A pragmatic approach to the analysis of diets of generalist predators: the use of next-generation sequencing with no blocking probes

Citation for final published version:

Pinol, Josep, San Andrés Aura, Victoria, Clare, E. L., Mir, G. and Symondson, William Oliver Christian 2014. A pragmatic approach to the analysis of diets of generalist predators: the use of next-generation sequencing with no blocking probes. Molecular Ecology Resources 14 (1) , pp.

18-26. 10.1111/1755-0998.12156 file

Publishers page: http://dx.doi.org/10.1111/1755-0998.12156 <http://dx.doi.org/10.1111/1755-0998.12156>

Please note:

Changes made as a result of publishing processes such as copy-editing, formatting and page numbers may not be reflected in this version. For the definitive version of this publication, please refer to the published source. You are advised to consult the publisher’s version if you wish to cite

this paper.

This version is being made available in accordance with publisher policies. See

http://orca.cf.ac.uk/policies.html for usage policies. Copyright and moral rights for publications made available in ORCA are retained by the copyright holders.

For Review Only

A pragmatic approach to the analysis of diets of generalist predators: The use of

1

next generation sequencing with no blocking probes

2

3

J. Piñol1,2,3, V. San Andrés3, E. L. Clare3,4, G. Mir5, W.O.C. Symondson3

4

5

(1) Univ. Autònoma Barcelona, Cerdanyola del Vallès, 08193, Spain

6

(2) CREAF, Cerdanyola del Vallès, 08193, Spain

7

(3) Cardiff School of Biosciences, Cardiff University, Sir Martin Evans Building, Museum Avenue,

8

Cardiff CF10 3AX, UK

9

(4) School of Biological and Chemical Sciences, Queen Mary, University of London, Mile End

10

Road, London E1 4NS, UK

11

(5) Centre de Recerca en Agrigenòmica (CRAG) CSIC IRTA UAB UB, Cerdanyola del Vallès

12

08193, Spain

13 14

Keywords: crop ecology, gut content analysis, linyphiid spiders, predator prey

15 interactions 16 17 Corresponding Author: 18 Josep Piñol 19

CREAF, Univ. Autònoma Barcelona, Cerdanyola del Vallès 08193, Spain

20

E mail: Josep.Pinol@uab.es

21

22

Running title: Diets through NGS with no blocking probes

For Review Only

24

Predicting whether a predator is capable of affecting the dynamics of a prey species in

25

the field implies the analysis of the complete diet of the predator, not simply rates of

26

predation on a target taxon. Here we employed the Ion Torrent Next Generation

27

Sequencing technology to investigate the diet of a generalist arthropod predator. A

28

complete dietary analysis requires the use of general primers, but these will also amplify

29

the predator unless suppressed using a blocking probe. However, blocking probes can

30

potentially block other species, particularly if they are phylogenetically close. Here, we

31

aimed to demonstrate that enough prey sequence could be obtained without blocking

32

probes. In communities with many predators this approach obviates the need to design

33

and test numerous blocking primers, thus making analysis of complex community food

34

webs a viable proposition. We applied this approach to the analysis of predation by the

35

linyphiid spider Oedothorax fuscus in an arable field. We obtained over two million raw

36

reads. After discarding the low quality and predator reads, the libraries still contained

37

over 61,000 prey reads (3% of the raw reads; 6% of reads passing quality control). The

38

libraries were rich in Collembola, Lepidoptera, Diptera, and Nematoda. They also

39

contained sequences derived from several spider species and from horticultural pests

40

(aphids). Oedothorax fuscus is common in UK cereal fields and the results showed that

41

it is exploiting a wide range of prey. Next Generation Sequencing using general primers

42

but without blocking probes provided ample sequences for analysis of the prey range of

43

this spider and proved to be a simple and inexpensive approach.

44

45 46

For Review Only

47

Critical questions in trophic ecology centre around the choices that predators and

48

herbivores make. For example, what a predator eats depends upon its ability to capture

49

and subdue a target prey species, with predation rates being affected by relative

50

densities of predators and prey, prey handling times and the shape of functional

51

response curves (Symondson et al. 2002). However, equally important is the availability

52

of ‘non target’ alternative prey (Harwood et al. 2004). Prey ‘choice’ in the field is highly

53

complex, where the densities and diversity of prey available to the predator are

54

constantly changing over time and space (Symondson et al. 2002). At its simplest, what

55

a predator eats in a given environment at any moment in time depends upon what is

56

available (quantitatively and qualitatively) and accessible to the predator.

57

58

Understanding and predicting whether a predator is capable of affecting the dynamics of

59

a prey species, and under what circumstances it manages to do so in the field, often

60

require techniques capable of analysing the complete diets of the predators, not simply

61

rates of predation on a target taxon, such as a particular crop pest. The introduction of

62

DNA based techniques (Asahida et al. 1997; Agustí et al. 1999; Zaidi et al. 1999) has

63

substantially improved the study of predation, and such approaches are now widely

64

used to study a range of trophic relationships (reviewed in Symondson 2002; King et al.

65

2008; Pompanon et al. 2012). This technology provides precise information on whether

66

a predator species feeds on a particular target prey or which parasitoid is present within

67

a host (Agustí et al. 2005; Traugott et al. 2008). However, the now classical approach,

68

based on prey specific primers, is of limited utility for generalist predators as the

For Review Only

analysis has to be repeated for a large number of potential, and often unknown, prey.

70

The use of multiplexing can make this process more efficient (Harper et al. 2005, King et

71

al. 2010, Traugott et al. 2012, Davey et al. 2013) but still limits the analysis to predicted

72

prey rather than neutrally screening for all prey consumed.

73

74

An alternative approach is to use general primers that amplify a range of dietary

75

components simultaneously. Initially this was followed by cloning and sequencing, and

76

has been applied to studies of both herbivory (e.g. Poinar et al. 1998, Passmore et al.

77

2006, Bradley et al. 2007) and predation (e.g. Sutherland 2000, Jarman et al. 2004,

78

Blankenship & Yayanos 2005, Deagle et al. 2007). The recent development of Next

79

Generation Sequencing (NGS) technologies provide a more efficient means of rapidly

80

gathering a mass of information on the dietary ranges of generalist predators and

81

herbivores (Pompanon et al. 2012). Using NGS it is possible, at least in theory, to

82

sequence all prey species present in predator gut or faecal samples simultaneously.

83

This technology has mainly been used to study the diets of vertebrate predators and

84

herbivores, such as seals (Deagle et al. 2009), penguins (Deagle et al. 2010), bats

85

(Bohmann et al. 2011), chamois (Raye et al. 2011), snow leopards (Shehzad et al.

86

2012), lizards (Brown et al. 2012) and bison (Kowalczyk et al. 2011), but has also been

87

tested on invertebrates (Valentini et al. 2009; Boyer et al. 2012).

