Open access journal: http://periodicos.uefs.br/ojs/index.php/sociobiology ISSN: 0361-6525 DOI: 10.13102/sociobiology.v66i2.3772Sociobiology 66(2): 227-238 (June, 2019) Effects of Fipronil on Non-target Ants and Other Invertebrates in a Program for Eradication of the Argentine Ant, Linepithema humile Introduction Biological invasion is a global environmental problem that harms biodiversity and ecosystem function (Clavero & Garcia-Berthou, 2005; Mack et al., 2000). Ants are some of the most successful invasive taxa in the world, with invasive ant species having become established on almost every continent (Suarez et al., 2010). The success of these species is related to a suite of characteristics that favor interactions with humans (Hoffmann et al., 2016; Holway et al., 2002; McGlynn, 1999). These tramp species cause serious harm to the environment, agricultural productivity, human health, and the economy (Holway et al., 2002; Williams, 1994), as Abstract Pesticides are frequently used to eradicate invasive ant species, but pose ecological harm. Previous studies assessed non-target effects only in terms of the increase or decrease of abundance or species richness after pesticide applications. Positive effects of the release from pressure caused by invasive ant species have not been considered so far. To more accurately assess pesticide effects in the field, the non- target effects of pesticides should be considered separately from the positive effects of such releases. Here, we used monitoring data of ants and other invertebrates collected in a program for the eradication of the Argentine ant, Linepithema humile (Mayr), using fipronil. First, we separately assessed the effects of L. humile abundance and fipronil exposure on non-target ants and other invertebrates using generalized linear models. The abundance of L. humile and the number of pesticide treatments were negatively associated with the total number of non-target individuals and taxonomic richness. We also noted negative relationships between the number of individuals of some ant species and other invertebrate taxonomic groups. The L. humile × pesticide interaction was significant, suggesting that the abundance of L. humile affected the level of impact of pesticide treatment on non-target fauna. Second, we evaluated the dynamics of non-target ant communities for 3 years using principal response curve analyses. Non-target ant communities treated with fipronil continuously for 3 years recovered little, whereas those treated for 1 year recovered to the level of the untreated and non-invaded environment. Sociobiology An international journal on social insects Y Sakamoto1, TI Hayashi1, MN Inoue2, H Ohnishi1, T Kishimoto3, K Goka1 Article History Edited by Gilberto M. M. Santos, UEFS, Brazil Received 08 October 2018 Initial acceptance 15 May 2019 Final acceptance 15 May 2019 Publication date 20 August 2019 Keywords Invasive alien species; pesticide impact; total number of individuals; species richness; community structure. Corresponding author Yoshiko Sakamoto National Institute for Environmental Studies 16-2 Onogawa, Tsukuba Ibaraki 305-0053, Japan. E-Mail: sakamoto.yoshiko@nies.go.jp reflected by the fact that five ant species are listed among the world’s 100 worst invasive alien species (IUCN ISSG, 2013). Invasive ants are typically controlled with pesticides, such as in bait carriers (Rabitsch, 2011; Williams, 1994). Pesticides have been used successfully in dozens of eradication programs targeting ant species, such as the little fire ant, Wasmannia auropunctata, on Santa Fe Island in the Galápagos (Abedrabbo, 1994; Causton et al., 2005), the African big-headed ant, Pheidole megacephala, and the tropical fire ant, Solenopsis geminata, within Kakadu National Park, Australia (Hoffmann & O’Connor, 2004), and the Argentine ant, Linepithema humile, on landfill islands in Japan (Sakamoto et al., 2017). However, the use of pesticides also harms non-target species (Pisa et 1 - National Institute for Environmental Studies, Tsukuba, Ibaraki 305-0053, Japan 2 - Tokyo University of Agriculture and Technology, Fuchu, Tokyo 183-8509 Japan 3 - Museum of Natural and Environmental History, Shizuoka, Suruga-Ku, Shizuoka 422-8017, Japan RESEARCH ARTICLE - ANTS Y Sakamoto et al. – Non-target effects of pesticide in invasive ant eradication program228 al., 2015; Prasifka et al., 2005), as toxic baits can attract non- target ants and other arthropods. Fipronil, hydramethylnon, pyriproxyfen, and methoprene, commonly used in invasive ant eradication programs, pose risks (Hoffmann et al., 2016). The impacts of these pesticides on non-target species in the field have been assessed only in terms of increases or decreases in the abundance or taxonomic richness of non- target species after invasive ant control. Plentovich et al. (2010) reported that hydramethylnon can be used to control S. geminata and Tetramorium bicarinatum but also noted its negative effects on non-target ants, cockroaches, and crickets. By contrast, Hoffmann (2010) documented the eradication of a small population of P. megacephala using hydramethylnon and the recovery of native ant abundance and species richness within the treated area. Inoue et al. (2015) reported the short- term recovery of non-target communities after application of fipronil to control L. humile and observed no non-target effects. However, the positive effects of release from pressure caused by invasive ant species have not been considered in previous research. To more accurately evaluate the effects of pesticides in the field, the non-target effects of pesticides must be considered separately from the positive effects of such releases. Linepithema humile, a native of South America, is one of the most significant pest ant species worldwide (Passera, 1994). In Japan, it was first discovered in 1993 (Sugiyama, 2000) and has