PCR-based detection of prey DNA in the gut contents of the tiger-fly, Coenosia attenuata (Diptera: Muscidae), a biological control agent in Mediterranean greenhouses

335 Final formatted article © Institute of Entomology, Biology Centre, Czech Academy of Sciences, České Budějovice. An Open Access article distributed under the Creative Commons (CC-BY) license (http://creativecommons.org/licenses/by/4.0/). EUROPEAN JOURNAL OF ENTOMOLOG Y EUROPEAN JOURNAL OF ENTOMOLOGY ISSN (online): 1802-8829 http://www.eje.cz lyses of a predator’s gut contents provide a practical method for the identifi cation of prey (Chen et al., 2000; Harper et al., 2005; Sheppard & Harwood, 2005; Pompanon et al., 2012). The tiger-fl y Coenosia attenuata Stein, 1903 (Diptera: Muscidae) has particular characteristics that make it an interesting biological control agent for use in Mediterranean greenhouses. This species is currently distributed worldwide and probably originated in the Mediterranean region (Hennig, 1964; P ohl et al., 2012; Seabra et al., 2015; Bautista-Martínez et al., 2017). The adult tiger-fl y preys on adult whitefl ies, leafminers, drosophilids, thrips, leafPCR-based detection of prey DNA in the gut contents of the tiger-fl y, Coenosia attenuata (Diptera: Muscidae), a biological control agent in Mediterranean greenhouses SOFIA G. SEABRA1 * , JOANA MARTINS 2, PATRÍCIA BRÁS 1, ANA M. TAVARES 1, INÊS FREITAS 1, ANTÓNIO BARATA1, MARIA TERESA REBELO 3 , CÉLIA MATEUS 4, OCTÁVIO S. PAULO 1 and ELISABETE FIGUEIREDO 2 1 Centre for Ecology, Evolution and Environmental Changes (cE3c), Departamento de Biologia Animal, Faculdade de Ciências, Universidade de Lisboa, 1749-016 Lisboa, Portugal; e-mails: [email protected], [email protected], [email protected], ifi [email protected], [email protected], [email protected] 2 Linking Landscape, Environment, Agriculture and Food (LEAF), Instituto Superior de Agronomia, Universidade de Lisboa, Tapada da Ajuda, 1349-017 Lisboa, Portugal; e-mails: joanar[email protected], [email protected] 3 Centre for Environmental and Marine Studies (CESAM), Departamento de Biologia Animal, Faculdade de Ciências, Universidade de Lisboa, 1749-016 Lisboa, Portugal; e-mail: [email protected] 4 Instituto Nacional de Investigação Agrária e Veterinária (INIAV), Av. da República, Quinta do Marquês, 2780-157 Oeiras, Portugal; e-mail: [email protected] Key words. Diptera, Muscidae, Coenosia attenuata, predation, gut contents, prey molecular identifi cation, Polymerase Chain Reaction, probit models, half-time molecular detection Abstract. The tiger-fl y Coenosia attenuata Stein (Diptera: Muscidae: Coenosiini) is a generalist predator that preys on several pests of greenhouse crops and is considered a biological control agent in the Mediterranean region. Previous behavioural observations identifi ed its preferred prey, but a more in-depth evaluation will benefi t from using Polymerase Chain Reaction amplifi cation of prey DNA remains in the gut of this predator. To evaluate the rate of decay and suitability of this method for use in the fi eld assessments, we carried out a laboratory feeding calibration experiment on 355 females of C. attenuata, which were killed at different intervals of time after ingestion (10 time points from 0 to 48 h). The prey species tested were: Trialeurodes vaporariorum (Westwood) (Hemiptera: Aleyrodidae: Trialeurodini), Liriomyza huidobrensis (Blanchard) (Diptera: Agromyzidae), Diglyphus isaea (Walker) (Hymenoptera: Eulophidae: Cirrospilini), Bradysia impatiens (Johannsen) (Diptera: Sciaridae) and Drosophila mercatorum Patterson & Wheeler (Diptera: Drosophilidae: Drosophilini). Based on a probit model, amplifi cation success of prey DNA declined exponentially with increasing time after ingestion. The half-time molecular detection differed between species, ranging from an average of 5 h for T. vaporariorum and D. isaea, 6 h for B. impatiens, 15 h for L. huidobrensis to more than 40 h for D. mercatorum. This study confi rmed the feasibility of using DNA based detection to identify prey species in the gut contents of C. attenuata and provided calibration curves for a better understanding of predation activity in this agroecosystem. * Present addresses: S.G. Seabra – Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Rua da Junqueira nº 100, 1349-008 Lisboa, Portugal; J. Martins – Ascenza Agro, Lda., Alameda dos Oceanos 1.06.1.1, 1990-207 Lisboa, Portugal; I. Freitas – CIBIO/InBIO, Research Center in Biodiversity and Genetic Resources of the University of Porto, Vairão, Portugal INTRODUCTION Generalist invertebrate predators may be important biological control agents against crop pests, as shown in a number of manipulative fi eld experiments (Symondson et al., 20 02). When considering introducing or enhancing a particular generalist predator species for pest control, its prey preference and effect on pests (target) and non-pests (non-target) should be studied (Stilin & Simberloff, 2000; Louda et al., 2003), even when native control agents are considered (Howarth, 2000). Field observations of predator preferences are not always feasible and molecular anaEur. J. Entomol. 