88

89

The use of NGS to study predator diets makes use of universal primers, able to amplify

90

a wide range of prey. As these primers also amplify predator DNA, especially when prey

91

and predator are phylogenytically close, the PCR product may be dominated by predator

For Review Only

instead of prey DNA. To overcome this problem predator specific blocking primers are

93

applied (Vestheim & Harman, 2008; Deagle et al., 2009, 2010). However, 'universal'

94

primers do not in practice amplify everything equally while blocking primers are rarely

95

specific. All 'universal' primers used in species rich mixtures are likely to have more

96

affinity for some species than for others. Similarly, 'specific' blocking primers could block

97

to some degree prey species in addition to the targeted predator, especially if prey and

98

predators are closely related (e.g. spiders feeding on other spiders). The difference in

99

affinity between the 'universal' primer and the 'specific' blocking primer for different

100

species may be small, but after a number of cycles the PCR product is going to be

101

enriched in some species and impoverished in others, in comparison to the original

102

mixture. Thus, the use of universal primers and predator specific blocking primers

103

introduces two different biases in the analysis of predator diets.

104

105

Here we propose the use of NGS to study animal diets with no predator specific

106

blocking primers. By doing so, we get rid of one of the two biases described above. The

107

final result will be dominated by reads of the predator, but the problem is overcome by

108

the formidable sequencing capacity of modern NGS platforms. This 'brute force' method

109

obviates the need to design and extensively test blocking primers. The latter may be

110

worthwhile for ecological studies involving single predator species and have been used

111

effectively (reviewed in Pompanon et al. 2012). However, our aim was to develop an

112

approach that would be applicable to complex invertebrate food webs involving multiple

113

predator species. Designing and testing blocking primers for dozens of different

114

predators would be both impractical and expensive.

For Review Only

116

Spiders, especially very small species such as the linyphiids, are a group for which

117

knowledge of their diets is often particularly difficult to obtain. They are fluid feeders,

118

precluding the slow and taxonomically challenging classical approach of microscopically

119

examining prey remains in predator gut samples. Many species hunt nocturnally, making

120

direct observations of predation events difficult if not impossible. Some species build

121

webs, from which prey can be collected, but not all captured prey are actually eaten.

122

However, many spiders are not web builders while others hunt away from their webs,

123

particularly at night (De Keer & Maelfait 1987; Alderweireldt 1994).There is a clear need

124

for objective analytical methods, both for a better understanding of spider ecology 125

generally and for applied research into the role of spiders as predators of major crop 126

pests. 127

128

Here we tested the ‘brute force’ approach through the analysis of the diet of the linyphiid 129

Oedothorax fuscus in an arable field. This small epigeal spider is widespread in Europe, 130

living amongst thick vegetation in a range of habitats including field margins and arable 131

crops. Most spiders are generalist predators of insects and other spiders (Pekár et al.

132

2012), making them appropriate targets for NGS of their complete dietary ranges. The 133

abundance of O. fuscus in crops makes it a potentially useful biocontrol agent, 134

particularly of aphids. As discussed in Pompanon et al. (2012), NGS provides an 135

excellent tool for initial screening of predators or herbivores. Though not quantitative 136

(Pompanon et al. 2012), it can provide an invaluable guide to the composition and range 137

of species consumed. Our aim was to evaluate the technology and to determine the 138

For Review Only

dietary range of this elusive predator, revealing the breadth of its feeding ecology and 139

hence its potential as a biocontrol agent in cereal fields. 140

141

Methods

142

Collection of samples and DNA extraction

143

Spiders were collected from a grass dominated boundary strip surrounding a crop of 144

winter barley at Burdons Farm, Wenvoe (51.439ºN, 3.271ºW), near Cardiff, Wales, UK.

145

We sampled spiders six times from August to November 2011 using between six and ten

146

quadrats (0.25 m2) 10 m apart on each sampling date. We collected spiders with an

147

entomological pooter and transferred them immediately to separate micro centrifuge

148

tubes. Back at the laboratory, the spiders were transferred to new Eppendorf tubes, in

149

absolute ethanol, and kept at 80ºC until needed.

150

151

The spiders were identified under a binocular microscope following Roberts (1996). We

152

extracted the DNA from whole homogenized O. fuscus according to the “animal tissue”

153

protocol of the Qiagen DNeasy tissue kit (Qiagen Ltd, Crawley, UK) following

154

manufacturer instructions. For each batch of extractions a negative control was

155

included. DNA quality was checked on 2% agarose gel. We then prepared a single

156

pooled sample consisting of 5 µL from each individual extraction (15 to 20 individuals

157

from each one of the six sampling events, a total of 109 individual extractions). The

158

analysis of this sample was intended to assess the diet breadth of O. fuscus at a

159

population level over a four month period.

For Review Only

161

Selection of universal invertebrate primers

162

We amplified arthropod DNA from spider extracts using the general invertebrate COI

163

primers ZBJ ArtF1c and ZBJ ArtR2c (Zeale et al. 2011; Bohmann et al. 2011). These

164

yielded a157 bp amplicon located within the COI barcode region (Folmer et al.1994) and

165

have been shown to amplify spiders and members of a wide range of insect orders

166 (Table S1). 167 168 Lab procedure 169

We prepared the samples with fusion primers following Ion Torrent recommendations for

170

bidirectional sequencing (Ion Amplicon Library Preparation, Fusion Method). Briefly, two

171

pairs of primers were designed (i) Ion Torrent primer A linked to the specific forward

172

primer with Ion Torrent primer trP1 linked to the specific reverse primer (from now on

173

“forward”), and the opposite (ii) Ion Torrent primer trP1 linked to the specific forward

174

primer with Ion Torrent primer A linked to the specific reverse primer (from now on

175

“reverse”) (Table S2).

176

177

The DNA was amplified in 50 l reaction volumes containing 3 l of template DNA, 45 l

178

of Platinum PCR Supermix High Fidelity (Invitrogen Corporation), and 1 l of each pair

179

of 10 M fusion primers. We ran 40 cycles at 94°C for 30 s, 45°C for 45 s and 68°C for

180

45 s following an initial denaturation step at 94°C for 5 min and before a final extension

181

step at 68°C for 10 min. The PCR product was purified using the QIA quick PCR

182

Purification Kit (Qiagen). Two PCR replicates of each forward and reverse reactions

For Review Only

184

185

We checked the PCR products from the forward and reverse design separately. We

186

selected fragments of the expected size (E Gel® Size Select 2% Agarose Gel, 187

Invitrogen), quantified the size selected DNA (DNA High Sensitivity kit, Bioanalyzer 188