since spread to 12 prefectures (National Institute of Environmental Studies, 2014). Since 2011, our group has conducted an eradication program using toxic baits containing fipronil (Inoue et al., 2015), and we have successfully eradicated two Tokyo populations on landfill islands (Sakamoto et al., 2017). The aim of the eradication program is to protect the indigenous invertebrate communities from the invasive alien ant species. Thus, non-target effects cannot be ignored when the method is applied to delicate natural areas. In this study, we evaluated the non-target effects of applying fipronil to eradicate L. humile in the two Tokyo populations. First, we separately assessed the effects of the pesticide and L. humile on non-target ant and non-ant invertebrates using generalized linear models (GLMs). Next, we used principal response curve (PRC) analyses to evaluate the dynamics of the non-target communities after successful eradication and cessation of the pesticide applications. Our findings will be useful for minimizing the risks to indigenous fauna as the eradication program moves to other areas. Materials and methods Study sites The Tokai site is on a landfill island 370 m west of the Oi Container Terminal, one of the largest international shipping ports in Japan, where 8.5 ha was invaded by L. humile (Fig 1a). The Jonan site is on the landfill island Jonan-jima, 1750 m southwest of the terminal, where 16 ha was invaded (Fig 1a). Pesticide Fipronil is a phenyl pyrazole insecticide and a potent disrupter of the arthropod central nervous system via interference through the chloride channel regulated by γ-aminobutyric acid (Rhône Poulenc, 1996). Fipronil acts slowly, allowing the pesticide in baits to be transferred from insect to insect (including queen and brood among social insects) by trophallaxis or contact (Vail et al., 2003), resulting in reproductive inhibition in colonies. Fipronil is effective at controlling invasive ant species, especially L. humile (Klotz et al., 2007). Pesticide treatment The eradication program began in April 2011 (Inoue et al., 2015). To evaluate the effects of the program, we established three monitoring plots each at Tokai (Plots i–iii; Fig 1b) and Jonan (Plots I–III; Fig 1c). We applied paste baits and sprays once a month (Fig 2). The paste bait, Aruzenchin Ari Ultra Sugoto-taiji (50 mg L–1 fipronil; Fumakilla, Ltd., Hiroshima, Japan), was placed every 5 to 10 m along the streets and buildings. The bait was applied in a given month only if L. humile had been found in the same plot at any time during the previous 6 months. If we found brood or queens in vegetation or under pavement during bait application, we sprayed them with a solution of 50 mg L–1 fipronil (Aruzenchin Ari Sugoto-taij Ekizai, Fumakilla, Ltd.). The estimated total rate of active ingredient applied was 137 mg/ha at Tokai and 1045 mg/ha at Jonan over the 3 years. Plot I at Jonan was not treated in the first year for comparison with treated areas. Fig 1. Maps of (a) Tokyo Bay area and monitoring plots at (b) Tokai and (c) Jonan. Solid line indicates transects in non-invaded plots (n, N); dashed line indicates transects in plots invaded by Linepithema humile (i–iii, I–III). Squares indicate locations of sticky traps. Sociobiology 66(2): 227-238 (June, 2019) 229 Sampling of ant species and non-ant invertebrates To monitor the abundance of ant species and non- ant invertebrates, sticky traps (8.8 cm × 19.5 cm × 2.2 cm; Monitoring PP Trap #J, Kankyokiki Co., Ltd., Osaka, Japan) were placed every 50 m or so along the perimeter of invaded plots (Plots i–iii and I–III) and non-invaded, untreated plots (Plot n in Tokai and Plot N in Jonan; Fig 1b, c). The traps were laid once a month from April 2011 to March 2014 in invaded plots and from April (Jonan) or May (Tokai) 2013 to March 2014 in non-invaded plots and were collected after 3 days. Up to 108 trap points were set per month. Captured ants and non-ant invertebrates were then identified and counted in the laboratory. Ants were identified to species. The other invertebrates were identified to order, except for Myriapoda and land snails, and Coleoptera were identified to superfamily because of the variety of beetle feeding habits. We define all species except for L. humile as “non-target”. Numerical and statistical analyses The effects of L. humile invasion and pesticide application on ants and non-ant invertebrates were statistically examined by two approaches. All analyses were conducted in R v. 3.1.1 software (R Development Core Team, 2013). First, we used GLMs to examine the relationship of L. humile and pesticide treatment, and their interaction, with the total number of individuals and species or taxonomic richness (number of species or taxonomic groups) of non- targets captured by each trap (McCullagh & Nelder, 1989). We created models in which the response variables were the log10(x + 1)-transformed total number of individuals of non- target ant species or invertebrate taxonomic groups per trap or the integral number of non-target ant species or invertebrate taxonomic groups per trap. These models assumed a Gaussian distribution in the response variable and used an identity-link function. The explanatory variables were the log10(number of L. humile per trap), number of pesticide treatments in the past 6 months, their interaction, and site (dummy variable). We were not interested in seasonal change, and so we selected the month with the largest number of individuals to avoid the effect of season. We therefore used the datasets of August 2011 and August 2013 in invaded plots (Plots i–iii and I–III) and that of August 2013 in non-invaded plots (Plots n and N). We previously confirmed that the number of L. humile workers was not correlated with the number of pesticide treatments (R2 = 0.06). We also analyzed the relationships of the explanatory