1961). Yes,pda Y& ter and begin flying in the spring at the onset of fair garis was recorded only from the island of Maui weather. A colony is founded by a single queen during the 1970's; populations were not widespread, (haplometrosis). She chooses a sheltered location to and the lack of collection or other records after 1980 initiate the nest, usually underground, but sometimes suggest a temporary infestation and failure to become in hollow spaces in buildings or tree trunks. Exposed established. In contrast, Y. has been aerial nests are rare. The queen engages in many present on the islands since the early twentieth cen- tasks during the early stages of colony development; tury, and may now be considered to be permanently she collects plant fibers to make paper used to build established. The first state record of Y.pensvlvanica the nest, lays eggs, collects food to feed herself and was on in 1919 (Nakahara 1980); there have the developing larvae, defends the nest against in- been continuous, but sporadic records from that is- truders if necessary, and may actively warm the a land up until the present. In 1936 it was recorded immatures during cold weather. The period from from Oahu, and the islands of Maui, Lana'i, nest initiation to the emergence of the fvst clutch of Moloka'i, and Hawai'i were invaded in the late workers is a time of great vulnerability for the colony. 1970's (Nakahara 1980). The persistence of popula- Once the first workers are reared and begin for- tions on Hawai'i and Maui through 1990 are clearly aging, they assume most of the work duties of the documented (Gambino 1991); informal oral comrnu- colony, and the queen does little more than feed, lay nications suggest that Y. pen&a&a is also still eggs, and regulate worker behavior through the ac- present on Oahu, Lana'i, Moloka'i, and Kahoolawe; tions of her pheromones. The colony enters a period there are no known reports from Niihau. of rapid growth. The nest, built of wasp paper, The Thomson can be subdivided consists of a series of horizontally oriented combs into a number of species groups; Y. pensvlvanica surrounded by an envelope with one or several small belongs to the group, with whose entrance holes. It expands downwards and laterally members it shares many biological characteristics as new combs are first added at the bottom and then (Carpenter 1987). The kquh life history is quite enlarged by the construction of new cells at the complex, as might be expected of a highly social periphery. Nests built in an easily excavated sub- species. General accounts of yellowjacket biology strate have a general spherical or ovoid shape, but are provided by Duncan (1939), Spradbery (1973), when the substrate cannot be removed by workers, Akre ct aL (1976), Roush & Akre (1978), Edwards the nest conforms to the shape of the cavity they are (1980), and Akre & MacDonald (1986). Numerous able to create, which may be highly irregular. Small aspects of y. biology, though worthy plant roots can be chewed through and removed, but of additional research, are not pertinent to the present larger roots are incorporated into the nest architecture study, and are omitted from consideration here. and may strengthen the nest by providing support. The development of an effective Y.pen&mh Genetically, females arise from fertilized (dip- abatement strategy in Hawai'i requires a good under- loid) eggs, while males develop from unfertilized standing of relevant aspects of its biology, such as (haploid) eggs. Early in the colony cycle, only small cells suitable for rearing workers (which are sterile produce some male offspring, but will eventually die females) are built. Later in the season large cells are out when the workers are lost through attrition. constructed; these are used to rear new queens. The difference in cell size is the only known basis for the MATERIALS & METHODS differentiation of female immature into workers or Similar methods were used to investigate Y. queens. The nonsverlap of sizes of females reared biology at Haleakala (HALE) and in small and large cells enables them to be easily Hawai'i Volcanoes (HAVO) National Parks. Four identified as workers or queens, respectively. Males basic aspects were studied: 1) fluctuations of worker can be reared in either size cell, yielding a bimodal populations; 2) nest sites; 3) colony analysis (struc- size distribution; however, in typical queenright ture and occupants); 4) queen physiology and behav- colonies most males are reared in small cells, and ior. used most large cells are for rearing queens. Newly Population Monitoring emerged queens normally do not participate coopera- Worker adults are the dhct cause of nearly the tively in colony activities; they feed while maturing Y. in the nest, mate, and disperse to seek out hibernation entire negative impact of pensvlvanica, both the sites. Males also perform little or no work to benefit stinging of people and the predation of native arthro- the colony; their function is to mate with new queens. pods. A monitoring system to quantify worker popu- lations is an important tool for identifying temporal When they leave the nest they engage in stereotypical "patrolling" flights, thought to be a component of and geographical variations, and to assess the effec- mating behavior. tiveness of control measures. This section describes The colony passes through a decline phase at the the use of monitoring to elucidate population trends; its use in the context of evaluation of abatement end of its cycle. Once colony resources have been channelled into reproductives, fewer workers are efforts will be discussed in Section 111. Heptyl produced, and the cohesive social organization char- butyrate, a synthetic chemical to which Y. pensvl- workers are attracted (Davis d aL 1%9), was acteristic of the early phases of the colony cycle used in traps deployed to monitor worker popula- breaks down. Many aspects of behavior, such as brood care, nest maintenance, foraging, and defense tions. The basis for Y. attraction to become erratic. As a result, the nest physically dete- heptyl butyrate is unknown; it is neither a food sub- riorates and the population drops. New queens and stance nor a pheromone naturally produd by the also males disperse, and colonies that are able to linger insects. Y. queens are attracted to through the fall are usually eventually shut down heptyl butyrate, and separate records of queen cap- completely by cold weather. tures were kept; these will be discussed in the section Although a successful colony may rear thou- on reproductive plasticity. sands of new queens, the chances of any individual Yellowjacket monitoring programs were con- (HALE) queen successfully producing queen offspring the ducted at Haleakala National Park on Maui next year are very small. Few queens are able to from 1981-1990, and Hawai'i Volcanoes National survive to establish colonies, and the majority of Park on Hawai'i (HAVO) from 1984-1990 (Table 1). colonies fail before rearing any queens. Premature Several sponsoring agencies operated these pro- colony decline may result from the death of the grams, using different techniques; consequently, queen, or from a queen unable to exert effective quantitative comparisons of population levels be- Nonetheless, some pheromonal control of her workers. In these cases, tween systems are not valid. qualitative population trends are evident regardless the ovaries of some workers, which are normally traps vestigial, may develop and produce viable eggs. of the monitoring system Two types of were used: a wet trap containing vegetable oil, water, and Since workers do not mate, eggs laid by them invari- d aL ably develop into males. A colony may continue to a wick of attractant (Oarnbino 1990), and a dry be active for some time without its queen, and may -- Table 1. Yellowjacket population monitoring programs on Maui and Hawaii.

Location Elevation (m) Dates Agency of Trap No. of Tkaps MAUI Haleakala Nat. June 1980- NPS Resources Wet 10 Park December 1988 Management Division HAWAII Kipuka Kulalio May 1984- Hawaii Dept. of Wet 1 December 1986 Health Kipuka Kulalio May 1988- NPS Research Dry 5 December 1989 Division Kipuka June 1988- NPS Research Dry 5 Mamiu November 1989 Division Kipuka Ki May 1984- Hawaii Dept. of Wet 1 December 1986 Health Kipuka Ki June 1988- NPS Research Dry 5 November 1989 Division Kipuka hulu May 1984- Hawaii Dept. of Wet 1 December 1986 Health Kipuka huh June 1988- NPS Research Dry 5 December 1989 Division Kilauea May 1984- Hawaii Dept. of Wet 1 (Park HQ) December 1986 Health Kilauea December 1988- NPS Research Dry 5 (Park HQ) December 1990 Division Kilauea December 1988- NPS Research Dry 5 (R-h) December 1990 Division Ola 'a August 1988- NPS Research Dry 5 October 1989 Division Ka'u Desert October 1988- NPS Research Dry 5 December 1989 Division Kipuka Nene May 1984- Hawaii Dept. of Wet 1 December 1986 Health trap (Yellowjacket Inn) containing just the attractant GLMprocedure of SAS Institute, Inc. (1985). To test wick for the effect of elevation on forager density, the total The main study site at HALE consisted of ap- annual number of yellowjackets captured at each trap proximately 1000 hectares on the northwest slope of was analyzed in a randomized block analysis of Haleakala Volcano, ranging from 2050-3050 m ele- variance (ANOVA) with elevation as the fmed vati- vation (Gambino d aL 1990). From 1981 to 1988, able; means were separated using Duncan's multiple ten traps were maintained by the HALE Resources range test. Management Division. To discern seasonal patterns At HAVO, a much larger and more diverse area in foraging worker populations at HALE,the number was sampled, ranging from 900 to 2165 m elevation. of workers captuned per week, pooled for all traps, Monitoring was not continuous at all sites, and was was calculated for each month from 1982- 1988 and suspended when the availability of staff time was analyzed in a randomized block ANOVA with month limited or when wasp populations approached zero as the fmed variable; means were separated using for extended periods during the winter or spring Duncan's multiple range test, calculated using the 4 months. HAVO monitoring data were not analyzed colonies. Thirty seven nests were excavated and statistically. brought to the laboratory for analysis. Data gathered from excavated nests included: date of excavation, Nest Sites general shape of the nest, number of combs, esti- A total of 101 active nests were discovered from mated total number of cells, number and location of 1986- 1990: 41 at HALE and 74 in or close to HAVO. large (queen) cells, approximate brood composition, * For each nest, a brief description of the nest site was and population of adult reproductives. We attempted recorded. Unless the nest was excavated, only the to extract as much information as possible from each surface features surrounding the opening(s) of the nest; however, given the time constraints of the pro- entrance tunnel could be described. For excavated ject, it was not practical to collect complete data from nests, certain subterranean features associated with all nests. Thus, the various data presented in the the nest were discovered only after digging. results section are from subsets of the total number At HALE there seemed to be an association with of nests. the native shtub Weliatarneimeiae (Chamisso & Comb areas were computed by tracing outlines Schlechtendal) Mueller, the dominant ground cover on heavy paper, weighing the cut out shapes, and species. To test if nests were preferentially associ- applying a conversion factor derived from the ated withS. tameiameiae,distribution of 34 nests was weights of paper pieces of known area. For some compared with a vegetation ground cover analysis. nests, direct cell counts (number of cells in 10 & One hundred points were sampled in a grid (2700 m were made for representative setions of combs con- by 200 m) straddling the 2620 m contour line, repre- taining small cells and, if present, large cells. In nests sentative of habitat where yellowjacket foragers and where cells were measured on only some combs, the nests were most abundant. At each sampling point a t 2 mean cells/cm2 value of measured cells was used to PVC-pipe square enclosing 400 cm was randomly calculate numbers of cells on combs whose cells thrown into the air, and the spot where it landed was were not measured. For colonies where no cells were - * examined. The substrate within the frame was as- measured, the mean cellslcm2 value of measured signed to one of two categories: &,pUa, if live cells for all nests from that island and year were used foliage of the plant was found anywhere within the to calculate cell numbers. For some nests the paper square; other, if it lacked SQ&dh foliage. Because of some combs had deteriorated prior to excavation; the category included some samples where in these cases comb numbers and areas were esti- S. meimeiae foliage did not fill the square, this mated from the distribution of meconial debris. Cell procedure likely overestimated the proportion of S. sizes could not be determined for these combs, and tameiameiae ground cover, thus providing a conser- estimates of numbers of cells were calculated using vative test of S. tameiameiae nesting association, the small cell conversion factor, a procedure that analyzed by a chi-square test. slightly underestimated numbers of large cells and Nests at HAVO occurred in a number of different overestimated numbers of small cells on combs con- habitats; no obvious nest site trends were noticed taining large cells. when all nests were pooled. To clarify trends in Nests types were classified according to the different areas, nests were grouped into the following following criteria: typical immature (TI) colonies geographical categories (represented by at least 5 were colonies collected before 1 October, prior to the colonies each): Kipuka Puaulu, Puhimau, Ola'a initiation of large cell construction; queen cell (QC) Tract, Keauhou, Namakani Paio, Ka'u Desert, and colonies were annual colonies that underwent a sin- Kilauea. gle episode of queen rearing, deduced from the pres- Colony Analysis ence in the nest of large cells containing meconla, no large cell (NLC)colonies were annual colonies that The mat complete picture of yellowjacket biol- remained active after 1 October, but had not initiated ogy can be obtained by making observations at active large cell construction; no large cell, queens (NLCQ) colonies were the same as NLC's except that there calendar year) was analyzed using the same methods was evidence of polygyny; overwintered (OW) colo- as for workers collected in the monitoring program. nies were colonies that underwent at least one epi- At HAVO, the trapping regime was more irregu- sode of queen rearing, and persisted at least into a lar than at HALE. To reduce the sampling bias, trap second summer. captures of queens were included for analysis only if the particular site was sampled for an entire calendar Queen Physiology and Behavior year. To supplement the data on trap captures of When present, subsets of pupae and mature queens at HMO, flying queens not associated with larvae from excavated nests were sexed. Male pupae any known nest were captured by net. Whereas the can be distinguished by their long antennae; male trap monitoring program consisted of a standardized larvae can be determined by the presence of testes, sampling regime, aerial queens were captured oppor- visible through the dorsal abdominal integument tunistically throughout 1989 in the course of frequent (Marchal 1896). In some cases adult queens and but irregular field trips taken while conducting this large workers were collected from nests; these were and other studies (Garnbino 1990) in the vicinity of stored frozen until they could be dissected under HAVO. A subset of netted queens was dissected to Heinz's diluent solution and examined for wing wear, evaluate their physiological status. Queens were gastral discoloration, size of largest egg, fat body designated as typical spring queens if they were development, and presence of sperm in the sper- inseminated and had greatly reduced fat body depos- matheca. A total of 143 queens from 23 nests were its. Ovarian status was evaluated, but not considered examined. Queens were designated "hibernation- in assigning them to the typical spring queen cate- ready" if they contained extensively developed fat gory, because for queens coming out of hibernation, body and undeveloped ovaries; "oviposition-ready" ovaries and eggs nodydevelop gradually over if they were inseminated, had reduced fat body, and several months during the colony initiation and pre- had developed ovaries with an egg at least 1.5 mrn emergence phases. Again, the status of some queens long; "mothers" if they were the only oviposition- was ambiguous, particularly if they were uninsemi- ready female in the colony or were clearly the most nated but lacked extensive fat body; most likely these worn (wing wear and gastral pigment loss) of a were queens that had not mated prior to hibernating, polygynous assemblage. Dissections sometimes and are more closely aligned with typical spring gave inconclusive or intermediate results, so that queens, though their ability to successfully initiate some queens could not be placed in any of these colonies is questionable. At HAVO, queens col- categories. lected by both methods were pooled for analysis of Extranidal behavior of queens falls into two ba- temporal patterns. sic categories. Hibernation-ready queens disperse from their natal nest to mate or to seek hibernation RESULTS sites. Collections of these queens outside of nests is The study sites (HALE and HAVO) differ in a rare because they neither engage in conspicuous number of physical characteristics, and gave consid- flight activities nor are attracted to heptyl butyrate. erably varied results. In this subsection, results from In contrast, post-hibernation queens (referred to here the two parks are presented separately. as spring queens, regardless of the date of capture) are attracted to heptyl butyrate and were collected in Population Monitoring monitoring traps; they may have an extended flight HALE: Fluctuations in forager populations period associated with the various tasks involved in above 2073 m, measured as pooled numbers of fora- establishing (or joining) a nest. Temporal occurrence gers captured weekly in all traps, are presented in of queens trapped at HALE (during 1982-1988, the Figure 1. Males were only rarely captured in moni- years for which data are available for the entire toring traps. The strong seasonal character of the cycles, with highest worker populations occurring Table 2. Seasonal activity of workers at Haleakala National Park, 1982-1987.