2100, Agilent Technologies) and prepared an equimolar pool of the forward and reverse 189

libraries. Then, we diluted, amplified (template amplification, Ion One Touch System) and 190

sequenced this pool on a PGM as described by the manufacturer (Ion Torrent). As a first 191

test of the technique we used the smallest Ion Torrent 314 chip, that should provide 192

approximately 105 DNA sequences. Based on the obtained results, we next ran a 318 193

chip in order to increase the sequencing depth and to lower the cost per read (the 318 194

chip can potentially increase the yield 50 times over the 314 chip). The two forward runs 195

were conducted with the product of two different PCR reactions using the same DNA 196

template and the two reverse runs with the product of a single third PCR reaction using 197

the same DNA template as in the forward reactions. The first run was amplified with the 198

Ion One Touch System whereas the second one was amplified with the upgraded 199

version Ion One Touch System DL. In all runs we used the sequencing chemistry for 200 200

bp read length and version 2.2 of the Torrent Suite software for base calling (Ion Torrent, 201

Life Technologies). 202

203

Processing and analysis of data

204

We divided the process into three steps: (1) Quality control and preprocessing; (2) 205

comparison of the reads with sequence databases; and (3) taxonomic assignment of 206

For Review Only

each individual read. 207

208

We separated the forward and reverse reads into two FASTQ files using a purpose 209

made perl script. We eliminated the primer sequence from the 5' end of each read using 210

TagCleaner (Schmieder et al. 2011). Then we trimmed (truncated to 150 bp at the end 211

opposite of the sequencing primer), discarded those with a mean quality score lower 212

than 25 or shorter than 150 bp, and downloaded the sequences in FASTA format (all 213

using PRINSEQ, Schmieder & Edwards 2011). Both TagCleaner and PRINSEQ were 214

run as web services at http://edwards.sdsu.edu/tagcleaner and

215

http://edwards.sdsu.edu/cgi bin/prinseq/prinseq.cgi, respectively.

216

217

To make downstream computation simpler, the FASTA files obtained above were first

218

visually inspected with a general purpose text editor for common sequences. We then

219

used another purpose made perl script to separate in two different files the common

220

sequence from the rest (all relevant FASTQ files and perl code are available in Dryad).

221

This step was iterated several times until the file containing the rest of the sequences

222

was small enough (< 5,000 reads) to be BLASTed directly at the NCBI website

223

(http://blast.ncbi.nlm.nih.gov/Blast.cgi). In particular, we used the nucleotide BLAST

224

(blastn; Zhang et al. 2000) optimized for very similar sequences (megablast) on the

225

nucleotide collection (nr/nt) that includes all GenBank+EMBL+DDBJ+PDB sequences.

226

The BLAST parameters were those provided by default.

227

228

We imported the output from the BLAST alignment into MEGAN (MEtaGenomics

For Review Only

ANalyzer; Huson et al. 2007) to compute and explore the taxonomical content of the

230

data set, employing the NCBI taxonomy to summarize and order the results. In order to

231

assign a taxon to each sequence in the BLAST files, MEGAN processes the BLAST

232

data of each sequence to determine all the hits. We discarded hits below the threshold

233

for the bit score of hits (min. score = 50). We also discarded hits due to the threshold for

234

the maximum percentage (top percentage = 10) by which the score of a hit may fall

235

below the best score achieved for the sequence. After collecting the hits that exceeded

236

the thresholds, MEGAN found the lowest node that encompassed all these hits using

237

the LCA assignment algorithm (LCA = Lowest Common Ancestor) to assign sequences

238

to taxa. We set the minimum support to a conservative value of 5, implying that a

239

minimum of five sequences had to be assigned to a taxon to appear in the final

240

cladogram.

241

242

The files (or a subset of them if the file was too big) containing the common sequences

243

were also BLASTed and then imported to MEGAN to explore their taxonomic content.

244

The reported result (Table 1) is the sum for each taxon of the number of reads provided

245

by MEGAN from each one of the files in which the original FASTA file was divided.

246 247 Results 248 Overall performance. 249

The 314 run produced a FASTQ file with 60,771 reads (52% forward and 48% reverse).

250

The quality control process reduced the number of “good” reads to 17,096 forward and

251

10,157 reverse reads. As expected, most reads belonged to the predator itself (14,020

For Review Only

forward and 6,905 reverse reads) and some produced no hit in BLAST or the result was

253

uninformative. Overall, the 314 run produced 2,362 forward and 827 reverse reads

254

useful to describe the diet of O. fuscus.

255

256

The 318 run produced a FASTQ file with 2,162,102 reads (50% forward and 50%

257

reverse). The quality control process reduced the number of “good” reads to 580,131

258

forward and 390,976 reverse reads. Again, most reads belonged to the predator itself or

259

were uninformative, so the overall result were 38,274 forward and 19,804 reverse reads

260

useful to describe the diet of O. fuscus.

261

262

Oedothorax fuscus diet

263

The sequences of the O. fuscus diet obtained with the forward sequencing on the 314

264

chip were dominated by Collembola (69%; Table 1). There were also many sequences

265

of spiders (other than O. fuscus), mostly of the thomisid Xysticus sp. There were also

266

reads of lepidopterans, aphids, and nematodes. The forward reads obtained with the

267

318 chip were also dominated by Collembola (91%; Table 1), but there were very few

268

spiders that could be assigned unambiguously to taxa that were different from O. fuscus.

269

Lepidopteran, dipteran, and nematode reads were abundant, but aphids were absent.

270

Finally, there were also a several reads of a cockroach and of an Auchenorrhyncha

271

hemipteran (probably a leafhopper).

272

273

The two reverse runs on the 314 and 318 chips were conducted on the same PCR

274

product, so they produced very similar results and are described together. Again, most

For Review Only

sequences belonged to Collembola (66%, considering both 314 and 318 runs; Table 1).

276

Spiders of three different linyphiid (other than O. fuscus) were also present:

277

Tenuiphantes tenuis, Bathyphantes gracillis, and Centromerita bicolor. Dipterans were

278

the second most abundant group in number of sequences (18%), but they were different

279

from those obtained in the forward runs. Nematodes, lepidopterans and

280

ephemeropterans were also present, but hemipterans and cockroaches were absent.

281

282

In summary, the four sequencing runs (two forward, two reverse) performed on the

283

product of three different PCR reactions from the same DNA template, showed that

284

Collembola, Nematoda, Araneae, and Lepidoptera were always present in the diet of O.

285

fuscus. The Collembola belonged to at least three families, Isotomidae, Tomoceridae,

286

and Sminthuridae, but most of them could only be resolved to Entomobryoidea. By

287

contrast, the common springtail Parisotoma notabilis (Isotomidae) was clearly identified

288

in all runs. Five genera of three different families of spiders (Thomisidae, Therididae,

289

and Linyphidae) were found in the diet of O. fuscus, but nematodes and lepidopterans

290

could only be resolved to higher groups. Dipterans were also present in all but one

291

reaction, whereas Aphididae, Auchenorrhyncha, Blattodea and Ephemeroptera were

292

only present in one reaction each.