variables with the number of individuals of each non-target ant species or each invertebrate taxonomic group because those explanatory variables were associated with the total number of individuals and species richness in the above analyses. Zero-inflated Poisson regression models with the function zeroinfl from the pscl package (Jackman, 2017) were used to analyze the relationships, because count data of the number of individuals of each species often include many zero observations. The explanatory variables and dataset were the same as above. We then tested whether the zero-inflated Poisson regression model fit the data better than an ordinary Poisson regression model by applying the Vuong test (using the function vuong from the pscl package). We do not present results that could not be calculated owing to small sample sizes of species or taxonomic groups. We did not use Bonferroni’s correction for multiple analyses because this would inflate the likelihood of a type II error. Instead, we used p < 0.025 for significance to decrease the likelihood of a type I error. Second, to analyze the temporal dynamics of ant and non-ant invertebrate communities under pesticide treatment, we conducted PRC analyses (Van den Brink & Ter Braak, 1999) using the vegan package (Oksanen, 2013) of R. The PRC method, which is based on the redundancy analysis ordination technique, can compare the temporal dynamics of treated communities with an arbitrarily prescribed “control” community (Van den Brink & Ter Braak, 1999). We performed the analyses of non-target ant and invertebrate community dynamics in Plots i and ii in Tokai from April 2011 to March 2014 with data from Plot n (never invaded) as a control. In Plot i the pesticide was discontinued after about 1 year, whereas in Plot ii it was used for almost 3 years (Fig 2). Species abundance data were ln (10x + 1)-transformed to down- Fig 2. Presence of Linepithema humile and fipronil pesticide use history in each plot at Tokai and Jonan over 3 years. Y Sakamoto et al. – Non-target effects of pesticide in invasive ant eradication program230 weight high abundance values (Lepš & Šmilauer, 2003). The significance of the overall treatment effect was tested using 1000 permutations and the first eigenvalue. The resulting PRC diagram displays the regression coefficient (Cdt, left axis) of the first principal component in the community pattern at each site d at each time t compared with the control, whose Cdt is always zero by definition. An advantage of PRC analysis is that it can detect taxon-level effects. The right axis indicates the species (or taxon) weight (bk). For a quantitative evaluation of PRC, the quotient exp (Cdt × bk) can be calculated for each species k at each site and each time. If the quotient is positive, species k is more abundant in the community than in the control. If it is negative, species k is less abundant. Therefore, species k is more abundant if bk is on the same side of Cdt on the vertical axis and is less abundant if bk is on the opposite side of Cdt. The greater the value of the quotient, the more different the abundance of species k is between treatment and control. Results The fauna Table 1 shows the total numbers of ants and non-ant invertebrates caught by traps. In total, we collected 51,307 ants belonging to 35 species, including L. humile, and 41,324 non-target invertebrates. The last observations of L. humile were in December 2012 in Tokai and December 2013 in Jonan. In the treated plots (Plots i, ii, iii, I, II, and III), L. humile was eradicated by the pesticide, as demonstrated with a statistical model (Sakamoto et al., 2017). Although the initial density of L. humile and number of fipronil treatments differed among plots (Fig 2), the number of non-target ant individuals also decreased in the first year but started to recover after fipronil treatment ceased (Fig 3). Table 1. Total numbers of each ant species and non-ant invertebrate taxonomic groups collected in monitoring traps at the Tokai and Jonan sites. Taxa Total Tokai plots1 Jonan plots1 i ii iii n I II III N Ant species Dolichoderinae Linepithema humile 18628 74 1372 21 0 14626 218 2317 0 Ochetellus glaber 330 237 63 0 15 2 1 12 0 Technomyrmex gibbosus 8 0 5 0 0 2 0 1 0 Formicinae Camponotus japonicus 562 122 85 88 241 12 3 11 0 Camponotus vitiosus 212 105 26 17 3 26 6 21 8 Formica japonica 1750 546 227 137 180 13 5 216 426 Lasius japonicus 50 6 9 2 1 9 7 4 12 Lasius fuji 1 0 0 0 0 1 0 0 0 Lasius productus 13 0 1 0 0 0 12 0 0 Lasius sakagamii 18 4 0 7 0 4 1 2 0 Lasius umbratus 1 0 0 0 1 0 0 0 0 Nylanderia amia 284 108 3 37 1 17 3 2 113 Nylanderia flavipes 35 5 1 8 10 5 6 0 0 Paraparatrechina sakurae 1911 11 465 471 47 311 259 339 8 Paratrechina longicornis 8 0 8 0 0 0 0 0 0 Ponerinae Brachyponera chinensis 1565 320 16 234 12 10 914 39 20 Hypoponera opaciceps 1 0 0 0 0 0 0 1 0 Myrmicinae Aphaenogaster osimensis 2 2 0 0 0 0 0 0 0 Crematogaster matsumurai 2145 685 31 1185 156 4 6 38 40 Crematogaster osakensis 84 0 2 0 66 13 3 0 0 Crematogaster teranishii 5 0 0 0 5 0 0 0 0 Crematogaster vagula 7 0 0 0 7 0 0 0 0 Myrmica kotokui 7 1 6 0 0 0 0 0 0 Monomorium chinense 26 5 21 0 0 0 0 0 0 Pheidole indica 1 1 0 0 0 0 0 0 0 Sociobiology 66(2): 227-238 (June, 2019) 231 Table 1. Total numbers of each ant species and non-ant invertebrate taxonomic groups collected in monitoring traps at the Tokai and Jonan sites. (Continuation) Pheidole noda 1135 31 209 0 889 2 1 0 3 Pristomyrmex punctatus 3339 41 246 112 173 32 3 2704 28 Pyramica membranifera 1 0 1 0 0 0 0 0 0 Solenopsis japonica 32 0 18 7 0 1 1 5 0 Strumigenys lewisi 7 0 0 6 0 1 0 0 0 Temnothorax anira 1 1 0 0 0 0 0 0 0 Temnothorax congruus 190 66 45 24 0 14 13 25 3 Temnothorax spinosior 5 3 2 0 0 0 0 0 0 Tetramorium bicarinatum 4 0 0 0 0 0 4 0 0 Tetramorium tsushimae 18939 1558 3974 2286 401 568 2151 6215 1786 Total 51307 3932 6836 