Month No. workers trapped per wk Jan. 1.60 a Feb. 2.07 a Mar. 3.63 a Apr. 3.43 a May 6.81 a June 11.99 ab July 23.17 abc Aug. 41.54 cd Sept. 54.59 d Oct. 32.97 bcd Nov. 9.86 a Dec. 3.27 a Means within columns followed by the same letter are not significantly different (P = 0.05; Duncan's multiple range test).

1 Table 3. Elevation effects on Y. forager populations at Haleakala National Park Val- ues are mean annual numbers of foragers captured per heptyl butyrate trap, 1982-1987

Elevation (m) Mean @ SD) I 2073 79.88 + 133.32 a I 2134 198.50 + 235.49 a 2256 313.88 + 423.04 ab 2347 955.75 2 8 10.79 ab 2438 63 1.25 + 624.27 ab 2591 2632.75 + 1614.39 c 2682 1210.75 + 1150.06 b 2835 973.00 + 948.08 ab 2957 353.75 + 527.59 ab 3018 218.50 + 304.72 a Means (+ SD) within a column followed by the same letter are not significantly different (P - 0.05; Duncan's multiple range test) from August through October, and low populations characteristic temperature inversion that occurs on occurring during the winter and early spring (F - East Maui (Blumenstock & Price, 1972). It generates 5.34; df - 1l,72; p 0.0001) (Table 2) corresponds a cloud layer that frequently produces dense ground well with typical mainland activity cycles. However, fog at elevations from 1350 to 2000 m, sometimes forager activity did not drop to zero in all years, extending up to 2300 m. The amount of insolation revealing the presence of overwintering colonies may be an important physical factor affecting Y. from 198 1-82,1982-83,1983- 84, and 1985-86. Yel- ga&mica distribution lowjacket distributions across elevations were non- HAVO.. As was true on Mad, worker popula- uniform (F - 7.18; df - 9,70; P 0.01), with a distinct tions were generally highest during the late summer peak at 2591 m (Table 3). Exclusion of yellowjack- and autumn. Colonies overwintered during some ets ftom the lower elevations of the study area may years, with evidence of this phenomenon in 1983-84 be due to weather conditions associated with the (Kipuka Kulalio, K. Ki, K. Puaulu), 1985-86 (K. 7 Table 4. Nest Sites of X on mainland North America and Hawaii. Figures indicate the per- centage of N nests in each site type.

Region* HALE HAVO N 34. 74 Overwinter 17.6 4.1 Subterranean 100 77.0 Soil Cavity 73.5 41.8 0 23.0 Under Rock 26.5 12.2 Under Building 0 0 On Ground 0 17.6 J-Wstump 0 17.6 D~@Sand/Sod Pile 0 0 Aerial 0 5.4 Building Walls 0 1.4 Tree Cavity 0 4.0 Tree/Exposed 0 0

*Key to regions and sources: OR: Oregon forest, Roush & Akre 1978 WA: Washington State, MacDonald et al. 1974 CA 1: Coastal central California, Garnbino unpublished CA2: Inland central California, Garnbino unpublished HALE: Haleakala National Park, Maui HAVO: Hawaii Volcanoes National Park & Vicinity, Hawaii Kulalio), and 1988-89 (K. Puaulu, Ka'u Desert) (Fig- lations of Y. -. In all populations, most ure 2). Continuous observations of wasp activity, or nests occured in subterranean situations. The inci- analysis of colony architecture, also gave evidence dence of nests at or above ground level generally of overwintering at K. Kulalio from 1987-88, and at increased with the availability of cavities in these Keauhou from 1989-1990. locations, mostly as hollow tree stumps (Oregon, Populations on Hawai'i were not tightly syn- HAVO) or voids in buildings (coastal California). chronized, and phenologies varied according to loca- HALE: Cavities at or above ground level are tion. At a number of sites there were extended extremely rare, accounting for the totally subterra- periods (3 months) of irregular population fluctua- nean nature of nests. Of 100 grid samples, 3 1 tion without cleatly defined peaks. Although the (31.0%) contained S. -, while 19 of 34 large variance inherent in using a single monitoring (55.9%) nests were located beneath S. tameiameiae trap per site @OH program: K. Kulalio 1986, K. bushes. Thus, there is a positive association with this Nene 1985 and 1986, K. Ki 1984 and 1986) may plant (chi-square - 9.84; df - 1; P 0.005).. However,. account for some of these irregularities, they were it was not determined whether S. tamelamelae is a also apparent when five traps per site were used (K. grefened site for queens initiating nests, or if nests Puaulu 1989, K. Nene 1989, Ka'u Desert 1989). are initiated more randomly, with those beneath S. tameiameiae becoming more successful and more Nest Sites likely to be detected. Accumulated organic matter P Table 4 provides a comparison of nest site data from the plant itself is easily excavated, and likely from mainland North American and Hawaiian popu- provides insulation from wetting and harsh weather. Table 5. Nest sites of Y. pa&aim at Hawaii Volcanoes National Park and vicinity.

sitea PUA PMA OLA KIL KHO NPA MU OTH N 2 1 10 7 7 7 5 8 9 Subterranean Soil Cavity 4 Rootsbg 3 Under Rock 1 On Ground, 0 WdStUmp aerial Building Walls 0 Tree Cavity 0

aKey to Sites: PUA: Kipuka Puaulu (Bird Park) PMA: Puhimau OM: Ola'a, Large Tract (HAVO) and State Forest Reserve KIL: Kilauea =cinity KHO: Keauhou Ranch NPA: Namakani Paio KAU: Ka'u Desert OTH: Other Furthermore, incorporation of S. tameiameiae roots spherical in shape, with numbers of small cells (x - into the nest architecture may enhance structural 3497; range - 558-8086) similar to the ranges re- stability. corded in Washington (MacDonald et al 1974) and I-IAVO: Among the various HAVO habitats there Oregon (Roush & Akre 1978). None of these colo- was considerable diversity in nesting sites (Table 5). nies had initiated large cell construction, which The Ka'u Desert, with terrain most similar to the would not be unusual for colonies collected during HALE study site, also had exclusively subterranean the summer (TI'S, n = 5), but atypical for colonies nesk, the majority were under large rocks. At the that remained active into the autumn (NLC's, n - 7; other end of the spectrum, the Ola'a rain forest, with NLCQ's, n = 2). a moist and densely packed soil substrate, had only Nests found in 1986 (two of these were exca- one of seven nests beneath the ground. The abundant vated during the spring of 1987, long after activity hollow interiors of old (3ibotium tree ferns provided had ceased during the previous winter) differed strik- cavities that were the most common nest site. In ingly from those of subsequent years. The number vegetationally complex, partly forested Kipuka of small cells per nest was approximately 20 times F'uaulu, most nests were associated with old logs or greater (n - 3; x - 65737; range - 58289-80508), and stumps, whether below, at, or above ground level. large (queen) cells were also constructed. In nest HAL8603 there were two distinct episodes of large Colony Analysis cell construction, separated by 3 combs containing HALE: A total of 17 colonies were excavated small cells exclusively. In nest HAL8606 only one (Table 6). Nests found in 1987 (n - 3) and 1988 (n episode of large cell construction was detected. The - 11) resembled typical prereproductive annual nests bottom two thirds of both of these nests were too of mainland Y. populations, and are deteriorated to provide an accurate record of cell considered together here. They were basically construction during late 1986 and early 1987. 9 Table 6. Characteristics of colonies from Haleakala National Park.

Date of No. of cells Queens Presentb Nest Excavation ma small large Mother Ovip Hiber HAL8601 21-Sep-86 ow 58414 2107 HAL8603 30-May-87 ow 58289 936 HAL8606 01-Jw-87 ow 80508 1503 HAL870 1 30-Aug-87 TI 4150 0 HAL8706 3 1-Aug-87 TI 1792 0 HAL8708 3 1-Aug-87 TI 2442 0 HAL8801 10-Sep-88 TI 558 0 HAL8802 10-Sep-88 TI 832 0 HAL8812 05-Oct-88 NLC 2423 0 HAL8813 05-Oct-88 NLC 4479 0 HAL8804 06-Oct-88 NLCQ 4452 0 HAL8810 06-Oct-88 NLC 4699 0 HAL8811 06-Oct-88 NLC 7644 0 HAL8807 14-NOV-88 NLC 2243 0 HAL8808 14-NOV-88 NLC 2776 0 HAL8814 14-NOV-88 NLC 2382 0 HAL88 16 14-NOV-88 NLCQ 8086 0

BNest lypes: TI - ?).pica1 immature QC-Queencell NLC - No large cells NLCQ - No large cells, queens OW - Overwintered

b~uem~yp~ (see text for descriptions) Ovip - Ovipition-ready Hiber - Hibemation-ready

Nest HAL8601 was retrieved in excellent condi- they were extended. Combs initiated separately tion, and its architecture received more detailed were sometimes fused together, as revealed by ir- study. Prior to excavation this colony showed many regularities of cell pattern along the fused margins. symptoms of decline. Dead larvae were being re- Some combs were even continuous, along a circui- moved from the nest, the colony contained an abun- tbus route, with the combs below them. The irregu- dant male population, and colony defense was not larities of the lower comb structure were vigorous. The nest was bell-shaped with a globular compounded by the substrate; by expanding down- top turret (diameter = 20 cm) that fanned out in the ward the nest had contacted irregularly surfaced bed- lower portions to a diameter of 50 cm. The top combs rock, and further downward excavation by the wasps (down to the ninth) were discrete units. Starting with was impossible. In all three queencell combs (W- comb # 10 the comb structure became more irregular; 9), large cells were clustered in the center, sur- lower combs were apparently initiated at several rounded by small cells. This arrangement, similar to distinct loci and engulfed obstacles such as rocks as that found in overwintered Y.gammb (E) nests in Table 7. Characteristics of colonies from Hawaii.

Date of NO. of cells ~uans-ntb Nest Excavation small large Ovip Hiber HV08810 17-Aug-88 15266 0 HVO88 19 18-Aug-88 9959 0 HV08815 15-0ct-88 5418 0 HVO88 16 26-0ct-88 43221 234 1 HV08828 27-Oct-88 15769 73 HV08823 10-NOV-88 8227 356 HV08826 10-NOV-88 3 1246 0 OLA8801 16-Dec-88 22720 806 + + HVO88 17 18-Jan-89 6282 829 + + HVO88 14 19-Jan-89 22967 5060 + + HVO8822 19-Jan-89 10155 1821 + - HVO883 1 24-Jan-89 4344 668 + - HV08902 16-Oct-89 835 0 + + KAU8901 25-Oct-89 4309 0 HV08908 20-NOV-89 48% 0 + HV08907 25-NOV-89 14857 645 + + HV08906 11 -Jan-90 2782 1 0 + - HV08904 17-Jan-90 26263 0 HV09001 26-Jan-90 14999 527 1 + - HV08903 10-Feb-90 546260 47229 + +

aNest I)pes: TI - 'Qpical immature QC - Queen cell NLC - No large cells NLCQ - No large cells, queens OW - Overwintered

b~ueenQpes (see text for full descriptions) Ovip - Oviposition-ready Hiber - Hiberna t ion-ready