293

294

Discussion

295

These results demonstrate that NGS technology using general primers without a

296

predator specific blocking primer can be an effective way to study the diet of

297

invertebrates such as spiders. This method is also simple and inexpensive and

For Review Only

particularly effective when used with the 318 Ion Torrent chip that provided more than 299

two million DNA sequences. As expected, most of these belonged to the predator itself, 300

but there were still ca. 58,000 sequences from prey, far more than enough to describe 301

the diet of this spider. 302

303

The analysis provided evidence of predation on pests, non pest prey and intraguild 304

predation (see below) while the species detected in the diet of O. fuscus were those 305

commonly found in UK cereal fields. Because the identification of these sequences 306

relies on the existence of taxonomically validated reference libraries, this work can only 307

be as precise as the sequence databases available. As such, more work needs to be 308

done on the barcoding of organisms found in arable crops to improve our ability to fully 309

identify these sequences. 310

311

Oedothorax fuscus diet

312

The diets of linyphiid spiders have been studied previously in cereal crops using other 313

techniques and these too show the importance of Collembola in their diets. Agustí et al.

314

(2003), using species specific primers, reported the consumption of Collembola by a 315

guild of linyphiid spiders that included Oedothorax spp. We confirmed here that 316

Collembola are common prey of O. fuscus. Amongst the Linyphiidae, the Erigoninae, 317

which includes O. fuscus, were shown to build their webs in areas of high Collembola 318

density (Harwood et al. 2003). When their webs were substituted with sticky traps, 319

Collembola were the dominant prey captured. 320

For Review Only

Of great interest was the extent of intraguild predation, a widespread phenomenon 322

amongst spiders (Pekár et al. 2012). Such intraguild effects are likely to negatively affect 323

control of a shared prey, such as aphids. San Andres et al. have confirmed that O. 324

fuscus preys on other spiders, as do other species in the Linyphiidae (unpublished 325

data). 326

327

There were some taxa that have to be considered with caution: (i) All runs showed high 328

numbers of nematode reads, a group of organisms certainly abundant in soil habitats 329

but which are unlikely prey of spiders. They could be internal or external parasites of the 330

spider or of some of their prey (Noordam et al. 1998). Another likely explanation may be 331

that this is secondary predation, as nematodes are major prey of many species of 332

Collembola (Read et al. 2006). (ii) The Auchenorryncha reported in Table 1 was in fact 333

attributed by BLAST plus MEGAN to the cicada Psithyristria crassinervis, but this genus 334

is endemic from Luzon, in the Philiphinnes (Lee & Hill, 2010). For this reason, we 335

reported the reads to the suborder level. (iii) Similarly, the Blattodea reported in Table 1 336

was unambiguously attributed (100% coverage; 100% identity; Genbank accession 337

code HM996892.1 ) to the German cockroach Blatella germanica. Whilst this species 338

can be found in Wales, its preferred habitat are buildings rather than cereal fields. For 339

this reason, we reported the reads to the order level. 340

341

The reason for not attributing many sequences to genus or species levels is the lack of 342

similar taxa in Genbank. This happened in this study with most nematodes and 343

collembolans. Considering the enormous amount of Lepidoptera sequences stored in 344

For Review Only

Genbank, and previous success assigning species level taxonomy to sequences from 345

this region (Razgour et al. 2011) it comes as a surprise that Lepidoptera reads could 346

only be resolved to order level and that those reads were very similar to many different 347

species of Lepidoptera. For example, a group of reads was very similar to Glyphodes nr. 348

stolalis (Pyraloidea: Crambidae; 97% coverage; 98% identity; Genbank accession code 349

JX970289.1), but also to Eois chrysocraspedata (Geometroidea: Geometridae; 97%; 350

97%; JX150916.1), and to Hippia pronax (Noctuoidea: Notodontidae; 97%; 97%; 351

JN806871.1), among others. In this case, MEGAN attributed the sequence to order 352

Lepidoptera. It remains unclear why we were unable to resolve species in MEGAN but it 353

may be caused by differences in BLAST vs. the HMM assignment algorithm of BOLD 354

(Ratnasingham & Hebert 2007). The approach of using short amplicons (necessary in 355

many molecular analyses of diet to ensure DNA survival in the guts of predators and 356

herbivores) could, paradoxically, be problematic in groups like Lepidoptera. In these 357

groups, a longer amplicon or more than one genomic region should be considered to 358

improve taxonomical resolution. Alternatively, a thorough knowledge of the local 359

arthropod fauna, plus construction of a local barcode database should help to assign 360

DNA sequences to species. 361

362

Methodological issues - blocking primers

363

As expected, a high proportion of the sequencing effort (>90%) was lost to co

364

amplification of the predator, but even so we obtained around 58,000 reads with the 318

365

chip that could be attributed to prey (or parasites). Clearly, the use of a blocking primer

366

of O. fuscus would have greatly increased the number of prey reads, but at the cost of

For Review Only

possible co blocking of other prey, especially other spiders. Suppressing amplification of

368

predator DNA may well improve detection of rare, low copy number prey (O’Rorke et al.

369

2012). This is not always an advantage, especially for invertebrate food webs, where

370

weak links may provide little nutrition to a predator and are unlikely to affect overall

371

predator prey dynamics.

372

373

The use of predator specific blocking primers in studies of diets has pros and cons and

374

so it is open to debate. In some instances, it can be highly beneficial. For example,

375

when the predator is phylogenetically distant from prey (seals or penguins eating fish or

376

cephalopods; Deagle et al., 2009, 2010), the use of predator specific blocking primers

377

increases the yield of the sequencing reaction without co blocking (it is hoped) any prey.

378

However, when prey and predator are phylogenetically close the risk of non specific co

379

blocking cannot be disregarded. Here we detected other spiders in O. fuscus diet, some

380

even from the same family (i.e. Tenuiphantes, Bathyphantes), which could have been

381

blocked. Careful design of the blocking primer considering all possible prey could solve

382

the problem, but in a field setting this is impossible because the range of potential prey

383

is often unknown. This also assumes that a sufficiently specific blocking site exists

384

adjacent to the primer region, which is by no means certain.

385

386

We obtained large numbers of prey sequences despite the fact that the whole spider

387

was homogenised. With larger invertebrate predators, such as carabid beetles or

388

earwigs from which the contents of the fore gut can be extracted, predator DNA is likely

389

to be less dominant. In a previous study using general invertebrate primers and TGGE

For Review Only

(temperature gradient gel electrophoresis) separation of PCR products on gels, prey

391

DNA was not swamped by predator DNA and dietary components were clearly

392

distinguished (Harper et al. 2006). Thus the whole spider approach provided a ‘worst

393

case’ scenario; if it can work on spiders it should work on other small predators that

394

cannot be dissected easily.

395

396

Another critical point for many is the cost per read. Here we conducted two runs, one

397

with the low end Ion Torrent chip (314) and one with the more powerful 318 chip. The

398

total cost (at internal rates of the University) was, approximately, 400 euros for the 314

399

and 900 euros for the 318. Considering the total number of prey reads obtained (Table

400

1), the price per read was 0.13 euros/read for the 314 and 0.02 euros/read for the 318.