4642 2208 15673 3617 11952 2447 Non-ant invertebrates (common name2) Isopoda (sowbugs) 33389 1452 9084 9669 507 5442 1571 5600 64 Myriapoda (centipedes/millipedes) 1343 171 155 581 8 103 122 184 19 Araneae (spiders) 1242 193 198 297 24 169 181 165 15 Orthoptera (grasshoppers) 239 25 35 28 9 73 13 44 12 Dermaptera (earwigs) 1708 338 78 355 149 121 315 340 12 Blattodea (cockroaches) 113 31 23 8 13 14 7 4 13 Mantodea (mantis) 1 0 0 0 0 1 0 0 0 Hemiptera (bugs) 820 89 96 306 19 94 100 107 9 Coleoptera (beetles) Byrrhoidea (pill beetles) 13 2 0 1 0 1 9 0 0 Cantharoidea (soldier beetles) 3 0 1 0 0 0 1 1 0 Caraboidea (ground beetles) 750 70 51 106 7 106 343 61 6 Chrysomeloidea (longhorn beetles) 62 7 10 19 0 4 10 9 3 Cucujoidea (darkling beetles) 326 67 84 81 7 11 50 24 2 Curculionoidea (weevils) 618 63 89 107 6 61 70 212 10 Dermestoidea (carpet beetles) 8 1 1 0 0 2 1 3 0 Elateroidea (click beetles) 107 10 11 50 5 4 16 9 2 Scarabaeoidea (gold beetles) 258 61 50 30 14 14 37 48 4 Staphylinoidea (rove beetles) 80 18 6 11 0 2 14 27 2 Land snails 244 5 74 56 25 14 21 46 3 Total 41324 2603 10046 11705 793 6236 2881 6884 176 1 Plots i–iii and I–III were invaded by Linepithema humile, and plots n and N were untreated, non-invaded plots. 2 Representative example. Effects of L. humile and pesticide on non-target species Both L. humile abundance and fipronil treatment had negative associations with total number of ant individuals, ant species richness, total number of non-ant invertebrates, and non-ant invertebrate taxonomic richness (Table 2). The L. humile × pesticide treatment interaction was significantly associated with total number of ant individuals. Total number of non-ant invertebrates and taxonomic richness were affected by sites Table 3 shows the relationships between L. humile abundance and fipronil treatment and the number of individuals in each ant species or each non-ant taxonomic group. Modeling the data with zero-inflated Poisson regression fit significantly better than (p < 0.025) or did not differ from the ordinary Poisson regression model. In the majority of the species/ taxonomic groups, both L. humile and pesticide treatment were not significant in the zero-inflated part of the model. In the Poisson part, the abundance of L. humile was negatively associated with the number of Pristomyrmex punctatus and Taxa Total Tokai plots1 Jonan plots1 i ii iii n I II III N Myrmicinae Y Sakamoto et al. – Non-target effects of pesticide in invasive ant eradication program232 Tetramorium tsushimae individuals, whereas the number of pesticide treatments was negatively associated with the number Formica japonica, Paraparatrechina sakurae, Pheidole noda, P. punctatus, and T. tsushimae individuals. The interaction between the two variables was significantly associated with the number of P. punctatus and T. tsushimae. Among the non-ant invertebrate groups, Isopoda had negative relationships with both L. humile and fipronil treatment and Blattodea had a negative relationship with pesticide treatment. Some species and taxonomic groups were affected by site. Community dynamics during chemical control and after eradication of L. humile The dynamics of non-target ant (Fig 4) and non-ant invertebrate communities (Fig 5) differed between pesticide usage histories. The deviations from the control were larger Fig 3. Mean total number of Linepithema humile (red line) and non-target ants (blue line) per trap by month in plots invaded by L. humile (i–iii, I–III) and untreated, non-invaded plots (n, N). Sociobiology 66(2): 227-238 (June, 2019) 233 from spring to fall (March–November) and smaller in winter (December–February). The structures of non-target ant communities in Plots i and ii clearly deviated from that in the non-invaded (control) plot (Fig 4). The ant community structure in 3-year-treated Plot ii responded to the treatment, in which the deviation from the control was larger in the second and third years than in the first year (Fig 4a), whereas that in 1-year-treated Plot i initially deviated from that in the non-invaded plot but was Fig 5. Principal response curve diagrams illustrating the shift of non-ant invertebrate communities over 3 years. (a) Plot ii, where no pesticide was applied after October 2013, and (b) Plot i, where no pesticide was applied after May 2012, relative to the control (untreated and non- invaded) plot over time. Left axis, regression coefficient; right axis, taxon weights (only taxa with a score of >0.5 or <−0.5 are shown). The first canonical axis explains 29.72% (p < 0.001) of the total variation in Plot ii and 62.97% (p < 0.001) in Plot i. Fig 4. Principal response curve diagrams illustrating the shift of non-target ant communities over 3 years. (a) Plot ii, where no pesticide was applied after October 2013, and (b) Plot i, where no pesticide was applied after May 2012, relative to the control (untreated and non-invaded) plot over time. Left axis, regression coefficient; right axis, species weights (only species with a score of >0.5 or <−0.5 are shown). The first canonical axis explains 56.63% (p < 0.001) of the total variation in Plot ii and 50.96% (p < 0.001) in Plot i. more similar in the third year (Fig 4b). These results indicate that the ant community structure recovered about a year after the eradication program ended. The structures of non-ant invertebrate communities in Plots i and ii did not deviate so clearly from that in the non-invaded (control) plot. The community structure in the 3-year-treated plot changed unidirectionally (i.e., Isopoda decreased) with time (Fig 5a), whereas that in the 1-year-treated plot did not respond clearly (Fig 5b). Discussion The number of L. humile was negatively associated with the total number of non-target ant individuals and species richness. As reported elsewhere in Japan (Miyake et al., 2002), in the USA (Heller, 2004; Suarez et