New Zealand (Thomas 1960), contrasts with the located and most colonies were more modest in size. more typical peripheral location of large cells on In 1988, none of the excavated colonies (n - 11) had rnixed-cell combs recorded elsewhere for annual Y. initiated large cell construction, even though 7 of 9 colonies (MacDonald et a1 1974; taken after September were queenright Roush & Akte 1978). IJAVO: Of 20 excavated nests, only 11 had The synchrony of forager populations at HALE, structures that might be considered typical (2 TI'S, 9 noted eatlier, is matched by the apparent synchrony QC's) (Table 7). A common feature of the atypical of nest development. In 1986, only six colonies were nests was the failure to construct large cells before found; all had apparently overwintered from 1985. October (4 NLC's, 3 NLCQ's). Five annual nests In subsequent years, no overwintering colonies were contained more than 20,000 small cells; of 88 nests recorded from Washington (MacDonald d g& 1974) had a separate entrance tunnel, and contained 1516 or Oregon (Roush & Akre 1978),the greatest number large cells in 5 combs. of small cells was 7397. Two overwintered nests were excavated. HVO Queen Physiology and Behavior 8816 was poisoned with insecticide on 1 September HALE There was a seasonal pattern in the 1988 and dug up approximately two months later. Its appearance of spring queens (Figure 3A), with most architecture was irregular due to the shallowness of appearing from May through July (F = 3.96; df = the soil; expansion along the contours of the rock 1l,72; p 0.0002). However, 32 (14.0%) of 228 surface had produced uneven lateral sections. This, queens were captured after 1 August, and in 1984 and the partial deterioration of some combs, made it queens were captured as late as October. Three impsible to determine which had been most re- foraging queens captured by net in late August 1987 cently active. All large cells were present on four had been inseminated and had moderately developed mixed large cell/small cell combs. ovaries. A fourth queen, also inseminated, was cap- Colony HV08903 was discovered in September tured on 1 September 1987 as she returned (without 1989; its peak traffic rate was estimated to be 1200 hesitation) to colony 8707. All four of these queens per minute. The nest, which had several irregularly had frayed wings, but lacked the dorsal gastral mark- shaped entrances, was within and beneath a large ings that would indicate repeated oviposition (Ross downed tree tnurk; to facilitate observations, the ref). entrances were closed or modified with structures of Regarding excavated colonies, at four no queen soil, cardboard, and wood. Males were often found was recovered; three of these showed symptoms of in the vicinity of the modified nest entrances. The queen loss (irregular pattern of brood in combs, high .. .. colony was not excavated until its t&c rate de- percentage of male brood). A single mother queen clined to zero on 10 February 1990. was recovered from eight colonies, including 5 The nest had been initiated within the hollow tree NLC's. Two colonies (NLCQ's) had a single insemi- @ tnmk,ffled the cavity within, and expanded beyond nated mother queen and smaller queens (nest into the soil in irregular lobes. It consisted of ap- HAL8804, n - 2; nest HA.8816, n = 1) that lacked proximately 35 comb layers containing about external signs of wear, were inseminated, and had 546,000 small cells and 43,000 large cells. The well developed ovaries (size of their largest eggs: in topmost comb contained 138 small cells, the second nest HAL8804,2.0 and 1.6 rnrn; in nest HAL8816, comb contained 459 small cells and 145 large cells, 2.2 rnrn). and the third comb contained 229 small cells and 48 HAVO: Data for trap captures of 160 queens large cells. If these were the first combs built they pooled from 1985,1986, and 1989 are presented in represent an extremely early switch to large cell Figure 3B. Most captures occurred from April to construction in the life of this colony. Combs #4-7 June, but there were a fair number of outliers, includ- contained small cells exclusively, followed by a se- ing some in November and December. Seventy-five ries of 12 combs containing cells of both sizes. Many queens were captured by net during 1989 (Figure lower combs contained both small and large cells. 3C), with a peak in March. Twenty-four netcap- Most large cells were spread in vertically aligned tured queens were dissected, and all but two were peripheral portions of up to eight combs. Many large typical spring queens; the exceptions were uninsemi- cells were capped; the contents of 10 large cells from nated, but still lacked the fat body deposits indicative each of 8 combs were examined, and contained 77 of queens seeking hibernation sites. queens and 3 males; all were dead and partly dessi- Of the 7 QC colonies in which queens (at least cated. 10 per colony) were examined, hibernation-ready A small %udn nest was constructed in a cavity queens were detected at only 4, whereas oviposition- adjacent to, but not connected to, the main nest. It ready queens were present in all 7. This contrasts sharply with the typical situation, where autumn queen production yields hibernation-ready queens expressed unless certain constraints are relaxed. On exclusively. Multiple oviposition-ready queens the North American mainland, most Y.gm&mh were found in all three NLCQ colonies. Queens in colonies follow the strict annual colony cycle pattern. NLCQ colonies, as well as some of the oviposition- Although peak numbers of foragers may fluctuate ready queens from QC colonies and the auxiliary widely between years for reasons that are not well queens of the polygynous HALE colonies, appeared understood, the degree of variation among colonies a smaller than typical queens. within a population is somewhat restricted. Thus, At the large overwintered colony HV08903 while it is diicult to predict the magnitude of the queens were fmt observed outside the nest on 26 forager population in any given year, the peak will January 1990. Their behavior was indistinguishable occur at a fairly predictable time, and appropriate from foraging workers; i.e., they flew in and out preventative or abatement measures can be sched- without hesitation (at a rate of about 0.5 per minute). uled accordingly. Three returning queens were captured, and all were The portion of V. mfls~Ivanica'smainland range oviposition- ready. During 30 minutes of observa- where reproductive plasticity is most fully expressed tion, 12 clumps of wasps were noticed in the vicinity is coastal California (Y. colonies also over- of one of the modified nest entrances. Clumps, con- winter in the region), whose Mediterranean climate sisting of 3- 10males clustered around a single queen, is characterized by nearly frost-free winters. Consid- seemed to be related to mating behaviour (Ono et aL ering the distribution of overwintered Yespula colo- 1985), although no actual copulations were ob- nies worldwide (Duncan 1939; Tissot & Robinson served. There were fewer than 150 live workers 1954; Thomas 1960, Vuillaume d aL 1%9; Sprad- present when the nest was excavated on 10 February, bery 1973; Akre & 1980; Jeanne 1980; Nakahara and approximately 400 live queens. Twelve queens 1980; Nakahara & Lai 1981; Ross & Matthews 1982; " were captured and dissected. Four were oviposition- Ross & Viher 1983; Oambino 1986; Plunkett el ready and inseminated, two were oviposition-ready aL 1989; Oambim d aL 1990), a mild winter is - Y but uninseminated, three were hibernation-ready but clearly prerequisite for overwintering. Winters at uninseminated, and three were intermediate regard- middle elevations of the Hawaiian Islands are also ing fat body and ovarian development, and were without sustained periods of freezing temperatures, inseminated. and it is hardly surprisii to find considerable plas- ticity here as well. Although V. pensylvanh might DISCUSSION be expected to thrive at sea level, this has not proven Fundamental to any effective pest management to be the case in the Hawaiian Islands. Perhaps the program is an understanding of the biology of the wasp has been unable to adapt to certain aspects of pest. For insects whose biology varies si@icantly the sub-tropical climate; alternatively, they may be under different environmental regimes, the need for excluded by the abundance of ants at lower eleva- on-site studies is essential. Data describing where tions. Interestingly, on the mainland, the only re- and when the pest occurs, and characteristics that corded colony at a latitude as far south as Hawai'i render it vulnerable, need to be incorporated into the was at Tancitaro, Mexico, elevation 2000 m (Miller program to permit fine tuning of control strategies. 1961). Research on Y.pmiqhnh in the Hawaiian Islands Colony cycles in Hawaiian Y. pauyhnh illustrates this point well. populations range from the predictable annual type An aspect of Y. perqhnhbiology particu- to erratic but massively productive two year nests, larly relevant to its pest managememt in Hawai'i with many intermediate variants. Overwintering might be described in general terms as reproductive colonies occur irregularly, with frequency varying plasticity; the most spectacular manifestation is the among years and localities. Though comprising a overwintered colony. Potential for plasticity likely small fraction of total colonies, they can contribute resides in all y. populations, but is not disproportionate numbers of workers (Plunkett et aL 1989) and repraductives to the overall population; in of fat body reserves remain essential components of some nests, many thousands of queens are produced development, or are wmetimes bypassed. over extended intervals. Joiner queens can be easily determined by their No two overwintering colonies are alike, but they presence in colonies lacking the large cells necessary appeat to share some characteristics. Requeening to rear them (the NLCQ colonies found on both seems to be a prerequisite. At least one episode of islands). In the native temperate range, the pool of large cell construction occurs, but only a fraction of potential nest joiners includes only pt-hibernation colony resources are diverted to queen rearing, and typical spring queens (Akre & Reed 1981). Where immatures are reared concurrently in large and small colonies overwinter, and especially where extended cells. A reversion to small cell construction follows, queen rearing occurs, the pool is enlarged by a set of presumably in the spring and summer of the second queens with an alternate phenology (Figure 3). The year. If the colony remains healthy, there may be a phenomenon of joiner queens may be even more second round of large cell construction. These crite- common than the few cases documented here. Some ria exclude late developing annual colonies that fail of the oviposition-ready queens at QC colonies may to construct large cells, or to revert to rearing work- well have been joiners rather than retained daughters. ers, regardless of whether their decline phase extends Also, the large size and high activity rates at some into the following spring. No three-year colonies fvst year colonies at I-IAVO suggest that polygyny (that is, active during thtee consecutive summers) may have preceded the opportunity for them to rear have been verified. their own new queens. Overwintered colonies are inextricably linked to Yellowjacket colonies were consistently concen- the formation of polygynous assemblages within trated in only a portion of the HALE study area them; these in turn arise from adjustments of queen (centered at 2591 m). Thus, colonies experience physiology and behavior. There are two proposed relatively uniform environmental conditions and are routes to polygyny: retention of daughtet queens, and well-synchronized in their development. From addition of joiner queens. The fvst has never been 1987- 1991, in the absence of overwintering colonies, unambiguously documented, though it is widely ac- forager populations peaked regularly from late Au- cepted among researchers familiar with overwinter- gust thtough September, providing a coherent win- ing bpdacolonies (Duncan 1939; Thomas 1960, dow of opportunity for abatement measures. The Spradbery 1973; Edwards 1980; Ross & Viher failure of any 1988 colonies to even initiate large cell 1983; Akre & MacDonald 1986; Gambino et al1990; construction supports the tantalizing suggestion that Gambino 1991). Among the 7 late-season QC colo- no new queens at all were produced in the study site nies excavated at HAVO, the composition of queen that year, and that recolonization the following year populations is telling; hibernation-ready queens resulted from spring queens immigrating from con- made up only 19.0% (15 of 79) of those examined, siderable distances. and these were distributed among only 4 of the \lesDula also occurs elsewhere in colonies. The majority of queens were already HALE (within the crater) outside of the study area. oviposition-ready, and were found in all of the colo- Clearly, conditions on the northwest slope are not nies. It may be that hibernation is not required for representative of the entire park, and a greater diver- successful reproductive development of new queens, sity of yellowjacket behavior is likely to be diov- though it has not been abandoned altogether. Al- ered by a more comprehensive investigation of Y. though hibernation-ready queens were not found in throughout the park some QC colonies, they may have dispersed or trans- pm&mh biology at the HAVO formed before the nests were excavated. Apparently, study site was much more variable than at HALE. some new queens mate, fail to disperse, develop their Populations in different regions were not synchro- ovaries, and commence laying eggs. It is not clear nized; the general pattern was for earlier population whether the accumulation and subsequent depletion buildups in drier regions such as Ka'u and Kipuka Nene. Some populations declined swiftly, while oth- answered, and a better understanding might provide ers lingered on erratically for several months. How- insights for adjusting management strategies. This ever, even within a region, colonies sometimes would be a challenging, though potentially reward- showed diverse patterns of development. The ir- ing, avenue for future research. regular occurrence of overwintered colonies made it difficult to predict population trends with accuracy SECTION II comparable to that possible at HALE. YELLOWJACKET PREDATION To understand the extreme variability of Y. biology at HAVO, the physical conditions INTRODUCTION of the study area must be considered. HAVO is Like other species in the Y. dgigh species composed of a highly diverse mosaic of habitat types group, Y. pemyhmb obtains proteinaceous food resulting from often steep gradients of elevation, through predation and scavenging (Akre & aL 1980). annual insolation, rainfall, and temperature, overly- Of the two food collection strategies, scavenging ing a variety of soil and vegetation substrates. aiven most often brings wasps in contact with humans, and such extreme environmental variation, different re- may acount for some overestimation of the impor- productive strategies may achieve varying degrees of tance of dead flesh in the nutrition of Y. success at different times and places; as a result, no vlvanica colonies. In California Y. pensyl- single pattern of colony development is fixed in the vanica can build to outbreak populations while rely- overall population. If there are genetic determinants ing almost exclusively on arthropods as food of the reproductive mode, and the males adqueens (unpublished data). can disperse between different habitat types (as To justify abatement research in Hawaii and would seem likely given the strong flying abilities of elsewhere, it is important to provide evidence of Y. reproductives of both genders), then extreme repro- pmqhnb's impact on native organisms. Two ductive plasticity may be adaptive for the species in approaches were employed to investigate this. First, Hawai'i. Colony variants that would seem to be a qualitative study was conducted at HALE and reproductive dead ends may even acquire adaptive HAVO to determine the identity of arthropods upon value under this scheme. For example, colonies that which Y. pms&anh successfully preys. Direct grow large (20,000 cells) without constructing large field observations of wasp predation, while helpful cells may instead contribute only males as input to in elucidating behaviors used to locate, capture, and the genetic pool, or may postpone production of new transport prey (Duncan 1939; Heinrich 1984), yield queens until assertive outside joiner queens can be limited information on the identity of the prey spe- recruited. cies. Furthermore, such studies of prey selection are Compated to HALE, a yellowjacket manage- biased in favor of easily ohwed wasp activities. ment program at HAVO would require a more com- Prey items in the yellowjacket diet can be objectively prehensive monitoring system and greater flexibility sampled by intercepting foragers carrying prey as in responding to asynchronously developing popula- they return to the nest (Kleinhout 1958; Broekhuizen tions, with diffuse forager peaks and the occasional & Hordijk 1%8; Archer 1977; aambino 1986; Harris unpredictable overwintered colony. Thus, the parti- 1989, although this method also has its drawbacks. tioning of the park into Special Ecological Areas is In the present study the forager interception strategy an important management technique facilitating ap- was used to collect and identify Y. prey propriate local responses to Y. m~lvanicapopula- items from Maui and Hawaii. tions as they develop. An additional study designed to quantitatively The dispersal and migration of yellowjacket assess the impact of Y. predation was queens are among the least well known aspects of conducted near Puhirnau in HAVO. This was the site their biology. Pertinent questions about movements of high Y. ~vlvanicaforager density during the of Y. pmyhrhqueens at both parks remain un- summer and fall of 1989. The short-term prey intake of several active colonies was measured. Observa- the outside. At the same time, a plug covering the tions away from colonies were also made, using upward facing apertute of the T was removed, allow- objective measures of yellowjacket forager popula- ing wasps exiting the nest to leave. Sampling ses- tions, and abundance of ohia (- sions lasted about 20 minutes, when an equilibrium pb& foliage &pods (potential prey). between wasps entering and escaping from the trap was reached. The trap was removed, and wasps were MATERIAL & METHODS anaesthetized with CO2 gas and dumped into a sort- Qualitative Studies ing tray. Prey items were collected with forceps into a vial; wasps recovering from anaesthesia were re- Two basic techniques were used to sample prey leased on the ground near the nest entrance. collected by Y. workers. The simplest In several instances yellowjackets not associated was to use a hand-held net to collect workers with any nest were observed engaged in attacks on as they returned to the nest. At moderate sized typical disabled arthropods, or carrying pieces of prey. Prey colonies, it was easiest to briefly inspect returning items recovered from these encounters are also in- foragers as they approached, and to capture only cluded in the analysis. those carrying visible prey items. By partially ob- Collected prey items were preserved by first structing the nest entrance, returning workers could steeping them for one minute in just-boiled water, be induced to hesitate before entering, increasing the and then transferring to vials of 70% ethanol. Indi- opportunity for inspection and capture. At very ac- vidual prey items were examined for identification in tive large colonies, an alternate strategy was to make the laboratory using a binocular dissecting scope. numerous sweeps with the net in the vicinity of the Because handling of prey by Vespd workers prior nest entrance and to retreat some distance once 40-60 to returning to the nest commonly includes removing workers had been captured. In either case, yel- appendages and other body parts, and chewing to lowjackets were separated from their prey items in form a bolus (Duncan 1939), the level of taxonomic the net bag. Wasps were allowed to escape; prey determination was highly variable. Specimens with items were retained and processed. clear diagnostic characters in good condition could In the second collection method a trap made of be assigned to species, while some mutilated pieces 5.1 cm (2")PVC pipe fixtures was placed over the could not be identified below phylum. Items were entrance of the tunnel leading to the subterranean considered identified if they could be determined to nest. The trap consisted of a flat wooden base, a order; comparisons between geographic regions T-shaped fmture set into a hole at the center of the were made at this level. base, a plug, and a laterally oriented removable el- Study areas included Haleakala National Park bow (Figure 4). The trap was placed over the tunnel (HALE) on Maui and Hawaii Volcanoes National entrance at night and anchored with spikes driven Park (HAVO) on Hawaii. pensvlvanica through the comers of the base; the base edges were Bpula colonies sampled at HALE occurred within a fairly covered with packed soil. During subsequent days, uniform habitat on the northwest slope of Haleakala foraging workers learned to pass thtough the unob- volcano (Garnbino 1990) and are considered as structed pipes to exit and return. Wasps sometimes a single geographic unit. The area sampled at HAVO constructed new tunnels around the trap base to was larger and more diverse; for comparative analy- bypass the modified entrance; this behavior was dm- sis of yellowjacket diets, HAVO nests were grouped couraged by covering the holes with soil. Once a into the following geographic units: Kipuka Puaulu, regular t&c pattern through the trap was estab- Ola'a Tract + Ola'a Forest Reserve, and Puhimau. lished, daytime sampling was initiated. To conduct tached to 3 colonies (HV08904, HV08905, and twenty trees were sampled at each plot. One limb per HV08906) in the Puhimau area. On 24 November tree was struck 10 times with a standard striker, and 1989, activity counts and prey samples were taken on arthropods landing on a square white sheet (area - a rotating basis from 0636 until 1746. During each 0.5 m) held directly beneath the limb were dumped sampling interval the number of yellowjacket sorties into a 15-1plastic tub and collected with an aspirator. per 2 minutes was counted; then the trap was inserted Limbs were selected so that approximately the same and left in place for ten minutes. The numbers of amount of foliage was directly over the sheet for ech yellowjacket captured, and the numbers of prey items sampling. Collembola numbers were estimated to they collected, were counted. For each interval, the the nearest five, but since this taxon was composed following parameters were estimated: wasp trips per of arthropods believed to be below the size threshold minute, total trips during the interval, prey items per for Y. gemyhmh predation, Collembola were not returning wasp, and total prey captured during the included in the quantitative analyses. W1 other ar- interval. For each colony, these quantities were thropods were transferred to plastic 1.7 ml tubules summed to give a total for the day (Table 8). Data and held in a cooler at approximately 5'~. from these trap captures were also included in the In the laboratory, samples were held in the closed qualitative prey sampling program. tubules for up to two days at -loOc.For each sample The study site for the evaluation of Y. ~nsyl- the total biomass was weighed (to the nearest 0.001 vanica predation was a section of land southeast of g); then the weight of isopods + millipedes was Kilauea Crater, 2.5 km long and 125 m wide, bisected subtracted, to give the mass of potential prey items by Chain of Craters Road, which descends gradually (isopods and millipedes are not collected by Y.gg~~- in a fairly straight line from 1120 m to 1055 m sylvanica foragers, and are not considered in the elevation. Four sampling plots were established, following analysis). After weighing, arthropods centered at 8 10 m intervals along the road. Vegeta- were transferred to vials containing 70% EtOH. tion at each of the plots was comparable, consisting To measure arthropod size, each specimen was of an open ohia forest with an understory of suspended in a 6-cm petri dish of 70% EtOH and hya, interspersed with more open areas vegetated.. by examined under a dissecting microscope with a scale native shrubs (- visc~~ia, -, . attached to the stage. Size was estimated by meas- and 7. . grasses ), alien .(& .. uring the length from the front of the head to the tip condensaturn and Andrompon. . of the abdomen, excluding appendages such as an- as), and native fern (vand tennae, mouthparts, legs, wings, and cerci. For case- NeDhroleDis spp.). bearing caterpillars the length of the case was Yellowjacket forager populations were sampled measured. Arthropods shorter than 4.5 rnm were using the Yellowjacket Inn trap, baited with a wick classified as small, others as large. Although this was containing 0.4 ml of heptyl butyrate. At each plot 5 a somewhat arbitrary demarcation, it effectively traps were suspended from tree limbs bordering the grouped many taxa into single categories; for exam- road, separated by at least 25 meters. After 24 hours, ple, all Psyllidae, , and Aloha were small, numbers of captured Y. pensylvanica workers (no and all adult and and were large. queens or males were taken) were tallied. Forager Other taxa, such as , crickets, and cock- densities were sampled in the fall on 29 November roaches, were well represented in both size classes. and 4 December 1989, and again on 4 June and 28 June 1990 for the early summer period. Numbers RESULTS & DISCUSSION were pooled to give mean estimates of forager den- Qualitative Studies sity (predation pressure) at each site for both late fall At HALE a total of 19 collection lots yielded 233 and early summer. identified specimens. Two specimens, an earthworm A beating sheet technique was used to sample and a centipede, are exceptional, and ate omitted arthropods from foliage-bearing limbs of ohia trees; from the following analysis. At HAVO,59 collection Native and introduced bees and wasps were lots (including 5 prey items taken from wasps not common at all sites, yet only two sp. associated with any known nest) yielded 29 1 identi- and four L. specimens were recov- fied specimens. Table 8 summarizes the yel- ered, suggesting that healthy aculeates are able to lowjacket diet according to geographic location and actively defend against predation better than other d. d. major taxa. The diversity of prey at the population taxa. MacDonald & d. (1974) also found live (geographic) and colony levels indicates the Y. aculeates to be relatively immune from Y. JEIISYL sykimka does not specialize in particular prey taxa, & predation. However, consti- although individual foragers may do so. The recov- tuted 10.4% of prey items at HALE. In this sample, ered prey items suggest that Y. foragers taken during the late summer of 1986 when foragers harvest arthropods at or near surfaces, and neither dig were very abundant (Gambino et d. 1990, Gambino in soil nor probe the interiors of plant parts to seek 1991), 13 (54%)of the Hymenoptera prey items were food. other Y. -. This most likely represents The yellowjacket diet is influenced by several scavenging on Y.pasykmh cadavers, which were characteristics of potential prey items: their presence not uncommon at the site, rather than predation within foraging range, their acceptability, and the It must be noted that the sampling/identification ease with which they can be located and captured. process may give an incomplete picture of yel- Two arthropod taxa, pillbugs (Isopoda) and milli- lowjacket predation, because not all prey items could pedes (Diplopoda) were common and fairly con- be identified. For instance, at the Puhirnau site, a spicuous at all sites, yet they were absent from the y. total of 322 items were collected, but only 57 (17.7%) diet. Immunity from predation may be could be identified, a rate typical for the entire study. due to a combination of biochemical defenses, which Thus, there may be bias in favor of taxa easily rec- make them unacceptable, and physical/behavioral ognized due to distinctive diagnostic characters that mechanisms that render them to subdue. In withstand processing by yellowjacket foragers; taxa 8 contrast, Araneae and were apparently that are soft bodied or lachg in good diagnostic ubiquitous, acceptable, and not to frnd and characters may be underrepresented in Table 8. subdue; these made up a relatively constant part of Another bias in the sampling procedure is related the yellowjacket diet regardless of location. Of the to prey size, as the smallest items ( 1 mm diameter) Lepidoptera, most were taken as larvae; the 100 carried by foragers were the most difficult to identify. specimens included 1 egg mass, 1 pupa, and 11 Nonetheless, there is a size threshold below which a adults. For other taxa, variations between sites in y. pa&mh forager will not collect potential the proportions of items taken may reflect circum- prey arthropods (Duncan 1939). Small arthropods stances peculiar to those sites and taxa. The lack of may be neglected due to difficulties in perceiving or Blattodea and Orthoptera in HALE samples reflects capturing them, but more likely, the cost/benefit ratio their near-absence from native habitats there. is much more favorable for Y.ggmyhmh to hunt Coleoptera were poorly reptesented in Y. & large prey. Smaller arthropods may thus be spared vanica prey samples, except at HALE, where 24 of because they cannot be efficiently harvested. How- 34 (70.6%) were of a single species, the introduced ever, if large prey items cannot be found in foraging weevil Pantomorus cervhs (Boheman). Perhaps range, Y. may adjust its threshold to this represented an abundance of this species unique accept smaller prey. In the present study, it seemed to the HALE site; alternatively, it may be a less that few of the boluses carried to colonies could have preferred species whose appeal increased after popu- derived from arthropods less than 3 rnm in length. lations of more desireable species had been depleted A total of 170 prey items could be determind to by a prolonged spell of intense area- wide Y.& genus or species level (Appendix 1). Of these, 112 vanica predation. (65.9%) were of endemic (to Hawaii) taxa, and 58 (34.1%) were of alien or non- endemic native tan. Table 8. Summary of prey taken by Y. pensvlvanica in Hawaii. Values shown represent percentages of N items. ~axa~ siteb N ARA ORT BLA HEM NEU COL DIP LEP HYM KP 120 25.0 18.3 6.7 4.2 1.7 3.3 26.7 12.5 1.7 PU 57 33.3 22.8 3.8 17.5 0.0 3.5 0.0 19.3 0.0 OT 30 36.7 0.0 10.0 6.7 6.7 3.3 6.7 30.0 0.0 HK 23 1 23.4 0.0 0.0 21.2 0.0 14.7 8.2 22.1 10.4 522 26.8 10.2 2.5 16.3 1.1 7.9 11.1 19.2 5.0 %EY TO SITES a~~~ TO TAXA KP - Kipuka Puaulu (I-IAVO) ARA - Araneae OL - Olaa Tract + Olaa Forest Reserve (HAVO) ORT - Orthoptera PU - Puhimau (HAVO) BLA - Blattodea OT - Other (HAVO) HEM - Hemiptera + Homoptera HK - Haleakala (HALE) NEU - Newoptera COL - Coleoptera DIP - Diptera LEP - Lepidoptera HYM - Hymenoptera

