Molecular identification of predation on the Dubas bug (Hemiptera: Tropiduchidae) in Oman date palms: density-dependent response to prey

Abstract

The date palm (Phoenix dactylifera L.) (Arecales: Arecaceae) is the most economically important crop in Oman with an annual production of >360,000 tons of fruit. The Dubas bug (Ommatissus lybicus de Bergevin) (Hemiptera: Tropiduchidae) is one of the major pests of date palms, causing up to a 50% reduction in fruit production. Across the course of 2 seasons, a variety of arthropod predators living in the date palm canopy were investigated for possible biological control of Dubas bugs, given the growing interest in nonchemical insect pest control in integrated pest management. We collected ~6,900 arthropod predators directly from date palm fronds from 60 Omani date palm plantations and tested them for Dubas bug predation using PCR-based molecular gut content analysis. We determined that ≥56 species of arthropod predators feed on the Dubas bug. We found that predatory mites, ants, and the entire predator community combined showed a positive correlation between predation detection frequency and increasing Dubas bug density. Additionally, there was a significant impact of season on gut content positives, with the spring season having a significantly higher percentage of predators testing positive for Dubas bug, suggesting this season could be the most successful time to target conservation biological control programs utilizing a diverse suite of predators.

Introduction

The date palm (Phoenix dactylifera L.) (Arecales: Arecaceae) is primarily cultivated in the Middle East, Northern Africa, the Horn of Africa, and southern Asia (WCSP 2019) for its sweet, edible fruit. Worldwide, approximately 8.7 million tons of dates are produced annually (FAO 2018) with about 70% of the worldwide 1.1-million-hectare production occurring in Arab nations (FAO 2018). As a result, dates have become a staple food and primary income source in the region (Chao and Krueger 2007, Saafi et al. 2008).

A total of 132 arthropod pests have been associated with the date palm (El-Shafie et al. 2017). However, 7 of these species are considered major pests on date palms. These include Dubas bug (Ommatissus lybicus de Bergevin) (Hemiptera: Tropiduchidae), the greater date moth (Arenipses sabella Hampson in Ragonot) (Lepidoptera: Pyralidae), the lesser date moth (Batrachedra amydraula Meyrick) (Lepidoptera: Batrachedridae), the red palm weevil (Rhynchophorus ferrugineus (Olivier)) (Coleoptera: Curculionidae), and the date palm stem borer (Jebusaea hammerschmidtii Reiche) (Coleoptera: Cerambycidae), the fruit stalk borer (Oryctes agamemnon Burmeister) (Coleoptera: Scarabaeidae), and the date stone beetle (Coccotrypes dactyliperda Fabricius) (Coleoptera: Curculionidae) (Howard 2001, Kinawy 2005, El-Shafie et al. 2017). In the Middle East and North Africa, the Dubas bug is one of the major pests of date palms, causing reductions in fruit production (Blumberg 2008, El-Shafie 2012, Bagheri et al. 2016). It is thought to have originated in the Tigris-Euphrates River Valley of Iraq and has spread throughout much of the Middle East, southeastern Russia and northern Africa since the 1970s (Hussain 1974, Klein and Venezian 1985, Blumberg 2008). Both adults and nymphs are sap feeders and cause chlorosis (loss of green color) in leaves; dense populations can kill whole fronds (Howard 2001). Indirect damage is caused when Dubas bug secretions coat leaves with honeydew, some of which damages understory growth when it drips from the palms. Honeydew production promotes the growth of sooty mold, clogging stomate openings, and reducing the photosynthetic process in the leaves, as well as reducing yield (Mokhtar and Al Nabhani 2013).

One Middle Eastern country in which dates are by far the most economically important crop is the Sultanate of Oman. With an annual production of over 361 thousand tons of fruit from the more than 350 varieties grown, it is second only to crude oil production in economic importance (MAF 2018). Date crop yield can be reduced by as much as 50% by heavy Dubas bug infestations in the region (Talhouk 1977) with dates being smaller and slower to ripen (Mokhtar and Ai-Mjeni 1999). There are 2 generations of Dubas bugs per year in Oman: an autumn generation hatching in September and a spring generation hatching in March–April. Eggs laid by the spring generation lay dormant through the 3 hottest summer months (Talhouk 1977, El Haidari and Al Hafidh 1986, Mokhtar and Ai-Mjeni 1999). For a summary of the life cycle of the Dubas bug in Oman, see Al-Khatri (2012). To control the Dubas bug, Oman has relied heavily on insecticides, particularly organophosphates and pyrethroids (Al Khatri 2012, Khan et al. 2019). Replacing the current insecticide-only controls with an integrated pest management (IPM) strategy will require not only investigating other insecticide modes of action for their effectiveness against Dubas bugs, but nonchemical means of control too (Al Khatri 2012).