401

These prices are massively lower than what could be achieved by cloning and

402

sequencing, and much lower still should labour costs be included in the calculations.

403

404

It is likely that the proportions of reads for each taxon only loosely reflect the actual

405

proportion of prey in the diet (Pompanon et al. 2012) but few empirical tests to

406

demonstrate this have been done. DNA analyzed from faeces of captive penguins

407

(Deagle et al., 2010) or seals (Deagle et. al., 2013) fed with a constant diet of three fish

408

species did not reproduce the original proportions of the three species after NGS.

409

Murray et al. (2011) found a good correspondence between the proportions of four

410

species of fish eaten by captive penguins obtained by NGS vs. qPCR, suggesting that

411

NGS does not necessarily introduce any new sources of bias (though qPCR has also

For Review Only

been problematic in some diet studies, see McCracken et al. 2012). Many of the sources 413

of bias (e.g. amplification efficiency, differences in mtDNA copy number) affect both 414

approaches, and indeed cloning and sequencing. The effect is likely to be exacerbated 415

as the diversity of prey increases. Even the short primer tags used to identify individual 416

samples and the stringency in the bioinformatic quality control step seems to induce 417

biases in the quantitative final result (Deagle et al., 2013). With increased prey diversity, 418

the chance of primer bias also increases, thus any attempt at quantification within a 419

sample/library should be avoided. In this context, the numbers of reads of each taxon in 420

Table 1 cannot be interpreted as the relative abundance of each prey in the diet of O.

421

fuscus, but mainly as a record of the richness of taxa in the diet. While there is likely to 422

be a trend such that higher numbers of reads will result from prey being eaten in greater 423

quantities, this can be influenced by the species specific survival of DNA during 424

digestion (e.g. soft bodied prey may degrade while hard bodied are better preserved), 425

the relative DNA content of a prey species (which can differ between genders, ages and 426

reproductive states) and the actual size of prey (larger prey may yield more DNA). 427

428

Methodological issues - repeatability

429

Here we sequenced three different PCR products obtained with the same template DNA. 430

All samples showed the importance of Collembola as well as a number of spiders (albeit 431

different in each sample) in the diet of O. fuscus. All samples also contained sequences 432

of lepidopterans and nematodes. However, dipterans, aphids, leafhoppers, cockroaches, 433

and ephemeropterans were absent in at least one sample. It might be predicted that a 434

greater diversity would be obtained with the 318 chip. This is generally true, but there 435

For Review Only

were exceptions such as the 157 reads for aphids and the 158 of Xysticus obtained with 436

314 forward sequences that did not appear in any other sample. In summary, the 437

method showed a degree of repeatability for the most abundant prey items, but there 438

was a lot of variation for other taxa. We attribute the observed differences to the 439

stochastic nature of the PCR, with some instability of PCR dominance between runs. 440

The results do suggest that (funds permitting) maximum information can be obtained by 441

running the same DNA through NGS more than once. 442

443

The forward runs produced approximately the same number of raw reads as the reverse 444

ones. However, the quality of the reads was poorer in the reverse runs, so the proportion 445

of forward:reverse reads increased to ca. 3:2 after the quality control step. The 446

elimination of predator and uninformative reads then increased the ratio to 2:1. Apart 447

from this fact, the reverse runs did not differ more from the forward ones than the two 448

forward runs (with the 314 and 318 chips) differed from each other. 449

450

Concluding remarks

451

Our proposal is therefore disarmingly simple. The protocol involves DNA extraction, PCR 452

amplification with universal primers, deep sequencing on a NGS platform, bioinformatic 453

disposal of predator reads, BLAST matching and MEGAN taxon assignment. There is no 454

need to sequence the region of interest in all prey to design blocking primers, saving 455

time and money. It is not even necessary to sequence the predator itself, as it is going to 456

appear many times in the output file after NGS. Consequently, when the price per read 457

comes down in the future, we envisage this method as a template to resolve real food 458

For Review Only

webs in which a high number of species interact, a goal that is today unattainable at a

459

reasonable cost in terms of time and money. This method has the advantage of

460

eliminating the danger that predator blocking probes co block DNA from other species.

461

The loss of sequencing power caused by co sequencing of dominant predator DNA is

462

compensated for by the extremely high capacity of modern NGS platforms. NGS of prey

463

remains is essentially an exploratory technique which, after application to pooled

464

samples, allows the primary prey to be identified. This can be followed by rapid

465

screening of large numbers of individuals using appropriate species specific primers

466

(discussed in Pompanon et al. 2012) and multiplexing (Harper et al. 2005) to obtain a

467

quantitative measure of the numbers of predator individuals testing positive for each

468

prey. The same thing could be done with individual tags, but where thousands of

469

predators need to be screened this would currently be very expensive in terms of primer

470

costs. Our data suggest that the ‘brute force’ approach for NGS is viable, but probably

471

not always so. Further work with other predators and general primers would be needed

472

to be sure that this approach could we applied more widely.

473

474

Acknowledgements

475

We thank several anonymous reviewers and Bruce Deagle for useful suggestions on the

476

initial version of the manuscript. Financial help was provided by the MICINN grant

477

CGL2010 18182. This work was developed during a sabbatical leave of JP to Cardiff

478

University financed by the Ministry of Education of Spain. Núria Agustí participated in the

479

early stages of the project. Enrique Blanco, Xavier Espadaler, and Laia Mestre kindly

480

commented on early versions of the manuscript. We thank Mr Robert Reader of Burdons

For Review Only

482

483

References

484

Alderweireldt M (1994) Day/night activity rhythms of spiders occurring in crop rotated

485

fields. European Journal of Soil Biology, 30, 55–61. 486

487

Agustí N, de Vicente MC, Gabarra R (1999) Development of sequence amplified 488

characterized region (SCAR) markers of Helicoverpa armigera: a new polymerase chain 489

reaction based technique for predator gut analysis. Molecular Ecology, 8, 1467 1474. 490

491

Agustí N, Shayler SP, Harwood JD, Vaughan IP, Sunderland KD, Symondson WOC 492

(2003) Collembola as alternative prey sustaining spiders in arable ecosystems: prey 493

detection within predators using molecular markers. Molecular Ecology, 12, 3467 3475. 494

495

Agustí N, Bourguet D, Spataro T, Delos M, Eychenne N, Folcher L, Arditi R (2005) 496

Detection, identification and geographical distribution of European corn borer larval 497

parasitoids using molecular markers. Molecular Ecology, 10, 3267 3274. 498

499

Asahida T, Yamashita Y, Kobayashi T (1997) Identification of consumed stone flounder, 500

Kareius bicoloratus (Basilewsky), from the stomach contents of sand shrimp, Crangon

