The high sensitivity and specificity of DNA-based diagnostic tools led to a new field of ecological research, molecular trophic interactions (González-Chang et al., 2016; King et al., 2008; Schmidt et al., 2021). The development of PCR-based diagnostic tools revolutionized studies of trophic interactions, given their advantage over previous approaches (i.e., gut content dissection and field observations) in detecting small DNA remains in arthropod digestive parts (Pompanon et al., 2012). Given their efficacy in providing detailed information on trophic linkages among species, molecular gut content analysis (thereafter MGCA) has been employed to determine herbivore–plant interactions (Avanesyan et al., 2021; Kheirodin et al., 2021; Wallinger et al., 2012), predator–prey interactions (King et al., 2008; Schmidt et al., 2021; Symondson, 2012), parasitoid–host interactions (Gariepy & Messing, 2012; Grab et al., 2018), intraguild predation (Traugott et al., 2012; Traugott & Symondson, 2008), and more recently to study plant–pest–predator interactions (Da Silva et al., 2019; Guenay et al., 2020).
Molecular diagnostic tools enable the detection of primary (i.e., feeding on target prey or plant DNA directly) and secondary feeding (i.e., indirect detection of the plant or insect DNA consumed by the prey in predators gut contents; Harwood et al., 2001) on target DNA remains in the gut contents of insect predators (Sheppard et al., 2005; Traugott & Symondson, 2008). The ability to detect target DNA post-feeding in predator gut contents can partly depend on the predator guild, mouthparts, and primer sensitivity. For instance, previous research indicates longer prey DNA half-life detectability in adult piercing-sucking predators relative to chewing predators (Greenstone et al., 2014; Kheirodin et al., 2022; Simmons et al., 2015) and primer-specific post-feeding target DNA detection patterns (Kheirodin et al., 2022). Therefore, these factors could influence the sensitivity of PCR-based methods in detecting primary and secondary DNA in predators. Using sensitive PCR-based or sequencing-based systems, DNA could be detected in the gut contents of predators from a variety of sources, including primary feeding on insect prey (Schmidt et al., 2021; Symondson, 2012) or secondary feeding on the insect prey inside the gut contents of primary prey (Sheppard et al., 2005; Traugott & Symondson, 2008). In addition, to prey insects, there are likely detectable sources of plant DNA within the gut contents of the consumed prey (Guenay et al., 2020). Often predator–prey interactions estimated from MGCA use the proportion of individuals positive for target prey species, assuming that predation occurred in the field where the predator sample was collected (e.g., Greenstone et al., 2010; Staudacher et al., 2016). However, generalist predators are known to move among fields to feed on alternative prey and plant carbohydrate and protein resources such as nectar and pollen (Samaranayake & Costamagna, 2018). Predation, therefore, could occur within the crop fields or in non-crop areas surrounding fields, and the sole presence of prey DNA in the gut contents of predators would not distinguish where predation occurred.
Given the known movement of generalist predators among habitats to forage on prey and non-prey resources, we aimed to develop a test case to track predation among fields using secondary plant feeding detection (e.g., plant DNA in the gut contents of prey eaten by predators). Such knowledge could lead to a better understanding of predator contributions to pest control services in the matrix of agricultural landscapes and further optimize the estimation of predation in specific crop fields. In addition, secondary plant feeding has been used to determine potential food web structural errors by distinguishing predatory carabids that directly feed on weed seeds from species that are testing positive (i.e., secondary feeding) due to feeding on herbivores that contain plant DNA in their gut contents (Guenay et al., 2020). While this knowledge can lead to a broader understanding of predators' role in landscape-wide pest control services, determining the source of the plant DNA remains in the gut contents of predators presents challenges given herbivore movement among fields (Kennedy & Storer, 2000; Mazzi & Dorn, 2012) and direct predator feeding on plant carbohydrates (He et al., 2021). Therefore, factors such as crop stage (i.e., flowering or not) and prey stage mobility could be essential in translating secondary plant feeding results by generalist predators. Overall, however, secondary plant feeding detection can enable testing hypotheses that MGCA proportion positive for prey calculated from a crop field could partially depend on the predator or pest movement from adjacent fields or non-crop habitats.
