Evaluation of responses of different ant species ( Formicidae ) to the scavenger deterrent factor associated with the entomopathogenic nematode-bacterium complex

According to previous observations, it was hypothesized that the feeding behavior of some ant species would be deterred by a scavenger deterrent factor (SDF), whereas for other species it would not. The effects of the SDF were studied on 11 ant species in three different subfamilies: Dolichoderinae Forel, 1878, Formicinae Latreille, 1809, and Myrmicinae Lepeletier de Saint-Fargeau, 1835. The experiments were conducted from 2014–2015 in Davis, California, United States, Aydin, Turkey, and Duzce, Turkey. Five-day-old Heterorhabditis bacteriophora (Poinar, 1976), (Hb)-killed and freeze-killed Galleria mellonella (Linnaeus, 1758) were exposed to ant colonies in the fi eld for 3 to 4 h. Seven ant species fed signifi cantly less on Hb-killed insects than freeze-killed insect. On the other hand, there was no signifi cant difference in cadaver consumption with fi ve species, but Liometopum occidentale Emery, 1895 did consume a higher rate of Hb-killed insects than freeze-killed insects and was not deterred by SDF. It was also observed that four ant species took Hb-killed insects into the nests, but two Myrmicinae species, Pogonomyrmex subdentatus Mayr, 1870 and Messor meridionalis (André, 1883) removed the cadavers after 30 min, whereas two Formicinae species, Cataglyphis nodus (Brullé, 1833) and Formica fusca Linnaeus, 1758, retained the cadavers in the nest. It was assumed that the latter two species consumed both Hb-killed and freeze-killed insects. Further studies are needed to explain why L. occidentale C. nodus and F. fusca are not deterred by SDF.


Nematodes and infection of larvae
The EPN, Heterorhabditis bacteriophora (Hb) (Bakersfi eld strain) (Poinar, 1976) (Nematoda: Rhabditida) was used in the experiments in USA and in Turkey.This nematode strain was maintained in Dr. Edwin E. Lewis' laboratory in the Department of Entomology & Nematology, Davis, California.The nematode was cultured using last instar Galleria mellonella (Linnaeus, 1758) (Lepidoptera: Pyralidae) larvae at 23-24°C according to Kaya & Stock (1997).The Hb-killed larvae were transferred to a White trap, and emerging IJs were harvested from the water (White, 1927) and stored at 15°C for no more than 3 weeks when they were used to infect the test insects.
To obtain the nematode-killed larvae for fi eld studies, 10 G. mellonella were exposed to 100 IJs/larva in a plastic petri dish (100 × 15 mm) lined with a fi lter paper.Petri dishes were placed in plastic bags to maintain moisture and kept at 25°C for 5 days.

Ant colonies
Twelve ant colonies representing 11 species in three subfamilies were observed for their response to nematode-killed insects (Table 1).Ant species from the USA and Turkey were identifi ed by Dr. Phillip S. Ward, University of California-Davis, Department of Entomology and Nematology, and Dr. Kadri Kiran, Trakya University, Department of Biology, respectively.Six of the colonies were from Davis, California, USA and the other six of were from Duzce, and Aydin, Turkey.
Of the 12 colonies, two were colonies of Tetramorium cf.caespitum (Linnaeus, 1758) (Hymenoptera: Formicidae); one collected in California and the other from Turkey (Table 1).