118: 335–343, 2021 doi: 10.14411/eje.2021.035 ORIGINAL ARTICLE 336 Seabra et al., Eur. J. Entomol. 118: 335–343, 2021 doi: 10.14411/eje.2021.035 contents of predators collected in greenhouses, whatever the method used to analyse the remains of the prey’s DNA. In this study, we determined the PCR detection time of prey DNA after ingestion in the gut contents of C. attenuata adults preying upon individual prey of fi ve different species in the laboratory. MATERIAL AND METHODS Laboratory feeding experiment Adults of C. attenuata used in this study came from a culture established at Instituto Superior de Agronomia (ISA) facilities in Lisbon, Portugal. The adults used for establishing the cultures were collected in greenhouses in the Oeste region of Portugal, mainly in Silveira, Torres Vedras municipality, on several occasions from 2010 to 2012, and periodically a few new adults from Silveira were added. Tiger-fl y culture was established following Martins et al. (2015): larvae were fed on fungus gnat larvae of Bradysia impatiens (Johannsen, 1912) (Diptera: Sciaridae) living in soil containing 90% organic matter and oats inoculated with Pleurotus ostreatus (Jacquin) Kummer, 1871 (Basidiomycota: Agaricales: Pleurotaceae); adults were fed Drosophila mercatorum Patterson & Wheeler, 1942 (Diptera: Drosophilidae). Cultures were kept in a room at 23–25°C, RH 65 ± 10% and a 12 h photoperiod. Species used as prey are important pests or biological control agents of protected vegetable crops in the region (Figueiredo et al., 2011): whitefl y Trialeurodes vaporariorum (Westwood, 1856) (Hemiptera: Aleyrodidae) (Tva), leafminer Liriomyza huidobrensis (Blanchard, 1926) (Diptera: Agromyzidae) (Lhu) and the eulophid Diglyphus isaea (Walker, 1838) (Hymenoptera: Eulophidae) (Dis). In addition, two other species kept in the culture unit, were also included: the fungus gnat B. impatiens (Bim) and drosophilid D. mercatorum (Dmrc). The whitefl y and leafminer were also reared in ISA facilities on tomato plants, under the conditions indicated above. D. isaea was acquired from Biobest (Diglyphus-system). Controlled feeding experiments were carried out during 2012 and 2013, with one adult female tiger-fl y, between 2–10 days old, put in a 10 × 10 × 10 cm transparent plastic cage with one or two adult individuals of a particular target species of prey (either Tva, Lhu, Dis, Bim or Dmrc). Prior to the beginning of each experiment, predators were starved for about 12 h. When each tiger-fl y ceased feeding (on one prey), the feeding time was recorded, and the tiger-fl y was placed in a box without prey and killed at a specifi c time after ingestion (0 min, 15 min, 30 min, 1 h, 2 h, 4 h, 8 h, 12 h, 24 h or 48 h) (Table S1). From previous observations, we knew that tiger-fl ies fed D. mercatorum would survive for at least 48 h (longest time period of the experiment), but if fed a smaller species of prey, they survive for 24 h. Bearing this in mind, we provided these tiger fl ies with an adult of D. mercatorum after 24 h so that they survived for 48 h (but not in the case of the prey D. mercatorum). We aimed at using 8 prey individuals in each experiment (for each period of time and species). However, depending on prey availability at the time of the experiments, the individuals initially included in the assays ranged from 2 to 23 (Table S1). We expected some mortality to occur during the experiments and tried to include more individuals whenever possible. Tiger fl ies (N = 407) were preserved in absolute ethanol and frozen. The temperature during the experiment was controlled at 25°C. Molecular analysis DNA extraction from each C. attenuata individual was done using the entire thorax and abdomen, instead of using only the dissected gut, in order to minimize the risk of cross-contaminahoppers, Psocoptera and fungus gnats, and the larva is also a predator, feeding on soil insects, such as, fungus gnat larvae (Kühne, 1998, 2000; Prieto et al., 2005; Martins et al., 2012; Mateus, 2012) . This species is adapted to high temperatures (Gilioli et al., 2005) , which is important for surviving inside Mediterranean greenhouses. It also kills more prey than it consumes, making it an effi cient biological control agent (Morris & Cloutier, 1987; Martinez & Cocquempot, 2000). Laboratory studies indicate that it prefers whitefl ies and leafminers to other species usually present in protected crops (Martins et al., 2012) . A method for mass rearing of C. attenuata has already been developed, which may be used to supplement tiger-fl y populations in commercial greenhouses (Martins et al., 201 5). To understand the effectiveness of this predator as a biological control agent in protected crops, it is necessary to identify the prey consumed under greenhouse conditions. This has already been established in the fi eld by collecting and identifying the remains of prey consumed by this predator ( Prieto et al., 2005; Mateus, 2012). This is, however, very time-consuming and biased in terms of what species of prey are identifi ed, which is dependent on their daily fl ight activity patterns (prey is captured by the predator during fl ight). The analysis of the gut content of arthropod insects by amplifi cation of prey DNA using Polymerase Chain Reaction (PCR) is an effi cient method of identifying prey (e.g. Chen et al., 2000; Foltan et al., 2005; King et al., 201 1), but this has not yet been done for Coenosia species. Detection times for DNA after ingestion depends on the predator, prey, rates of consumption and digestion and the length of the DNA fragment amplifi ed (Hoogendoorn & Heimpel, 2001; Gagnon et al., 2011). Thus, it is important to evaluate the rate of decay of DNA of different prey after ingestion and the suitability of this method for subsequent fi eld assessments. There are also several challenges in detecting degraded DNA after digestion using PCR methods: lack of sensitivity of primers, short post-ingestion detection periods and cross-amplifi cation of non-target DNA (King, 2008) . Although new methods of mass sequencing are now available and are increasingly used in metabarcoding and metagenomics (e .g., Pompanon et al., 2012; Paula et al., 2016; Macías-Hernández et al., 2018), PCR amplifi cation of prey DNA remains in the gut of a predator followed by gel visualization is less costly and continues to be useful, especially when the potential species of prey are relatively well-known, as is the case in biocontrol, and when calibration based on a large number of specimens is possible. In such experiments, the individual predators, after a period of starvation, are fed the target prey and are then killed at known time intervals after feeding and preserved for molecular gut content analysis. The proportion of predators for which the prey is detected is expected to decline exponentially with time since feeding and the half-life of detectability can be estimated using probit or logistic models (Greenstone & Hunt, 1993; Payton et al., 2003; Greenstone et al., 2014). This information will be important for interpreting future studies on the DNA of prey in the gut 337 Seabra et al., Eur. J. Entomol. 118: 335–343, 2021 doi: 10.14411/eje.2021.035 tion in such minute samples. From each predator, we also extracted DNA separately from the head, used as a negative control to ensure that the amplifi cations obtained from the thorax plus abdomen were not of the predator DNA. DNA extractions from individual predators and prey were done using the EZNA®Tissue DNA Isolation Kit (Omega Bio-tek). DNA of each prey was obtained from fresh prey insects and the mitochondrial cytochrome c oxidase subunit I gene (COI), a DNA barcode region widely used in insects (H ebert et al., 2003), was amplifi ed and sequenced. COI was amplifi ed using primers LEP-F and LEP-R (H ajibabaei et al., 2006), yielding a fragment of 620 bp. Polymerase Chain Reaction (PCR) was carried out in a Perkin Elmer 2700 thermocycler. PCR reaction volume of 12.5 μl contained 1 buffer (Promega), 2 mM of MgCl2, 0.1 mM of dNTPs, 0.4 μM of each primer, 0.25 U of GoTaq Flexi DNA polymerase (Promega) and approximately 10 ng of DNA. PCR conditions were: 94°C for 1 min, 5 cycles of 94°C for 30 s, 45°C for 1 min and 72°C for 1 min, followed by 30 cycles of 94°C for 1 min, 50°C for 1.5 min and 72°C for 1 min and a fi nal extension of 5 min at 72°C. PCR products were purifi ed with SureClean (Bioline) and Sanger sequencing of the forward sequence was done on an ABI3730XL, at Macrogen Europe. DNA sequences were checked and edited using Sequencher version 4.0.5 (Gene Codes Corporation). For aligning several sequences of this COI fragment from the predator and from different species of target prey, either already available in Genbank or sequenced by us (Genbank accession numbers MT428362 for Tva, MT428363 for Bim and MT428364 for Dmrc), we designed specifi c primers for each prey species with the help of Amplicon version 02 (Jarman, 2004). We aimed at amplifying short fragments between 100 and 250 bp, which are more likely to be detected in digested samples, in regions that differed between species. Primers for two different fragments for each species of prey (three for Tva) were