To advance methods for studying tri-trophic interactions under complex field conditions, we employed previously designed insect primers (Sint et al., 2014) and newly designed plant primers (i.e., current study, see methods) to isolate primary and secondary feeding of generalist predators. Currently, only one other study traced secondary plant DNA in predator gut contents and concluded that secondary plant DNA detection is rare in omnivorous predators (Guenay et al., 2020). However, plant identity (Wallinger et al., 2013) and primer-specific post-feeding detection times (Kheirodin et al., 2022) could affect plant DNA detection probabilities. The overall trend suggests that prey DNA post-feeding detection time is longer in predators with piercing/sucking mouthparts (e.g., predatory bugs and spiders) relative to chewing predators (e.g., beetles; Greenstone et al., 2014; Kheirodin et al., 2022), which warrants exploration of how different predator mouthparts could influence primary and secondary plant feeding. Therefore, our study focused on determining the sensitivity of MGCA in detecting secondary plant DNA in the gut contents of three contrasting generalist predators. Here, we hypothesized that (1) detecting secondary plant feeding can aid in tracing predators to source habitats where they consumed the prey, (2) secondary feeding detection probability is dependent on the predator mouthpart types, and (3) simultaneous detection of insect prey and plant DNA might aid in distinguishing primary and secondary plant feeding in predators. However, determining the source of plant DNA consumed by predators is challenging in agricultural landscapes or diversified cropping systems, given no direct way of distinguishing primary and secondary plant feeding is available. In this paper, we report results from a series of hierarchical laboratory feeding experiments to make methodological advances in studies of trophic interactions, using secondary plant feeding detection. The implications of this study and caveats around secondary plant DNA detection under field conditions are discussed.
MATERIALS AND METHODS Study systemThis experimental study provides the first methodological approach to simultaneously estimating generalist predator primary and secondary plant feeding. Through secondary plant feeding detection, we aim to show the possibility of linking consumed insect prey to the habitat where it was preyed upon by the predators. As a model system, we used the collard plant (i.e., collard leaves and pollen solution), a major Brassica crop grown in the Southeastern USA (USDA, 2019). We selected the diamondback moth (DBM), Plutella xylostella, given that (1) it is one of the most common herbivores of brassicaceous crops and (2) its ability to cause significant yield loss worldwide and in the United States (Furlong et al., 2013). To ensure the approach will capture plant DNA movement in a wide variety of predator taxa, we selected three representative predators with chewing and piercing-sucking mouthparts, given the significant effect of predator mouthparts (i.e., piercing/sucking and chewing) on post-feeding detection time (Greenstone et al., 2014; Kheirodin et al., 2022). We selected C. septempunctata as a representative of chewing predators, given its (1) ecological plasticity and adaptations likely resulting in widespread populations (Hodek & Michaud, 2008), (2) generalist nature that feeds on various insect pests (Hodek et al., 2012; Michaud et al., 2012), and (3) known feeding on various plant nectar and pollen (Ricci et al., 2005). To represent piercing/sucking predators, we selected Geocoris punctipes (Say) (Hemiptera: Geocoridae) and Pardosa spp. (Araneae: Lycosidae), given their high abundance in our region's agricultural fields (Kheirodin et al., 2022) and their significant contribution to the control of major insect pests such as whiteflies (Vandervoet et al., 2018) and DBM (Huang et al., 2018; Muckenfuss, 1992). Under laboratory-controlled conditions, we determined these three predator species' primary and secondary feeding on DBM and collard DNA.
Primer designWe sought specific gene regions for amplifying B. oleracea var viridis. Due to the unavailability of this plant's genome in public databases, we leveraged genomes from other Brassica species to identify conserved regions suitable for Brassica genus-specific primers. Selected candidate genes for Brassica species-specific amplification encompassed chloroplast DNA (cd) genes, such as maturase (matK), transfer RNA genes (trnT and trnF), as well as ribulose bisphosphate carboxylase large subunit (rbcL) gene. We also examined the external transcribed spacer (ETS) region, 5S rDNA region, phenylalanine ammonia-lyase (PAL), transferase family protein, and pectin esterase genes. DNA sequences for these genes were obtained from the National Centre for Biotechnology Information (NCBI) and subjected to BLASTn searches using CLC genomic workbench version 11.0.1 (Qiagen Bioinformatics) against different Brassica species' whole genome sequences (Table 1). Following alignment using the MAFT online tool (Multiple Alignment using Fast Fourier Transform; Katoh et al., 2002), we identified conserved portions exclusive to the Brassica species selected for primer design. The PAL region was chosen for genus-specific amplification due to its notable variation among various Brassica species. Subsequently, we designed PCR primers within the PAL region using CLC genomics workbench 20.0 (Qiagen) to yield 216–277 bp products suitable for gut content analysis (King et al., 2008). In total, five primer pairs were designed, with three specific to B. oleracea (not reported) and two specific to the Brassica genus (Table 3). For DBM detection, we employed preexisting Lepidoptera-specific primers (Sint et al., 2014), including Lep-gen-S274 (5′-GCAAGCCGTATTAAGGCGAT-3′) and Lep-gen-A275 (5′-CCCATCGCTGGTCAGAGTT-3′).
