Fruit flies (Diptera: Tephritidae) are considered one of the main fruit pests worldwide (Enkerlin 2005). To reduce pest populations, various control tactics have been developed among which the augmentative release of parasitoids has arisen as one sound and well oriented strategy against these pests (Sivinski et al. 1996, Montoya et al. 2007).

Parasitoids are insects whose larvae develop by feeding in or on the body of other arthropods, usually insects; larval feeding almost always results in the death of the host (Godfray 1994). Parasitoids are immersed in a multitrophic context (Hassell and Waage 1984, Vet and Dicke 1992), where foraging for nutrients and hosts is performed at different scales (Kramer 2001, Gingras et al. 2002). In general, it is the female parasitoid that locates a suitable host. Because parasitoid development is dependent on limited resources (the host), adult preference and larval performance should be correlated to maximize fitness (Harvey et al. 2015) and the host acceptance procedure is considered the definitive step in host searching behavior (Vinson 1984). The hosts are often hidden in the interior of stems, leaves or fruits (Richerson and Borden 1972), consequently parasitoid females must detect and respond to a number of indirect signals where chemical-sensorial information plays a fundamental role (Vinson 1976, 1998, van Alphen and Vet 1986, Vet and Dicke 1992). In addition to chemical stimuli, parasitoids are also capable of identifying vibrations emitted by their hosts through the substrates in which they develop (van Alphen and Janssen 1982, Vet and van Alphen 1985, Meyhöfer et al. 1997).

Once female parasitoids have located their hosts, they have the capacity to distinguish between parasitized and not parasitized hosts, a strategy known as discrimination ability (van Alphen and Visser 1990). This ability can occur at three levels: (1) self-discrimination, (2) conspecific discrimination and (3) heterospecific discrimination (Mackauer 1990). This ability has been observed in many species of hymenoptera parasitoids (Vinson 1976), and is particularly important in the case of potential biocontrol agents, since these are expected to be efficient in host searching and to have the ability to discriminate between parasitized and non-parasitized hosts (van Lenteren et al. 1978). The latter helps females to avoid superparasitism, reducing the time and energy spent in searching behavior (Mackauer 1990, Godfray 1994).

Host location and host acceptance behavior has been widely studied in the generalist fruit fly parasitoid Diachasmimorpha longicaudata (Ashmead) (Greany et al. 1977, Lawrence 1981, Carrasco et al. 2005). This species is exotic in the Americas where it has been successfully reared for augmentative biological control of Anastrepha (Schiner) fruit flies in Mexico (Montoya et al. 2000, 2007) and in Florida USA (Sivinski et al. 1996); and for Ceratitis capitata (Wiedemann) in Argentina (Sanchez et al. 2016). However, there is a guild of native opiine braconid parasitoids (Sivinski et al. 2000, 2001) with potential as biocontrol agents, for which little information exists regarding their foraging and host acceptance behavior. This is the case of Doryctobracon crawfordi (Viereck), Utetes anastrephae (Viereck) and Opius hirtus (Fischer), all of which are solitary, larval-pupal endoparasitoids of Anastrepha spp. (López et al. 1999), that coexist in different regions of America (Sivinski et al. 2000). It has been postulated that differences in ovipositor size, as well as specific foraging behaviors, serve to prevent direct competition among these species (Ovruski et al. 2000, Sivinski et al. 2000). Under laboratory conditions, the three species can develop in the third instar larvae of Anastrepha ludens (Diptera: Tephritidae) (Aluja et al. 2009).

Doryctobracon crawfordi is native in habitats above 600 masl from Mexico to Argentina (Ovruski et al. 2005); possess a long ovipositor 5.39 ± 0.08 mm and attacks Anastrepha spp. mainly in citrus fruits and is sensitive to both high temperature and low humidity (Sivinski et al. 2000). Utetes anastrephae is characterized by a short ovipositor (1.57 ± 0.04 mm, Sivinski et al. 1997, 2001) and can be found associated with small fruits such as those of Spondias spp. (Anacardiaceae), with 2-5 cm of diam and 4 to 33 g weight (Avitia 2000). Opius hirtus is a more specialized parasitoid being recovered from Anastrepha obliqua (Macquart) in Spondias mombin L. and from Anastrepha alveata (Stone) infesting Ximenia americana L. (Olacaceae) (Sivinski et al. 2000). The three species are synovigenic (Sivinski et al. 2001).

