Introduction
The leaf-cutting ants Atta spp. and Acromyrmex spp. are hosts for 38 species of parasitoids from the Phoridae family, mainly from the genera Neodohrniphora spp., Apocephalus spp. and Myrmosicarius spp. (Feener & Moss, Reference Feener and Moss1990; Disney, Reference Disney1996; Disney & Bragança, Reference Disney and Bragança2000; Brown, Reference Brown2001; Disney et al., Reference Disney, Elizalde and Folgarait2006). These parasitoids can be observed flying along foraging trails of the ants (Borgmeier, Reference Borgmeier1928, Reference Borgmeier1931; Disney, Reference Disney1996; Disney & Bragança, Reference Disney and Bragança2000; Brown, Reference Brown2001).
Atta sexdens rubropilosa, which is an important agricultural and forestry pest in the Neotropical region, is frequently parasitized by Neodohrniphora spp. (Diptera: Phoridae) (Fowler et al., Reference Fowler, Pagani, Silva, Forti, Silva and Vasconcelos1989). Parasitism levels of these flies are relatively low, but their presence over nests and trails also modifies the ants' behaviour and reduces the foraging ability of the colonies (Orr, Reference Orr1992; Feener & Brown, Reference Feener and Brown1993; Bragança et al., Reference Bragança, Tonhasca and Della Lucia1998).
Neodohrniphora spp. are suggested as potential agents of biological control of ants (Tonhasca, Reference Tonhasca1996; Bragança et al., Reference Bragança, Tonhasca and Della Lucia1998); however, little is known about their host specificity and diversity (Brown, Reference Brown2001). In the few studies where the Neodohrniphora-A. sexdens rubropilosa interaction was investigated, the identification of flies was limited to the genus level (Tonhasca, Reference Tonhasca1996; Bragança et al., Reference Bragança, Tonhasca and Della Lucia1998).
The morphology of Neodohrniphora species is similar, except for their ovipositors, which apparently are modified in relation to theirs hosts (Brown, Reference Brown2001). Nonetheless, there is evidence that some species parasitize more than one species of leaf-cutting ant (Brown, Reference Brown2001; Bragança et al., Reference Bragança, Tonhasca and Moreira2002). In the work reported here, we surveyed the occurrence of species of Neodohrniphora phorids in the proximity of A. sexdens rubropilosa nests; and, on the basis of the attacking behaviour of phorids, we discuss the possibility that these flies parasitize other Atta and Acromyrmex species.
Material and methods
Phorid sampling
Phorids were sampled two to four times every month between May 2002 and April 2004 in a wooded patch located in Viçosa (20° 45′ S; 42° 51′ W), Brazil. Ten A. sexdens rubropilosa nests were initially selected on the basis of accessibility of foraging trails and abundance of forager ants. Sampling was conducted between 8:00 and 11:00 h and between 15:00 and 18:00 h. An observer walking at a constant pace captured flies hovering over the trails using a glass tube. The specimens were prepared in the laboratory and identified with the most recent key (Brown, Reference Brown2001).
To analyze if there is a variation in the phorids occurrence throughout the year, we compared the number of phorids collected by bimesters (the monthly data were grouped into bimesters because two monthly samplings were lost). The data were analyzed using the Kruskal-Wallis test because they were not normally distributed.
Attack behaviour specificity
To investigate the specificity of attack behaviour, females of Neodohrniphora spp. were released in a glass chamber (50×50×50 cm) containing a single worker ant and filmed for five minutes. Four ant species were tested consecutively with each phorid: A. sexdens rubropilosa, Atta laevigata, Acromyrmex crassispinus and Acromyrmex subterraneus molestans. The phorids that showed no response to the four ant species were not used in the analysis. The order of ant species in the test was randomized for each phorid tested (n=20). After the tests, the flies were killed and identified at the species level. The tapes were analyzed with help of the software, The Observer, version 4.1 (Noldus Information Techonology). We considered three behavioural acts associated with host attack: inspection flights, attacking bouts and attacks (Orr, Reference Orr1992; Feener & Brown, Reference Feener and Brown1993; Silva et al., Reference Silva, Bailez, Viana-Bailez and Tonhasca2007). Inspection flight was considered when the phorids hover at no more than 5 cm above a host. Attacking bouts are when the phorids make contact with any body part of the host without ovipositor penetration. Finally, attacks are contacts with ovipositor penetration in the posterodorsal extremity of the ant's head.
Results
We collected 303 flies of the genus Neodohrniphora. Fifteen were males and, therefore, could not be identified at the species level. These numbers underestimated the actual abundance of phorid because the rate of capture was about 60% (phorids captured/phorids pursued×100).
The phorids females (288 individuals) belonged to three species: Neodohrniphora elongata Brown (54% of individuals), Neodohrniphora declinata Borgmeier (31%) and Neodohrniphora tonhascai Brown (15%). Individuals of each phorid species were observed attacking ants of A. sexdens rubropilosa in the foraging trails.
The three phorid species were found throughout the year and did not differ in seasonal patterns of activity (fig. 1), but the mean daily number of N. elongata (3.32±0.34) was higher than for N. declinata (1.77±0.30) and N. tonhascai (0.87±0.14) (χ2=14.60; P<0.001). The number of phorids captured did not differ significantly between the two years nor among bimesters (P<0.05).