designed. These primers were tested on the target species and the predator. Since all of them amplifi ed, the two (or three) primer sets were kept for this study, allowing replication within each species of prey in order to assess the consistency of amplifi cation. Polymerase Chain Reactions (PCR) were carried out using positive controls (DNA of prey) and negative controls (DNA from the head of the predator) and a blank. PCR reaction volume of 12.5 μl contained 1 × buffer (Promega), 1.8 mM of MgCl2, 0.1 mM of dNTPs, 0.24 μM of each primer, 0.25 U of GoTaq Flexi DNA polymerase (Promega) and approximately 10 ng of DNA. PCR conditions were: 94°C for 3 min, 30 cycles of 94°C for 1 min, 50°C for 45 s and 72°C for 45 s and a fi nal extension of 7 min at 72°C. PCR products were visualized on 1% agarose gels for checking amplifi cation success (Fig. S1). Data analysis Probit models were fi tted to the PCR detection success against time after ingestion (as in e.g. Ch en et al., 2000; Ma et al., 2005; Greenstone et al., 2007). Probit analysis is a type of regression applied when response variables are binomial and the relationship between the response and the predictors is sigmoid. The fi t of each model was tested against a null model (with just the intercept) by calculating the difference between the residual deviance of our model and the null model, and doing a χ2 test with 1 degree of freedom. The half-time molecular detection, i.e., the time after ingestion that corresponds to 50% of the predators testing positive for the DNA of the target prey, was calculated using the probit model coeffi cients: Probit(Y) = aX + b; a – decay rate, b – Y-intercept, X – time after ingestion; Y – probit of detection success. Solving for the value of X when Y = 0 (inverse normal of 0.5), gives the half-time. Probit analyses were done using generalized linear models (GLM) with a probit link function (scripts available in https://github.com/seabrasg/probit). Testing if the regression lines differed between different PCR fragments for the same prey species (PCR-1 and PCR-2, as well as PCR-3 for Tva) and between different species of prey, was done by comparing GLM models with and without the interaction term PCR × Time or the interaction Prey × Time (Robertson et al., 2017). In order to explore the relationship between the body sizes of the prey and half-time molecular detection, we calculated nonparametric Spearman correlations between average body length (values for each species from Figueiredo, unpubl. results) and average half-time molecular detection obtained here (average of the two or three primer sets of each species). In our experiments, we recorded the time each predator spent feeding on each prey, and correlated the average value for each species with both body length and half-time molecular detection. Regression models and correlations were done in R version 3.4.0. RESULTS A total of 407 tiger-fl ies was used in the predation calibration experiments. Two (or three) pairs of primers were designed to amplify two (or three) different fragments of the COI region of each species of prey (Table 1). When the amplifi cation was ambiguous for a particular primer set, it was considered to be missing data (Fig. S1), reducing sample size for some particular primer sets, particularly Tva-1 that was left with only one individual for some periods of time. In the case of D. mercatorum, the samples from the shorter time periods were used initially for PCR using other sets of primers designed to amplify D. melanogaster (designed from sequences obtained in GenBank). This occurred because we initially thought we were working with that species. However, the poor amplifi cation obtained with those primers made us suspect we were dealing with a different species. After sequencing COI gene for this prey (as we did for all the other prey) and blasting in NCBI, the Table 1. Sequence of primers specifi c for each species of prey and amplicon sizes for each pair of primers. Tva – Trialeurodes vaporariorum, Lhu – Liriomyza huidobrensis, Dis – Diglyphus isaea, Bim – Bradysia impatiens, Dmrc – Drosophila mercatorum. Prey species Fragment Primer 5’-3’ Sequence of primer Size of amplicon Trialeurodes vaporariorum Tva-1 Tva-125F CTTATGGTTGGGGCTCCTG 171 Tva-295R ATTCCCCCTATGTGACAGC Tva-2 Tva-113F TGGCTGGTTCCTCTTATGG 144 Tva-256R AACAGTTCAACCCGTGCCGC Tva-3 Tva-216F GACTTTTGATTGATAGGGGAAGC 192 Tva-407R CCATTTTTATGCCCAACGCC Liriomyza huidobrensis Lhu-1 Lhu-277F ACCTCCAGCTCTTACCC 202 Lhu-478R CGTATATTAATAATTGTTG Lhu-2 Lhu-327F CTGGGACAGGATGAACGG 219 Lhu-545R AAGAACTGCAGTAATTAG Diglyphus isaea Dis-1 Dis-384F ATTTAGGTCATAGAGGGCC 168 Dis-551R TGCAGTTAATAATATAGCCC Dis-2 Dis-323F CTAGAATATTTATTGGATC 138 Dis-460R ACCTATAATTGAAGAAGCC Bradysia impatiens Bim-1 Bdi-411F CACTTTTATTAACTAGAAGG 110 Bdi-520R ATAGATCAACAGAGGCCCCTGA Bim-2 Bdi-426F GAAGGTTAGTAGAAAGAGGTAC 199 Bdi-624R TCAAAAGATATTCCTGGGGC Drosophila mercatorum Dmrc-1 Dmrc-239F TAGTTCCATTAATACTTGG 176 Dmrc-414R TCCACCATGAGCAATTCCTG Dmrc-2 Dmrc-379F TACCCACCTTTATCTTCAGG 190 Dmrc-569R GCTGTAATTACTACAGATC 338 Seabra et al., Eur. J. Entomol. 