Experimental design
Four G. mellonella cadavers were pinned on a piece of cardboard (10 cm × 12 cm) and placed nearby ant nests or trails.(See Fig. 1 for arrangement of cadavers on cardboard.)The color of cardboards was brownish which is similar to the color of the soil.Two cadavers were fi ve-day-old Hb-killed and two were freezekilled larvae.Five-day-old EPN-killed cadavers were used because in the previous studies, the highest SDF activity was reported after four days (Gulcu et al., 2012).The experiments were repeated three times for each ant species.The cadavers were placed out in the fi eld for 3 to 4 h between 0600 and 1100 h, or in the late afternoon between 1500 and 2000 h in a shaded location in summer of 2014 and 2015.Ant activity was observed hourly over the 3 h period, but in four cases, we observed that Pogonomyrmex subdentatus Mayr, 1870 (Hymenoptera: Formicidae), Messor meridionalis (André, 1883) (Hymenoptera: Formicidae), C. nodus and Formica fusca Linnaeus, 1758(Hymenoptera: For-al., 2003;Boemare & Akhurst, 2006).After infection, the nematodes develop to adults and it takes a week or more before the newly produced IJs emerge to seek out new hosts.
Ants (Hymenoptera: Formicidae) are one of the most successful groups of eusocial insects in the world (Fittkau & Klinge, 1973).They are important as scavengers, herbivores, or predators and show highly specialized behaviors like farming fungi, harvesting seeds, herding and milking of other insects (e.g., aphids), communal nest weaving, cooperative hunting in packs, social parasitism and slavemaking (Hölldobler & Wilson, 1990).Other species are omnivores feeding on a variety of food sources.Studies conducted by Baur et al. (1998), Zhou et al. (2002), Gulcu et al. (2012), and Ulug et al. (2014) showed that several ant species do not feed on Steinernema-or Heterorhabditis-killed insects that were more than 2-days old and heterorhabditids produce a stronger deterrent factor than steinernematids.Some scavenger species, such as ants and crickets, partly consume 2-day-old Steinernema-killed insects but not Heterorhabditis (Gulcu et al., 2012;Ulug et al., 2014).Recently, it was observed that two ant species, Cataglyphis nodus (Brullé, 1833) (Hymenoptera: Formicidae) and Liometopum occidentale Emery, 1895 (Hymenoptera: Formicidae), did consume Heterorhabditis-killed insects.Accordingly, it was hypothesized that the response of ants to SDF varied by species and we report herein a quantitative study to determine differential responses of several ant species to SDF in Heterorhabditis-killed insects.micidae) workers demonstrated different behaviors compared with the other six ant species.Therefore, these ant species were observed for an additional hour.In the experiments, each cadaver had its weight recorded using a precision scale before pinning on cardboard.After 3 or 4 h, the cadavers that were not completely consumed were brought to the laboratory, re-weighed and examined under the dissecting microscope to determine the extent of ant feeding activity.
To measure the level of potential deterrent effect on each ant species, we compared the larval weight loss of Hb-killed and freeze-killed cadavers.Statistical difference between these two groups indicated that a certain ant species responded to SDF.If there was no statistically signifi cant difference, it was accepted that the ant species was unaffected by SDF produced.

Statistical analysis
Response of ants to Hb-killed or freeze-killed insects were performed by using independent sample t-test with IBM, SPSS Statistics version 22.0 within each species.Weight reductions were compared at the P = 0.05 level (SPSS, 2013).Percent weight reduction data were arcsine transformed before statistical analyses.