Employing generalist predators from local communities can benefit IPM programs (Symondson 2002, Harwood and Obrycki 2005, Chapman et al. 2009, Kheirodin et al. 2020, Bordini et al. 2021), and evidence is mounting that spiders, the most abundant predator taxon in date palms in Oman, play a key role in pest suppression (Michalko et al. 2019). Pest control is a valuable ecosystem service provided by generalist predators (Power 2010), and it has been estimated that utilizing natural enemies to control pests in the United States results in an annual savings of $4.5 billion (Losey and Vaughan 2006). Furthermore, there is a push towards organic production in Oman, as the largest farm (100,000 palms) in the One Million Date Palm Trees Project began the transition to organic cultivation in 2018. As organic insecticides can be expensive and often require many applications, effective natural enemies incorporated into the date palm plantations may be a key component of IPM programs for Dubas bug control.

Molecular gut content analysis is an invaluable tool in identifying which predators are preying on a given pest (Gariepy et al. 2007, Furlong 2015, Athey et al. 2019). Molecular techniques facilitate the analysis of field-collected predators where hundreds to thousands of individual predators can be screened in a relatively short time. Prior to the development of these techniques, visual observations of predation and/or dissection of gut contents and attempts to identify chewed remains were required. This severely limited the amount of data that could be collected. Furthermore, because over 80% of predaceous arthropod families feed by liquid ingestion of extra-orally digested prey (Cohen 1995), examination of gut contents is futile. PCR-based molecular gut content analysis has its limitations, including the inability to differentiate primary predation from secondary predation or scavenging, and is qualitative in the sense that it cannot differentiate between one and multiple prey items in a predator’s gut. However, molecular gut content analysis does yield a lower bound on the predation rate, especially since the detectability half-life of pest DNA in predator gut contents is usually less than 48 h (Greenstone et al. 2014). Also, the integration of data on prey abundance and molecular gut content can provide information about the possible effects of predators on pest populations (Romeu-Dalmau et al. 2012, Boreau de Roincé et al. 2013, Chapman et al. 2013, Firlej et al. 2013, Furlong 2015, Krey et al. 2021).

One last consideration is seasonality. Control of insect pests in IPM programs usually concentrates on the most damaging life stage or a specific time of year when the pests are most vulnerable, such as in the early season when the predator: pest ratio is the highest (Settle et al. 1996, Landis and Van Der Werf 1997). As there are 2 generations of Dubas bugs per year, we wanted to determine if natural enemies were having a greater effect on either generation, which may help us make recommendations to growers about what time of the year biological control may be most effective.

In this study, we used molecular gut content analysis to determine which of the nearly 7,000 arthropod predators collected from 60 date palm plantations in Oman were consuming Dubas bugs during the autumn and spring sampling periods. We tested the effect of prey availability and seasonality on Dubas bug predation. The main objective of this study was to determine which predators were consuming Dubas bugs most frequently and if either autumn or spring would be a more effective time to concentrate biological control efforts.

Materials and Methods

Collection Sites

Predator collection for molecular gut content analysis occurred in November 2016 (autumn collection) and February–March 2017 (spring collection). All sites were between 27 and 207 km from Muscat, Oman (Fig. 1). In autumn 2016, 3,156 specimens were collected from 29 locations (Fig. 1). In spring 2017, 3,915 specimens were collected from 38 locations (Fig. 1) for a grand total of 7,071 specimens. Of these, ~6,900 were predators (Supplementary Fig. 1). Specimens were collected directly from date palm fronds using beat sheets and were individually aspirated into 2 ml microcentrifuge tubes containing 99% ethanol. Although contamination with rough collecting methods has been shown (Greenstone et al. 2011), this study used individual collections with aspirators creating fewer opportunities for cross-contamination (Harwood 2008, Chapman et al. 2010, Greenstone et al. 2011, Athey et al. 2017). Predator collection for feeding trials occurred in November 2017 and February–March 2018 at date palm plantations and was collected directly from date palm fronds using beat sheets and placed individually into 2 ml microcentrifuge tubes.

Fig. 1.

Map showing the localities of the 60 date palm plantations sampled during this study. See Supplementary Table 1 for precise locality information and sampling dates.

Dubas bugs for prey density estimates were collected in 2017 at 36 locations (Table 1).GPS coordinates for each site were obtained using hand-held GPS units in the field and verified using Google Earth.

Table 1.

