Olives (Olea europaea L.) have been a representative crop of the Mediterranean Basin since they were domesticated some 6000 years ago (Besnard et al., 2013). Currently, the European Union produces 71.7% and 30.5% of the world's olive oil and table olives respectively ( http://www.internationaloliveoil.org/), with Spain being the main producer and exporter ( http://www.prosodol.gr/).

The olive fruit fly, Bactrocera oleae (Rossi, 1790), is the single most destructive olive pest. Its distribution is limited to areas with a Mediterranean climate and to where olive trees (domesticated or wild) are found. The females lay their eggs under the olive epidermis, and when the larvae hatch, they feed on the pulp. The galleries they leave behind also provide an environment in which microorganisms can grow. These insults reduce crop yields and the quality of the olive oil that can be produced (Daane & Johnson, 2010).

Unfortunately, controlling B. oleae with chemical insecticides can have undesired effects on other members of the arthropod fauna, and even on human health and can result in insecticide resistance (Pereira-Castro et al., 2015). Alternative, more environmentally friendly control methods are available, such as mass trapping or spraying with processed kaolin, but these methods can also negatively affect the arthropod community (Pascual et al., 2010).

Biological control may offer a solution to manage B. oleae populations. Attempts to use natural enemies such as parasitoids have been ineffective up to date (Daane & Johnson, 2010).The use of generalist predators, such as carabids that feed on the pupae of B.oleae, could be an alternative and promising approach to handle this pest (Dinis et al., 2015). However, selecting the most appropriate predator is difficult in the field, not only because the food web is complex, but also due to the presence of researchers making observations could alter natural behaviour (Rosenheim et al., 1999). Post-mortem visual examination of the gut contents of candidate predators might appear to provide another method of gathering information on the prey species chosen. Given the feeding habits of arthropods, and the size of any prey remains, such analyses are generally unsuccessful (Symondson, 2002). Currently it is possible to study the predator gut contents using PCR-based techniques. The reliable identification of a prey species requires that only prey DNA sequences are detected. The design of species-specific primers is therefore essential (O’Rorke et al., 2012). Since digestion of the prey results in degradation of its DNA, the mitochondrial cytochrome oxidase I (COI) gene appears as a suitable target for amplification because there are several copies of the gene per organelle and also several organelles per cell. The variability of this gene makes it suitable for distinguishing even very closely related species. This feature makes this gene to be considered as the universal marker for animal species identification (Pentisaari et al., 2016). In addition, the Barcoding of Life Database (BOLD), that contains a large library of eukaryote COI sequences, facilitates the design of species-specific primers (Ratnasingham & Hebert, 2007).

The aim of the present work was to develop a large primer set to detect small COI fragments belonging to B. oleae and to test them in specificity and sensitivity trials and also in the gut of carabid predators. This information will be essential to identify future candidate predatory species in the field for controlling this pest.

All the DNAs extracted from tephritids, carabids and other arthropods present at the olive agroecosystem were firstly used for PCR controls with the universal COI primers. Amplified fragments of around 700 bp confirmed the presence of suitable DNA in all the specimens for the subsequent analyses to be performed in this study.

The different combinations of the primers developed to specifically detect DNA of the olive fly allowed the amplification of seven fragments. Amplicon sizes varied from near 100 bp for the combination 253F-334R, to 300 bp for 334F-635R (Fig. 1). All the primer pair combinations were highly specific in the cross-species amplification tests. They only amplified DNA from B. oleae and not from other fruit fly species or potential predators. Fig. 2 shows an illustrative example with the primer combination 334F-440R.

Suitability and sensitivity of the primers were assessed in mixes of heterospecific DNA. Several dilutions of B. oleae in O. barbarus DNA were used for this purpose. All the primer combinations amplified the B. oleae DNA, even when it was present at a very low concentration compared to the predator DNA (1 pg/µL of B. oleae DNA per 5000 pg/µL of O. barbarus DNA) (Fig. 3).

Feeding trials revealed that pairs 253F-334R and 334F-440R exhibited the highest percentage of detection success 48 hours after predation and were able to detect B. oleae DNA even 72 hours after feeding (Table 2).

These two primer combinations were then tested in triplicate on independent PCR reactions (Table 3) to determine the ID50. The B. oleae DNA was amplified at 36 hours post feeding. The percentage of detection was over the 95% (98.98% for primer pair 253F-334R and 96.87% for 334F-440R). After 48 hours, the detection efficiency decreased to 70.3% in the three replicates and for both primer pairs and it was further reduced at 72 hours post feeding, as expected (41.6% and 50% for primer pairs 253F-334R and 334F-440R, respectively). The regression equations for the percentage of positives per replicate against time were Y _253F-334R = -0.719X+105.96 ( R ²=0673, p=0.001) and Y _334F-440R = -0.693X+104.255 ( R ²=0.556, p=0.001). Calculated ID50 were around 78 hours for both primer pairs.