RESULTS AND DISCUSSION
The loss of cadaver weight at the end of each experiment was determined to evaluate ant feeding activity.This approach differed from previous research in which Baur et al. (1998) and Zhou et al. (2002) classifi ed the cadavers as intact or not intact according to breaks in the integument.They compared the number of intact cadavers belonging to each group in the treatments after 24 h.Furthermore, Gulcu et al. (2012) and Ulug et al. (2014) examined the extent of cadaver consumption after 12 and 3 h, respectively, by visual examination using a dissecting microscope at 20 × and estimating the percentage of each cadaver that was consumed.In our study both visual examination and weight loss observed in cadavers were considered.
Using the loss of cadaver weight, we found signifi cantly more of the freeze-killed insects were consumed than the Hb-killed insects in the following eight ant species (Figs 2-4).Interestingly, our data with T. cf.caespitum colony from Davis differed from the T. cf.caespitum colony from Duzce.That is, signifi cant differences were observed between Hb-killed and freeze-killed insects for T. cf.caespitum colony Davis, but no signifi cant differences were observed for Hb-killed or freeze-killed insects for T. cf.caespitum colony Duzce.Yet, the weight loss for Hb-killed insects for the Davis and Duzce colonies were similar.In this case, statistical differences or lack of it may be an artifact.On the other hand, Zhou et al. (2002) suggested that the difference in cadaver consumption might originate from colony size and/or weather conditions.
Different feeding behaviors with C. nodus, F. fusca, P. subdentatus and M. meridionalis which took the Hb-killed insects into their nests were observed.But P. subdentatus and M. meridionalis removed and discarded the Hb-killed  insects a short distance outside the nests usually within 30 min, whereas C. nodus and F. fusca also took Hb-killed insects into the nest but did not remove the cadavers even after 3 h.We made another observation at 4 h after experimental set up and no cadavers were found outside the nest of C. nodus and F. fusca.It was assumed that these cadavers were consumed by C. nodus and F. fusca within the nest (Fig. 3).As far as we are aware, the only other invertebrate scavenger that consumed all heterorhabditid-and steinernematid-killed insects was the mite, Sancassania polyphyllae (Acari: Acaridae) (Ekmen et al., 2010a, b).Oi & Pereira (1993) classifi ed the removal behavior of ants as hygiene of nest and pathogen avoidance.For instance, Solenopsis invicta, Buren, 1972 (Hymenoptera: Formicidae) (Storey,   ) indicates the signifi cant difference between nematode-killed and freeze-killed groups for each ant species.This symbol also shows which ant species respond to SDF.The weight lost was determined by recording the weights of cadavers before and after treatments.Hb = H.bacteriophorakilled insects; Control = freeze-killed insects.A -C. nodus (t = 0.000; df = 1; P = 1.00);B -F. fusca (t = 0,00; df = 1; P = 1.00).
We also observed that L. occidentale showcased an unusual feeding behavior.The parasitoid phorid fl ies (see Porter et al., 1995a;Feener, 2000) interrupted the feeding activity of L. occidentale foragers.The foraging ants assumed a defensive posture by elevating their front legs and mandibles.The foraging ants not under attack moved the cadavers under the cardboard during the 3 h experimental period.Furthermore, these ants were unable to complete their feeding activity on Hb-killed and freeze-killed insects even though they attacked both groups.
In this respect, variable responses were observed with Dolichoderinae ants in our study (Fig. 2).For example, D. insanus and T. erraticum consumed less of the Hb-killed insects compared to the freeze-killed insects.There was no statistical difference in consumption between Hb-and freeze-killed insects for the Argentine ant (L.humile), and the velvety ant (L.occidentale) was the only species that consumed more Hb-killed insects than the freeze-killed insects.Baur et al. (1998) reported that several ant species including L. humile, V. andre, Pheidole vistana Forel, 1914 (Hymenoptera: Formicidae), F. pacifi ca, and Monomorium ergatogyna Wheeler W.M., 1904 (Hymenoptera: Formicidae) scavenged nematode-killed insects.And they observed that L. humile scavenged 10-20% of heterorhabditid-killed insects.But in our study, there was no signifi cant difference between consumption of heterorhabditid-killed (50.1%) and freeze-killed insects (59%).This discrepancy in the response of L. humile to Hb-killed insects might be derived from using different nematode-bacterium strains or evaluation methods.Some ant colonies might have already had a contact with H. bacteriophora killed insects and probably were "conditioned" to react to different cues they come across.This could also explain the different colony responses observed for the same species.
In conclusion, it was observed that SDF does infl uence the feeding behavior of some ant species and not in other species.Besides, different colonies of the same ant species have varying responses to SDF and this may be due to prior experience.It was determined that eight ant species fed signifi cantly less on Hb-killed insects than freezekilled insects, whereas only L. occidentale consumed signifi cantly more Hb-killed insects than freeze-killed insects.However, It was assumed that attack of the phorid fl ies affected it's response to cadavers.On the other hand, there was no signifi cant difference in cadaver consumption with four species.It was also observed that four ant species took Hb-killed insects into the nests but P. subdentatus and M. meridionalis removed the cadavers after 30 min, whereas C. nodus, and F. fusca retained the cadavers in the nest.Further studies need to be conducted to explain why L. occidentale, C. nodus and F. fusca are not deterred by SDF and whether other ant species in the Formicinae as well as other species in other subfamilies have similar responses.data: EEL.Contributed reagents/materials/analysis tools: BG.Wrote the paper: BG, SH, HKK.

Fig. 3 .
Fig.3.Larval weight reduction of H. bacteriophora-killed and freeze-killed insects by two species of subfamily Formicinae.(*) indicates the signifi cant difference between nematode-killed and freeze-killed groups for each ant species.This symbol also shows which ant species respond to SDF.The weight lost was determined by recording the weights of cadavers before and after treatments.Hb = H.bacteriophorakilled insects; Control = freeze-killed insects.A -C. nodus (t = 0.000; df = 1; P = 1.00);B -F. fusca (t = 0,00; df = 1; P = 1.00).

Table 1 .
Ant species (common names) and colony locations used in the experiments.
* Accepted common name of ants is shown in parentheses.** Pavement ant occurred in Davis, CA USA and Duzce, Turkey.