List of taxa positive for Dubas bug DNA, including the number of sequences analyzed and the highest percent match on the BOLD database. Species in bold were unequivocally matched to a single species on BOLD, and the remainder should be viewed with some uncertainty. The BOLD database was accessed the week of 12 December 2021

Higher taxon	Family	Species (best match on BOLD or determination via morphology as noted)	No. sequences	Percent match on BOLD
Acari	Anystidae	Anystis agilis	47	100
Insecta	Formicidae	Polyrhachis dives a	5	100
		Paratrechina longicornis	2	100
		Iridomyrmex suchieri a	1	100
		Tapinoma sp.	2	99.49–100
		Crematogaster spb	5	97.52–97.72
	Coccinellidae	Cheilomenes sexmaculata	6	100
		Pharoscymnus flexibilis c	7	99.13–100
		Undetermined species 1 (larva)	1	100
		Undetermined species 2 (larva)	1	97.48
	Nabidae	Nabis capsiformis	3	99.85–100
	Chrysopidae	Chrysoperla pudica	7	100
		Chrysoperla zastrowi/carnea/annae	1	100
		Chrysoperla sp.	35	96.23–96.61
		Chrysopa sp.	1	99.85
		Pseudomallada sp.d	2	99.64
	Mantidae	Nilomantis floweri e	38	87.98–88.3
Pseudoscorpiones	Undetermined	Undetermined	2	98.46–98.77
Araneae	Sparassidae	Olios mahabangkawitus	1	98.93
		Olios sp 5GAB PAK	4	99.85–100
		Undetermined sp. 1	2	91.71–91.94
		Undetermined sp. 2	1	91.22
	Undetermined	Undetermined	1	88.95
	Gnaphosidae	Undetermined	1	98.46
	Oecobiidae	Oecobius putus	1	100
		Undetermined	1	97.69
	Uloboridae	Uloborus plumipes	3	100
		Uloborus tetramaculatus	9	98.46–100
	Lycosidae	Hogna ferox	1	98.77
	Thomisidae	Undetermined	43	90.21–91.13
	Dictynidae	Undetermined sp. 1	49	88–89
		Undetermined sp. 2	67	88–89
	Araneidae	Eriovixia excelsa or laglaizei	10	100
		Undetermined sp. 1	1	98.68
		Undetermined sp. 2	4	90.39–91.16
		Neoscona polyspinipes	1	99.69
		Neoscona subfusca	1	100
		Neoscona theisi	2	99.38–99.54
		Cyrtophora citricola	12	98.77–99.23
	Linyphiidae	Erigone prominens	2	99.69–99.54
	Theridiidae	Emertonella taczanowskii	14	99.22–100
		Theridion sp. 1	1	97.55
		Theridion sp. 2	10	99.39–99.85
		Theridion melanostictum	8	99.85–99.69
		Undetermined sp.	2	92.59
		Theridion sp 1GAB PAK	12	99.54–100
	Cheiracanthiidae	Cheiracanthium sp.	1	99.38
		Cheiracanthium insulanum/inclusum	52	98.61–100
	Philodromidae	Undetermined	30	88.79–89.71
	Salticidae	Rudakius ludhianaensis	1	100
		Thyene semiargentea	1	99.5
		Undetermined sp. 1	7	97.83–99.16
		Undetermined sp. 2	21	91.05–91.50
		Undetermined sp. 3	6	94.62–95.24
		Undetermined sp. 4	1	98.14
		Plexippus paykulli	68	99.63–100

^aNot found in the ant records from the Arabian Peninsula, and Oman in particular, https://www.antwiki.org/wiki/Oman; further taxonomic research is imperative to confirm its identity.

^bSequences match Crematogaster aegyptiaca at 97.72%.

^cIdentified via morphology by Dr. John Leavengood, Jr. (USDA-APHIS, Tampa, FL, USA).

^dSequences match Pseudomallada clathratus at 96.94%.

^eIdentified via morphology by Dr. Gavin Svenson (Cleveland Museum of Natural History, Cleveland, OH, USA).

Identified undetermined species of Araneae and Pseudoscopiones via morphology.

Feeding Trials

Because we conservatively estimate that a minimum of 56 species feed on the Dubas bug in date palms, it is impractical to attempt feeding trials for all of them. Therefore, we attempted to get DNA decay rate data across different orders. In doing so, we collected mantids (Nilomantis floweri Werner) (Mantodea: Mantidae), lacewing larvae (Chrysoperla sp.) (Neuroptera: Chrysopidae), lady beetle larvae and adults in separate trials (Cheilomenes sexmaculata Fabricius) (Coleoptera: Coccinellidae), jumping spiders (Plexippus paykulli Audouin) (Araneae: Salticidae) and crab spiders (Araneae: Thomisidae) (a single, undetermined species was found in data palms through COI sequencing).