501

affinis (De Haan) using mitochondrial DNA analysis. Journal of Experimental Marine

502

biology and Ecology, 217, 153 163. 503

For Review Only

Blankenship LE, Yayanos AA (2005) Universal primers and PCR of gut contents to study 505

marine invertebrate diets. Molecular Ecology, 14, 891 899. 506

507

Bohmann K, Monadjem A, Noer CL, Rasmussen M, Zeale MRK, Clare EL, Willerslev E, 508

Gilber MTP (2011) Molecular dietary analysis of two African free tailed bats (Molossidae) 509

using high throughput sequencing. PLoS One, 6, e21441. 510

511

Boyer S, Brown SDJ, Collins RA, Cruickshank RH, Lefort M CD, Malumbres Olarte J, 512

Wratten SD (2012) Sliding window analyses for optimal selection of mini barcodes, and 513

application to 454 pyrosequencing for specimen identification from degraded DNA. 514

PloSOne,7, e38215. 515

Bradley BJ, Stiller M, Doran Sheehy DM, Harris T, Chapman CA, Vigilant L, Poinar H 516

(2007) Plant DNA sequences from faeces: potential means for assessing diets of wild 517

primates. American Journal of Primatology, 69,699–705. 518

519

Brown DS, Jarman SN, Symondson WOC (2012) Pyrosequencing of prey DNA in reptile 520

faeces: analysis of earthworm consumption by slow worms. Molecular Ecology

521

Resources, 12, 259 266. 522

523

Davey JS, Vaughan IP, King RA, Bell JR, Bohan DA, Bruford MW, Holland JM, 524

Symondson WOC (2013) Intraguild predation in winter wheat: prey choice by a common 525

For Review Only

epigeal carabid consuming spiders. Journal of Applied Ecology, 50, 271 279. 526

527

Deagle BE, Chiaradia A, McInnes J, Jarman SN (2010) Pyrosequencing faecal DNA to 528

determine diet of little penguins: is what goes in what comes out? Conservation

529

Genetics,11, 2039–2048.

530

531

Deagle BE, Gales NJ, Evans K, Jarman SN, Robinson S, Trebilco R, Hindell MA (2007)

532

Studying seabird diet through genetic analysis of faeces: a case study on macaroni

533

penguins (Eudyptes chrysolophus). Plos One, 2(9), e831.

534

535

Deagle BE, Thomas AC, Shaffer AK, Trites AW, Jarman SN (2013) Quantifying

536

sequence proportions in a DNA based diet study using Ion Torrent amplicon sequencing:

537

which counts count? Molecular Ecology Resources,13, 620 633.

538

539

Deagle BE, Kirkwood R, Jarman SN (2009) Analysis of Australian fur seal diet by

540

pyrosequencing prey DNA in faeces. Molecular Ecology, 18, 2022–2038.

541

542

De Keer R, Maelfait JP (1987) Life history patterns of Oedothorax fuscus (Blackwall

543

1834) (Araneida, Linyphiidae) in a heavily grazed pasture. Revue d Ecologie et de

544

Biologie Du Sol, 24, 171 185.

545

546

Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R (1994) DNA primers for amplification

547

of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates.

For Review Only

549

550

Harper GL, King RA, Dodd CS, Harwood JD, Glen DM, Bruford MW, Symondson WOC

551

(2005) Rapid screening of invertebrate predators for multiple prey DNA targets.

552

Molecular Ecology, 14, 819 827.

553

554

Harper GL, Sheppard SK, Harwood JD, Read DSM, Glen DM, Bruford MW, Symondson,

555

WOC (2006) Evaluation of temperature gradient gel electrophoresis for the analysis of

556

prey DNA within the guts of invertebrate predators. Bulletin of Entomological Research,

557

96, 295 304.

558

559

Harwood JD, Sunderland KD, Symondson WOC (2003) Web location by linyphiid

560

spiders: prey specific aggregation and foraging strategies. Journal of Animal Ecology,

561

72, 745 756.

562

563

Harwood JD, Sunderland KD, Symondson WOC (2004) Prey selection by linyphiid

564

spiders: molecular tracking of the effects of alternative prey on rates of aphid

565

consumption in the field. Molecular Ecology, 13, 3549 3560.

566

567

Huson DH, Auch AF, Qi J, Schuster S (2007) MEGAN analysis of metagenomic data.

568

Genome Research, 17, 377–386.

569

570

Jarman SN, Deagle BE, Gales NJ (2004) Group specific polymerase chain reaction for

For Review Only

DNA based analysis of species diversity and identity in dietary samples. Molecular

572

Ecology, 13, 1313 1322.

573

574

King RA, Read DS, Traugott M, Symondson WOC (2008) Molecular analysis of

575

predation: a review of best practice for DNA based approaches. Molecular Ecology, 17,

576

947 963.

577

578

King RA, Vaughan IP, Bell JR, Bohan DA, Symondson WOC (2010) Prey choice by

579

carabid beetles feeding on an earthworm community analysed using species and

580

lineage specific PCR primers. Molecular Ecology, 19, 1721 1732.

581

582

Kowalczyk R, Taberlet P, Coissac E, Valentini A, Miquel C, Kami ski T, Wójcik JM 583

(2011). Influence of management practices on large herbivore diet—Case of European 584

bison in Biaøowie a Primeval Forest (Poland). Forest Ecology and Management, 261, 585

821–828. 586

587

Lee YO, Hill KBR (2010) Systematic revision of the genus Psithyristria Stal (Hemiptera: 588

Cicadidae) with seven new species and a molecular phylogeny of the genus and higher 589

taxa. Systematic Entomology, 35, 277–305. 590

591

McCracken GF, Westbrook JK, Brown VA, Eldridge M, Federico P, Kunz TH (2012) Bats 592

track and exploit changes in insect pest populations. PLoS ONE, 7, e43839. 593

For Review Only

Murray DC, Bunce M, Cannell BL, Oliver R, Houston J, White NE, Barrero RA, Bellgard 595

MI, Haile J (2011) DNA Based Faecal Dietary Analysis: A Comparison of qPCR and 596

High Throughput Sequencing Approaches. PloS ONE, 6, e25776. 597

598

Noordam AP, Samson RA, Sudhaus W (1998) Fungi and Nematoda on Centromerus

599

sylvaticus (Araneae, Linyphiidae). In Selden PA (ed.). Proceedings of the 17th European

600

Colloquium of Arachnology, Edinburgh 1997. 601

602

O'Rorke R, Lavery S, Jeffs AG (2012) PCR enrichment techniques to identify the diet of 603

predators. Molecular Ecology Resources, 12, 5–17. 604

605

Passmore AJ, Jarman SN, Swadling KM, Kawaguchi S, McMinn A, Nicol S (2006) DNA 606

as a dietary biomarker in Antarctic krill, Euphausia superba. Marine Biotechnology, 8, 607