118: 335–343, 2021 doi: 10.14411/eje.2021.035 best hit was with D. mercatorum. We designed the new primers for this species and used them in this study. Since at 1h and 2h every sample was detected, and later on the amplifi cation rate continued to be high, the shorter periods would likely give full or very high proportion of positive PCRs and therefore we did not perform them. Although the estimates of the probit model may be adversely affected by very low sample sizes, as seen in the large confi dence interval estimated for Tva-1, there was a good model fi t for most of the cases (see below). Comparing the regression lines obtained in PCR-1 and PCR-2 for each species of prey (and PCR-3 for Tva), they were not signifi cantly different (Ftests, p > 0.1), except for Lhu (p = 0.041) (Fig. 1; Table S1). Amplifi cation success of the DNA of prey by PCR declined exponentially with increasing time after ingestion and was well described by the probit models in all cases (highly signifi cant model fi t, p < 0.0001), except for D. mercatorum (Dmrc-1, p = 0.011, Dmrc-2, p = 0.087) (Table S1), for which after 48 h there was still a substantial detection (Fig. 1; Table S1). According to the fi tted models, the estimated half-time molecular detection ranged from 3 h to 6 h for T. vaporariorum, B. impatiens and D. isaea (lower limits of confi - dence intervals ranged from 0.42 to 2.63 h and the upper limits ranged from 10.76 to 18.76 h). The half-time was estimated to be 17.04 h (95% CI 8.03–37.86) and 12.13 h (95% CI 4.05–30.67) for L. huidobrensis’s Lhu-1 and Lhu-2, respectively. For D. mercatorum, the estimated half-times were longer (37.23 h, 95% CI 10.02–259.19 for Dmrc-1 and 47.90 h, 95% CI 9.93–567.24 for Dmrc-2). The wider confi dence intervals refl ect the lack of data for after 48 h. The regression lines differed signifi cantly between species (PCR-1 was used for each species) (Ftests, p < 0.05), except between Dis and Tva (p = 0.075), between Bim and Tva (p = 0.332) and between Dis and Lhu (p = 0.118). The time each predator spent feeding on prey ranged from 1 to 33 min and the median time differed signifi cantly (at 0.05 level) between species of prey (Kruskal-Wallis = 126.64, p < 2.2e-16, Fig. S2). Pairwise comparisons using Wilcoxon rank sum test revealed signifi cant differences between all species (p < 0.001), except between Dis and Bim (p = 0.14). The predator spent longer feeding on the drosophilid prey (Dmrc). The average half-time molecular detection of each species of prey was positively and signifi cantly correlated with the average time spent feeding on each prey (rS = 1.0, p = 0.017, N = 5) (Fig. S3). These two variables were positively but not signifi cantly correlated Fig. 1. Probability of detection by PCR of a single specimen of prey over time after ingestion. Continuous lines are fi tted probit models for each fragment amplifi ed (two fragments for each prey, except Tva with three fragments). Dashed lines are 95% confi dence intervals. Points are recorded values. Filled circles and orange lines – PCR fragment 1 (e.g. Lhu-1); crosses and blue lines – PCR fragment 2; empty diamonds and pink lines for Tva – PCR fragment 3. Tva – Trialeurodes vaporariorum, Lhu – Liriomyza huidobrensis, Dis – Diglyphus isaea, Bim – Bradysia impatiens, Dmrc – Drosophila mer catorum. 339 Seabra et al., Eur. J. Entomol. 118: 335–343, 2021 doi: 10.14411/eje.2021.035 (at 0.05 level) with average body size of each species of prey (rS = 0.9, p = 0.083, N = 5 in both cases) (Fig. S3). DISCUSSION The calibration feeding experiments carried out in this study allowed the determination of the probability of detecting the DNA of prey over time in the predator C. attenuata. The calibration regression lines between the two (or three) fragments of DNA analysed for each species of prey were consistent, which gives support to the results. For these particular sizes of DNA fragments, from 140 to 220 bp, most species of prey were detected up to 8, 12 or 24 h after ingestion. Only D. mercatorum was still detected after 48 h in about half of the predators, which means that it is likely that it would remain detectable for longer than two days. The ability to detect prey remains (DNA included) in the gut content of predators depends on several factors, such as the time after ingestion, meal size and digestion rates, which are infl uenced by ambient temperature and predator activity levels (e.g., Hagler & Na ranjo, 1997; Hoogendoorn & Heimpel, 2001; Greenstone et al., 2014). In this study there was a positive but non-signifi cant correlation between body size (at 0.05 level) and half-time detection, with the substantially bigger D. mercatorum having much longer half-time detection than the smaller species. However, size is clearly not the only factor involved since L. huidobrensis is slightly smaller on average than B. impatiens and it has longer detection times (average 15 h in Lhu against 6 h in Bim) and D. isaea is smaller than B. impatiens and they have similar detection times (6h for Bim and 5 h in Dis). In fact, a predator may not consume an entire prey and so the amount of time spent feeding may not be related to prey size. The strong positive correlation (non-parametric Spearman correlation = 1) here recorded between time spent feeding and half-time detection may indicate that feeding time is indeed a more reliable indicator of the time DNA will remain detectable in the gut contents of a predator. Longer or shorter detection intervals may each have advantages and disadvantages when analysing feeding periodicity in the fi eld (Hagler & Naranjo, 1997). If prey DNA remains detectable for long periods, the chances of detecting DNA of prey are increased, but the power to identify timing patterns of feeding is reduced. For example, in the case of a tiger-fl y sampled in a greenhouse for which D. mercatorum DNA is detected in its gut content, it may have fed at any time between several days to just before being collected. For smaller prey detection intervals, it is possible to have a better discrimination and identifi cation of feeding periodicity. Based on previous observations this predator is very voracious and able to capture large numbers of prey. In some situations, only some of them are actually consumed (Morris & Cloutier, 1987; Moreschi & Süss, 1998) and this can reduce detection times. In a 24-h period, tiger-fl y females can each consume up to 12.0 adult sciarids, or 23.8 Bemisia tabaci (Gennadius) whitefl ies, or 17.25 adult Liriomyza trifolli (Burgess in Comstock) leafminers (Tellez et al., 2009) or 5.7 adult D. mercatorum, 10.7 Trialeurodes vaporariorum whitefl ies or 11.2 Diglyphus isaea (Figueiredo et al., 2016). The higher the number of prey consumed the expectation is that the detectability times for particular prey species is increased, which means that the detection times obtained in this study are likely underestimated comparing to what happens in fi eld settings. Another factor to be considered is the ambient temperature, which is usually higher inside greenhouses than in the laboratory, which may have the opposite effect in reducing the detection time because of the resultant increased rate of digestion. Our aim in this study was to test DNA detectability and decay. In order to use these specifi c primers in fi eld studies we would additionally need to test for cross-amplifi cations, in order to guarantee the specifi city of the primers. These primers may then be used to study prey preferences and prey-switching behaviour. Despite all the possible factors affecting detectability, this study proved the feasibility of a DNA based detection and identifi cation of preyed species in the fi eld, which could be used to increase our knowledge about the predator’s diet (a more complete prey list) and feeding periodicity, more effi ciently than observations of behaviour in the fi eld. Even if future studies on the diet of C. attenuata in several greenhouse settings and over time will use DNA metagenomics shot-gun sequencing of gut contents (Paula et al., 2016), without the need for specifi c PCR primers for each prey species, the information that we obtained here on the detectability and decay times of DNA provide the basis for better planning and interpreting such studies. This approach will be dependent on the availability of reference sequence databases for prey taxa. ACKNOWLEDGEMENTS. This work was funded by national funds through FCT – Fundação para a Ciência e a Tecnologia in the frame of the projects PTDC/AGR-AAM/099723/2008, UID/BIA/00329/2019 (cE3c), UIDP/04129/2020 (LEAF), and UIDP/50017/2020 (CESAM), and the co-funding from FEDER, within the PT2020 Part nership Agreement and Compete 2020. We would like to thank the growers J. Firmino and L. Canuto who allowed us to collect samples in their greenhouses. We also thank the anonymous reviewers that contributed to an improvement of the manuscript. REFERENCES BAUTISTA-MARTÍNEZ N., ILLESCAS-RIQUELME C.P. & GARCÍA-ÁVILA C.J. 2017: First report of “hunter-fl y” Coenosia attenuata (Diptera: Muscidae) in Mexico. — Fla Entomol. 