Dubas bug colonies were maintained in the lab at room temperature on small date palms in mesh cages. Predators were starved for a minimum of 48 h in an incubator at a constant 25 °C under a photo-period regime of light: dark = 16:8 before feeding to ensure that their digestive tracts were empty. These incubator settings were selected because they are similar to field temperatures and lighting conditions during the collecting periods. For these trials, predators (n = 10/taxon/time period) were fed a single Dubas bug and preserved immediately in >95% ethanol and stored at −20 °C. An additional 10 predators were fed and stored in an incubator at 25 °C for each of the following time periods (1, 2, 4, 8, 12, 18, 24, and 48 h) before preservation. Larger predators (e.g., mantises) were fed adult Dubas bugs, whereas smaller predators were fed nymphs. Predators fed until they finished consuming the specimen or no longer showed interest in continued feeding. DNA was extracted from these predators and tested for Dubas bug DNA. The detectability of Dubas bug DNA in predator guts over time was calculated using a Probit model (PROC PROBIT) in SAS version 9.4 (SAS Institute, Cary, NC, USA) (Payton et al. 2003, Greenstone et al. 2007, 2014).

Prey Density Tests

To test for the effect of Dubas bug density on the proportion of predators testing positive for Dubas bug DNA, we used beat sheets to collect Dubas bugs from 20 random date palm fronds at all but 2 sites during 2017 (Supplementary Table 1) on the same day as predator collection. A sample consisted of all the Dubas bugs from 20 fronds collected at each location in 95% ethanol. In the laboratory, the sample was poured into a 500 ml beaker, and the total ethanol volume was brought up to either 100 ml or 200 ml, enough to calibrate the liquid amount relative to the density of Dubas bugs in the sample. A stir bar was placed in the beaker on a magnetic stir plate and set to stir at approximately 300 rpm. This put the samples in a state of constant mixing. A 5 mL Eppendorf pipette was used to draw up 3 ml of the solution being stirred. The tip of the pipette tip was cut off to obtain an opening ~8 mm in diameter to allow Dubas bugs to flow freely into the tip as the ethanol was drawn up. Each 3-ml subsample was placed in a small (5.5 cm diameter) petri dish, and the Dubas bugs were counted using a dissecting microscope. Ten subsamples (30 ml total) were counted from each sample, and the total number of Dubas bugs per frond was extrapolated as follows:

$${\displaystyle \begin{array}{lll}& \mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}\mathrm{o}\mathrm{f}\mathrm{s}\mathrm{p}\mathrm{e}\mathrm{c}\mathrm{i}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{f}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{d}=& \frac{\mathrm{N}\mathrm{o}.\mathrm{D}\mathrm{u}\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{b}\mathrm{u}\mathrm{g}\mathrm{s}\mathrm{i}\mathrm{n}30\mathrm{m}\mathrm{l}\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}\times \mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}\mathrm{v}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{m}\mathrm{e}(\mathrm{m}\mathrm{l})}{30\mathrm{m}\mathrm{l}\times 20\mathrm{f}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{s}}\end{array}}$$

This gave us an estimate of the number of Dubas bugs per frond. To ensure that our density estimate subsamples were representative, we calculated the coefficient of variation (Pearson 1896). The coefficient is expressed as a percent, and we subsampled until the coefficient fell below 20% (Sokal and Rohlf 1981), with 10 subsamples usually being enough.

To test whether Dubas bug densities show significant effects on the proportion of predators testing positive for Dubas bug predation, we used a generalized linear model (GENMOD) with a gamma distribution and a log link function in SAS 9.4 (SAS Institute, Cary, NC, USA). To determine the distribution, we used AIC to compare the normal distribution with the gamma distribution. For these analyses, we tested the most common taxa collected: jumping spiders, philodromid crab spiders, crab spiders (Thomisidae), mesh web weavers, cobweb spiders, mantids, ants, and mites. We also performed 2 additional analyses of all spiders combined and all predators combined to look for overall predator effects. We also tested for an effect of season and predator identity on the proportion of predators testing positive for Dubas bug DNA using a GENMOD with season and family as factors and the interaction between both. Only predator groups where more than 100 individuals per taxa (family) were tested in at least one of the seasons were included in the analysis. These were considered common enough in the environment that they could be recommended for possible use in biological control programs.