The purpose of this study was to compare the foraging and host acceptance behaviors of the parasitoid species D. crawfordi, U. anastrephae and O. hirtus on previously parasitized and non-parasitized larvae of A. ludens, using video recording equipment under laboratory conditions. This knowledge should allow an improved understanding of the oviposition performance and potential of these parasitoid species as biocontrol agents against fruit fly pest species.

The general behavioral sequences of the three parasitoid species on the two host types were identified. The operational definitions for the observed behaviors are presented in Table 1. The most common sequence for any of the three-braconid species included: 1. Walking (W), 2. Searching for a host (S), 3. Detection of a host (D), 4. Oviposition attempt (OP), 5. Oviposition (O), 6. Rejection (RE), and 7. Failure (F) with some variants occurring depending on species (Figs 1a, b; 2a, b; 3a, b).

Knowledge on host acceptance behavior in insect parasitoids is fundamental to improve our understanding on the plant-herbivore-natural enemy tritrophic relations (Vet and Dicke 1992), as well as the population dynamics and their possible implications in pest biological control programs (Minkenberg et al. 1992).

Several studies have indicated that responses of natural enemies are mediated mainly by chemical signals detected in the environment (Vinson 1998, Vet and Dicke 1992, van Alphen and Visser 1990), by host-generated vibrations in its microhabitat (Meyhöfer et al. 1997, Vet and van Alphen 1985), and by the individual learning experiences of foraging females. Its stated that parasitoids perceive stimuli about host quality once direct contact has been made with the host, influencing the host acceptance process (Brodeur and Boivin 2004, Wajnberg et al. 2008). Here we established the patterns of host acceptance by D. crawfordi, U. anastrephae and O. hirtus in the presence of both non-parasitized and previously parasitized hosts. We further characterized the time spent in different behaviors, and the duration of the paralysis induced through parasitism.

The three studied species presented typical behavior of antennal contact with the surface of the oviposition unit during the process of searching for the host larvae, which is an important step for host detection (Leyva et al. 1991, Gonzalez et al. 2010). All of the species presented a similar foraging pattern, beginning the search for the host by walking and touching the oviposition surface with the antennae. Once a larva was detected, the females adopted an alert position that consisted of remaining immobile for some seconds with the antennae extended to the front, skimming the surface of the parasitization unit. On locating a larva, the females performed small turns on their axis until positioning their first pair of legs towards the front and arranging their ovipositor to form a 90° angle to the contact surface. On initiating the process of oviposition in the parasitization unit, as reported for D. longicaudata (Montoya et al. 2003), the females make various attempts to insert the ovipositor until contact was made with a host, which then was accepted or rejected.

According to our results, the three parasitoid species have a high discrimination ability in the form defined by van Alphen and Visser (1990), given that females rejected most of the previously parasitized hosts compared to those with no previous parasitization. However, U. anastrephae notoriously presented the highest frequency of rejection of parasitized hosts (79%). This suggests that this species possesses a high performance avoiding superparasitism and saving time and energy when foraging for their hosts (Godfray 1994, Mackauer 1990). This corroborate previous findings by Aluja et al. (2013) , who showed that this species avoids ovipositing on previously parasitized hosts under conspecific and heterospecific situations, although it also has been noted that superparasitized hosts yielded relatively more daughters (Alvarenga et al. 2016), as referred for D. longicaudata (Montoya et al. 2011). The host acceptance behavior presented by the three species was similar to that reported for D. longicaudata by Montoya et al. (2003), who observed that the previously parasitized hosts experienced a lower number of ovipositions than the hosts with no previous parasitization.