The roughly 2: 1 ratio of enderniclnon- endemic taxa Calculating prey consumption for single colo- may underestimate the true proportion of native prey, nies or populations requires extrapolation from the considering that: 1) some prey items were identified limited data available. One way to estimate a col- only to endemic subfamily level; 2) many native taxa ony's total input is to calculate the number of wasps lack fonnal taxonomic designations due to insuffi- reared and to multipy it by the number of prey needed cient study; and, 3) many alien species are diitinctive to rear one wasp. Assuming that 1) most prey are and easily identified. small in relation to a final instar larva; 2) only a portion of each prey is used as food; and 3) biomass Quantitative Studies would ahinclude the saliva fed back to aduIts Table 9 summarizes forager and prey intake pa- during the duration of the larval stage and the voided rameters for the three monitored colonies. The colo- meconium, one final instar might conservatively be nies differed substantially in total numbers of trips expected to account for 20 prey items. A colony and prey per wasp; the most active colony rearing a total of 20,000 individuals (also wnserva- (HV08905) also collected the most prey per wasp, tive for the sizes and population structures of annual harvesting in one day about 16 times as many prey Y.pensvlvanica colonies in Hawaii) would consume items (22,055) as the least active nest (HVO8904). about 400,000 prey items. For colony HV08905 this These results, and subsequent observations of colony would represent less than three weeks of foraging at activity, size, and longevity, suggest that colonies its rate of 24 November (which was probably lower HV08904 and HV08906 were well into their &- than an earlier maximum rate). For this colony, an cline phase when sampled, while HV08905 was estimate of 60,000 wasps reared and 1,200,000 prey more vigorous. During the course of the day there consumed would still be conservative. The con- were within-colony fluctuations within each colony sumption of a single overwintered colony such as in foraging rate (Figure 5) and prey per wasp (Figure HVO8903) (600,000 cells) would probably be in the 6), but no clear patterns emerged. tens of millions, spread out over two yervs! The impact of Y. predation is hardly trivial. Table 9. Prey Intake at Thtee Colonies at Puhirnau, HAVO, 24 November 1989.

The'hap Foraging Rate Interval Total Tkips Total Prey Off (Sorties1 (Minute) Minute) 06:46 90 46 4140 0 08:M 76 78 5928 97 09:18 150 74 11100 89 10:33 146 75 10950 175 l2:O5 150 92 13800 425 13:07 170 62 10540 88 14:22 150 75 11250 225 15:31 170 69 11730 201 16:41 92 70 6440 8 1 TOTAL 85878 1381

07:08 68 8568 498 08:27 79 17301 1977 09:39 72 27000 2925 10:52 73 26937 2950 1218 86 20640 4034 13:25 67 18090 3536 14:40 75 18000 2242 15:50 70 21630 3183 17:00 70 11130 710 17:46 46 3864 0 TOTAL 173160 22055

07:43 73 14308 84 09:00 77 18942 0 10: 12 72 18000 900 11:27 75 21000 477 12:48 8 1 14580 1166 14:00 72 11520 576 15: 12 72 15840 1485 16:22 70 15400 324 17:30 68 13600 199 TOTAL 143190 521 1

Ohia foliage sampling was scheduled to compare contrast, the spring sample was taken after several the effects of different levels of 1. pensvlvanica months of negligible Y. population lev- predation presswe. The initial late fall ~plewas els; colonies had died out during early 1990, and no taken following several months of high forager popu- foragers were captured on the fmt monitoring date. lations. Furthermore, there was a gradient in preda- Newly founded colonies were in their early develop- tion pressure, with highest numbers of foragers in the mental stages, and predation pressure would have Kokoolau area, and lowest numbers at Lua Manu. In Table 10. Effects of Yellowjacket Predation on the Standing Crop of Ohia Foliage Arthtopods (Potential Prey).