Fig. 1. Daily mean number of individuals of three Neodohrniphora species in Viçosa, Brazil (□, N. tonhascai;
, N. declinata; ■, N. elongata).
The three Neodohrniphora species were found together at 37% of the samplings. They were found during both sampling times (in the morning and in the afternoon). Two species were found at 36% of samplings (23% N. elongata and N. declinata; 12% N. elongata and N. tonhascai; 1% N. declinata and N. tonhascai). N. elongata was found alone in 19% of the samplings, N. declinata in 7%, and N. tonhascai in 1%.
In laboratory tests, inspection flights were observed for the three phorid species with the four ant species. The mean duration and the 95% confidence interval of the inspection flights of N. elongata, N. declinata and N. tonhascai were 60.47 s (45.29–75.65 s) (n=50), 46.49 s (28.01–64.98 s) (n=16), and 67.86 s (32.07–103.65 s) (n=14), respectively. Except for A. subterraneus molestans, all ant species were attacked by the three phorid species (table 1). The mean frequency of attacking bouts and attacks on ants were 1.18±0.25 (n=80) and 0.76±0.10 (n=80), respectively. The mean duration and the 95% confidence intervals of attacking bouts and attacks were 0.32 s (0.18–0.47 s) (n=80) and 0.92 s (0.58–1.26 s) (n=80), respectively.
Table 1. Attacking bouts (A.bouts) and attacks (means±s.e) carried out by trhee phorids species in four species of the leaf-cutting ants.
Discussion
Differences in the frequencies of phorids were not detected over the year. However, the largest numbers were captured in September–October, which represents the onset of the rainy season. Erthal (Reference Erthal1999) found this seasonality for Neodohrniphora spp. attacking A. laevigata.
The higher frequencies of N. elongata observed may be due to either a relatively high abundance of this species in relation to N. tonhascai and N. declinata or to a higher preference of this parasitoid for A. sexdens rubropilosa, since N. declinata is described as a parasitoid of other ant species (Brown, Reference Brown2001).
In laboratory tests, the confidence intervals of inspection flight durations suggest a similar host inspection mechanism for the three phorid species. The duration of attacking bouts was significantly lower than the duration for attacks, and the absence of ovipositor penetration during these bouts may indicate that this behaviour allows flies to test the suitability of the potential host.
The attacks recorded on different ant species show that the three parasitoid species can attack non-host leaf-cutting ants. However, this plasticity is restricted because the absence of attacking bouts and attacks after inspections by flies on A. subterraneus molestans indicates that Neodohrniphora spp. are not very selective to locating hosts, but they have other mechanisms for host-recognition. Morehead & Feener (Reference Morehead and Feener2000) found similar results in relation to the phorid Apocephalus paraponerae (Formicidae: Ponerinae), which locate Paraponerae clavata and Ectatomma ruidum (Formicidae: Ponerinae) with equal efficiency but only attacks P. clavata. This plasticity in the host location mechanism probably serves to prevent premature rejection of individuals that belong to polymorphic host species.
N. declinata is described by Brown (Reference Brown2001) as a parasitoid of A. laevigata, but in this work we found females of this phorid all year long over foraging trails of A. sexdens rubropilosa in areas where there were neither trails nor nests of A. laevigata. Moreover, we observed that this parasitoid attacks these ants in field and laboratory conditions. Therefore, these results suggest that this phorid may be also a parasitoid of A. sexdens rubropilosa. However, to confirm this host-parasitoid relationship, it should be determined if a new generation of flies may be obtained from the attacked ants (Gilbert & Morrison, Reference Gilbert and Morrison1997; Porter, Reference Porter1998; Porter & Alonso, Reference Porter and Alonso1999).
As well as N. declinata, N. elongata and N. tonhascai also have behavioural plasticity to attack other non-host ants, including other genera. This plasticity of attack behaviour of Neodohrniphora spp. has also been reported by Bragança et al. (Reference Bragança, Tonhasca and Moreira2002), who verified, in laboratory conditions, parasitism of N. tonhascai on A. laevigata with phorids collected from A. sexdens rubropilosa trails. However, in these situations, it is difficult to suggest a host-parasitoid relationship because the attacks were only observed in a laboratory in a no choice situation, and we do not know if these phorids find the same ants in the field (Porter, Reference Porter1998; Orr et al., Reference Orr, Seike, Benson and Dahlsten2001).
The data presented in this paper demonstrate that N. declinata, as well as N. elongata and N. tonhascai, was present all year long over foraging trails of A. sexdens rubropilosa and attacks this ant in laboratory and field conditions. Further investigations must be carried out to determine if such attacks of N. declinata also produce a new generation of phorid flies in order to prove a host-parasitoid relationship.
Acknowledgements
We thank Brian V. Brown for helping with phorid identifications, M. Bragança for suggestions to improve the manuscript, and T.R. Souza, Denise D.O. Moreira and A. de Fátima for helping with laboratory work. V.S.G. Silva was supported by a grant from FAPERJ. This research was supported by the International Foundation for Science.