TABLE 1 Isolate numbers and genome sequence information used for
Species | Common name | GenBank accession number | References |
Brassica rapa | Field mustard | AENI00000000 | Wang et al. (2011) |
Brassica oleracea | Wild cabbage | JJMF00000000 | Parkin et al. (2014) |
Brassica juncea | Major oil seed crop | JAATWR000000000 | Yang et al. (2021) |
Brassica napus | Rape seed | JAGKQM000000000 | Chalhoub et al. (2014) |
Brassica nigra | Black mustard | JAEACV000000000 | Paritosh et al. (2020) |
For in silico primer specificity testing, we used Primer-BLAST (Ye et al., 2012) to assess the designed primers against all available plant and arthropod genomes in the NCBI database. Parameters were set to ignore non-target species with more than four total mismatches and the target species with more than one mismatch. In the laboratory, we conducted in-vivo primer specificity tests on four target plants, screening 21 non-target plants and seven non-target insect species to ensure no cross-amplification occurred (Table 2). These non-target selections were based on their prevalence in the Southeast USA (plants) and Georgia agricultural fields (insects) (USDA, 2019). The primer set was individually tested on each non-target DNA and on a mixed non-target DNA sample to confirm no amplification of non-target species in either single or mixed DNA scenarios. Sensitivity testing involved diluting the pure collard extracted DNA stock into four 10-fold series dilutions (1:10–1:10,000), each replicated three times. The initial collard DNA stock concentration was estimated with a NanoDrop One/Onec (ThermoFisher Scientific), and the dilution series was created accordingly (for more details, see Kheirodin et al., 2021). For DNA extraction protocol and PCR optimization, see supplementary files (S1).
TABLE 2 The list of target and non-target insect and plant species used to test in-vivo specificity of
Non-target plant species tested | Non-target insect species tested |
Solanum lycopersicum; Tomato | Bemisia tabaci; Whitefly adult |
Vigna unguiculate; Blue lake snap peas | Acrosternum hilare; Stink bug adult |
Gossypium hirsutum; Cotton | Melanoplus ponderosus; Grasshopper |
Solanum melongena; Eggplant | Orius spp.; Minute pirate bug |
Sorghum bicolor; Sorghum | Aphis gossypii; Cotton aphid |
Prunus persica; Peach | Geocoris spp.; Big-eyed bugs |
Rudbeckia hirta; Susan | Coccinella septempunctata; 7 spotted lady beetle |
Cucumis melo; Athena cantaloupe | |
Allium cepa; Onion | |
Glycine max; Soybean | Target plants species tested |
Arachis hypogaea; Peanut | Brassica napus; Oilseed rape |
Cucumis sativus; Cucumber | Brassica oleracea; Collard |
Clitoria ternatea; Butter peas | Brassica rapa; Oilseed, Mustard |
Capsicum annum; Jalapenos pepper; Bell pepper | Brassica spp.; Wild mustard |
Ipomoea batatas; Sweet potato | |
Phaseolus vulgaris; Beans | |
Zea mays; Maze, Corn | |
Abelmoschus esculentus; Okra | |
Solanum tuberosum; Potato | |
Citrullus lanatus; Watermelon | |
Cucurbita pepo; Squash |
Note: None of the non-target plant and insect species tested were amplified by the Brassica-specific primer under singleplex tests of cross-reactivity. All four tested Brassica targets were amplified by the Brassica genus-specific primer.
Laboratory experiments to trace primary and secondaryThe individual C. septempunctata, G. punctipes, and Pardosa spp. were collected from the University of Georgia (UGA Tifton Campus) Horthill experimental farm (31.4722613 N, −83.5283317 W, Tift County) from non-crop vegetations, including wild mustard and buckwheat, using a sweep net and aspirator. Predators were immediately transferred to plastic containers containing a lid covered by fine mesh to allow airflow and were stored in a growth chamber (25°C, L16: D8). We added wet filter paper to each container to provide water for predators. The DBM larvae were purchased from Frontier Agriculture Sciences (Frontier Agri-Science) and housed inside rearing cages (Bugdorm®, W60 × D60 × H60) with eight pots with collard (B. oleracea) plants. The fourth instar DBM larvae were used for the feeding trial. The collard seeds were purchased from Seed Barn® and planted in pots (W8 × D10) 2 weeks before starting the colony and trials. Predators were starved for 48 h before the trial to ensure that they were ready to feed on the given food. DBM larvae were provided with (1 cm length × 1 cm width) collard leaf and were kept under constant observation to ensure they fed the entire collard leaf before the trial. Wild mustard flowers were collected from Horthill UGA experimental plots at the UGA Tifton campus. All flowers were inspected in the laboratory to ensure no other insect was present. The pollen solutions were created thoroughly grinding 120 grains of wild mustard pollen with a pestle in a 1.5 mL Eppendorf tube containing 1000 μL of PCR-grade water. We used 1000 μL of PCR-grade water with 0.1 g of sugar for the control solution.