100: 174–175. CHEN Y., GILES K.L., PAYTON M.E. & GREENSTONE M.H. 2000: Identifying key cereal aphid predators by molecular gut analysis. — Mol. Ecol. 9: 1887–1898. FIGUEIREDO E., MEXIA A., MATEUS C., GODINHO M. & RODRIGUES S. 2011: Integrated pest management in vegetable protected crops in the Oeste region. — Acta Hortic. 917: 93–101. FIGUEIREDO E., MARTINS J., NUNES R., GARCIA A., FRANCO J.C. & MATEUS C. 2016: The tiger-fl y Coenosia attenuata as a potential biological control agent in vegetable protected crops. In Gerling D., Steinberg S. & Coll M. (eds): Proceedings of the 9th Meeting “Pest Management in an Ever-Changing World” at the Robert H. Smith Faculty of Agriculture, Food & Environment, The Hebrew University of Jerusalem, Rehovot (Israel), October 11-15, 2015. — IOBC-WPRS Bull. 119: 32–33. 340 Seabra et al., Eur. J. Entomol. 118: 335–343, 2021 doi: 10.14411/eje.2021.035 FOLTAN P., SHEPPARD S., KONVICKA M. & SYMONDSON W.O.C. 2005: The signifi cance of facultative scavenging in generalist predator nutrition: detecting decayed prey in the guts of predators using PCR. — Mol. Ecol. 14: 4147–4158. GAGNON A.E.Á., DOYON J., HEIMPEL G.E. & BRODEUR J. 2011: Prey DNA detection success following digestion by intraguild predators: infl uence of prey and predator species. — Mol. Ecol. Resour. 11: 1022–1032. GILIOLI G., BAUMGARTNER J. & VACANTE V. 2005: Temperature infl uences on functional response of Coenosia attenuata (Diptera: Muscidae) individuals. — J. Econ. Entomol. 98: 1524– 1530. GREENSTONE M.H, ROWLEY D.L., WEBER D.C., PAYTON M.E. & HAWTHORNE D.J. 2007: Feeding mode and prey detectability half-lives in molecular gut-content analysis: an example with two predators of the Colorado potato beetle. — Bull. Entomol. Res. 97: 201–209. GREENSTONE M.H., PAYTON M.E., WEBER D.C. & SIMMONS A.M. 2014: The detectability half-life in arthropod predator-prey research: what it is, why we need it, how to measure it, and how to use it. — Mol. Ecol. 23: 3799–3813. HAGLER J.R. & NARANJO S.E. 1997: Measuring the sensitivity of an indirect predator gut content ELISA: detectability of prey remains in relation to predator species, temperature, time, and meal size. — Biol. Contr. 9: 112–119. HAJIBABAEI M., JANZEN D.H., BURNS J.M., HALLWACHS W. & HEBERT P.D.N. 2006: DNA barcodes distinguish species of tropical Lepidoptera. — Proc. Nat. Acad. Sci. USA 103: 968–971. HARPER G.L., KING R.A., DODD C.S., HARWOOD J.D., GLEN D.M., BRUFORD M.W. & SYMONDSON W.O.C. 2005: Rapid screening of invertebrate predators for multiple prey DNA targets. — Mol. Ecol. 14: 819–827. HEBERT P.D.N., CYWINSKA A., BALL S.L. & DEWAARD J.R. 2003: Biological identifi cations through DNA barcodes. — Proc. R. Soc. (B) 270: 313–321. HENNIG W. 1964: Muscidae: Gattung Coenosia Meigen, section 63b. In Lindner E. (ed.): 7: Die Fliegen der Palaearktischen Region. Schweizerbart’sche Verlagsbuchhandlung, Stuttgart, pp. 518–619. HOOGENDOORN M. & HEIMPEL G.E. 2001: PCR-based gut content analysis of insect predators: using ribosomal ITS-1 fragments from prey to estimate predation frequency. — Mol. Ecol. 10: 2059–2067. HOWARTH F.G. 2000: Non-target effects of biological control agents. In Gurr G. & Wratten S. (eds): Biological Control: Measures of Success. Springer, Dordrecht, pp. 369–403. JARMAN S.N. 2004: Amplicon: software for designing PCR primers on aligned DNA sequences. — Bioinformatics 20: 1644– 1645. KING R. 2008: Molecular analysis of predation: a review of best practice for DNA-based approaches. — Mol. Ecol. 17: 947– 963. KING R.A., MORENO-RIPOLL R., AGUSTÍ N., SHAYLER S.P., BELL J.R., BOHAN D.A. & SYMONDSON W.O.C. 2011: Multiplex reactions for the molecular detection of predation on pest and nonpest invertebrates in agroecosystems. — Mol. Ecol. Resour. 11: 370–373. KÜHNE S. 1998: Open rearing of generalist predators: a strategy for improvement of biological pest control in greenhouses. — Phytoparasitica 26: 277–281. KÜHNE S. 2000: Räuberische Fliegen der Gattung Coenosia Meigen, 1826 (Diptera: Muscidae) und die Möglichkeit ihres Einsatzes bei der biologischen Schädlingsbekämpfung. — Stud. Dipterol. (Suppl.) 9: 1–78. LEORA SOFTWARE 1987: POLO-PC: A Users Guide to Probit or Logit Analysis. LeOra Software, Berkeley. LOUDA S.M., PEMBERTON R.W., JOHNSON M.T. & FOLLETT P.A. 2003: Nontarget effects – The Achilles’ heel of biological control? Retrospective analyses to reduce risk associated with biocontrol introductions. — Annu. Rev. Entomol. 48: 365–396. MA J., LI D., KELLER M., SCHMIDT O. & FENG X. 2005: A DNA marker to identify predation of Plutella xylostella (Lep., Plutellidae) by Nabis kinbergii (Hem., Nabidae) and Lycosa sp. (Aranaea, Lycosidae). — J. Entomol. Nematol. 