al., 1998), and in Europe (Oliveras et al., 2005), L. humile reduces the diversity of indigenous ants. Therefore, it is reasonable to conclude that the invasion by L. humile harmed non-target ants in our study area as well. Our GLM results for each ant species indicated decreases in P. punctatus and T. tsushimae abundance. The incidence of T. tsushimae was also reported to be clearly lower where L. humile had increased over time in several parks in Japan (Park et al., 2014). Two mechanisms have been proposed to explain the displacement of indigenous ant fauna by L. humile invasion: exploitative competition and interference Y Sakamoto et al. – Non-target effects of pesticide in invasive ant eradication program234 L. humile on other invertebrate taxonomic groups are also not universal, and various studies have reported negative, positive, or no relationship. Factors underlying negative relationships may include direct feeding by L. humile on adult or immature organisms or spatial competition for limited habitats (Cole et al., 1992; Dreistadt et al., 1986), while factors underlying positive relationships may include feeding by invertebrates on dead and immature L. humile individuals or on the remains of prey items brought to the nest area by foraging L. humile (Cole et al., 1992). Negative effects of pesticide treatment were found in ants as well as in non-ant invertebrates. The greatest factor underlying the negative effects is that toxic baits are typically attractive to a wide range of non-target species (Buczkowski, 2017). Our GLM analyses showed that the L. humile × pesticide treatment interaction also affected the total number of ants and non-ant invertebrates. This result can be interpreted in two ways, based on different biological scenarios. First, the effect of L. humile on indigenous invertebrates decreases when there is pesticide treatment, which kills the invasive ants. Second, the effect of pesticide treatment on non-target fauna may decrease when the L. humile population is large and it may increase when the population is small. This idea reflects our observations and those of previous studies (Abedrabbo, 1994; Hoffmann & O’Connor, 2004) that the invasive alien ants were eradicated first and non-target ants were not eradicated. This likely occurred because non-target fauna was deprived of opportunities to eat the bait because of greater consumption by L. humile (Buczkowski & Bennett, 2008; Hoffmann, 2010; Holway, 1999; Human & Gordon, 1996). We compared the PRC results for 3 years in this study with those for only the first year in the previous study (Inoue et al., 2015) in the same eradication program. It should be noted that the plots with low-density L. humile were selected for PRC analyses and the control plot was untreated and never invaded in this study, whereas the plots with high-density L. humile were selected and the control plot was untreated but invaded in the previous study (Inoue et al., 2015). Inoue et al. (2015) concluded that non-target populations recovered within the first year of pesticide treatment, and they found no non-target effects of pesticide in the first year, we suspect because the number of L. humile was large. By contrast, our PRC results showed that the non-target community structure recovered about a year after the eradication program ended. That is, until L. humile was eradicated, the negative effects of pesticides on non-targets increased as L. humile decreased, suggesting that non-target effects cannot be ignored. Ensuring that indigenous ants and other invertebrates remain after pesticide treatment is crucial for ecosystem recovery after the eradication of invasive species. In fact, the invertebrate community recovered to a similar structure as in the non-invaded plot in this study. Such recovery eventually can be achieved by the continuous recruitment of immigrants of indigenous species from non-invaded sites (Holway, 1998). Variable Estimate SE z-value p-value Ant species Total number of individuals Intercept 1.911 Linepithema humile –0.651 0.101 –6.443 <0.001 * Pesticide treatment –0.127 0.027 –4.650 <0.001 * LH × PT1 0.137 0.036 3.850 <0.001 * Site –0.262 0.126 –2.080 0.040 Species richness Intercept 4.155 Linepithema humile –1.205 0.259 –4.654 <0.001 * Pesticide treatment –0.326 0.070 –4.640 <0.001 * LH × PT1 0.161 0.091 1.762 0.081 Site 0.300 0.324 0.927 0.356 Non-ant invertebrates Total number of individuals Intercept 2.202 Linepithema humile –0.499 0.080 –6.205 <0.001 * Pesticide treatment –0.099 0.022 –4.539 <0.001 * LH × PT1 0.071 0.028 2.513 0.014 * Site –0.333 0.101 –3.307 0.001 * Taxonomic richness Intercept 5.892 Linepithema humile –1.143 0.283 –4.044 <0.001 * Pesticide treatment –0.182 0.077 –2.367 0.020 * LH × PT1 0.022 0.100 0.216 0.829 Site –1.724 0.353 –4.881 <0.001 * 1 Linepithema humile × pesticide treatment interaction. * p < 0.025 Table 2. Results of generalized linear models examining the effects of Linepithema humile, pesticide treatment, their interaction, and site on total number of individuals and species or taxonomic richness per trap. competition (Holway, 1999; Human & Gordon, 1996). However, although negative associations were reported between several other ant species (i.e., F. japonica and Crematogaster matsumurai) and L. humile in previous studies (Miyake et al., 2002; Park et al., 2014), no associations were observed between them in this study. The difference in results may be due in part to seasonal or temporal factors and/or small sample sizes. Likewise, the number of L. humile was negatively associated with the total number of non-ant individuals and taxonomic richness. The analysis for each taxonomic group, however, showed that L. humile had negative associations with abundance of only isopods. Almost all isopods we found were Armadillidium vulgare, which can reproduce in urban areas (Hornung et al., 