Molecular Methods

DNA Methods for Primer Design and Molecular Gut Content Analysis

Dubas bug specimens and all field and feeding trail predators were crushed using a pipette tip in 180 µl Buffer ATL and 20 µl proteinase K (Qiagen Inc, Chatsworth, CA, USA), which were incubated in Eppendorf Thermomixer 5350 (Eppendorf North America, Enfield, CT, USA) overnight at 56 °C. Total DNA was extracted (Read et al. 2006, Chapman et al. 2010) using QIAGEN DNeasy Blood & Tissue kits (QIAGEN Inc., Chatsworth, CA, USA) following the manufacturer’s animal tissue protocol. For primer design and predator identification, we amplified cytochrome c oxidase subunit I (COI) sequences from Dubas bugs and predators using general primers: LCO-1490 (Folmer et al. 1994) and HCO-700ME (Breton et al. 2006). Polymerase chain reactions (PCR) (25 µl) consisted of 1× Takara buffer (Takara Bio Inc., Shiga, Japan), 0.2 mM of each dNTP, 0.2 mM of each primer, 0.625U Takara Ex Taq, and template DNA (1 µl of total DNA). PCR was carried out in Bio-Rad PTC-200 and C1000 thermal cyclers (Bio-Rad Laboratories, Hercules, CA, USA). The PCR cycling protocols were 94 °C for 1 min, followed by 35 cycles of 94 °C for 50 s, 40 °C for 45 s, and 72 °C for 45 s. Reaction success was determined with electrophoresis of 10 µl of PCR product in 1.5% SeaKem agarose (Lonza, Rockland, ME, USA) stained with gel red (stock solution diluted 1:10,000 as per manufacturer’s recommendations; Biotium Inc., Freemont, CA, USA). PCRs that yielded significant products were purified with the QIAGEN MinElute PCR purification kit according to the manufacturer’s guidelines. Cycle sequencing reactions were carried out in both the forward and reverse directions in an ABI 9700 thermal cycler using the ABI Big-Dye Terminator mix (v. 3.0; Applied Biosystems, Foster City, CA, USA) at the University of Kentucky’s Genomics Core Laboratory.

To facilitate the design of Dubas bug primers, a data matrix was constructed that contained COI sequences from every Tropiduchidae species available from GenBank in early 2016 and 28 Dubas bug sequences we generated from Oman specimens (GenBank accession numbers KP719890–KP719917). Forward and reverse COI sequences from the same individual were aligned using Geneious Pro (v. 6.1.5; GraphPad Software, LLC, Boston, MA, USA)  and multiple alignments were assembled using MAFFT (v. 5; Katoh et al. 2005) using the default settings and refined manually. Primers were designed by visually inspecting the alignment in Mesquite (v. 3.04; Maddison and Maddison 2017) and the optimal primer properties were confirmed in Primer3 (Rozen and Skaletsky 2000). The primers are OL298-F (5ʹ-CTGACTTTTACCACCTTCATTACT-3ʹ) and OL473-R: (5ʹ-GTTAATTGCTCCTAAGATTGAA-3ʹ), and they generate a 184 bp amplicon.

Dubas bug primer sensitivity was determined by testing the primers on a series of dilutions of Dubas bug extractions. Initial concentrations of DNA extractions were determined using a CLAIROstar microplate reader (BMG Labtech, Ortenberg, Germany). Dilutions of 100, 50, 25, 12.5, 6.25, 3.13, 1.56, and 0.78 pg/µl were used as starting material for PCR, which was conducted the same way as for molecular gut content analysis (below).

PCR reagents using the Dubas bug primers were the same as above with the following exceptions: (i) total reaction volume was 12.5 µl and (ii) 0.25 µl of bovine serum albumin (20 µg/µl) was added to the PCRs to overcome any unknown PCR inhibition (Juen and Traugott 2006, Penn et al. 2016). The PCR cycling protocols were 94 °C for 1 min followed by 35 cycles of 94 °C for 45 s, 62 °C for 45 s, and 72 °C for 30 s. The primers were tested against 137 nontargets for cross-reactivity (Supplementary Table 2).

Predator Identification

In the laboratory, 2 pictures, typically of a dorsal and ventral angle, were taken of each predator prior to DNA extraction with a Gigastone 8MP digital camera (Gigastone Corp., Irvine, CA, USA). For predators we could not identify from images, we amplified the barcode region of COI (Hebert et al. 2003) and submitted 658 amplifications for sequencing. Of the specimens sequenced, we obtained 617 usable sequences. Sequences were submitted to the identification engine on the BOLDSYSTEMS database (http://www.boldsystems.org).