Wasp Forager Weight Number of Small Number of Large Density Items Items 1989 Lua Manu I'uhimau Kokoolau Hilina Pali 1990 Lua Manu 25.50 2.20 2.25 l'uhimau 25.85 6.15 2.05 Kokoolau 23.35 4.55 1.75 Hilina Pali 18.60 3.00 1.70 been insignificant, and relatively uniform for the four may have a subtle effect of altering the composition sites. of local arthropod communities rather than causing Table 10 summarizes the seasonal indices of Y. wholesale depletion of arthropod biomass. By re- abundance, and the three measures of moving massive numbers of larger arthropods from arthropod abundance, at the four sites. These figures ohia foliage, Y.m~lv& may also affect native must be interpreted with caution, because baseline birds, for which insects and spiders are important data on arthropod abundances are lacking. Here, it components of the diet. Birds also likely have a size is possible only to identify apparent trends and to threshold for prey, and to the extent that Y. suggest links to Y. predation. For 1989 &harvests the same items as the birds, there is the correlations between Y. pmsybnh forager direct competition for these resources, and a possi- density and all three measures of prey abundance are bility of additional stress on native bird populations. striking; the lowest values were at Kokoolau, where The threat to native Hawaiian ecosystems posed Y.mlvanica was most abundant. However, there by Y. -'s invasion is exacerbated by a is also substantial variability in values from the fol- number of factors. Many endemic tam ate highly lowing spring, in the absence of Y. pensvlvanica precinctive within native Hawaiian habitats, and lo- predation at all four sites. Factors besides predation cal populations may be unable to recover from per- effects, especially seasonal fluctuations of arthropod turbations arising from intense predation pressure. populations, may play an important role here, but The y.pensvlvanica monitoring programs show that because of the lack of information, they cannot be areas of high yellowjacket abundance may also be systematically accounted for. extremely localized, and predation pressure may Only one measure of arthropod abundance gave vary substantially over relatively short distances (1-2 consistent evidence of predation effects over both km). Hawaiian arthropods have evolved in the ab- years: numbers of large (4.5 mm) arthropods. At all sence of predation pressure from social Hymenop- four sites values were lower in 1989 than in 1990. tera, and may thus lack antipredator mechanisms This accords well with the concept of a size threshold selected for elsewhere (Oagne & Christensen 1985). for Y. pensvlvanica prey; if larger atthropods are Ycqzh has penetrated into a vatiety of more preferentially harvested, then small arthropods native Hawaiian habitats (Howarth 1985; Garnbho would be expected to be less affected. Furthermore, ef d. 1990, Gambho 199 1) where, under favorable when large arthropods ate predatom, theii removal circumstances, it can form very large annual and from the habitat may actually release populations of overwintered colonies. The latter, which occur un- smaller arthropods. Thus, Y.pasyhka predation predictably, reduce the effectiveness of seasonality as a temporal refuge from Y.pensylvanica predation. overall environmental effects of the pesticide toxi- Although Y. is a generalist predator cant, risks to applica!ors, and ease of use. (chcteristic of the genus), there is evidence that Sometimes an outbreakis caused by a population species adjust their foraging habits to focus of colonies, making it impractical to find them all. In on abundant prey species (Broekhuizen & Hordijk this situation, the scavenging habit of Y. 1%8; Oambino 1986). Thus, species which have & can be exploited by deploying a toxic bait pronounced peaks of abundance, even of short dura- distribution system. The effectiveness of toxic baits tion, may stimulate a functional response in Y. was first demonstrated by Grant & aL (1968), who svlvanica that would reduce the effectiveness of used a mixture of a horsemeat bait base and the predator satiation (May 1981) as an antipredator toxicant chlordane to reduce forager populations of mechanism. Y. and Y. &a& (L.) in California. Since then, the strategy has been modified for use SECTION m against scavenging species in a number of YELLOWJACKET ABATEMENT situations (Keh ct aL 1968; Ruddock & Rohe 1968; Wagner & Reierson 1969,1971; Rogets 1972; Ennik INTRODUCTION 1973; Perrott 1975; Chang 1988). In Hawaii, Chang A prime objective of this project is the develop- (1988) conducted an abatement program against Y. ment of a practical pest management program for in sugar cane fields on O'ahu, investi- Hawaii's National Parks. Because Y. pensvlvanica gating the effects of different baits and toxicants, is only one of a large number of pest organisms that dispenser designs, and bait distribution schedules. the Parks' Resources Management Divisions must Forager behavior, including food preference, varies deal with, every effort was made to consider their among bpda species (MacDonald & aL 1976; staff and time limitations in the design of the control Akre & MacDonald 1986), and possibly among geo- program. The &tails of Y. pmyhmka biology in graphic strains within species (Akre & aL 1980). Hawaii, outlined in Section I, need to be incorporated Thus, the design of an optimal pest management into the program. The persistence, widespread dis- program incorporates characteristics of the specific tributions, and dispersal potential of Y.pensvlvanica pest in the area of concern. populations on Hawaii and Maui suggest that eradi- This section is subdivided into three subsections: cation is either impractical or impossible, and that 1) results of nest treatments with two pesticides, management efforts ought to focus on abatement. Ficam D and PT 5 15 Waspfreeze; 2) investigation of Additional studies reported here address relevant factors affecting the toxic baiting program; and, 3) practical and logistical details of an effective abate- results of control efforts at HALE and HAVO com- ment program. bining nest treatment and toxic baiting strategies. The most feasible abatement methods involve applications of pesticides to kill wasps, and hope- NEST TREATMENTS fully, entire colonies. These are applied according to Materials & Methods strategies designed to limit the impact on non-target The effectiveness of Ficam D (dust formulation, organisms. Pesticides can be effectively used against active ingredient bendiocarb) and Whitmite PT5 15 Y. pm&mka in two ways. Fitst, when an active - Waspfreeze (aerosol spray can, active ingredients nest is located, pesticide is applied diitlyto it. This - pyrethrins) were evaluated by monitoring activity at fairly obvious technique no doubt has a long history; active nests before and after their application. For spot treatment of colonies probably started as soon Ficam, dust was dumped directly into the entrance as early man could associate wasps with the nests tunnel leading to the yellowjacket nest. For Wasp- from which they came. From the point of view of the freeze, varying amounts, up to one 12-ounce can, National Park Setvice, other factors beside effective- were sprayed into the entrance tunnel. ness against wasps must be weighed, such as the & foragers in a specialized habitat, the Ola'a Results & Discussion Tract ohia- rainforest, was studied. The results of pesticide tratments of individual Bait:Nine baits were evaluated for Y. colonies are summarized in Tables attraction of Y. foragers: Swanson's 11 (Ficam) and 12 (PT515). Neither treatment was Premium White and Dark Chicken (CHK), Hormel always effective in destroying active colonies with a Chunk Turkey (TKY), Hormel Chunk Ham (HAM), .. single application. Reasons for failure probably in- Sea Alaska Fancy Sockeye Red Salmon (SAL), cluded large colony size (and the ability to recover as Spam (SPM), Coral Tuna in Water (TNA), Swan- new adults emerged within the nest) and difficulty in son's Chunky Chicken Spread (SRD), Figaro Tuna applying the pesticide to ensure exposure of wasps Cat Food (CAT) and Nebraska Brands Feline Food (allowing them to avoid exposure when exiting or (NBF). With the exception of NBF,all were canned entering the nest). The advantage of Ficam is that as meat- or fish-based products available at local retail a dust, it is perhaps more easily picked up on the legs stores. Preiiminary screening showed that CHK was of workers and spread among colony members, highly attractive, so a mixture of CHK and the toxi- whereas the aerosol PT5 15 would tend to soak into cant Knoxout 2Fh4 formulated according to label the soil. Although toxicity data were not available, instructions (3 ml per 100 g bait) was also included Ficarn seemed to be more lethal to Y. -. as a tenth bait (KOC). For each test ca. 25 grams of R This is an advantage in terms of efficacy, but its broad bait was put into a Yellowjacket Inn trap. Ten traps spectrum toxicity is a disadvantage in terms of expo- were suspended at intervals of 0.75 m in random sure of nontarget organisms. Three of the Ficam- order from a line 1.5 m above ground, oriented across treated nests were subsequently attacked by the prevailing wind. Numbers of wasps were vertebrates (probably rats or mongooses) and combs counted after 24 hours. Tests were run at four loca- * were dug up and scattered. These attacking animals tions within HAVO: Kipuka Nene (15 Sep, 6 Oct, 11 were thus also likely exposed to the pesticide, and Oct 1989), Kipuka Puaulu ( l5 Sep), Ka'u Desert (15 - * there is potential for the poisons to move up the food Sep), and Puhimau (13 Oct). Data were log trans- chain (to endangered predator/ scavenger bitds, for formed to correct for unequal variances, and ana- example). As a dust, Ficarn is also cumbersome to lyzed as six independent trials in a randomized block store and apply safely, and exposure of applicators or analysis of variance (ANOVA) with means separated other people is a potential hazard. The National Park by Duncan's test, using the GLM procedure of SAS Service is justitled in seeking a substitute for Ficam Institute (1985, 183-260). D for treatment of nests, but the results of trials with Bait:The four most attractive baits Whitmire PT5 15 Waspfreeze indicate considerable (CHK,KOC, TKY, HAM) were selected for tests of room for improvement. bait acceptance. For each test, 25 g of bait was placed in an 88-ml paper cup; a fifth cup, containing bait FACTORS AFFECTING TOXIC BAITING covered by a screen to keep out yellowjackets, was included as a control for each replicate. Since baits Material & Methods were similar in density and water content, CHK was Several factors affecting the baiting program selected as the representative control bait. Cups were evaluated. First, baits were screened to find a were suspended at intervals of 0.5 m from a horizon- suitable bait base by testing nine different foods to tal pole 1.5 m above ground; the middle cup was the determine relative attraction. The thtee most attrac- control, and other baits were placed randomly. The tive were then tested to determine which one yel- dependent variable was amount of bait taken; to lowjackets preferred to collect. The use of heptyl correct for weight loss through evaporation, the butyrate to enhance bait collection, and the effects of amount was calculated as (total weight loss - weight aging on bait acceptance were also tested in the field. loss from screened cup). Tests were replicated at Finally, the attraction of heptyl butyrate to Y. pen& three sites in the Ka'u Desert (where foragers were Table 11. Tests of Ficam D treatments on nests of Y. peuyhnh .

Nest Amount (oz.) Results HV08802 1 Activity terminated; vertebrate feces found Activity terminated Activity terminated Activity terminated Activity terminated; combs dug up and scattered Activity terminated Activity terminated; combs dug up and scattered Activity terminated Activity terminated Activity terminated Required additional application Activity terminated Activity terminated; combs dug up and scattered Required additional application Required additional application Activity terminated Required additional application

Table 12. Tests of Whitmire PT5 15 Waspfreeze treatments on nests of Y.pa&mka . Nest Amount Results HVO9002 1 can Additional treatment necessary HV09006 1 can Activity terminated HAL9001 112 can Additional treament necessary HAL9002 112 can Activity terminated most abundant) on 3 July, 24 July, and 3 1 July 1990. heptyl butyrate with a paper towel wick, affixed. Data were analyzed in a two-way ANOVA with baits Dispensers were put out on 2 Aug 1990 at approxi- and dates as treatment factors. mately 0800. Bait remaining was weighed after eight p:The chemical hours; amounts of bait taken (uncorrected for evapo- heptyl butyrate is one of a series of synthetic esters ration) were analyzed by t-test (two-tailed). attractive to Y. foragets (Davis a aL of F- In preliminary 1%9). Its effect on the amount of KOC collected by toxic baiting experiments using KOC, it was noticed wasps was tested in the context of an ongoing abate- that when baits were put out eatly in the day, Y. ment program at HAVO. The test area was approxi- pensylvanica foragers at dispensers were more abun- mately 26 acres at Namakani Paio. Dispensers dant in the morning than in the afternoon. It was not similar to that designed by Chang (1988) were sus- clear if this was due to decreasing attraction of the pended from trees ca. 1.5 m above ground at intervals bait as it aged, or to incapacitation of foragers result- of ca. 70 m around the perimeter. Each dispenser ing from their contact with the insecticide. To decide held 60 g bait; alternating dispensers had an added between these competing hypotheses, the accept- open 1.8 ml microfuge tubule, containing 0.4 ml ability of aged bait was tested. Fresh KOC was put out on 20 Aug 1990, using the same sites and dispens- (consistent with the fmdings of Reid & MacDonald, ers as the heptyl butyrate bait enhancement test. Bait [1986]); if anything, attraction was slightly en- was collected after eight hours and stored overnight hanced. at 5' C. The next day the aged KOC and fresh CHK Bait:Regardless of inherent attrac- were offered to naive yellowjackets using the tech- tiveness, a toxic bait will be ineffective unless yel- niques and sites of the previous bait acceptance trials. lowjackets carry it away from the dispensers. Reid Because the aged KOC had already lost some of its & MacDonald (1986) demonstrated that acceptance moisture through evaporation the previous day, sepa- of different preparations of the same bait base may rate screened cups were put out for both it and fresh vary according to texture. The four baits chosen for chicken at each site. Amount of bait taken was the acceptance test (CHK, MOC, TKY, HAM) corrected by subtracting the loss through evaporation showed some difference in the amounts removed by at the control from the total weight loss of the test 1. foragers (F - 5.89; df - 324; P bait. Data were analyzed using a t-test (one-tailed). 0.01). In paitwise comparisons (Table 14), CHK and of tyl but- in Ola'a 'Bad: KOC did not differ significantly, but less ham was HAVO contains a diversity of habitats that vary in taken than chicken with or without pesticide. The suitability for supporting a baiting program. The close proximity of baits to each other was Intended Ola'a Tract is a rainforest with large areas of dense to reduce the effect of differential attractiveness by ground-level vegetation where yellowjacket foragers having baits generate a single pooled odor plume were rarely seen. The discovery of a series of four effective over long distances, but some fine discrimi- active colonies within a span of 170 meters along a nation by wasps nearing the baits was inevitable. fence line provided an opportunity to investigate However, increasing reliance on visual cues at close foraging behavior. On 1 Dec 1988, foragers at the range would minimize this effect. There was a sig- L four colonies were marked by dumping 25 grams of nificant influence of date (F - 18.84; df - 2,24; P < colored fluorescent powder into the nest entrance 0.01), likely due to a combination of seasonal vari- - C tunnel. A series of 14 heptyl butyrate-baited Yel- ation in the wasp populations and diierent weather lowjacket Inn traps were suspended from a 430 m conditions on the test dates. The interaction of bait section of fence line (height - 1 m) that overlapped with date did not produce a significant effect. with the locations of all four nests. Ten of the traps w:Of the scav- were within 50 m of at least one nest; the farthest trap enging species, Y. is excep- was 250 m from the nearest nest. Traps were put out tional in that effective synthetic attractants are at 1030 and checked five hours later. available. So far, heptyl butyrate has proven useful for monitoring Y. pens&&a populations (Mac- Results & Discussion Donald a 1973; Chang, 1988; Oarnbino & Bait Attraction: Baits differed significantly in 1990) and for population depletion trapping (Davis their attractiveness to Y. pasyhhforagers (F - & aL 1973). At I-IAVO, wasps took significantly 5.75; df - 9,40; P < 0.01), with two chicken formu- more KOC bait from dispensers with heptyl butyrate lations (KOC, CHK) most preferred (Table 13). The wicks (means of 16.9 g versus 7.4 g; t - 3.14; df - texture of chicken had a substantial effect; CHK 16; P < 0.01). Thus, the potential for chemical tna- consisted of fairly large chunks of flesh, whereas the nipulation of Y.pensvlvanica forager behavior in the paste SRD, presumably diiering only in the process- context of a toxic baiting program, shown for heptyl ing, drew in significantly fewer yellowjackets. This crotonate (Wagner & Reierson, 1969), is also dem- ., contrasts with the results of Reid & MacDonald onstrated for heptyl butyrate. (1986), who found the number of Y. gamah(F.) v of F- The microen- visits to be relatively uninfluenced by grinding treat- capsulation process for Knoxout 2FM reduces the ments of NBF. The addition of Knoxout 2FM to repellency of its active ingredient, diazinon, to yel- CHK (to make KOC) did not reduce its attractiveness lowjackets (Ennik, 1973). The stab'ity of the formu- Table 13. Numbers of Y. pensyhhworkers attracted to ten baits. No. of wasps caught Bait 'ISpe of food a x log(x+l) 2 SD~ KOC Chunk chicken + Knoxout 60.8 1.65 5 0.28 a 2FM CHK Chunk chicken 51.3 1.55 2 0.42 ab TKY Chunk turkey 35.3 1.23 5 0.66 abc HAM Ham 28.0 1.18~0.52abc SPM SP- 19.2 1.05 5 0.17 abc SAL Salmon 22.5 0.97 50.16 bc TNA Chunk tuna 7.5 0.78 5 0.13 cd SRD Chicken spread 13.0 0.77 2 0.13 cd NBF Horsemeat 1 .O 0.23 5 0.04 d CAT Tuna cat food 1.2 0.19 2 0.03 d a See text for full descriptions of baits. Means within a column followed by the same letter are not significantly different (P - 0.05, Duncan's multiple range test).

Table 14. Amounts of baits collected by Y.pemyhnb workers.