2007). Linepithema humile has been reported to cause both significant decreases (Stanley & Ward, 2012) and increases in isopod abundance (Cole et al., 1992; Human & Gordon, 1997; Walters & Mackay, 2003). The impacts of Sociobiology 66(2): 227-238 (June, 2019) 235 Table 3. Results of zero-inflated Poisson regression models examining the effects of Linepithema humile, pesticide treatment, their interaction, and site on the number of individuals of each ant species or non-ant invertebrate taxonomic group per trap. Species/Taxonomic group1 Variable Poisson model Zero-inflated model Estimate SE z-value p-value Estimate SE z-value p-value Ants Formica japonica Linepithema humile –38.02 NA NA NA –89.79 448.9 –0.200 0.842 Pesticide treatment –0.226 0.023 –10.03 <0.001 * 0.253 0.107 2.377 0.018 * LH × PT3 7.462 NA NA NA 18.20 89.78 0.203 0.839 Site –0.665 0.107 –6.194 <0.001 * –0.33 –0.687 0.492 Paraparatrechina sakurae Linepithema humile –24.22 NA NA NA –108.3 379.9 –0.285 0.775 Pesticide treatment –0.195 0.050 –3.885 <0.001 * –0.308 0.126 –2.443 0.015 * LH × PT3 4.96 NA NA NA 21.43 75.97 0.282 0.778 Site 0.02 0.230 0.099 0.921 –0.34 0.51 –0.655 0.513 Pheidole noda Linepithema humile –30.00 443.9 –0.068 0.946 –235.9 541.5 –0.436 0.663 Pesticide treatment –1.419 0.143 –9.92 <0.001 * –18.84 NA NA NA LH × PT3 9.967 88.77 0.112 0.911 107.7 NA NA NA Site –62.84 995.6 –0.063 0.950 –520.2 1201 –0.433 0.665 Pristomyrmex punctatus Linepithema humile –2.333 0.825 –2.827 0.005 * –1.4755 1.608 –0.918 0.359 Pesticide treatment –0.769 0.234 –3.284 0.001 * –0.404 0.482 –0.838 0.402 LH × PT3 0.757 0.193 3.929 <0.001 * 0.493 0.365 1.352 0.176 Site –3.100 1.197 –2.591 <0.001 * –2.233 2.507 –0.891 0.373 Solenopsis japonica Linepithema humile –50.56 6945 –0.007 0.994 58.39 1001000 0.000 1.000 Pesticide treatment –9.585 387.2 –0.025 0.980 –7.570 NA NA NA LH × PT3 10.50 1389 0.008 0.994 –11.82 200100 0.000 1.000 Site 4.128 0.985 4.192 <0.001 * 7.438 19 0.383 0.702 Tetramorium tsushimae Linepithema humile –1.240 0.155 –8.022 <0.001 * 1.502 0.524 2.866 0.004 * Pesticide treatment –0.116 0.008 –13.91 <0.001 * 0.164 0.135 1.214 0.225 LH × PT3 0.303 0.032 9.509 <0.001 * –0.298 0.166 –1.797 0.072 Site –0.480 0.042 –11.54 <0.001 * 0.568 0.569 0.999 0.318 Non-ant invertebrates Isopoda (sowbugs2) Linepithema humile –0.996 0.082 –12.09 <0.001 * 0.637 0.575 1.109 0.268 Pesticide treatment –0.140 0.010 –13.63 <0.001 * 0.274 0.159 1.729 0.084 LH × PT3 –0.081 0.031 –2.581 0.010 * 0.059 0.175 0.340 0.734 Site –0.292 0.054 –5.447 <0.001 * 1.497 0.657 2.279 0.023 * Araneae (spiders2) Linepithema humile –0.355 0.985 –0.360 0.719 2.042 1.690 1.208 0.227 Pesticide treatment 0.155 0.080 1.949 0.051 0.242 0.381 0.634 0.526 LH × PT3 0.055 0.200 0.275 0.784 –0.359 0.351 –1.023 0.306 Site –0.084 0.252 –0.333 0.739 1.953 1.048 1.864 0.062 Orthoptera (grasshoppers2) Linepithema humile 0.070 0.280 0.249 0.804 –10.785 47406 0.000 1.000 Pesticide treatment –0.092 0.109 –0.839 0.401 0.336 0.198 1.699 0.089 LH × PT3 1.068 0.648 1.648 0.099 22.274 142.6 0.156 0.876 Site 1.016 0.572 1.775 0.076 31.430 14269 0.002 0.998 However, our data revealed temporary negative effects of pesticides on non-target communities. It is also important to reduce the impacts on indigenous communities, because their restoration inhibits successful re-invasion into the ecological gap by invasive alien species (Hoffmann, 2010; Hoffmann & O’Connor, 2004; Plentovich et al., 2009; Tschinkel & King, 2017). When an eradication program comes close to achieving success, the end should be judged by using a statistical model (Sakamoto et al., 2017), for example, so as to avoid the unnecessary prolongation of pesticide treatment. Moreover, in delicate infested habitats and in the presence of sensitive wildlife, traditional eradication methods with toxic baits may be inappropriate. To reduce the impacts on non-target fauna, target-specific approaches should be developed, such as using ribonucleic acid interference (Campbell et al., 2015; Gould, 2008) or prey-baiting (Buczkowski, 2017). Y Sakamoto et al. – Non-target effects of pesticide in invasive ant eradication program236 Species/Taxonomic group1 Variable Poisson model Zero-inflated model Estimate SE z-value p-value Estimate SE z-value p-value Blattodea (cockroaches2) Linepithema humile –0.496 0.376 –1.318 0.188 –51.85 2490 –0.021 0.983 Pesticide treatment –0.353 0.141 –2.508 0.012 * –0.054 0.276 –0.197 0.844 LH × PT3 1.744 0.707 2.467 0.014 * 23.71 146.9 0.161 0.872 Site 0.842 0.653 1.290 0.197 20.75 716.3 0.029 0.977 Hemiptera (bugs2) Linepithema humile –0.687 0.379 –1.813 0.070 0.215 1.095 0.196 0.845 Pesticide treatment –0.042 0.045 –0.925 0.355 0.187 0.187 1.001 0.317 LH × PT3 –0.051 0.115 –0.442 0.658 –0.034 0.289 –0.117 0.907 Site –0.896 0.241 –3.723 <0.001 * –0.596 0.891 –0.670 0.503 Caraboidea (ground beetles2) Linepithema humile 0.053 0.268 0.196 0.844 –1.927 3502 –0.001 1.000 Pesticide treatment –0.071 0.100 –0.708 0.479 0.190 0.200 0.948 0.343 LH × PT3 –0.305 0.186 –1.642 0.101 –0.323 700.4 0.000 1.000 Site 0.802 0.499 1.608 0.108 14.747 617.5 0.024 0.981 Cucujoidea (darkling beetles2) Linepithema humile –0.595 0.616 –0.965 0.335 2.921 4624 0.001 0.999 Pesticide treatment 0.287 0.183 1.563 0.118 10.39 116.8 0.089 0.929 LH × PT3 0.100 0.166 0.604 0.546 –0.710 924.8 –0.001 0.999 Site –2.167 1.270 –1.706 0.088 0.616 2.0 0.308 0.758 Curculionoidea (weevils2) Linepithema humile 0.787 0.575 1.370 0.171 2.122 1.842 1.152 0.249 Pesticide treatment 0.002 0.184 0.013 0.989 0.05 0.35 0.13 0.896 LH × PT3 –0.327 0.211 –1.550 0.121 –8.343 117.7 –0.071 0.943 Site 0.808 0.721 1.120 0.263 5.612 4.35 1.291 0.197 Elateroidea (click beetles2) Linepithema humile –15.68 NA NA NA –0.442 25990 0.000 1.000 Pesticide treatment 41.77 NA NA NA 12.12 393 0.031 0.975 LH × PT3 3.650 NA NA NA 0.490 5198 0.000 1.000 Site –3.188 0.920 –3.464 0.001 * –30.93 1661000 0.000 1.000 Scarabaeoidea (gold beetles2) Linepithema humile 0.143 0.336 0.426 0.670 0.050 0.723 0.069 0.945 Pesticide treatment –0.030 0.168 –0.178 0.858 –0.037 0.391 –0.096 0.924 LH × PT3 –0.207 0.173 –1.195 0.232 –0.416 0.639 –0.650 0.515 Site –0.763 0.769 –0.993 0.321 –1.932 2.128 –0.908 0.364 1 Species or taxonomic groups that could not be calculated because of insufficient data are not shown. 