To estimate the total number of species, we used the COI sequences to build a phylogenetic tree (Supplementary Fig. 2). We conducted maximum likelihood (ML) analyses on the COI data set using Garli (v. 2.01; Zwickl 2006). The data were partitioned by codon position (total of 3 partitions). We applied the most complex model available (GTR + I + G; Rodriguez et al. 1990) to each partition as per the recommendations of Huelsenbeck and Rannala (2004) for likelihood-based analyses. Garli generates and applies separate parameter estimates to each partition. A 100-replicate ML analysis was conducted using the default settings. The tree with the highest (least negative) log-likelihood is presented in Supplementary Fig. 2. Clades sharing similar morphology with very short internal branches are conservatively interpreted to be the same species.

Results

Primer Performance

The Dubas bug primers had 100% amplification success for Dubas bugs with no amplification from nontarget arthropods (Supplementary Table 2). The Dubas bug primers in our PCR assays were estimated to have primer sensitivity to approximately 1.56 pg/µl for target DNA detection.

Predator Taxa Feeding on the Dubas Bug

We estimate that at least 56 species of predators tested positive for Dubas bug DNA (Table 1). We base this estimate on how the 617 COI sequences were grouped into clades in the phylogenetic tree in Supplementary Fig. 1 (GenBank accession numbers MK950154–MK950770). Clades having very short internal branches in conjunction with very similar-appearing specimens were conservatively estimated to be the same species. Submitting the sequences to the BOLD database enabled the identification of 15 taxa for described species and one for a known but undescribed species (Theridion sp. 1GAB PAK) (Araneae: Theridiidae). The remaining taxa did not produce an unequivocal match to a single species. These taxa are presented in Table 1 in the same order in which they occur on the tree (Supplementary Fig. 1). Images taken of each predator prior to DNA extraction were cross-checked with molecular identifications to confirm that the molecular determinations were reasonable. DNA barcode-based identifications are tentative and should be considered a first step in identifying the predators of species. Statistical analysis were all based on higher taxonomic groupings that we could confirm through morphology.

Predation Detection Frequencies Among Predator Taxa

Molecular gut content analysis of field-caught predators shows that a variety of predators are feeding on the Dubas bug with relatively high frequencies. Supplementary Table 3 shows the predator families that tested positive, arranged by percent positive from high to low. Of the groups that had at least 100 specimens in 1 season, several had over 10% specimens testing positive for Dubas bug DNA: mites (Acari: Anystidae) at 47.8%, followed by long-legged sac spiders (Araneae: Cheiracanthiidae), 39.3%, mantids (Mantodea: Mantidae), 33.9%, running crab spiders (Araneae: Philodromidae), 25.4%, and jumping spiders (Araneae: Salticidae), 18.6% (Fig. 2). There was a strong effect of season on the proportion of predators testing positive for Dubas bug DNA (χ² = 31.9, P < 0.0.0001) with the spring sampling having consistently higher positives for Dubas bug DNA. There was no significant interaction between season and predator group (χ² = 16.0, P = 0.2) (Fig. 2). The percentage testing positive for Dubas bugs was significantly higher in the spring compared to the autumn for crab spiders, mesh web weavers (Araneae: Dictynidae), long-legged sac spiders, ants, and mites.

Fig. 2.

Percent of predators testing positive for Dubas bug DNA by season. Only groups with ≥100 specimens in at least 1 season were included. Significant differences are denoted by *P ≤ 0.05, ns = not significant. The numbers above the bars indicate the sample size for molecular gut content analysis.

Feeding Trials

We conducted feeding trials on select taxa (Fig. 3). None of the crab spiders would feed on Dubas bugs in the lab. However, we were successful with the remaining taxa. The predators tested showed very different DNA decay rates. Lacewing larvae had a very short detection period with a DNA detectability half-life of just over 1 h with no specimen testing positive after 18 h. Jumping spiders had by far the longest DNA detectability half-life of almost 26 h, with 1 specimen testing positive 48 h postfeeding (Table 2). None of the crab spiders would feed on Dubas bugs in the lab.

Fig. 3.

Feeding trial data for predator groups. A) praying mantises; B) lacewing larvae; C) lady beetle adults; D jumping spiders; E) lady beetle larvae. DNA detectability (solid line), 95% confidence interval estimates (dashed lines), and detectability half-life were calculated using a probit model.

Table 2.