Bait Grams taken x 2 sDa Chicken 9.9 5 4.7 a Chicken + Knoxout 2FM 8.4 2 3.2 ab Turkey 7.0 L 5.0 bc Ham 4.3 5 2.7 c a Means within a column followed by the same letter are not significantly different (P - 0.05, Duncan's multiple range test). lation under field conditions merits evaluation, since particles before carrying them in flight back to the an increase in repellency will reduce the usefulness nest. This is likely to disrupt the capsules containing of Knoxout 2FM toxic baits. In the present study, the pesticide, exposing the workers to it even though they acceptability of KOC bait after eight hours of expo- do not consume the food. sure to field conditions was about two thirds that of ction of he~tdbutvrate m Ola'a Tract: fresh CHK bait, a significant decrease (means of 16.2 Only one (unmarked)yellowjacket was caught in one g versus 10.7 g; t - 2.51; df = 4; 0.05 p 0.025), but of the fourteen traps, which was 200 meters from the not sufficient to indicate a major increase in repel- nearest nest. This result comborated the informal lency. Another explanation must be sought for the observation that wasps exiting nests in vegetationally near total absence of Y. pmsykmh foragers at dense areas of rainforest flew straight up to forage in baits after ca. four hours of exposure. Most plausible well-lit regions of the canopy, and rarely flew at is that wasps carrying bait succumb to the effects of ground level. It suggests that the effectiveness of the insecticide after one or several trips, depleting heptyl butyrate as an attractant is very reduced under within four hours the portion of the forager popula- rainforest conditions, probably because a coherent tion attracted to KOC bait. A greater delay in the odor plume that wasps can follow to its source does death of wasp foragers, allowing them to make more not form. A combination of high humidity, low wind, trips to collect bait, would enhance toxic bait effec- and restricted straight lines of vision and flight might tiveness. One aspect of worker behavior may be contribute to this effect. These results would also call difficult to overcome: foragers trim and reshape food into question the attractiveness of food bait bases favoring well lit areas of the canopy may be more at under rainforest conditions at Ola'a. risk This study underscores the importance of screen- ing a variety of bait bases before selecting the one to YELLOWJACKET use in a toxic baiting program. Past studies of yel- ABATEMENT TRIALS lowjacket feeding preferences have concentrated on Area-wide baiting research in 1988 and 1989 - canned pet foods and fish (Ruddock & Rohe 1968; was conducted concurrently with research described Wagner & Reiemn 1969,1971;Ennik 1973; Perrott in previous sections. Consequently, the selection of 1975; Ross & & 1984; Chang 1988); in the present sites and schedules for baiting did not always coin- study, such baits ranked low in attractiveness. Ne- cide with the most favorable opportunitiesto demon- braska Brands Feline Food, which showed potential strate the baiting technique. By 1990, a better for incorporation into a Y. gtmdm management understanding of factors affecting baiting permitted program (Ross & 1984) would be totally unsuit- greater confidence in the effectiveness of the tech- able for Y. pensvlvanica control in Hawaii, consid- nique. The 1990 trials were conducted as demonstra- ering its inability to attract workers and its tions for Resources Management Division staff, to specialized storage requirements. Of the three most provide hands-on experience. This, combined with attractive bait bases, only ham has been compara- the Resource Manager's Guide (Appendix I), repre- tively evaluated in past studies (Reid & MacDonald sented the final technology transfer aspect of the 1986). project. From a land manager's point of view, acceptance Materials & Methods of a bait by yellowjackets is only one of several For each yellowjacket abatement trial there was considerations in implementing a baiting program; +# a treatment area and an untreated area of similar size. ease of bait base acquisition and storage are also Although we tried to control for such factors as important. Thus, for the sake of the intended users altitude, vegetation type, and other environmental of the present research results, the focus was on -. conditions, matches were only approximate. Asepa- widely available canned meat and fuh products, from ration of at least 800 m was necessary to minimize which several suitable candidates were selected. spillover effects of baiting on populations outside the Overall, canned chunk chicken was unsurpassed as a bait distribution area (especially the untreated bait base in both attraction and acceptability. Fonnu- in areas). However, non-uniformity of environmental lation with Knoxout 2FM did not diminish these qualities. The problem of reduced acceptability of conditions and of yellowjacket populations in matched areas increased with distance. This was not aging bait remains, but a more limiting factor may be a trivial consideration, because in both parks fairly the knockdown of bait-collecting foragets. Per- steep gradients of a number of environmental factors formance of the baiting system is also affected by occur. physical factors at the baiting site; success is more For each trial, populations each area were likely in relatively open areas where yellowjackets in monitored with a battery of five heptyl butyrate can easily find the baits. If competing attractions baited Yellowjacket Inn traps. Data from at least two draw workers away from the baits, and especially if sampling intervals were gathered prior to bait distri- baits cannot be easily placed where yellowjackets bution. An arbitrary action threshold of five wasps prefer to forage, it may be difficult to justify the per trap per day (5 W/T/D) was assigned. There was expenditure of resources on a baiting program. Re- no fmobjective justification for this value; rather, garding the conservation of native arthropods, the it was chosen because at this level, yellowjackets reluctance of Y. pasyh&a to hunt in well-shaded were observed to be fairly conspicuous and some- sections of the rainforest suggests that this may offer what bothersome to people. Furthennore, above this a refuge for arthropods that occur there, while those level, the effects of y. predation on some populations of arthropods may be detectable. Baiting programs were conducted in conjunction Thus, the objective was to maintain forager popula- with destruction of individual nests in the treatment tions below 5 W/T/D in treated areas, in contrast to ares, and it was impossible to separate the effects of untreated areas. the two techniques. The extent to which results may Table 15 summarizes area-wide abatement re- be confounded will be discussed for each specific search from 1988- 1990. A total of seven trials were trial. In 1988 and 1989 nests were treated with Ficam conducted. Generally, 28 grams (one ounce) of toxic D; in 1990 they were treated with PT5 15. bait per dispenser was sufficient; only in a few in- stances was this amount depleted. The amount of Results & Discussion bait can easily be adjusted upwards in proportion to Each baiting trial represents a unique situation higher density populations; two ounces might be (Table 15); results from each trial will be discussed more appropriate where the W/T/D exceeds 10. Dur- separately. Then conclusions from the series of trials ing 1988and 1989 dispensers similar to that of Chang will be summarized. s88. HAVO.K' (1988) (See Appendix 1) were distributed in a regular -: These grid pattern (slightly irregular for Kipuka Puaulu, sites are mature native mesic forests intermixed with 1988) throughout treatment areas; numbers of bait- open grasslands. They have substantial well shaded ing applications ranged from one to five. Baits were areas, difficult to penetrate due to dense ground level put in dispensers around sunrise, left out for at least vegetation. Baiting at Kipuka Puaulu was fairly la- one day, and collected as soon as practical. bor-intensive, and despite a moderately high yel- In 1990several modifications were made. Anew lowjacket population, very little bait was taken from dispenser, consisting of a suspended open paper cup, dispensers in well-shaded areas. The forager popu- was used (Appendix 1). At HALE,based on success- lation was definitely depressed due to control meas- ful demonstrations the previous two years, a large- ures (Figure 9), but the destruction of 18 colonies scale baiting (1100 hectares = 2700 acres), without a may have been more influential than the baiting comparable untreated area, was conducted. Bait dis- program. One nest that was missed apparently over- pensers were deployed along the paved road from wintered and developed into the massive HV08903, 2275 m to 2850 melevation, with an additional 23 the only colony discovered in Kipuka. .Puaulu in 1989. dispensers along transects away from the road. For 1988: 8 evaluation, the population in the treated area (moni- These sites are mostly open areas dominated by a tored with single Yellowjacket Inn traps at 2438, mixture of scrub and grasslands, with some sections 259 1, and 2693 m elevations) was compared with the of dense old growth koa (Acacia).Due to the population upslope from the treatment atea (traps at large size (140 acres), it was very labor-intensive to 2835 and 2957 m elevation), where populations are distribute baits throughout the treatment area. The usually less dense than those around 2591 m (the passage of a hurricane one day after the initial bait center of the treatment area). application confounded the results, as populations Placement of dispensers along the mad rather dropped in both treated and untreated areas. How- than in a grid created an opportunity to investigate ever, subsequent developments indicated good suc- the effects of distance from the nearest dispenser. cess of baiting (Figure 10). No colonies were lko upslope-downslope transects were established discovered within the treatment area. However, and monitored with 10 traps per transect. For each HV08816, presumably overwintered from the pre- transect we designated the four central traps as "dis- vious year, was found slightly upslope of the treat- tant" (minimum distance to the nearest dispenser: ment area, and could easily have contributed many Leleiwi-Halemau transect, 350 m; Motley Gulch foragers to the monitored population. The colony transect, 300 m) and compared Wp/D counts with was destroyed on 1 September, which probably &- the six peripheral traps (three at each end of the pressed the forager population within the treatment transect) designated 'hear". area; this was after the positive effects of baiting had Table 15. Yellowjacket abatement trials, 1988-1990.

'hated/ Year Site Area (Acres) No. of No. Nests Bait Dates Untreated Dispensers Destroyed U 1988 Kipuka Ki 60 T 1988 Kipuka 80 58 18 29 Jul Puaulu 1 Aug 2 Aug 20 Oct Kipuka Maunaiu Kipuka 5 Aug Kulalio 23 Aug 26 Aug 29 Aug HALE2 HALEl 9 Aug 13 Aug 16 Aug 7 Sep 9 Sep Rt 1l/Kabu Desert Footprints 19 Aug 4 Oct HALE2 HALE1 19 Sep Kipuka Puaulu Namakani 2 Aug Paio been demonstrated. Forager populations were ex- 1989: HAVO: F- 11 II(a'u De- tremely low during 1989 (consistent withacatryover a:These sites are basically rocky desert with effect of abatement in 1988 into the following year), stunted scrub vegetation; the Footprints site contains while they reached moderate levels in untreated a few small sand dune kipukas of several acres with Kipuka Maunaiu. somewhat larger trees. These sites were chosen after u88: HALE::These are both Y. pns&anh populations failed to develop in open subalpine scrub habitats, although HALEl also preferred sites during 1989. The forager peak prob- includes a small ( acres) grove of introduced trees ably occurred early in the summer; by the time moni- (mostly eucalyptus and conifers). Baiting results toring was initiated the population was very high, but were good (Figute 11); probably, not all five baitings about to ctash Although forager populations in the were necessary. The two colonies destroyed were treatment ate.declined after each baiting, they also both quite small, and did not substantially confound dropped (thoughnot so dramatically) in the untreated the effect of the baiting program. area (Figure 12). This trial was not a cleat demon- 29 stration of the success of the baiting program, prob- and distant populations were not great, with all sites ably due to the confounding independent population remaining below threshold of 5 W/T/D. We interpret declines. this as evidence that baits are effective over distances 1989:The baiting pro- reaching to the most isolated monitoring trap (360 gram was repeated at the same sites used in 1988. m). Only one bait application was used, and it produced the desired population reduction (Figure 13). How- ACKNOWLEDGEMENTS ever, the effects of baiting might have been more clearly demonstrated had it been started earlier. The J. Beardsley and A. Medeh provided much of destruction of two small colonies in the treatment the impetus for this work T. Gavin was instrumental area probably had little influence on the effect of in obtaining funding. R. Alexander, D. Coon, R. baiting. ... Hart, C. Hodges, S. Jessel, A. Medeiros, W. Miyard, 1990: HAVO: N- P. Rasfeld, M. Rose, and J. Wakeley assisted in The Namakani Paio site is a small, eucalyptus-domi- gathering data. Heptyl butyrate monitoring over a nated grove (26 acres) with a fairly open understory, 10-year period was conducted by the Resource Man- surrounded by mixed desert/scrub vegetation. A por- agement Staff of HALE under direction of R. Nagata. tion of Kipuka Puaulu, similar to Namakani Paio in E Brittain kindly provided data collected by the density of vegetation (and different from the section Department of Health. R. Gillespie, B. Kumashito, used for treatment in 1988) was monitored as the and A. Samuelson assisted with identifications. untreated area. This trial was a successful demon- Valuable suggestions in the course of the work were stration of the revised baiting technique (Figure 14). provided by S. Anderson, J. Beardsley, B.K. Clapper- HALE: This was the largest scale baiting ton, L. Cuddihy, W. Gagne, F. Howarth, G. Johnston, trial attempted (2700 acres). The treatment area en- J. Kliejunas, G. Kornatsu, A. Medeh, J. Neisess, J. compassed both HALE1 and HALE2 from the pre- Rosenheim, P. Wtousek, and J. Wenz. Major logis- vious years. The logistics of baiting were simplified tical support was provided by the Research and Re- considerably by putting dispensers along the road. source Management Divisions of HAVO and HALE Results were very good; the success shown in Figure and by the Cooperative National Park Studies Unit 15 is likely a conservative estimate, because the of University of Hawaii at Manoa. The study would treatment area included areas where yellowjacket not have been possible without the support of C. forager populations are usually most dense (Table 3). Stone and staff and C. Smith. Eunding was provided The baiting seemed to kill off two active colonies primarily by the U.S. Forest Service. Support was within the treatment area (the fmt time this had been also received from the National Park Service and the observed), while the destruction of 3 other colonies Nature Conservancy. G. Kornatsu and C. Smith pro- probably had little influence on the overall effect of vided helpful suggestions on the manuscript. We baiting. apologize to many who contributed but have been The effect of distance ftom baiting stations is inadvertently left out of this enumeration. summarized in Figure 16. Differences between near LITERATURE CITED

Akre, R.D., W.B. Garnett, J.F. MacDonald, A. Oreen & P. Landolt. 1976. Behavior and colony develop- ment of hpda and Y. atmpiba (Hymenoptera: ). J. Kansas Entomol. Soc. 49(1): 63-64. - Akre, R.D., A. Greene, J.F. MacDonald, P.J. Landolt, & H.G. Davis. 1980. Yellowjackets of America North of Mexico. USDA Agriculture Handbook #552,102 pp. Akre, R.D., W.B. Hill, J.E MacDonald, & W.B. Garnett. 1975. Foraging distances of pen& &workers (Hymenoptera: Vespidae). J. Kansas Entomol. Soc. 48(1): 12-16. Akre, R.D., & J.F. MacDonald. 1986. Biology, economic importance and control of yellowjackets, pp. 353-412. In Vhson, S.B. (ed.), Economic impact and control of social insects. Praeger, N.Y. Akre, R.D., & H.C. Reed. 1981. A polygynous colony of Jkq& (Sawsure) (Hymenop- tera: Vespidae). Entomol. News 92(1): 27-3 1. Archer, M.E 1977. The weights of forager loads of (Linn.) (Hymenoptera: Vespidae) and the relationship of load weight to forager size. Ins. Soc. 24(1): 95-102. Blumenstock, D.I. & S. Price. 1972. Climates of the states: Hawaii, pp. 155-203. In Kay, E.A. (ed.),A natural history of the Hawaiian Islands. Univ. Hawaii Press, Honolulu.