Doryctobracon crawfordi presented the longest time spent on oviposition compared to the other two species. Host acceptance may depend on extrinsic factors such as host availability and quality, as well as intrinsic factors such as the quantity of eggs in the females, the age and their nutritional state (Vet et al. 2002, Bernstein and Jervis 2008). In the case of D. crawfordi, availability of eggs in the females can be an important limiting factor (Iwasa et al. 1984). Females of this species may tend to be more selective, avoiding oviposition on previously parasitized hosts or those considered to be of poor quality (Rosenheim 1996, Ayala et al. 2014). This could explain the large numbers of oviposition attempts (host probing) observed and the greater time on selection of non-parasitized hosts. The native parasitoids D. crawfordi, U. anastrephae and O. hirtus invest more time in the process of oviposition (2.2 ± 0.8, 1.7 ± 0.1 and 1.52 ± 0.75 min, respectively) than exotic species such as D. longicaudata (0.49 ± 0.2 min; Montoya et al. 2003) and D. tryoni (0.69 ± 0.065 min; Ramadan et al. 1994) under laboratory conditions. This could be related to the level of host discrimination ability, since D. longicaudata has a strong tendency to superparasitize (Montoya et al. 2003) while the native species here studied seem to avoid superparasitism. In O. hirtus, the duration of oviposition when parasitizing previously parasitized hosts was significantly smaller (1.2 ± 0.1 min) than with non-parasitized l hosts (1.6 ± 0.1). The time invested in oviposition can vary according to the particular species and host size (Rivero 2000). In our study, females with experience that had contact with previously parasitized hosts, proved to be the most insistent and inserted their ovipositor a second time in order to conduct contact (D. crawfordi 16/30, U. anastrephae 22/30 and O. hirtus 16/30).

Though koinobionts do not arrest host development, some species can induce transient host paralysis (temporary paralysis after being stung by the female wasp; e.g. Desneux et al. 2009, Chau and Maeto 2009). Our data show that the three braconids studied here temporarily paralyzed their hosts, with 100 percent of hosts undergoing transient paralysis. Interestingly, the duration of paralysis was species specific. Larvae parasitized by U. anastrephae remained immobile for 13 ± 1 minutes, and thus presented this state for the shortest time, compared to the immobility presented by host larvae parasitized by D. crawfordi and O. hirtus (20.9 ± 1.1 and 23.8 ± 1 min, respectively). The duration of immobility caused by oviposition of the native parasitoids exceeds the time of immobility experienced by larvae parasitized by D. longicaudata (4.8 ± 27 min; Montoya et al. 2003).

The factors associated with host immobility are toxic substances in a mixture such as venom, as well as polydnaviruses (PDVs) that function as regulatory elements and disrupt the host metabolism (Moreau and Guillot 2005, Kaeslin et al. 2010), affecting the immune system (Richards and Parkinson 2000, Cai et al. 2004). Two hypothesis have been advanced to explain the adaptive value of transient host paralysis: 1) facilitation of oviposition by interfering with host defensive behaviors; and 2) self-superparasitism avoidance. Support for the latter hypothesis comes from the work of Desneux et al. (2009) with two species of aphidiine braconids of the genus Binodoxys that attack aphids, and from the work of Chau and Maeto (2009) with Meteorus pulchricornis also a braconid (Euphorinae) that attacks a wide range of lepidopteran larvae. Transient paralysis caused by Binodoxys spp. lasted up to 15 min and paralyzed aphids were accepted at a significantly lower rate than control aphids (Desneux et al. 2009). Likewise, in M. pulchricornis, host movements remained at a low level for approximately 1h after oviposition, and additional ovipositions on paralyzed hosts were not observed (Chau and Maeto 2009). We further hypothesize that transient host paralysis may also be a means to avoid host detection by conspecifics and heterospecific competitors, reducing the risk of larval competition not only from superparasitism but also from multiparasitism. In solitary endoparasitoids only one adult emerges per host, all other larvae are eliminated through direct (intrinsic) competition. Intrinsic competition in the guild of opine braconids that attack Anastrepha spp. has been demonstrated in U. anastrephae and D. areolatus, with the first instar larva of U. anastrephae being a superior competitor (Aluja et al. 2013). The duration of paralysis of the host in the three species studied here may allow some advantage to the developing embryo, delaying additional attacks. The first eclosed first instar larva might have more chances to win when competing with second laid individuals.

There are few studies regarding the oviposition behavior of opiine parasitoid species native to the Neotropical region, which makes our data of valuable importance. Our study reveals that behavioral differences among the studied parasitoid species lay mainly in the time of parasitization and in the time for which the parasitized hosts remained immobile, which could delay or minimize superparasitism. The three species were significantly capable of discriminating previously parasitized hosts, suggesting that this strategy is commonly present in the guild of fruit fly parasitoids attacking larvae in the Neotropics. Finally, our data also suggest that the studied species have the potential to be considered as suitable biological control agents. However, more studies are necessary to better understand the advantages and limitations that each one presents as natural enemies of fruit fly pests under field conditions.