Half-life detectability experimentsThis experiment was carried out to determine the target DNA half-life detectability in the gut contents of predators and herbivores. Our objective was to determine how long after feeding the primary and secondary plant DNA is detectable in the gut contents of generalist predators with a 50% probability. Four independent feeding experiments were conducted to estimate the half-life detectability of the primary (DBM larvae, pollen solution) and secondary feeding (collard DNA in DBM larvae) in the gut contents of C. septempunctata under controlled environmental conditions (25°C, L16: D8). Predators and DBM were starved 48 h before the trial to ensure their guts were empty and ready to feed on the given prey/plant. We randomly assigned a time interval to each Petri dish representing the time post-feeding at which the individual predator/DBM should be transferred to ethanol. One adult C. septempunctata was placed inside each Petri dish and, depending on the objective, provided with either one-fourth instar DBM larvae that fed on collard (0.5 cm length × 0.5 cm width) on top of a piece of a collard leaf (1 cm length × 1 cm width), or 20 μL of collard pollen solution (120 pollen grains in 1000 μL of water) with 10 replications. The collard leaf was placed inside the Petri dish to ensure that larvae had continued access to the target DNA (collard). The subsequent feeding trial was performed using fourth instar DBM larvae that fed on a piece of a collard leaf (0.5 cm length × 0.5 cm width) to estimate the half-life detectability of collard DNA in the gut of DBM. Geocoris punctipes and Pardosa spp. were provided with DBM larvae on top of a collard leaf to investigate the possibility of secondary feeding detection in piercing/sucking predators and estimate the decay of collard in their gut. At the designated time intervals of 0, 2, 8, 16, 32, and 48 h (only spider) post-feeding, all individuals were transferred to 95% ethanol and stored at −20°C till extraction. After extraction, using Brassica and Lepidoptera-specific primers, we screened the predators for the presence of target DNA in their gut contents.
Unrestricted foraging and secondary feeding detection experimentOur goal in this experiment was to provide predators with simultaneous primary and secondary plant food resources and, without observation, test the possibility of primary and secondary plant DNA detection within 12 h trial window to mimic field-collected predators (i.e., unknown time or incident of feeding). To generate predators exposed to prey and plant DNA sources simultaneously, we allowed predators to feed freely on primary (i.e., DBM, pollen solution, and wild mustard flowers) and secondary DNA sources (i.e., collard DNA in gut contents of DBM) under controlled conditions (L:D 16:8, 25°C, ~40% RH). This experiment was carried out inside Petri dishes (60 × 15 mm, Falcon®) for 12 h post-release, without continuous observation or recording of the time/incident of feeding. Unlike the half-life feeding trials, in this experiment, predators had access to plant and insect materials simultaneously. Given our findings (Section 3.2) that suggest secondary plant DNA detection success was negligible in G. punctipes and Pardosa spp., we excluded these predators from this feeding experiment and conducted this experiment with C. septempunctata. Four treatments were employed with 10 replications to test primary and secondary insect and plant DNA detection in C. septempunctata gut contents. The treatments were: (1) C. septempunctata adult with DBM larvae fed on collard leaf, which was placed on the piece of collard leaf (1 cm length × 1 cm width), to provide continuous access to the target plant species (i.e., secondary plant feeding), (2) C. septempunctata adult with DBM larvae fed on collard leaf, 20 μL of mustard pollen solution, and one mustard flower (i.e., primary and secondary plant feeding), (3) C. septempunctata adult with only 20 μL of mustard pollen solution, and one mustard flower (i.e., primary plant feeding), and (4) C. septempunctata adult with sugar solution (1000 μL of PCR-grade water with 0.1 g of sugar) as a control. Predators were allowed to feed freely on food items offered for 12 h, at which point, each was transferred to their own vial containing chilled 95% ethanol and stored in a − 20°C freezer until DNA extraction. We then used BS1 and Lepidoptera-specific primers to screen the predators for primary and secondary plant and insect feeding. Finally, we tested the possibility of differentiating primary and secondary plant feeding based on amplified plant DNA concentration using the QIAxcel Advanced System (i.e., amplification signal strength calculated by the relative fluorescent units).
Statistical analysisWe used probit regression to estimate the half-life detectability of collard DNA in DBM gut contents, and for primary (DBM larva, pollen solution, and flowers) and secondary (collard DNA in DBM gut) feeding by C. septempunctata, G. punctipes, and Pardosa spp. This analysis shows how long after feeding the plant and insect prey DNA is traceable, with a 50% probability in the predator's gut contents (Greenstone et al., 2014). The binary presence/absence of target DNA was the dependent variable, and post-feeding time intervals were the independent variable. To assess the effect of predator mouthparts on primary (DBM) and secondary (collard) DNA detection half-lives, we merged the binary positive/negative data of the three predator species for each target DNA. Then, to assess predator-specific decay curves, we fit a binomial generalized linear model (GLM) with probit link function, with binary positive/negative PCR results for target DNA as the dependent variable and predator species as the independent variable. Likewise, we fit binomial GLMs to test the effect of the diet treatments (i.e., only pollen solution, pollen solution and collard-fed DBM larva, and only collard-fed DBM) on collard DNA detection probabilities.
Furthermore, to test the diet treatment effect on PCR product concentration, we used a generalized linear model with a negative binomial distribution, with PCR product concentration as a dependent variable, and treatments (i.e., only pollen solution, pollen solution, and collard-fed DBM larva, and only collard-fed DBM) as the independent variable. We calculated the probit model Wald χ2 using the “Anova” function of the car package to assess the significant effect of predator species and treatments on target DNA detection frequency and concentration (Fox & Weisberg, 2018). The significant effects were then followed by pair-wise tests among independent variables (i.e., predator species and feeding trial treatments) using the function “wald.test” of the aod package (Lesnoff & Lancelot, 2020). All statistical analyses were conducted in R version 4.1.2 “Bird Hippie” (R Development Core Team, 2021).