Broekhuizen, V.S. and C. Hordijk 1968. Untersuchungen uber die Beute von y&p& L. (Hym., Vespidae) und ihre Abhangigkeit von der Beutetierdichte. Z. Ang. Entomol. 62: 68-77. Carpenter, J.M. 1987. Phylogenetic relationships and classification of the Vespinae (Hymenoptera: Vespi- dae). Syst. Entomol. 12: 413-43 1. Chang, V. 1988. Toxic baiting of the western yellowjacket (Hymenoptera: Vespidae) in Hawaii. J. Econ. Entomol. 81(1): 228-235. Davis, H.G., G.W. Eddy, T.P. McGovern, & M. Beroza. 1%9. Heptyl butyrate, a new synthetic attractant for yellow jackets. J. Econ. Entomol. 62: 1245. Davis, H.G., R.W. Zwick, W.M. Rogoff, T.P. Mcaovern & M. Beroza. 1973. Perimeter traps baited with synthetic lures for suppression of yellowjackets in fruit orchards. Environ. Entomol. 2569-571. Duncan, C.D. 1939. A contribution to the biology of North American vespine wasps. Stanford Univ. Publ., Biological Sciences 8( 1): 1-272. Edwards, R. 1980. Social Wasps. Rentold Ltd., East Orinstead. Ennik, F. 1973. Abatement of yellowjackets using encapsulated formulations of diazinon and rabon. J. &on. Entomol. 66(5): 1097-1098. Gagne, W.C. and C.C. Christensen. 1985. Conservation status of native terrestrial invertebrates in Hawaii, pp. 105-141. In C.P. Stone and J.M. Scott (eds.). Hawaii's terrestrial ecosystems: preservation and management. Univ. Hawaii Press, Honolulu, Hawaii. Gambino, P. 1986. Winter prey collection at a perennial colony of ydgalis (L.) (Hymenop- tera: Vespidae) Psyche 93: 33 1-34. Gambimo, P. 1990. Mark recapture studies on pemyhhqueens (Hymenoptera: Vespidae). Pan-Pac. Entomol. 66(3): 227-23 1. Gambino, P. 1991. Reproductive plasticity of Vesph pasyhub(Hymenoptera: Vespidae) on Maui and Hawaii Islands, U.S.A. New Zealand J. Zool. 18: 139-149. Garnbino, P., A.C. Medeh, and L.L. Loope. 1990. Invasion and colonization of upper elevations on East Maui (Hawaii) by (Hymenoptera: Vespidae). Ann. Entomol. Soc. Arner. 83: 1088-1095. Grant, C.D., C.J. Rogers & T.H. Lauret. 1%8. Control of ground-nesting yellowjackets with toxic baits - a five year program. J. Econ. Entomol. 6 1: 1653-1656. Harris, R.J. 1989. An entrance trap to sample foods of social wasps (Hymenoptera: Vespidae). New Zea- land J. Zd. 16: 369-371. Heinrich, B. 1984. Strategies of thennoregulation and foraging in two vespid wasps, meand &s,p.&i dgak. J. Comp. Physiol. B 154: 175-180. Howarth, F.G. 1985. Impacts of alien land arthropods and mollusks on native plants and animals in Ha- waii, pp. 149-179. In C.P. Stone and J.M. Scott (eds.). Hawaii's terrestrial ecosystems: preservation and management. Univ. Hawaii Press, Honolulu, Hawaii. Jeanne, J.L. 1980. Evolution of social behavior in the Vespidae. Ann Rev. Entomol. 25: 371-396. Keh, B., N.T. Brownfield & M.E. Person. 1968. Experimental use of bait with mirex lethal to both adult and immature Vesph (Hymenoptera: Vespidae). Calif. Vector Views 15: 115-1 18. Kleinhout, J. 1958. Het vemelen van prooien van sociale wespen. De Levende Natuur 6 1: 179- 182. Loope, L.L. & P.G. Scowcroft. 1985. Vegetation response within exclosures in Hawai'i: a review, pp. 377- 402. In C.P. Stone and J.M. Scott (eds.). Hawaii's terrestrial ecosystems: preservation and manage- ment. Univ. Hawaii Press, Honolulu, Hawaii. MacDonald, J.F., R.D. Akre, and W.B. Hill. 1974. Comparative biology and behavior of bpdaatm& . haand Y. (Hymenoptera: Vespidae). Melanderia 18: 1-66. MacDonald, J.F., R.D. Akte, and W.B.Hill. 1973. Attraction of yellowjackets (kqu,la spp.) to heptyl butyrate in Washington State (Hymenoptera: Vespidae). Environ. Entomol. 2(3): 375-379. MacDonald, J.F., R.D. Akre and R.W. Matthews. 1976. Evaluation of yellowjacket abatement in the United States. Bull. Entomol. Soc. Arner. 22(4): 397-401. Marchal, P. 18%. La reproduction et l'evolution des guepes sociales. Arch Zool. Exp. Gen. Third series (4): 1-100. May, R.M. 1981. Models for two interacting populations, pp. 78-104. In R.M. May (ed.). Theoretical Ecology, 2nd Edition. Sinauer Associates, Inc., Sunderland, Mass. Miller, C.D.F. 1%1. and distribution of Nearctic &q&, Can. Entomol, Suppl. 22: 1-52. Nakahara, L. and P. Lai. 1981. Hawaii pest report. Hawaii Department of Agriculture, Plant Pest Control Branch l(3). Ono, M., M. Sasalci & I. Okada. 1985. Mating behavior of the giant hornet, mandarinia Smith and its pheromonal regulation. Proc. of the 30th Int'l Apicultural Congr.: 255-259. Pemtt, D.C.E 1975. Factors affecting the use of rnirex-poisoned protein baits for control of European wasp (hmxsphgmxm&z$ in New Zealand. N. Zealand J. Zool. 2(4): 491-508. Plunkett, O.M., H. Moller, C. Hamilton, B.K. Clapperton, & C.D. Thomas, C.D. 1989. Overwintering colonies of German (kqdagmna,nh) and common wasps (Yespula vulnaris) (Hymenoptera: Vespi- dae) in New Zealand. New Zealand J. Zool. 16: 345-353. Reid, B.L. & J.F. MacDonald. 1986. Influence of meat texture and toxicants upon bait collection by the Oennan yellowjacket (Hymenoptera: Vespidae). J. Econ. Entomol. 79: 50-53. Rogers, C.E 1972. Field testing of 'yellow jacket stoppers" and population depletion trapping for the con- trol of ground-nesting yellowjackets. Proc. Calif. Mosq. Contr. Assoc. 40: 132-134. 32 Ross. D.R., R.H. Shukle & J.F. MacDonald. 1984. Meat extracts attractive to scavenger in east- ern North America (Hymenoptera: Vespidae). J. Econ. Entomol. 77: 637-642. Ross, K.O. 1983. Cuticular pigment changes in worker yellowjackets (Hymenoptera: Vespidae). J. N.Y. Entomol. Soc. 91(4): 394-404. Ross K.0.& R.W. Matthews. 1982. Two polygynous overwintered colonies from the southeastern U.S. (Hymenoptera: Vespidae). Fla. Entomol. 65(1): 176-184. Ross, K.O. & P.K. Viher. 1983. Reproductive plasticity in yellowjacket wasps: a polygynous, perennial colony of ksgda maculifrons. Psyche 90: 179-191. Roush, C.F. & R.D. Akre. 1978. Nesting biologies and seasonal occurrence of yelhwjackets in north- eastern Oregon forests (Hymenoptera: Vespidae). Melanderia 30: 57-94. Ruddock, J.C. & D.L. Rohe. 1968. A bait station system for controlling the ground nesting yellowjacket kkqah pmsghnh (Saussure). Calif. Vector Views 15(1): 3-6. SAS Institute, Inc. 1985. SASISTAT guide for personal computers, Version 6 Edition Cary, N.C. Spradbery, J.P. 1973. Wasps. Univ. Washington Press, Seattle. Stone, C.P. 1985. Alien animals in Hawai'i's native ecosystems: towatd controlling the adverse effects of introduced vertebrates, pp. 25 1-297. In C.P. Stone and J.M. Scott (eds.). Univ. Hawaii Press, Honolulu Hawaii. Stone, C.P. 1989. Non-native land vertebrates, pp. 88-95. In C.P. Stone and D.B.stone (eds.). Conserva- tion Biology in Hawai'i. Univ. Hawaii Press, Honolulu, Hawaii. Thomas, C.R. 1960. The European wasp (- gma&a Fab.) in New Zealand. New Zealand Dept. Sci. & Industrial Res. Information Series 27: 1-74. Tissot, A.N. & F.A. Robinson. 1954. Some unusual insect nests. Ha. Entomol. 37(2): 73-92. VuiUaume, M., J. Schwander & C. Roland. 1969. Note preliminare sur l'existeme de colonies perennes et polygynes de lkmeqda w.C.R. Acad. Sci. Paris 269: 237 1-2372. Wagner, R.E. and Reierson, D.A. 1969. Yellow jacket control by baiting. 1. Influence of toxicants and at- tractants on bait acceptance. J. Econ Entomol. 62(5): 1192-1197. Wagner, R.E. and Reierson, D.A. 1971. Yellowjacket control with a specific rnirex-protein bait. Calif. Agric. 25(4): 8-10. Williams, EX. 1927. Notes on the habits of bees and wasps of the Hawaiian Islands. h.Hawaii. Ento- mol. Soc. 6(3): 425-464. Appendix 1.

Arthropods identified to genus among Y. prey items. Asterisks indicate endemic taxa.

Order Family sitesa Araneae Lycosidae I-WQ=m* Simon HK,OT Salticidae Sandalodes spp.* PU,OT Tetragnathidae TetraPnatha SPP.* KP, PU, OT Theridiidae Theridion sp. OL Thomisidae Misumupvitellinus* (Simon) HK Musp.* HK Blattidae Allacta sirnilis (Saussure) KP, OT Gryllidae Anaxigba sp.* KP, OL LeDtomvllus SP.* KP Alydidae Ithamar hawaiiensis* (Kirkaldy) HK Flatidae Siphnh ash (Walker) OT Nabidae luisspp.* PU,HK Lepidoptera Sphingidae H&s wilsoni* (Rothschild) KP Geometridae lhpih& spp.* PU,HK Coleoptera Cerarnbycidae funebris* Sharp HK - Ig&h@s sp.* HK Coccinellidae Qlhabdominalis (Say) HK &pniamb ammgem (Guerin- HK Meneville) Curculionidae Pantomorus cervinus (Boheman) Hymenoptera Colletidae NesoDt.osoDis sp.* Apidae BpismelliferaL. Vespidae VesDula Densvlvanica (Saussure) Diptera Tachinidae Gonia hgipddi Tothill Trichowda SP.

'KEY TO SITES KP - Kipuka Puaulu (HAVO) OL - Olaa Tract + Olaa Forest Reserve (HAVO) PU - Puhimau (HAVO) OT - Other (I-IAVO) HK - Haleakala (HALE) Appendix 2. RESOURCE MANAGER'S GUIDE TO YELLOW JACKET ABATEMENT IN HAWAII'S NATIONAL PARKS