2 Representative example 3 Linepithema humile × pesticide treatment interaction * p < 0.025 Table 3. Results of zero-inflated Poisson regression models examining the effects of Linepithema humile, pesticide treatment, their interaction, and site on the number of individuals of each ant species or non-ant invertebrate taxonomic group per trap. (Continuation) Acknowledgments We are grateful to Takashi Sugiyama and Katsuo Sugimaru (Fumakilla Ltd.), Mitsuhiko Toda and Hideaki Mori (Japan Wildlife Research Center), Sachiko Moriguchi (Niigata University), and Kazutaka Suzuki, Takuji Nomura, and Hiromoto Agemori (National Institute for Environmental Studies, NIES) for conducting the eradication program and identifying species, and to Makihiko Ikegami and Naoki H. Kumagai (NIES) for providing useful comments. We also greatly appreciate the support of Fumakilla Ltd. in providing its products. This research was supported by the Environment Research and Technology Development Fund (No. 4-1401) of the Ministry of the Environment, Japan. References Abedrabbo, S. (1994). Control of the little fire ant, Wasmannia auropunctata, on Santa Fe Island in the Galapagos islands. In: D.F. Williams (Ed), Exotic Ants: biology, impact, and control of introduced species (pp. 219-227). Boulder, Colorado: Westview Press. Buczkowski, G. (2017). Prey-baiting as a conservation tool: selective control of invasive ants with minimal non-target effects. Insect Conservation and Diversity, 10: 302-309. Buczkowski, G., Bennett, G.W. (2008). Detrimental effects of highly efficient interference competition: Invasive argentine ants outcompete native ants at toxic baits. Environmental Entomology, 37: 741-747. Sociobiology 66(2): 227-238 (June, 2019) 237 Campbell, K.J., Beek, J., Eason, C.T., Glen, A.S., Godwin, J., Gould, F., Holmes, N.D., Howald, G.R., Madden, F.M., Ponder, J.B., Threadgill, D.W., Wegmann, A.S., Baxter, G.S. (2015). The next generation of rodent eradications: Innovative technologies and tools to improve species specificity and increase their feasibility on islands. Biological Conservation, 185: 47-58. doi: 10.1016/j.biocon.2014.10.016 Causton, C.E., Sevilla, C.R., Porter, S.D. (2005). Eradication of the little fire ant, Wasmannia auropunctata (Hymenoptera: Formicidae), from Marchena Island, Galapagos: On the edge of success? Florida Entomologist, 88: 159-168. doi: 10.1653/0015-4040(2005)088[0159:Eotlfa]2.0.Co;2 Clavero, M., Garcia-Berthou, E. (2005). Invasive species are a leading cause of animal extinctions. Trends in Ecology and Evolution, 20: 110. doi: 10.1016/j.tree.2005.01.003 Cole, F.R., Medeiros, A.C., Loope, L.L., Zuehlke, W.W. (1992). Effects of the Argentine ant on arthropod fauna of hawaiian high-elevation shrubland. Ecology, 73: 1313-1322. doi: 10.2307/1940678 Dreistadt, S.H., Hagen, K.S., Dahlsten, D.L. (1986). Predation by Iridomyrmex humilis (Hym., Formicidae) on eggs of chrysoperla- carnea (Neu., Chrysopidae) released for inundative control of Illinoia liriodendri (Hom., Aphididae) infesting Liriodendron tulipifera. Entomophaga, 31: 397-400. doi: 10.1007/bf02373157 Gould, F. (2008). Broadening the application of evolutionarily based genetic pest management. Evolution, 62: 500-510 Heller, N.E. (2004). Colony structure in introduced and native populations of the invasive Argenine ant, Linepithema humile. Insectes Sociaux, 51: 378-386. doi: 10.1007/s00040-004-0770-0 Hoffmann, B.D. (2010). Ecological restoration following the local eradication of an invasive ant in northern Australia. Biological Invasions, 12: 959-969. doi: 10.1007/s10530-009- 9516-2 Hoffmann, B.D., Luque, G.M., Bellard, C., Holmes, N.D., Donlan, C.J. (2016). Improving invasive ant eradication as a conservation tool: A review. Biological Conservation, 198: 37-49. doi: 10.1016/j.biocon.2016.03.036 Hoffmann, B.D., O’Connor, S. (2004). Eradication of two exotic ants from Kakadu National Park. Ecological Management and Restoration, 5: 98-105 Holway, D.A. (1998). Effect of Argentina ant invasions on ground-dwelling arthropods in northern California riparian woodlands. Oecologia, 116: 252-258. doi: 10.1007/ s004420050586 Holway, D.A. (1999). Competitive mechanisms underlying the displacement of native ants by the invasive Argentine ant. Ecology, 80: 238-251 Holway, D.A., Lach, L., Suarez, A.V., Tsutsui, N.D., Case, T.J. (2002). The causes and consequences of ant invasions. Annual Review of Ecology and Systematics, 33: 181-233. doi: 10.1146/annurev.ecolsys.33.010802.150444 Hornung, E., Tóthmérész, B., Magura, T., Vilisics, F. (2007). Changes of isopod assemblages along an urban-suburban- rural gradient in Hungary. European Journal of Soil Biology, 43: 158-165. doi: 10.1016/j.ejsobi.2007.01.001 Human, K.G., Gordon, D.M. (1996). Exploitation and interference competition between the invasive Argentine ant, Linepithema humile, and native ant species. Oecologia, 105: 405-412. doi: 10.1007/Bf00328744 Human, K.G., Gordon, D.M. (1997). Effects of Argentine ants on invertebrate biodiversity in northern California. Conservation Biology, 11: 1242-1248. doi: 10.1046/j.1523- 1739.1997.96264.x Inoue, M.N., Saito-Morooka, F., Suzuki, K., Nomura, T., Hayasaka, D., Kishimoto, T., Sugimaru, K., Sugiyama, T., Goka, K. (2015). Ecological impacts on native ant and ground-dwelling animal communities through Argentine ant (Linepithema humile) (Hymenoptera: Formicidae) management in Japan. Applied Entomology and Zoology, 50: 331-339. doi: 10.1007/s13355-015-0338-7 IUCN