RESULTS Primer specificity and sensitivityThree initial primers were designed to specifically amplify B. oleracea (collard, data not shown) but failed to amplify non-diluted collard DNA (95 ng/μL). Therefore, we excluded them from this study. Given the strong similarities in Brassica genomes and the low sensitivity of primers designed specifically for B. oleracea, we designed two primer sets specific to the genus Brassica (Table 3). Both genus-specific primers (i.e., BS1 and BS2) successfully amplified all four tested target Brassica species and none of the non-target plant and insect species (Table 2). The NCBI Primer-BLAST further confirmed the specificity of the designed BS1 and BS2 primers for the tested four Brassica species with 100% similarities. Further, the Primer-BLAST result indicated the possibility of amplifying two other Brassica species not tested in this study (Raphanus sativus (reddish) and Eutrema salsugineum) using BS1 and BS2 primers with up to 97% similarities. In addition, the Primer-BLAST confirmed the specificity of these primers and indicated that no non-target species exist for the designed primer pairs. Both BS1 and BS2 Brassica genus-specific primers were sensitive and detected Brassica in samples at concentrations of 0.1 ng/μL. However, we observed higher quality bands (i.e., darker bands with a higher concentration of amplicons, as determined by capillary electrophoresis using the Qiaxcel Advanced System, Qiagen) using the BS1 primer set and used this primer for secondary plant feeding screening. Since Sint et al. (2014) performed extensive primer sensitivity and specificity tests on Lepidoptera-specific primers, we did not perform any additional testing on Lep-gen primers.
TABLE 3 List of designed primers from the phenylalanine ammonia-lyase (PAL) gene, along with detailed information regarding their target plant species, primer sequences, expected band sizes, and optimized annealing temperature (°C).
Target plant species | Primer names | Primer sequences (5′-3′) | Annealing temperature (°C) | Primer size (bp) |
B. oleracea, B. rapa, B. napus, Raphanus sativus, Eutrema salsugineum |
BS-F1 BS-R1 |
GAAGAGGATGGTTGAGGAGTA GATGAGTTCCGTTTGCAATG |
56.17 55.65 |
277 |
B. oleracea, B. rapa, B. napus, R. sativus |
BS-F2 BS-R2 |
TCCACTCCGGTGGACCTTAT CACCGCTCATGCAAAACACA |
59.96 59.97 |
216 |
A series of feeding trials were conducted to estimate the detectability of target DNA half-life in the gut contents of C. septempunctata, G. punctipes, Pardosa spp., and DBM larvae. We aimed to determine how long after feeding, plant DNA (primary and secondary) remains detectable in the gut contents of these predators. These feeding trials were necessary for designing an experiment within the detection possibility of our primers (section 3.4; predators freely feeding on plant and prey resources). The slope of the probit regression significantly differed from zero on all estimated half-lives in C. septempunctata and DBM larvae. The detectability of mustard pollen solution, DBM DNA, and secondary collard feeding in the C. septempunctata gut were 15.5 h (slope = −0.05521, z-value = −2.20, df = 49, p = 0.017; Figure 1), 24.2 h (slope = −0.07387, z-value = −3.74, df = 49, p < 0.0001; Figure 1), and 5.5 h (slope = −0.05514, z-value = −2.80, df = 49, p = 0.004; Figure 2) post-feeding, respectively. The detectability of collard DNA half-life in the DBM gut was 24.3 h (slope = −0.09281, z-value = −4.31, df = 49, p < 0.0001; Figure 1). The detectability of DBM DNA in the gut of G. punctipes and Pardosa spp. was 32.5 h (slope = −0.06113, z-value = −2.84, df = 49, p = 0.004) and 41.5 h (slope = −0.05839, z-value = −3.03, df = 49, p = 0.0022) post-feeding, respectively (Figure 1). However, the slope of the probit regression was not significantly different from zero for secondary collard feeding in G. punctipes (slope = −0.1777, z-value = −0.95, df = 49, p = 0.3340; Figure 2) and Pardosa spp. (slope = −0.2634, z-value = −1.49, df = 59, p = 0.136; Figure 2). Therefore, we could not produce half-life detectability curves for these two species because only 4% of G. punctipes and 8.3% of Pardosa spp. tested positive for secondary feeding on collard DNA.