686–696. 608

609

Pekár S, Coddington JA, Blackledge TA (2012) Evolution of stenophagy in spiders 610

(Araneae): evidence based on the comparative analysis of spider diets. Evolution, 66, 611

776–806. 612

613

Poinar HN, Hofreiter M, Spaulding WG, Martin PS, Stankiewicz BA, Bland H, Evershed 614

RP, Possnert G, Paabo S (1998) Molecular coproscopy: dung and diet of the extinct 615

ground sloth Nothrotheriops shastensis. Science, 281, 402 406. 616

For Review Only

Pompanon F, Deagle BE, Symondson WOC, Brown DS, Jarman SN, Taberlet P (2012) 618

Who is eating what: diet assessment using next generation sequencing. Molecular

619

Ecology, 21, 1931 1950. 620

621

Ratnasingham S, Hebert PDN (2007) BOLD: the barcode of life data system 622

(http://www.barcodinglife.org). Molecular Ecology Notes, 1, 355 364. 623

624

Raye G, Miquel M, Coissac E, Redjadj C, Loison A, Taberlet P (2011) New insights on 625

diet variability revealed by DNA barcoding and high throughput pyrosequencing: 626

chamois diet in autumn as a case study. Ecological Research, 26, 265–276 . 627

628

Razgour O, Clare EL, Zeale MRK, Hanmer J, Schnell IB, Razmussen M, Gilbert TP, 629

Jones G (2011). High throughput sequencing offers insight into mechanisms of resource 630

partitioning in cryptic bat species Ecology and Evolution, 1 556 570. 631

632

Read DS, Sheppard SK, Bruford MW, Glen DM, Symondson WOC (2006) Molecular 633

detection of predation by soil microarthropods on nematodes. Molecular Ecology,15, 634

1963 1972. 635

636

Roberts M (1996) Spiders of Britain and Northern Europe (3rd Ed.). Harper Collins, 637

London. 638

639

Schmieder R, Edwards R (2011) Quality control and preprocessing of metagenomic 640

For Review Only

641

642

Schmieder R, Lim YW, Rohwer F, Edwards R (2011) TagCleaner: Identification and 643

removal of tag sequences from genomic and metagenomic datasets. BMC

644

Bioinformatics, 11, 341 645

646

Shehzad W, Riaz T, Nawaz MA, Miquel C, Poillot C, Shah SA, Pompanon F, Coissac E, 647

Taberlet P (2012). Carnivore diet analysis based on next generation sequencing: 648

application to the leopard cat (Prionailurus bengalensis) in Pakistan. Molecular Ecology, 649

8, 1951 1965 650

651

Sutherland RM (2000) Molecular Analysis of Avian Diets. PhD Thesis, University of 652

Oxford (UK). 653

654

Symondson WOC (2002). Molecular identification of prey in predator diets. Molecular

655

Ecology, 11, 627 641. 656

657

Symondson WOC, Sunderland KD, Greenstone MH (2002) Can generalist predators be 658

effective biocontrol agents? Annual Review of Entomology, 47, 561 594. 659

660

Traugott M, Bell JR, Broad GR, Powell W, van Veen FJF, Vollhardt IMG, Symondson 661

WOC (2008) Endoparasitism in cereal aphids: molecular analysis of a whole parasitoid 662

community. Molecular Ecology, 17, 3928 3938. 663

For Review Only

664

Traugott M, Bell JR, Raso L, Sint D, Symondson WOC (2012) Generalist predators 665

disrupt parasitoid aphid control by direct and coincidental intraguild predation. Bulletin of

666

Entomological Research,102, 239 247 667

668

Valentini A, Miquel C, Nawaz MA, Bellemain E, Coissac E, Pompanon F, Gielly L, 669

Cruaud C, Nascetti G, Wincker P, Swenson JE, Taberlet P (2009) New perspectives in 670

diet analysis based on DNA barcoding and parallel pyrosequencing: the trnL approach. 671

Molecular Ecology Resources, 9, 51–60. 672

673

Vestheim H, Jarman SN (2008) Blocking primers to enhance PCR amplification of rare 674

sequences in mixed samples – a case study on prey DNA in Antarctic krill stomachs. 675

Frontiers in Zoology, 5, 12. 676

677

Zaidi RH, Jaal Z, Hawkes NJ, Hemingway J, Symondson WOC (1999) Can the detection 678

of prey DNA amongst the gut contents of invertebrate predators provide a new 679

technique for quantifying predation in the field? Molecular Ecology, 8, 2081 2087. 680

681

Zeale MRK, Butlin RK, Barker GLA, Lees DC, Jones G (2011) Taxon specific PCR for 682

DNA barcoding arthropod prey in bat faeces . Molecular Ecology Resources, 11, 236– 683

244. 684

685

Zhang Z, Schwartz S, Wagner L, Miller W (2000) A greedy algorithm for aligning DNA 686

For Review Only

687

688

Authors contribution

689

JP prepared the DNA for NGS, conducted the bioinformatics and drafted the manuscript. 690

VSA designed and carried out the field sampling, identified the spiders and extracted the 691

DNA from the spiders. ELC was an advisor on the bioinformatics. GM was responsible 692

for the Ion Torrent sequencing. WOCS was overall supervisor of the project, coordinated 693

the group and co drafted the manuscript. All authors played a role in editing the final 694

version of the paper. 695

696

Data Accessibility

697

Perl code and sequence data (FASTQ): Dryad doi:10.5061/dryad.qv633. 698

699

Supporting information

700

Additional supporting information may be found in the online version of this article. 701

702

Table S1. Alignment of the Zeale et al. (2011) primers with some representative insects, 703

spiders, and collembolans. 704

705

Table S2. Fusion primers for the forward and reverse sequencing in the Ion Torrent 706

PGM. 707

For Review Only

Both forward (fwd) and reverse (rvr) reads using the 314 or the 318 chips are given, as well as the sum of all them. 710

Broader group Intermediate group Family Genus or species 314 fwd 318 fwd 314 rvr 318 rvr TOTAL Nematoda Chromadorea 257 873 9 206 1345 Nematoda Rhabditida 24 81 36 1353 1494 Collembola 10 220 0 0 230 Collembola Entomobryoidea 1527 32594 491 11445 46057 Collembola Entomobryoidea Isotomidae 0 2 8 164 174 Collembola Entomobryoidea Isotomidae Parisotoma notabilis 102 2186 71 1417 3776 Collembola Entomobryoidea Sminthuridae Sminthurus viridis 2 0 0 0 2 Collembola Entomobryoidea Tomoceridae 0 4 2 6 12 Araneae Thomisidae 27 0 0 0 27 Araneae Thomisidae Xysticus 158 0 0 0 158 Araneae Therididae Theridion 0 8 0 0 8 Araneae Linyphiidae Tenuiphantes tenuis 1 0 11 303 315 Araneae Linyphiidae Centromerita bicolor 0 0 0 5 5 Araneae Linyphiidae