Feeding trial statistics and calculations using a probit model. The chi-square test and associated P-value show that in all cases, DNA degradation in the gut contents were significantly different from zero, i.e., prey DNA was degrading in predator guts

Taxon	χ 2 (df, N)	P	Half-life (hours)	95% Confidence interval
Praying mantises	11.42(1, N = 90)	0.001	8.2	0–16.1
Lady beetle adults	26.44(1, N = 90)	<0.001	7.1	5.1–9.5
Lady beetle larvae	19.15(1, N = 91)	<0.001	15.1	9.5–22.3
Lacewing larvae	13.51(1, N = 91)	0.001	1.1	0–4.0
Jumping spiders	18.82(1, N = 92)	<0.001	25.9	21.5–42.8

Effects of Prey Density on Proportion of Predators Testing Positive

Most of the predator groups showed no significant association between the proportion testing positive for Dubas bug predation and Dubas bug density (Table 3). For predatory mites (χ² = 3.8, P = 0.05), ants (χ² = 7.2, P = 0.007), and all predators combined (χ² = 6.6, P = 0.01) there was a significant association between the proportion of predators positive for Dubas bug DNA and Dubas bug density with more predators tested positive for Dubas bug DNA as the population of Dubas bugs increased (Fig. 4). Mesh web weavers (F_(1,32) = 2.79, P = 0.09), and all spiders combined (χ² = 2.2, P = 0.1), showed a positive but not significant relationship between Dubas bug density and proportion of predators testing (Table 3).

Table 3.

Statistics for the GLM model for all predator groups tested for a relationship between prey availability and gut content positives

Taxon	Wald χ2	P
Jumping spiders (Salticidae)	χ 2 (1, N = 37) = 0.11	0.74
Running crab spiders (Philodromidae)	χ 2 (1, N = 30) = 0	0.96
Mesh web weavers (Dictynidae)	χ 2 (1, N = 35) = 0.3	0.59
Crab spiders (Thomisidae)	χ 2 (1, N = 36) = 1.14	0.29
Cobweb spiders (Theridiidae)	χ 2 (1, N = 34) = 0.25	0.62
All spiders	χ 2 (1, N = 37) = 2.24	0.1
Praying mantises (Mantidae)	χ 2 (1, N = 33) = 0	0.95
Ants (Formicidae)	χ 2 (1, N = 33) = 7.18	0.007*
Mites (Anystidae)	χ 2 (1, N = 24) = 3.77	0.05*
All predators	χ 2 (1, N = 37) = 6.60	0.01*

Significant differences are denoted by *p ≤ 0.05.

Fig. 4.

Relationship between Dubas bug density and proportion of predator taxa testing positive for Dubas bug DNA that was statistically significant, A) All predators, B) Ants, C) Mites (see Table 3).

Discussion

A diverse suite of predatory taxa living on date palms tested positive for Dubas bug DNA (Table 1; Supplementary Fig. 1). We collected nearly 6,900 potential predators and identified predators from 22 arthropod families that tested positive for Dubas bug DNA (Supplementary Table 3). Overall, the percentage of predators testing positive for Dubas bug DNA was higher in spring than in autumn. The primary reason for this was probably the timing of predator collections in each season. For the autumn season, predator collection took place in November when Dubas bugs were predominately late instar nymphs or adults, whereas spring collections took place in late February and early March, earlier in the Dubas bug developmental season, when they were eggs and/or early instar nymphs (Mokhtar and Al Nabhani 2013). Eggs and early instar nymphs are smaller and less mobile, likely making them easier for the predators to find and consume. Based on their lifecycle, there may have been more vulnerable prey available in the spring, but this is unknown as we do not have Dubas bug density estimates for the autumn sampling period.

If we therefore consider our autumn sampling a late-season sampling period and our spring sampling an early season sampling period, we can apply an early season versus late-season dynamic to our results. While this dynamic is frequently applied to a growing season, we could also apply it to pest development here since the early season was represented by spring collections, which occurred when the eggs and early instar nymphs were present, and the late season was represented by the autumn collections, which occurred when the late instar nymphs and adults were present.

Our results seem to support the early season versus late-season dynamic. For example, the most abundant predator families were mesh web weavers and crab spiders. For the mesh web weavers, 509 were collected in the autumn, and 989 were collected in the spring with 1.9% and 11% positive for Dubas bug DNA, respectively. For the crab spiders, 802 were collected in the autumn, and 738 were collected in the spring, with 2.9% and 11.8% positive for Dubas bug DNA, respectively. Both groups had large sample sizes, with more Dubas bug consumption taking place in the early season sampling period. This trend was observed in almost every group of predators with over 100 specimens tested, with the largest disparity in mites where 15.9% were positive in the late season sampling (autumn) versus 65.8% positive in the early season sampling (spring). This could suggest that the predator groups, both individually and collectively, would be more efficient in the early season compared to the late season. Early season predation can delay pests from reaching outbreak levels by consuming them when predator: pest ratios are the highest in favor of the predators (e.g., Chiverton 1987, Landis and Van Der Werf 1997, Athey et al. 2016). For early season predation to be most effective, the pest group needs to have the potential for exponential population growth, such as aphids (Chiverton 1987, Birkhofer et al. 2008, Boreau de Roincé et al. 2013). Since other fulgoroid pests are known to exhibit exponential growth, e.g., the citrus flatid planthopper (Lee et al. 2019) and Prokelisia marginata (Van Duzee) (Hemiptera: Delphacidae) (Harkin 2016), Dubas bugs may be capable of exponential growth and thus would be good candidates for conservation biological control using early season predation.