ISSG (2013). International Union for the Conservation of Nature Invasive Species Specialist Group: 100 of the world’s worst alien invasive species. http://www.issg.org/ database/species/search.asp?st=100ss. Jackman, S. (2017). pscl: Classes and methods for R developed in the political science computational laboratory, Stanford University. Stanford, California: Department of Political Science, Stanford University. Klotz, J.H., Rust, M.K., Greenberg, L., Field, H.C., Kupfer, K. (2007). An evaluation of several urban pest management strategies to control Argentine ants (Hymenoptera: Formicidae). Sociobiology, 50: 391-398 Lepš, J., Šmilauer, P. (2003). Multivariate Analysis of Ecological Data using Canoco. Cambridge, UK: Cambridge University Press. Mack, R.N., Simberloff, D., Lonsdale, W.M., Evans, H., Clout, M., Bazzaz, F.A. (2000). Biotic invasions: causes, epidemiology, global consequences, and control. Ecological Applications, 10: 689-710. doi: 10.2307/2641039 McCullagh, P., Nelder, J.A. (1989). Generalized linear models, second edition. Abingdon, UK: Taylor & Francis. McGlynn, T.P. (1999). The worldwide transfer of ants: geo- graphical distribution and ecological invasions. Journal of Bioge- ography, 26: 535-548. doi: 10.1046/j.1365-2699.19 99.00310.x Miyake, K., Kameyama, T., Sugiyama, T., Ito, F. (2002). Effect of Argentine ant invasions on Japanese ant fauna in Hiroshima Prefecture, western Japan: A preliminary report (Hymenoptera : Formicidae). Sociobiology, 39: 465-474 National Institute of Environmental Studies (2014). Invasive Y Sakamoto et al. – Non-target effects of pesticide in invasive ant eradication program238 species of Japan http://www.nies.go.jp/biodiversity/invasive/ DB/detail/60090e.html. Oksanen, J. (2013). Multivariate analysis of ecological communities in R: vegan tutorial. Oliveras, J., Bas, J.M., Casellas, D., Gomez, C. (2005). Numerical dominance of the Argentine ant vs native ants and consequences on soil resource searching in Mediterranean cork-oak forests (Hymenoptera : Formicidae). Sociobiology, 45: 643-658 Park, S.H., Hosoishi, S., Ogata, K. (2014). Long-term impacts of Argentine ant invasion of urban parks in Hiroshima, Japan. Journal of Ecology and Environment, 37: 123-129 Passera, L. (1994). Characteristics of tramp species. In: D.F. Williams (Ed), Exotic Ants: biology, impact, and control of introduced species (pp. 23-43). Boulder, Colorado: Westview Press. Pisa, L.W., Amaral-Rogers, V., Belzunces, L.P., Bonmatin, J.M., Downs, C.A., Goulson, D., Kreutzweiser, D.P., Krupke, C., Liess, M., McField, M., Morrissey, C.A., Noome, D.A., Settele, J., Simon-Delso, N., Stark, J.D., Van der Sluijs, J.P., Van Dyck, H., Wiemers, M. (2015). Effects of neonicotinoids and fipronil on non-target invertebrates. Environmental Science and Pollution Research, 22: 68-102. doi: 10.1007/s11356- 014-3471-x Plentovich, S., Hebshi, A., Conant, S. (2009). Detrimental effects of two widespread invasive ant species on weight and survival of colonial nesting seabirds in the Hawaiian Islands. Biological Invasions, 11: 289-298. doi: 10.1007/s10530-008-9233-2 Plentovich, S., Swenson, C., Reimer, N., Richardson, M., Garon, N. (2010). The effects of hydramethylnon on the tropical fire ant, Solenopsis geminata (Hymenoptera: Formicidae), and non-target arthropods on Spit Island, Midway Atoll, Hawaii. Journal of Insect Conservation, 14: 459-465. doi: 10.1007/s10841-010-9274-6 Prasifka, J.R., Hellmich, R.L., Dively, G.P., Lewis, L.C. (2005). Assessing the effects of pest management on nontarget arthropods: The influence of plot size and isolation. Environmental Entomology, 34: 1181-1192. doi: 10.1603/0046-225x(2005)034[1181:Ateopm]2.0.Co;2 R Development Core Team (2013). R: A language and environment for statistical computing http://www.r-project.org/. Rabitsch, W. (2011). The hitchhiker’s guide to alien ant invasions. Biocontrol, 56: 551-572. doi: 10.1007/s10526- 011-9370-x Rhône Poulenc (1996). ‘Fipronil’ worldwide technical bulletin. Lyon, France: Agrochimie, 20 p. Sakamoto, Y., Kumagai, N.H., Goka, K. (2017). Declaration of local chemical eradication of the Argentine ant: Bayesian estimation with a multinomial-mixture model. Scientific Reports, 7. doi: 10.1038/s41598-017-03516-z Stanley, M.C., Ward, D.F. (2012). Impacts of Argentine ants on invertebrate communities with below-ground consequences. Biodiversity and Conservation, 21: 2653-2669. doi: 10.1007/ s10531-012-0324-0 (in 295) Suarez, A.V., Bolger, D.T., Case, T.J. (1998). Effects of fragmentation and invasion on native ant communities in coastal southern California. Ecology, 79: 2041-2056 Suarez, A.V., McGlynn, T.P., Tsutsui, N.D. (2010). Biogeographic and taxonomic patterns of introduced ants. In: L. Lach, C.L. Parr, and K.L. Abbott (Eds), Ant Ecology (pp. 233-244). Oxford: Oxford University Press Sugiyama, T. (2000). Invasion of argentine ant, Linepithema humile, into Hiroshima Prefecture, Japan. Japanese Journal of Applied Entomology, 44: 127-129 (in Japanese) Tschinkel, W.R., King, J.R. (2017). Ant community and habitat limit colony establishment by the fire ant, Solenopsis invicta. Functional Ecology, 31: 955-964. doi: 10.1111/1365- 2435.12794 Vail, K.M., Bailey, D., McGinnis, M. (2003). Perimeter spray and bait combo. Pest Control Technology, 31: 96 Van den Brink, P.J., Ter Braak, C.J.F. (1999). Principal response curves: Analysis of time-dependent multivariate responses of biological community to stress. Environmental Toxicology and Chemistry, 18: 138-148. doi: 10.1897/1551-5028(1999)018<0138:prcaot>2.3.co;2 Walters, A.C., Mackay, D.A. (2003). The impact of the Argentine ant, Linepithema humile (Mayr) on native ants and other invertebrates in South Australia. Records of the South Australian Museum: 17-24 Williams, D.F. (1994). Exotic ants: biology, impact, and control of introduced species. Boulder, Colorado, USA: Westview Press.