FIGURE 1. Results from DNA detectability half-life experiments. The proportion of Geocoris punctipes positive for (a) diamondback moth (DBM) DNA; the proportion of Pardosa spp. positive for (b) DBM DNA; the proportion of Coccinella septempunctata positive for (c) DBM DNA, and (d) collard DNA from direct feeding on collard pollen solution; and (e) the proportion of diamondback moth larvae positive for (a) direct feeding on collard leaf DNA at the intervals of 0, 2, 8, 16, 32, and 48 h (only spider) post-feeding, using the Brassica genus-specific (BS1) and Lepidoptera-specific primers (Sint et al., 2014). The solid horizontal line indicates the fitted probit model with 95% confidence intervals (dashed lines). The vertical line indicates the detectability of the target DNA half-life (Probability of DNA amplification with 50% chance) in the gut content of C. septempunctata, G. punctipes, Pardosa spp., and DBM. The black dots represent proportion positives for diamondback moth, collard and collard pollen solution at each post-feeding time interval, calculated using 10 replications at each time interval (number tested/number positive).
FIGURE 2. Results from DNA detectability half-life experiments. The proportion of Geocoris punctipes positive for (a) collard DNA in the gut contents of diamondback moth (DBM); the proportion of Pardosa spp. positive for (b) collard DNA in the gut contents of DBM, and the proportion of Coccinella septempunctata positive for (c) collard DNA in the gut contents of DBM at the intervals of 0, 2, 8, 16, 32, and 48 h (only spider) post-feeding, using the Brassica genus-specific (BS1) primers. The solid horizontal line indicates the fitted probit model along with its 95% confidence intervals. The vertical line indicates the detectability of the secondary collard DNA half-life in the gut contents of G. punctipes, Pardosa spp. and C. septempunctata. The black dots represent the proportion positive for secondary collard DNA at each post-feeding time interval. The “n.s” indicate that the slope of the fitted probit model is not significantly different from zero, and no half-life/confidence interval is estimated.
We aimed to provide the initial evidence of the effect of predator mouthparts (i.e., chewing and piercing/sucking mouthparts) on post-feeding primary (DBM and collard DNA) and secondary plant (collard) DNA detection possibilities in three generalist predators. There were no significant overall differences in DBM DNA half-life detectability among predators with different mouthparts (χ2 = 2.99, df = 2, p = 0.22). On the contrary, we found significant effects of predator mouthparts on post-feeding secondary collard detection (χ2 = 25.28, df = 2, p < 0.0001). The collard detectability half-life was significantly longer in C. septempunctata (i.e., chewing predator) relative to piercing/sucking predators, including G. punctipes (χ2 = 15.0, df = 1, p < 0.0001) and Pardosa spp. (χ2 = 14.8, df = 1, p = 0.0001).
Unrestricted foraging and secondary feeding detectionThis experiment was designed to simulate field-collected predators, where predators have access to prey and plant carbohydrate sources, and the time and incident of feeding are unknown. The 12 h duration of this trial was selected based on secondary collard half-life detectability (Figure 2) in C. septempunctata. Our findings suggest that the Brassica genus-specific primer pair (BS1) is sensitive to detect both primary and secondary plant feeding by C. septempunctata over 12 h of this laboratory experiment. We found evidence of primary and secondary feeding simultaneously, where the C. septempunctata had both DBM and collard DNA in their gut contents in five out of 10 replications. In all treatments with DBM larvae, we traced DBM DNA in the gut of C. septempunctata. Out of 10 replications with only collard-fed DBM, 50% of C. septempunctata individuals tested positive for secondary collard DNA. The prevalence of primary collard DNA detection was slightly higher, where out of 10 replications with collard pollen solution and flower, 60% of individuals tested positive for collard DNA (Table 4). Out of 10 replications with the simultaneous presence of collard-fed DBM and pollen solution, 100% of individuals tested positive for collard DNA. We found a significant effect of treatments (i.e., excluding sugar solution from the analysis) on collard DNA detection probability (χ2 = 8.71, df = 2, p = 0.01). While the overall model was significant, and the probability of collard detection was higher when collard DNA was present in both primary and secondary forms simultaneously, it was not significantly different from when collard was present as primary-only (χ2 = 0.01, df = 1, p = 0.99) or secondary-only form (χ2 = 0.83, df = 1, p = 0.39). Comparably, while PCR product concentration was higher when collard was present as a primary-only form, it was not significantly different from other treatments (χ2 = 4.19, df = 2, p = 0.12). Overall, we found evidence that C. septempunctata consumes both plant pollen material (primary feeding) and insect prey simultaneously and that prevalence of secondary plant feeding detection is significant and promising (Table 4).
TABLE 4 The list of treatments and results for testing primary and secondary trophic interactions in laboratory feeding trials. The results include the percentage positive for primary (insect and plant) and secondary target DNA in the gut contents of
Treatmenta/Percentage positives (PP)b | DBM PP/Conc.c (ng/ul) | COL PP/Conc. (ng/ul) | Joint DBM and COL PPd |
DBM fed on COL | 100%/ 7.31 | 50%/ 0.46 | 50% |
WMS + WMF + DBM fed on COL | 100%/ 8.58 | 100%/ 0.83 | 100% |
WMS + WMF | NPe | 70%/ 1.11 | NP |
SS | NP | NP | NP |
aThe treatment abbreviations include COL, collard plant; DBM, diamondback moth; WMF, wild mustard flower; WMS, wild mustard pollen solution; SS, sugar solution.
bPercentage positive was calculated by dividing the total number of samples tested positive for the target plant and insect DNA by the total number of screened samples, multiplied by 100.
cConcentration of the amplified DNA, determined by capillary electrophoresis using the Qiaxcel Advanced System, Qiagen.
dPercentage of the tested C. septempunctata with both DBM and collard DNA in their gut contents.
eGroup of samples with no positive hit to the target DNA.