Our results also suggest the predator community is responding to changes in prey availability. For the spring season, when Dubas bug population densities were determined, we found a significant relationship between the detection of Dubas bug DNA in predator guts and Dubas bug density. Although there are many factors that influence the rate at which a predator feeds on a given prey item, including prey mobility (e.g., Eubanks and Denno 2000) and nutritional quality (e.g., Schmidt et al. 2012), an obvious factor is the availability of that prey item. As most generalist predators can forage selectively (Greenstone 1979, Bilde and Toft 1994, Toft 1995, Mayntz et al. 2005, Welch et al. 2013), it is not a safe assumption that as a given pest population rises, so does the frequency that the predator population feeds on it. Although many molecular gut content studies report a decoupling of prey availability and the detection of prey DNA in the guts of predators (Kerzicnik et al. 2012, Chapman et al. 2013, Visakorpi et al. 2015, Eitzinger et al. 2019), our findings indicate that the predator community, as a whole, may be positively responding to Dubas bug density changes.

Additionally, a diverse suite of predators attacking Dubas bugs suggests that conservation biological control may be a practical component of an IPM program. Not only did we find taxonomic diversity in the predators that tested positive for Dubas bug DNA, but we also found diversity in hunting behavior. For the spiders, especially, we had 4 different guilds of spiders (ambushers, foliage running spiders, space web builders, and stalkers (Uetz et al. 1999)) frequently positive for Dubas bug DNA. Members of these groups might be good candidates for conservation biological control.

Although we did not experimentally manipulate the predator diversity, other authors have (Wilby et al. 2005, Snyder et al. 2006, Straub and Snyder 2006, Greenop et al. 2018, Alhadidi et al. 2019), and we may be able to make inferences based on previous studies. Predator diversity has been studied extensively for its effect on biological control with studies showing both a strong positive effect (reviewed in Letourneau et al. 2009) and no effect (Straub and Snyder 2006, Letourneau et al. 2009) or a negative effect (Rosenheim et al. 1993, Finke and Denno 2004). Sometimes, authors found both. For example, in a study comparing organic to conventional potato production, increasing predator biodiversity was positive in conventional production but negative in organic production (Lynch et al. 2021). This could suggest that diversity for its own sake is not the answer, but diversity in feeding habits, such as spider feeding guilds or niches, would be a better target for biological control practitioners. In a meta-analysis utilizing 51 studies where predator species richness was experimentally manipulated, it was found that functional diversity was the most important variable overall (Greenop et al. 2018). This included habitat, diet breadth, and hunting strategy. In our results, within the spiders alone, we had 4 different hunting strategies, so conservation biological control programs looking to maximize functional diversity could target this group as part of their IPM program. In addition to predator identity, the optimal time to use biological control will be early in the life cycle of Dubas bugs, when eggs and early instar nymphs are prevalent in the environment. This will likely result in maximum pest suppression using generalist predators.

In summary, we found a diverse assemblage of predators in date palms preying on Dubas bugs with the predator community responding significantly to Dubas bugs and preying on them more frequently when prey densities are higher. Targeting the early season in both spring and fall for an IPM program using both biological control and other traditional methods will likely result in more effective Dubas bug control, and our study serves as a first step by identifying predator groups that are likely to be most useful in a conservation biological control program. In the future, uncovering the exact spider species among a group of cryptic species that consume Dubas bugs will be beneficial for conservation biological control practitioners.

Author contributions

Kacie Athey (Conceptualization [supporting], Data curation [equal], Formal analysis [lead], Investigation [supporting], Methodology [equal], Validation [lead], Visualization [lead], Writing—original draft [equal], Writing—review & editing [lead]), Eric Chapman (Data curation [equal], Formal analysis [supporting], Investigation [lead], Methodology [equal], Validation [supporting], Visualization [supporting], Writing—original draft [equal], Writing—review & editing [supporting]), Salem Al-Khatri (Conceptualization [equal], Funding acquisition [equal], Investigation [equal], Writing—review & editing [supporting]), Abdel Moktar (Conceptualization [supporting], Funding acquisition [equal], Resources [supporting]), and John Obrycki (Project administration [lead], Supervision [lead], Writing—original draft [supporting], Writing—review & editing [supporting])

Funding

We thank the Research Council of Oman for making this study possible (contract number TRC/SRG/DB/13/005).