by & Parker Gambino opment and the atypical (but not uncommon) Hawai- Introduction ian variants is necessary to design an effective abate- As of 1990, the western yellowjacket ment program. It is convenient to classify colonies pat&ah (Saussure) was the only vespine wasp as annual (typical) or overwintering (atypical), al- established in the Hawaiian islands. Detailed de- though some colonies do not fit clearly into either scriptions of its biology in its native range (western pattern. Queens, males, and workers all play their North America) and in the Hawaiian Islands are role in the colony cycle, but the negative impacts on available elsewhere. Information presented here is human activities and native arthropod faunae are limited to what is pertinent for local abatement pro- almost exclusively due the activities of workers. g-. Annual Colonies: A typical colony is started in Yellowjacket abatement in Hawaii is desirable the spring by a solitary queen It takes several for two basic reasons. Theit stinging behavior can months for her to generate the fvst clutch of workers; cause pain and discomfort; for allergic individuals, a once they start foraging, the queen does not ventute sting may result in medical complications such as outside the nest. For the next few months, colony anaphylactic shock and even death. High concentra- resources go into increasing the worker population, tions of foragers may intimidate people without and the colony grows, slowly at first, and then expo- Thus, stinging. when populations are high, they can nentially. Later in the summer, colony production - +& interfere with recreational, research, and manage- shifts to rearing reproductives (new queens and ment activities in the park Fortunately, legal com- males). The worker population of the colony stabi- plications arising from stinging incidents have not lizes and then declines. The function of the repro- been a problem thus far, although this possibility ductives is to mate and disperse, and they contribute cannot be dismissed indefinitely. little to the welfare of theit natal colony, which The second rationale for yellowjacket abatement eventually declines and is abandoned. Fertilized derives from their feeding behavior; they are gener- queens hibernate individually and emerge the follow- alist predators of other arthropods. In native Hawai- ing spring to initiate new colonies. The total number ian ecosystems, predation pressure on native of wasps produced by a successful mual colony is arthropods may cause problems for the conservation usually less then 40,000 (mainland colonies are sel- of biotic resources. The survival of local populations dom even half that productive). Only a very small of some arthropod species may be seriously chal- proportion of queens from any year produce success- lenged. The total number of arthropods consumed ful colonies the following yeat. by a vigorous overwintered colony is surely in the Overwintered Colonies: Overwintered colonies millions, possibly in the tens of millions. start out as annual colonies, but instead of dying out in their fvst year, they continue into a second year. Biology of Y, b Although there is no documented case of a colony Since its invasion of the Hawaiian Islands, X lasting thtee or more years, there are no obvious -'s -'s adaptations to local conditions have consttaints against it, and this possibility cannot be resulted in variations of the life cycle that complicate excluded. Whereas annual colonies phase out both management efforts. Some understanding of worker production when reproductives are reated, the typical pattern of X pens~lvanicacolony devel- overwintering colonies produce workers continu- dictable than at HAVO. Prime yellowjacket habitat ously through the reproductive phase. Instead of is consistently centered near the 8500 ft (2591 m) dying out, the colony remains active, acquiring new bend in the road. Wethere is annual variation in queens to replace the original foundress. The colony the magnitude of yellowjacket populations, they are then has several or many egg-laying queens and an well synchronized, peaking regularly from late Au- abundant worker force as it entm its second year. gust to mid-October. Colony production revert.to workers, at a rate orders of magnitude greater than annual colonies just get- Population Monitoring ting started. Additional reproductives may be gener- Monitoring of pest populations is a basic tenet of ated during the second season. The largest analyzed integrated pest management. The attraction of X colony from Hawaii (Kipuka Puaulu) had about gm&anim foragers to the synthetic chemical hep- 560,000 small cells and 40,000 large (queen) cells, tyl butyrate (which is a non-toxic substance some- and produced at least one million wasps. times used as a food additive) provides a convenient The general pattern of yellowjacket forager mechanism for measuring populations. populations in Hawaii resembles that in its native The trap chosen for monitoring is the Seabright range; worker activity is low (or more likely, com- Yellowjacket Inn, baited with a heptyl butyrate wick pletely absent) during the winter and early spring, To load the trap, pipette 0.4 ml heptyl butyrate into a rises during the summer, peaks during the late sum- 1.5 ml polyethylene microfuge tubule. Add a wick, mer and fall, and then crashes. The presence of which can be made from a rolled up piece of paper foraging workers (and also large numbers of males) towel. The wick should not extend so far as to during the early spring is a good indication of over- interfere with closing the tubule. For each area to be wintering colonies. monitored, it is recommended that 5 traps be set out, Weather exerts a dominant influence on yel- separated by at least 25 meters, but preferably dis- lowjacket phenology and abundance. Highly local- tributed evenly throughout the area. Traps should be + r ized variations in yellowjacket populations can be suspended between 1-2 meters above ground, and not traced to the extremely variable and sometimes un- placed in dense shade. Metal shower curtain rings predictable Hawaiian weather patterns. Thus, at Ha- work very well to suspend traps. Heptyl butyrate waii Volcanoes National Park (HAVO), forager tubules can be carried closed into the field, and populations may peak during July in the western Kau opened just prior to replacing the old tubule + wick Desert, while further east, the peak does not occur Numbers of yellowjackets captured are counted until September or October. During the same year, when the attractant is replaced. An interval of one yellowjackets may be abundant in dry localities and week is recommended; at intervals longer than two neatly absent from wetter areas, especially if the weeks, counts are less valid because the heptyl period of nest initiation during the spring was par- butyrate evaporates and attraction decreases. Popu- ticularly wet. The dominant influence at HAVO lation levels are expressed as wasps per trap per day, seems to be the rain. calculated by adding the number of wasps for all 5 At Haleakala National Park (HALE)there is a traps and then dividing by 5 times the monitoring fairly steep elevational gradient on the northwestern interval. Forager density can then be plotted against slope of Haleakala volcano, and a moderate east-west time, demonstrating the seasonal nature of popula- moisture gradient. The mid-Pacific summer tem- tion fluctuations. Populations can also be compared perature inversion commonly produces a dense layer between different monitoring areas. of fog that is thickest below the park boundary, but Two factors have been found to interfere with also extends up to the lower elevations within the accurate counts of trapped yellowjackets. Ants park The fog apparently inhibits development of sometimes enter the trap and remove dead wasps. ' yellowjacket colonies. Population trends on the This problem is solved by suspending the trap + northwest slope of Haleakala volcano are more pre- curtain ring from a second curtain ring (or a bent paper clip will also do) that has been coated with to avoid spreading the attractant around on your Tacktrap or some other sticky substance. Tacktrap clothes, body, or the interior of your vehicle. may need to be reapplied after several months. It is also necessary to ensure that the suspending ring or Spot btmentof Colonies wire is the only ant access to the trap, If leaves or When an active nest is located it can be treated twigs touch the sides of the trap, ants will use these directly. Wo pesticide treatments have been ap- as bridges, thus avoiding the sticky barrier. proved for NPS use: Ficam D dust and Whitmire The second problem is due to yellowjackets PT5 15 WaspFreeze II. Successful treatment of a nest themselves. Some will enter the trap seeking to requires that the insecticide be applied into the nest harvest the dead wasps as food. They cut up the dead entrance and that good coverage is achieved. Most wasps into smaller fragments, and some can even of the time a single application (1 oz of Ficarn D successfully exit the trap to carry wasp fragments poured into the nest entrance; one can of WaspFreeze back to their nest. This both lowers the accuracy of sprayed into the entrance) will work When the the trap catch and makes counting more =cult. entrance is irregular, or upward-sloping, it is very When there are wasp fragments instead of intact difficult to get the pesticide in to where it will do the wasps in the trap, it is best to count abdomens (or most good, in such cases, the cumbersome Ficam more properly, gasters), as these are the least desir- duster (a gadget that uses a piston to force dust into able pieces for foragers to remove. the nest entrance) may give better results. It is pref- Additional notes of caution: erable to treat a colony at night and to cover the 1. Sometimes there are live yellowjackets in the entrance with soil after the pesticide has been ap- trap when you wish to count them. There are two plied. This is not logistically feasible for most re- options for dealing with this. The trap can be re- source managers; daytime treatments are usually moved, tape placed over the openings, and held until satisfactory. the wasps die (a few hours in a freezer will do it). Toxic Baiting Alternatively, the trap can be quickly opened and placed on the ground, allowing the wasps to escape The toxic baiting strategy takes advantage of (they should be counted as they fly out). The latter yellowjacket feeding behavior to deliver pesticide to method is more convenient; it is slight more hazard- colonies within the baiting area without locating the ous, but can be done safely with a sensible degree of nests. Foraging workers collect the toxic bait from caution. Live wasps in the trap are much more eager bait dispensers and carry it back to the colony, where to escape than to aggressively attack it is distributed among colony members. During 2. Heptyl butyrate can quickly attract horde. of times of colony growth, the greatest nutrient demand yellowjackets, so all effort should be made to mini- is for protein; thus protein-based baits are most useful mize contamination of any extraneous surfaces. Pre- in attacking populations that are increasing. Most of pare all tubules in advance, and carry them in a closed the protein is fed to immatures, so a successful toxic container. Styrofoam or polystyrene containers are bait program has an immediate effect by killing fora- inappropriate; heptyl butyrate diisolves these plas- gers that collect it, and a delayed effect by killing tics. It is best to test the container to make sure it larvae that might become foragers several weeks doesn't dissolve before carrying it into the field and after the baiting episode. being surprised. Try to minimize exposure of heptyl A number of protein-based baits have been butyrate where it is not needed in the field (i.e. don't evaluated by other researchers, including some that are have open vials in the back of your truck, or inside a recommended elsewhere in the literature. The used vehicle with the windows open). Despite your best criteria in Hawaii included local availability, efforts, heptyl butyrate will end up on your fingers. easy storage, attractiveness to X (they Carry paper towels and wipe your fingers constantly come to it), acceptability to Y. pensvlvanica (they carry it away), and acceptability of the bait/toxicant mixture. The best bait is canned Swanson's Premium along upper Mauna Loa Strip Road. Less satisfac- Chunk White and Dark Chicken in Water, available tory results can be anticipated in shaded and densely in 5 ounce cans (larger sizes are only sometimes forested areas, such as Olaa Tract and much of available). Somewhat less effective substitutes are Kipuka Puaulu. While this is somewhat discourag- (canned) Hormel Turkey and Hormel Ham. ing from the standpoint of using a baiting program to The toxicant used in the baiting program is protect native arthropods in some of the most biologi- Knoxout 2FM, a microencapsulated flowable formu- cally diverse forested areas, it also suggests that the lation of diazinon. This is the only toxicant regis- sub-canopy layers may offer some degree of refuge tered for this use in Hawaii; although there is mom from yellowjacket predation. for improvement, there are currently no legal alter- The density and arrangement of bait dispensers natives. This situation may change in the future, as in a treatment area depends on several factors, such at least two mainland research teams are making as the distribution of open sunny spots, numbers of progress on developing alternative bait/toxicant sys- foragers present at the time of baiting, and manpower tems. The main problem with Knoxout is that as it and logistical considerations. A density of one dis- ages, the microcapsules break down, releasing the penser per acre (which corresponds to a grid arrange- toxicant. Free diazinon is foul-smelling to humans ment with dispensers at 64 m [=209 ft] intervals) and makes baits less acceptable to yellowjacket fora- provides good results. If optimum baiting conditions gers. Experience on Hawaii shows that local sup- are met, one dispenser per two acres is probably plies of Knoxout may vary in age and quality; it is satisfactory. hard to know how long the container has been sitting Before baits can be put out, the treatment area in a hot Kona warehouse. In two and a half years we needs to be surveyed and bait dispensers put out. have had one unsatisfactory batch and one good This is more labor intensive than the actual distribu- batch tion of baits. In accordance with the previous discus- Yellowjacket foragers searching for food use sion of yellowjacket foraging habits, the best location their senses of smell and sight. To locate dead meat for a bait dispenser is in a conspicuous sunny spot. (such as canned chicken), foragers follow an odor In the placement of dispenser, it is preferable to plume to the general atea, and then rely increasingly deviate slightly from a strict grid (or other) arrange- on vision to find the specific source. To hunt live ment than to put one in a spot where it will not be arthropods, vision probably plays the primary role well attended. Each bait dispenser should have a thtoughout the entire process. Yellowjackets can tubule of heptyl butyrate attractant (similar to that remember the location of a site where they have used in the monitoring program), which draws in successfully collected food, and returned to the same yellowjackets and increases the amount of bait taken. site to continue foraging. Thus, foragers from a Do not mb the attractant in to&b& colony do not become uniformly distributed locally; Dispensers should be blatantly visible, marked with basically, they either follow an odor plume to a dead conspicuous flagging if necessary, so that they can meat source, or hunt in sunny well-lit areas. Condi- be easily found on baiting day; it is frustrating and tions at ground level in densely forested areas inter- counterproductive to waste time playing hide-and- fere with both senses used in seeking food; dense seek with bait dispensers. vegetation may prevent the formation of an odor We used two different types of dispensers; each plume that can be followed to its source, and yel- had its shortcomings. Future yellowjacket manage- lowjacket vision is poor at low light densities. Fora- ment effort should include the design of a better bait ". gers concentrate their attentions in open areas or the dispenser. Initially we used a dispenser modified tops of tree canopies; consequently, baiting programs from the version designed by Chang (Journal of do not work equally well in all these areas. The best Economic Entomology 81:228-235, 1988). It con- results are obtained in fairly open areas, such as the sisted of a 4-inch diameter white PVC pipe segment northwest slope of Haleakala or Kipuka Kulalio 12 inches long (Figure 1). To hang the dispenser, nylon cord was passed through 4 holes drilled at each When mixing and distributing baits, precautions end of the dispenser. A slip knot was tied in the against pesticide exposure need to be taken. In the middle of the cord and used to suspend it from a metal spectrum of pesticide hazards, Knoxout is relatively shower curtain ring hung from the vegetation. A bent safe for humans. Recommended gear for mixing is metal paper clip attached to one of the drilled holes a respirator with cartridges and filters appropriate for held the heptyl butyrate tubule + wick; the free end pesticides, goggles, and rubber gloves; for distribut- of the paper clip was passed through the tubule cap. ing the mixed baits, just rubber gloves. Bait was apportioned into 8-cm plastic weigh boats. To mix bait, fvst determine how much bait is to This dispenser proved to be cumbersome to lug be made. Empty enough cans of chicken into a bowl, around in the field, and it also may confuse the and break up the chunks by mixing it thoroughly yellowjackets, making it difficult for them to locate before adding toxicant. Knoxout is added at the rate the bait. Another problem with this style of dispenser of 4.2 ml per five ounce can of chicken (or 5 ml per is that the weigh boat containing the bait is unstable; 6-02can of other baits). Stir it thoroughly to ensure a strong gust of wind can tip it over, causing bait to even distributionof Knoxout in the chicken. Capped fall to the ground where non-target organisms may plastic l-0~cups are the easiest way to carry baits feed on it. into the field. After cups are filled and capped, they A simpler dispenser was designed and used at can be carried in a latge covered plastic tub (Rubber- HALE during 1990. It consisted of a 3-ounce waxed maid or similar style). paper cup, with wires (28-gauge or thicker galva- Results have suggested that foragers removing nized steel wite works well) attached at three points toxic bait die after making several trips, and the along the rim Figure 2). It was suspended from numbers of foragers at baits drop off after several vegetation by wrapping the free end of the wire hours. Our experience in Hawaii has shown that one around a branch. Bait was apportioned into l-ounce ounce of bait per dispenser is usually sufficient. For plastic cups; these fit neatly into the paper cup, did heavy populations (10 per trap per day), 2 ounces per not become dislodged even in windy areas, and left dispenser may be recommended. The State of Ha- enough room in the paper cup to also put in a heptyl waii Knoxout registration information suggests 3 butpte tubule + wick At the end of the baiting ounces per dispenser, but I think that at a rate of 1 session, the plastic cup could be easily removed and dispenser per acre, this could only be justified by a the dispenser left in place. This dispenser was less total inundation of yellowjackets. durable than the PVC model, but was cheaper and Baits left in the field longer than 2 days probably easier to build and transport. don't affect yellowjackets very much, and increase Baits should be distributed early in the day on the risk of exposure of non-target organisms, so after sunny days (good luck predicting weather!). We each baiting a followup session should be planned to have delayed baiting until after populations reach a remove baits. threshold of 5 per trap per day, and recommend this For any toxicant-based pest control program as the action level until better evidence suggests it be there is legitimate concern about exposure of non-tar- raised or lowered. Thus the monitoring program get organisms. The Knoxout,/protein bait in dipens- provides information for scheduling the baiting. em is a potential site for non-target organism Baiting is most successful when applied while popu- exposure. We made the following observations, ad- lations are still rising. This is the time when foragers aptations, and compromises in the baiting program are most hungry for protein, and also when the popu- with regatd to the following groups of organisms. .r. lation peak can be most effectively truncated. As Invertebrates: Several types of muscoid £lies with any preemptive preventative program, it is hard were attracted to both fresh and aged baits; to document effectiveness without a matched un- staphylinid beetles were attracted to baits older than treated area for comparison. several days. Both groups suffered some mortality, which we considered unfortunate but acceptable. Birds: We saw no evidence that birds were at- Humans: Warnings were written onto the out- tracted to or feeding on baits at either HAVO or side of bait dispensem. The PVC units had WARN- HALE. ING - Pesticide - Do Not Disturbwstenciled in red. Mammah: We did not see mammals at any of The small paper cups did not have room for such an the bait stations, but suspected that they would be elaborate message, so we just marked them "POI- interested. To reduce the risk of exposure we tried to SONw. For applicators mixing bait, standard per- suspend stations above ground level on vegetation sonal protection measures were taken (respirator, where it would be difficult for rats, mongooses, cats, goggles, gloves). People distributing baits into dis- or dogs to reach it. At one point baits were put out in pensers wore rubber gloves. Haleakala National the vicinity of Volcano House, where the manager Park is in need of a proper facility to mix pesticides was concerned about the safety of the house cats. and clean soiled equipment; construction of such a PVC dispensers were used, slots were sawed at each facility is recommended. end of the dispenser, and a 3" x 4-112" rectangle of 112" hardware cloth was inserted, making it quite tamper-proof. SOURCES FOR MATERIAL USED IN Densvlvsnica MANAGEMENT PROGRAM

MONITORING PROGRAM 7. Baits 1. Yellowjacket Inn Traps Swanson's Premium White & Dark Chunk Chicken in Water (canned) Seabright Enterprises (Suboptimal substitutions are [canned] Hormel 4026 Harlan St. Ham or Hormel Turkey) Emeryville, CA 94608 Available locally (KTA in Hilo) 415-655-3 126 8. PVC pipe for dispensers 2. Heptyl butyrate 4" gravity drain pipe (thinner walls, cheaper than Aldrich Chemical Co. Inc. regular PVC pipe) 1101 West St. Paul Ave. Island Supply Milwaukee, WI 53233 30 Halekauila 1-800-227-2463 Hilo, HI 3. Dropper/Pipettor 935-288 1 Nothing special needed; 9. Paper Cups calibrate ordinary eyedropper to 0.4 rnl. S303 - 3-02., w/o lids ("Reflections" variety 4. 1.5 rnl Microfuge tubules best) Hawaii Paper Products, Inc. Brinkmann Instruments, Inc. 167 Makaala St. Cantiague Road Hilo, HI Westbury, NY 11590 935-9796 5' 1-800-645-3050 10. Plastic Cups 5. Metal shower curtain rings us1 - l-ozsize Available locally (Woolworth's, etc.) LUS 1 - lids for US 1 6. Knoxout 2FM Hawaii Paper Products, Inc. Van Waters & Rogers OTHER 3160 Ualena St. Honolulu, HI 968 19 836-136 1 Schedule of availability in Hawaii not certain. Company representative Jonathan Berger 3 14- 225-5371 Bremer Chemical Co. 808-244-3761 Van Water & Rogers 808-836-1361 1981 1983 1984 1985 1980 YEAR Figure 1. pensylvanica seasonality at Haleaksllr National Park, 1981-1 988 50 - A Kipuka KuWi 10 - 6-

lW B. Klpuka Maunaiu 16 - A

C. Kipuka

10 -

D. Kipuka Puaulu SO - 20 - 10 -

YJ JALOND JfYAYJ JAOOWO JFYAYJJA~OWD lee4 I loss 1986 E. Kilauoa: Hemdqumrters 30 - 2D - 10 -

YJ JAOOWDJfYAYJJAbOWDJfYAYJJA8OWD 1984 I 1985 1QM F. Kilauecl: Fiesoarch 6-

YJJAOOYDJFYAYJJA ObWD I I 1sw G. Ola'a Tract

YJJA*OW, IYAY+JA*O*D

H. Ka'u Desert loas IOU 10 - I

I. Kipuka Nene 10 -

Figure 2. Vespula pensylvanica seasonality at

Hawaii Volcanoes National Park, 1984-1989. 100 - 80 - A Haleakala - Traps

a, a, 60 - 3 - 40- B Hawaii - Traps

Month

Figure 3. Vespula pensylvanica queen capture data. Figure 4. Yellowjacket prey sampling trap. Base (A) with attached T-unit (C) is affixed to ground with opening (B) over nest entrance tunnel and top plug (E) inserted into T-unit. Wasp traffic flows through elbow (D). To sample, top plug is removed, and elbow is replaced with trap chamber (F), which has removeable proximal plug (G) and removable distal inward facing screen cone (H). Time

Figure 5. Foraging rates at three Vespula pensylvanica colonies at Puhimau, 24 November 1989. Time

Figure 6. Foraging success (prey per wasp) at three Vespula pensylvanica colonies at Puhimau 24 November 1089. I 0: II 11111111111~~~~~1~ 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Log ((Mean WaspsTTraplDay) + 1)

Figure 7. Effect of Vespula pensylvanica predation on ohia arthropod abundance - weights.

Kokoolau A Hilina Pali 0 SITES Puhimau * Lua Manu Log ((Mean Wasps/Trap/Day) + 1)

Figure 8. Effect of Vespula pensylvanica predation on ohia arthropod abundance - numbers.

Small Large Kokoolau A Puhimau >t< SITES - Hilina Pali 0 LuaManu 0 m Figure 9. Yellowjacket abatement: HAVO, 1988

Untreated: Kipuka Ki ------.

Treated: Kipuka Puaulu Baiting Colony Treatments Figure 10. Yellowjacket abatement: HAVO, 1988 Untreated: Kipuka Maunaiu ------Treated: Kipuka Kulalio Baiting t: Colony Treatments 7 Figure 11. Yellowjacket abatement: HALE, 1988 Untreated: HALE2 ...... Treated: HALE1 Baiting Figure 12. Yellowjacket abatement: HAVO, 1989 Untreated: Rt 11/@'u ------Treated: Footprints Baiting J B

-

-. -

JUN I JUL AUG I SEP I OCT I NOV I DEC I

Figure 13. Yellowjacket abatement: HALE, 1989 Untreated: HALE2 ...... Treated: HALE1 Baiting Figure 14. Yellowjacket abatement: WO,1990 Untreated: Kipuka Puaulu ------Treated: Narnakani Paio Baiting Figure 15. Yellowjacket abatement: HALE, 1990 Untreated: HALE Upslope ------Treated: HALE1 + HALE2 Baiting Motley Gulch - 8500' T~UISOC~ B

Figure 16. Effect of distance from toxic baits: HALE, 1990. Near traps Figure 1. Toxic Bait Dispenser: PVC pipe, after Chang, 1988 Figure 2. Toxic Bait Dispenser - 3-oz. paper cup