DISCUSSIONThis study is the first to develop a methodology sensitive to detecting primary and secondary predators feeding on plant materials. We showed that the designed primers could detect primary and secondary plant feeding for up to 15.5 and 5.5 h post-feeding, respectively. Our results suggest that the chance of post-feeding secondary plant detection depends on predator mouthparts. Our results further indicate that MGCA with sensitive plant-specific primers can detect both primary and secondary plant feeding in predators. Such secondary plant feeding detection has implications for future studies of arthropods and vertebrate diet analysis for cautious interpretation of plant DNA detection in individuals (i.e., primary or secondary feeding). This approach can be used in future studies to help refine the estimation of predation in crop fields and to explore the delivery of landscape-wise pest control services by generalist predators.
Few studies explore plant DNA detectability in the gut contents of insect predators (Pumarino et al., 2011; Sint et al., 2018; Wallinger et al., 2015), and currently, only one other study was designed for detecting secondary plant feeding in predators (Guenay et al., 2020). Wallinger et al. (2015) reported that seed predation by carabids (chewing predators) could be traced using plant-specific primers and estimated detectability of Taraxacum officinale (Asterales: Asteraceae) and Lolium perenne (Poales: Poaceae) seed DNA half-lives in Harpalus rufipes (Coleoptera: Carabidae) gut contents for up to 23.2 and 25.9 h post-feeding, respectively. Our results provide additional support and indicate plant DNA half-lives of 24.5 and 15.5 h post-feeding from direct feeding on a collard plant leaf tissue and a pollen solution in the gut contents of DBM and C. septempunctata, respectively. More recently, Guenay et al. (2020) reported limited secondary raspberry DNA detection in fruit flies in the gut contents of carabids. These authors found that the secondary plant feeding is only detectable when carabids fed on recently raspberry-fed fruit flies (i.e., 0 h post-feeding). Indeed, our results confirm that plant-specific primers can detect primary and secondary plant feeding in generalist predators (for C. septempunctata and at low frequency for Pardosa spp.). However, our result provides the first evidence that secondary plant DNA feeding detection in C. septempunctata (i.e., a chewing predator) for up to ~5.5 h with a 50% chance and remained detectable for up to 30 h post-feeding, which contradicts Guenay et al. (2020) findings. The contrary result could be partly due to the pest species tested in their study and ours, as it is evident that the DBM is larger than a fruit fly and can carry a larger portion of plant DNA material in their gut, which can alter detection success. Furthermore, our results indicate that collard DNA in the gut contents of DBM remains detectable with a 90% chance for up to 8 h post-feeding (Figure 1). Therefore, secondary plant feeding detection in C. septempunctata and likely other chewing predators should be viable within this time frame post-feeding. However, future work is needed on other chewing predators to confirm this conclusion. Overall, our findings reveal the sensitivity of MGCA with plant-specific primers in detecting direct pollen and secondary plant DNA consumed by C. septempunctata, paving the way for future efforts to advance studies of trophic interactions by unraveling predation in the matrix of agricultural landscapes. Our findings may have further implications for vertebrate diet analysis using molecular approaches, suggesting part of plant DNA found in vertebrate diet could be due to their secondary feeding on arthropods that consumed plant tissue rather than the plant itself.
Predator mouthpart types, primer-specific half-lives, and plant and insect identity can significantly influence post-feeding DNA detection time in generalist predators (Greenstone et al., 2010; Kheirodin et al., 2022; Sint et al., 2018; Wallinger et al., 2015). The overall pattern suggests longer prey DNA half-life detectability in adult piercing/sucking and spiders relative to chewing predators (Greenstone et al., 2014; Kheirodin et al., 2020, 2022; Simmons et al., 2015). Our result is consistent with previous findings, suggesting that DBM DNA post-feeding time is longer in sucking predators relative to chewing predators. However, we found an opposite pattern for secondary plant DNA post-feeding detection. Surprisingly, C. septempunctata had a significantly higher chance of secondary plant DNA detection than G. punctipes and Pardosa spp. However, more work is required to confirm this pattern, as our conclusion is based on three predatory species. The lower chance of detection in piercing/sucking predators could be due to (1) their feeding on prey hemolymph without fully extracting prey gut contents material or (2) their inability to suck/ingest plant leaf parts from the gut contents of herbivores. In addition, other factors such as plant identity (Sint et al., 2018; Wallinger et al., 2015) and primer-specific post-feeding detection time (Kheirodin et al., 2022) can affect post-feeding detection chances and may influence secondary feeding detection probability. For instance, Wallinger et al. (2015) found that while DNA half-lives were similar among different plant species using plant-specific primers, they were significantly different using general plant primers, indicating primer-specific outcomes. Further, in a recent study, Kheirodin et al. (2022) revealed that using different species-specific primers for the same target species could yield a significantly different rate of prey DNA detection in generalist predators, regardless of the primer sensitivity towards the target DNA. Overall, our results provide initial evidence that the prevalence of secondary plant feeding detection is promising only in chewing predators. However, future work is needed using different predators, prey, plant species, and primers to enable an overall conclusion regarding the effect of predator mouthparts on secondary plant DNA detection.