129: 330–335. MACÍAS-HERNÁNDEZ N., ATHEY K., TONZO V., WANGENSTEEN O.S., ARNEDO M. & HARWOOD J.D. 2018: Molecular gut content analysis of different spider body parts. — PLoS ONE 13(5): e0196589, 16 pp. MARTINEZ M. & COCQUEMPOT C. 2000: La mouche Coenosia attenuata, un nouvel auxiliaire prometteur en culture protégée. — PHM Rev. Hort. 414: 50–52. MARTINS J., DOMINGOS C., NUNES R., GARCIA A., RAMOS C., MATEUS C. & FIGUEIREDO E. 2012: Coenosia attenuata (Diptera: Muscidae), um predador em estudo para utilização em culturas protegidas. — Revta Ciênc. Agrár. 35: 229–235. MARTINS J., MATEUS C., RAMOS A.C. & FIGUEIREDO E. 2015: An optimized method for mass rearing the tiger-fl y, Coenosia attenuata (Diptera: Muscidae). — Eur. J. Entomol. 112: 470–476. MATEUS C. 2012: Bioecology and behaviour of Coenosia attenuata in greenhouse vegetable crops in the Oeste region, Portugal. — Bull. Insectol. 65: 257–263. MORESCHI I. & SÜSS L. 1998: Osservazioni biologiche ed etologische su Coenosia attenuata Stein e Coenosia strigipes Stein (Diptera: Muscidae). — Boll. Zool. Agrar. Bachicolt. 30: 185– 197. MORRIS D.E. & CLOUTIER C. 1987: Biology of the predatory fl y Coenosia tigrina (Fab.) (Diptera: Anthomyiidae): reproduction, development, and larval feeding on earthworms in the laboratory. — Can. Entomol. 119: 381–393. PAULA D.P., LINARD B., CRAMPTON-PLATT A., SRIVATHSAN A., TIMMERMANS M.J.T.N., SUJII E.R., PIRES C.S.S., SOUZA L.M., ANDOW D.A. & VOGLER A.P. 2016: Uncovering trophic interactions in arthropod predators through DNA shotgun-sequencing of gut contents. — PLoS ONE 11(9): e0161841, 14 pp. POHL D., KÜHNE S., KARACA I. & MOLL E. 2012: Review of Coenosia attenuata Stein and its fi rst record as a predator of important greenhouse pests in Turkey. — Phytoparasitica 40: 63–68. POMPANON F., DEAGLE B.E., SYMONDSON W.O.C., BROWN D.S., JARMAN S.N. & TABERLET P. 2012: Who is eating what: diet assessment using next generation sequencing. — Mol. Ecol. 21: 1931–1950. PRIETO R., FIGUEIREDO E. & MEXIA A. 2005: Coenosia attenuata Stein (Diptera: Muscidae): prospecção e actividade em culturas protegidas em Portugal. — Bol. Sanid. Veget. Plagas 31: 39–45. ROBERTSON J.L., JONES M.M., OLGUIN E. & ALBERTS B. 2017: Bioassays with Arthropods. 3rd ed. CRC Press, Taylor & Francis Group, Boca Raton, 212 pp. SEABRA S.G., BRÁS P.G., MARTINS J., MARTINS R., WYATT N., SHIRAZI J., REBELO M.T., FRANCO J.C., MATEUS C., FIGUEIREDO E. & PAULO O.S. 2015: Phylogeographic patterns in Coenosia attenuata (Diptera: Muscidae), a widespread predator of insect species associated with greenhouse crops. — Biol. J. Linn. Soc. 114: 308–326. SHEPPARD S.K. & HARWOOD J.D. 2005: Advances in molecular ecology: tracking trophic links through predator – prey foodwebs. — Funct. Ecol. 19: 751–762. STILIN P. & SIMBERLOFF D. 2000: The frequency and strength of nontarget effects of invertebrate biological control agents of 341 Seabra et al., Eur. J. Entomol. 118: 335–343, 2021 doi: 10.14411/eje.2021.035 plant pests and weeds. In Follett P.A. & Duan J.J. (eds): Nontarget Effects of Biological Control. Springer Science + Business Media, New York, pp. 31–44. SYMONDSON W.O.C., SUNDERLAND K.D. & GREENSTONE M.H. 2002: Can generalist predators be effective biocontrol agents? — Annu. Rev. Entomol. 47: 561–594. TELLEZ M.D.M., TAPIA G., GAMEZ M., CABELLO T. & VANEMDEN H. 2009: Predation of Bradysia sp. (Diptera: Sciaridae), Liriom yza trifolii (Diptera: Agromyzidae) and Bemisia tabaci (Hemiptera: Aleyrodidae) by Coenosia attenuata (Diptera: Muscidae) in greenhouse crops. — Eur. J. Entomol. 106: 199–204. Received February 20, 2021; revised and accepted September 24, 2021 Published online October 28, 2021 Table S1. Proportion of PCR positive detections (Prop) for each fragment amplifi ed for each species of prey recorded in each time period after ingestion. N – number of C. attenuata tested. Results are shown for: Probit models; signifi cance of fi t compared to a null model (with just an intercept); half-time molecular detection; and comparisons between regression lines for each PCR fragment. Tva – Trialeurodes vaporariorum, Lhu – Liriomyza huidobrensis, Dis – Diglyphus isaea, Bim – Bradysia impatiens, Dmrc – Drosophila mercatorum. 342 Seabra et al., Eur. J. Entomol. 118: 335–343, 2021 doi: 10.14411/eje.2021.035 Fig. S1. Agarose gel images of PCR products after electrophoresis. Primer pairs are indicated at the top of the wells and sample identifi - cation below each well. The DNA ladder on the right of each fi gure ranges from 100 bp to 1000 bp fragments. 343 Seabra et al., Eur. J. Entomol. 118: 335–343, 2021 doi: 10.14411/eje.2021.035 Fig. S2. Distributions of time each predator spent feeding on each species of prey. Tva – Trialeurodes vaporariorum, Lhu – Liriomyza huidobrensis, Dis – Diglyphus isaea, Bim – Bradysia impatiens, Dmrc – Drosophila mercatorum. Fig. S3. Scatterplots and estimated Spearman correlation indices (rho) and corresponding p-values between pairs of the following variables: body length, feeding time and half-time molecular detection.