Understanding the soil food web of olive orchards is essential to detect the potential of different arthropods as natural olive fruit fly control agents.The studies describing the insect fauna in the olive agroecosystem have inferred potential predators of B. oleae mainly based on trophic guild classification and relative abundance (Dinis et al., 2015; Gkisakis et al., 2016), and only recently a molecular method has been reported (Rejili et al., 2016).

In our study, we present a set of primers to detect the DNA from the olive fly in the gut of its putative predators. All the primers designed successfully amplified the B. oleae mitochondrial COI gene fragments, with amplicon sizes ranging from 81 bp to 302 bp. Given the DNA degradation during digestive processes, amplifying small fragments of multi-copy genes improves detection rates of prey DNA in predators’ gut (Brown et al., 2015; MacDonald et al., 2014). The amplicons obtained are within the size range for post-digestion amplification products reported in previous surveys with other primers and other pest species: 78-242 bp (Harper et al., 2005), 160-281 bp (Harwood et al., 2007), 101-274 bp (King et al., 2010), 130-330 bp (Monzó et al., 2010), and 150-345 bp (Moreno-Ripoll et al., 2012). The absence of cross reactivity shows that the primer pairs used are highly specific (Fig. 2).

The primers were also very sensitive in heterospecific mixes of DNA. Even though some tests where predator DNA was in great excess, all primer pair combinations returned a detection limit of 1 pg/µL of prey DNA (Fig. 3). This outcome surpasses that obtained for the same gene in a similar work on sharpshooters (6 pg/µL, León et al., 2006).

The specificity and sensitivity tests revealed that all primer combinations were appropriate for molecular detection of predation under laboratory conditions (O’Rorke et al., 2012). However, it was also necessary to test their reliability under real conditions of digestion. Therefore, feeding trials of a single pupa of B. oleae were carried using O. barbarus, one of the most abundant carabid species in the olive growing area of south-eastern Madrid, Spain. As these analyses were performed with the carabid species O. barbarus, post feeding intervals were chosen based on previous works that employed carabids and spanned the 0-72 hours periods (Harper et al., 2005; King et al., 2010; Sint et al., 2011; Monzó et al., 2011).

After a long time post feeding, the low concentration of prey DNA template compromises the success of its detection by PCR. Likewise, half-time of detection of prey DNA varies considerably among predators due to their feeding habits or the number of preys ingested (King et al., 2008; Gagnon et al., 2011). For ID50 determination, three independent post feeding replicates were performed in order to discard false negatives related to handling errors, small quantities or bad quality of prey DNA. The two primer combinations 334F-440R and 253F-334R were used given their detection efficiency.

Robust ID50 was obtained for both primer combinations, specifically 77.82 hours for 253F-334R and 78.29 hours for 334F-440R. These values were similar to those obtained in other works using a different kind of predators such as spiders: 78 h (Monzó et al., 2010) or 79.2 h (Sint et al., 2011). Nevertheless, they were higher than the estimated value for other species of carabids like Calathus fuscipes, 23.8-39 h (Boreau de Roincé et al., 2012); Oreonebria castanea, 30 h (Sint et al., 2011) or Pseudophonus rufipes, 32.33 h (Monzó et al., 2011). Our amplicon sizes were small (106 bp and 81 bp, respectively), which may explain their longer survival in the gut, consistent with previous studies (King et al., 2008).

A work closely related to ours has been published recently (Rejili et al., 2016). Three specific primers of B. oleae to amplify 108 and 214 bp COI fragments were developed, and only one of them, Sbo1R, overlaps with one of our primers, 635R. However, given the success of the pairs 334F-440R and 253F-334R, 635R was not further employed after the specificity and sensitivity assays. Diagnostic PCR assays on the guts of the carabid P. globosus at 0, 2, 4, 6, 8, 10 and 16 hours after feeding two B. oleae pupae in Rejili et al. (2016), determined an optimal DNA level of 100 ng for primers SBo1F-SBo1R and 50ng for the combination SBo2F-SBo1R. In contrast, the sensitivity of our whole primer set allowed amplifying at very low template concentrations (1pg/µL). In fact, our study goes a step further regarding the limits of detection and ID50 values. This not withstanding, all reports about predation of B. oleae by soil arthropods are very welcome for creating an effective synergy towards reducing insecticide and pesticide treatments.

The aim of integrated pest management programmes is to find new, environmentally friendly approaches to control pest populations. In this context, natural enemies can be a good complement to the use of insecticides. Despite the big efforts made studying the fauna at the olive agroecosystem until now, it is only possible to determine if one species is a potential predator or not, according to its trophic guild. The proposed PCR-based method can overcome this issue. The two primer pairs selected are highly specific and sensitive, ensuring the reliable detection of B. oleae prey remains in predator guts. Thus, possible candidacies of soil arthropods as biological agents for control of the olive fruit fly can now be reliably assessed and confirmed.

The authors thank Sonia A.P. Santos and Sergio Pérez-González for the taxonomic identification of the carabids and Manuel González, Marta Ortega and Ismael Sánchez for their efforts sampling in the field.