Detection of primary and secondary plant DNA in generalist predator diets can elevate our understanding of trophic interactions in agricultural fields. In agricultural landscapes, predators frequently move among habitats to feed on insect prey and plant carbohydrate resources (Samaranayake & Costamagna, 2018). Therefore, we hypothesize that secondary plant detection in generalist predator gut contents can link the predation to the source habitat where the predator consumed the prey. While our results indicate the prevalence of secondary plant feeding detection in chewing predators is promising, the challenge remains to distinguish it from primary plant feeding in agricultural landscapes. Therefore, the secondary plant feeding data should be interpreted cautiously for several reasons. First, similar to generalist predators, polyphagous herbivores can also move among habitats to feed on multiple crop and non-crop plants (Kennedy & Storer, 2000; Mazzi & Dorn, 2012). As a result, the plant DNA remains in predator gut contents could be due to their feeding on polyphagous herbivores that recently moved into target crop fields from adjacent habitats rather than predator movement among habitats. However, given the lower dispersal of immature prey/sessile adults (e.g., scale insects) (Hagstrum & Subramanyam, 2010), plant DNA detection due to prey movement across fields could be ruled out for immature prey stages and sessile adults. Second, for omnivorous predators, plant DNA detection may indicate direct feeding on crop pollen, followed by their movement to the field where they were collected. However, during early and late growing seasons, when crops are not flowering (or anthesis), the possibility of direct predator feeding on plant pollen will be minimal and can be ruled out. Therefore, the time of the season and prey mobility are two major factors affecting the interpretation of secondary feeding results. Our study provides a fairly simple cost-effective molecular approach to advance the current understanding of food webs in agricultural landscapes by linking prey plant feeding to predator diets (i.e., for sessile insect pests and non-flowering crop stages) and opens a window of opportunity for future research to test secondary plant DNA detection in generalist predators to untangle tritrophic interactions in agricultural landscapes.
AUTHOR CONTRIBUTIONSA.K: Conceptualization, Methodology, Data curation, Formal analysis, Writing – original draft, Writing – review & editing. M.S: Conceptualization, Methodology, Writing – review & editing. J.S: Conceptualization, Methodology, Data Validation, and Analysis review, Writing – revising original draft, Writing – review & editing, Funding acquisition.
ACKNOWLEDGMENTSWe thank Melissa Thomson for helping with DNA extraction and PCR in the laboratory. We thank Pedro F.S. Toledo for his artistic contributions. Grace Padgett, Winston Cornish, and Olivia Centanni for sample collection. We thank The University of Georgia for providing laboratory and office spaces. We also thank the two anonymous reviewers whose comments greatly improved the quality of this manuscript.
FUNDING INFORMATIONThis work was supported by USDA-ARS Non-Assistance Cooperative Agreement #58–6080–9-006 (Managing Whiteflies and Whitefly-transmitted Viruses in Vegetable Crops in the Southeastern U.S.). The funders had no role in the study or design. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the University of Georgia or the USDA.
CONFLICT OF INTEREST STATEMENTThe authors have no conflict of interest to disclose.
DATA AVAILABILITY STATEMENTThe data associated with this research will be deposited in Dryad upon acceptance.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
© 2024. This work is published under http://creativecommons.org/licenses/by-nc-nd/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
Conservation biological control efforts depend on accurately estimating predator roles in crop fields, and knowledge of plant resources generalist predators utilize in agricultural landscapes. Generalist predators move among habitats to feed on insect prey and some predators feed on plants for non-prey nutrients. Studying predation with molecular gut content analysis (MGCA), provides estimates of within field frequencies of predation on target pests and alternative prey. However, prey DNA takes time to move through the predator digestive system, so a portion of the observed predation likely occurs in adjacent crops or semi-natural habitats. Therefore, we tested a strategy to estimate recent secondary feeding to help trace predation back to the source habitat. We selected the diamondback moth and three common predators with different mouthparts:
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Details

1 Department of Entomology, University of Georgia, Tifton, Georgia, USA; Department of Entomology, Texas A&M AgriLife Research and Extension Center, Dallas, Texas, USA
2 Department of Plant Science, University of Manitoba, Winnipeg, Manitoba, Canada
3 Department of Entomology, University of Georgia, Tifton, Georgia, USA