Bathyphantes

gracillis 0 0 3 127 130

Insecta Diptera 0 0 74 1322 1396 Insecta Diptera (Schizofora) 0 0 42 1041 1083 Insecta

Diptera

(Acalyptratae) 0 1066 0 0 1066 Insecta Diptera Cecidomyiidae 0 0 48 1009 1057 Insecta Diptera Anthomyiidae Delia 0 0 0 148 148 Insecta Lepidoptera 97 693 32 1251 2073 Insecta Hemiptera Aphididae Aphis 157 0 0 0 157 Insecta

Hemiptera

(Auchenorrhyncha) 0 274 0 0 274 Insecta Blattodea 0 273 0 0 273 Insecta Ephemeroptera Heptageniidae Afronurus 0 0 0 7 7

1

accession number, species, and family of all the reported sequences are the following ones:

Accesion # Species Family

Collembola 1 JN298134 Parisotoma notabilis Isotomidae Collembola 2 HM398042 Tomocerus minor Tomoceridae Araneae 1 HQ924416 Erigone atra Linyphiidae Araneae 2 HQ979349 Philodromus rufus vibrans Philodromidae Hemiptera EU701907 Sitobion avenae Aphididae Diptera HQ979110 Drosophila simulans Drosophilidae Coleoptera JF296222 Coccinella septempunctata Coccinellidae Lepidoptera JF859814 Cydia pomonella Tortricidae Psocoptera HQ978921 Psocoptera sp

Hymenoptera JN300351 Aphidius sp. Braconidae Ephemeroptera JN297844 Baetis tricaudatus Baetidae

2 Diptera AACTTTATATTTTATCTTTGGAGCTTGAGCTGGGATAGTCGGAACATCATTAAGAATTTT Coleoptera AACATTATATTTCTTATTCGGAATATGAGCCGGAATAATTGGGACCTCTTTAAGAATTTT Lepidoptera AACATTATATTTTATTTTTGGTATTTGAGCCGGAATAGTAGGAACTTCTCTAAGATTACT Psocoptera AACCATATATTTCATTTTTGGAATTTGAGCTGGAATAGTAGGTTCTAGATTAAGTATACT Hymenoptera AATTTTATATTTTATTTTTGGTATATGATCAGGAATAGTTGGGTTATCAATAAGATTAAT Ephemeroptera TACTCTGTATTTTATTTTTGGTGCTTGGTCGGGTATGGTGGGCACTTCTCTTAGTTTGTT ZBJ-ArtF1c AACWTTATATTTTATTTTTGG Collembola 1 AATCCGGTTAGAATTGGGACAAACAGGATCTTTTATCGGAGATGACCAAATT-TACAATG Collembola 2 TATTCGACTTGAACTAGGACAACCAGGTAGATTTATTGGTGATGATCAAATT-TATAATG Araneae 1 AATTCGTATTGAGTTAGGACAAACTGGAAGATTGTTAGGGGATGATCA-ATTGTATAATG Araneae 2 AATTCGAATAGAATTAGGACAAGTAGGTAAATTTTTAGGTGATGATCA-TTTGTATAATG Hemiptera TATTCGTCTTGAATTAAGACAAATTAATTCAATTATTAATAATAATCA-ATTATATAATG Diptera AATTCGAGCCGAATTAGGACATCCTGGAGCATTAATCGGAGATGACCAAATT-TATAATG Coleoptera AATTCGTCTTGAATTAGGAACTACTAATAGATTAATTGGAAATGACCAAATT-TATAATG Lepidoptera TATTCGAGCAGAATTAGGAAATCCAGGATCTTTAATTGGTGATGATCAAATT-TATAATA Psocoptera AATTCGTTTAGAATTAAGTCAACCAGGCTTACTCATAGAAGATGACCAAACA-TATAATG Hymenoptera TATTCGAATAGAATTAAGAATTACTGGTACTTTTATTGGTAATGATCAAATT-TATAATA Ephemeroptera AATTCGGGCTGAGTTGGGTAATCCTGGCTCACTTATTGGGGATGACCAGATT-TATAACG Collembola 1 TTGCAGTGACTGCCCATGCTTTTATTATAATTTTTTTCATAGTTATACCTATTATGATTG Collembola 2 TAATAGTTACGGCCCACGCTTTTATTATAATTTTTTTTATAGTTATACCAATCATAATTG Araneae 1 TTATCGTTACGGCGCATGCTTTTATTATAATTTTTTTTATAGTTATACCTATTTTAATTG Araneae 2 TTATTGTTACTGCGCATGCATTTGTTATAATTTTTTTTATAGTAATACCTATTTTAATTG Hemiptera TAATTGTTACAATCCATGCTTTTATTATAATTTTTTTTATAACTATACCAATTGTTATTG Diptera TAATTGTAACTGCACATGCTTTTATTATAATTTTTTTTATAGTTATACCTATTATAATTG Coleoptera TAATTGTAACAGCTCATGCCTTCATTATAATTTTTTTTATAGTTATACCAATTATAATTG Lepidoptera CTATTGTAACTGCTCATGCTTTTATTATAATTTTTTTTATAGTAATACCTATTATAATTG Psocoptera TAATTGTTACTGCACATGCTTTCATTATAATTTTCTTCATAATTATACCAATTATAATTG Hymenoptera GTATTGTTACTGCACATGCTTTTGTAATAATTTTTTTTATAGTTATACCTATCATAATTG Ephemeroptera TTATTGTTACTGCTCATGCGTTTATTATAATCTTTTTTATAGTGATACCAATTATAATCG ZBJ-ArtR2c G Collembola 1 GAGGATTCGGAAACTGATTAGTTCCTTTAATAATTGGAGCCCCGGATATAGCGTTCCCTC Collembola 2 GAGGGTTTGGAAATTGATTAGTACCATTAATAATTAGAGCCCCAGATATAGCATTTCCAC Araneae 1 GGGGATTTGGTAACTGATTAGTTCCTTTAATATTAGGGGCTCCTGATATAGCTTTTCCTC Araneae 2 GTGGATTTGGAAATTGATTAGTTCCTTTAATATTAGGTGCTCCAGATATAGCATTTCCTC Hemiptera GTGGTTTTGGAAATTGATTAATTCCTATAATAATAGGATGTCCTGATATATCATTCCCAC Diptera GTGGATTTGGAAATTGATTAGTGCCTTTAATATTAGGTGCCCCTGATATAGCATTCCCGC Coleoptera GAGGATTTGGAAATTGACTTGTTCCTTTAATAATTGGAGCACCTGACATAGCTTTCCCTC Lepidoptera GTGGATTTGGTAATTGATTAGTACCACTAATATTAGGAGCTCCTGATATAGCTTTTCCTC Psocoptera GAGGATTTGGAAATTGATTAGTTCCTTTAATACTAGGAGCCCCCGATATAGCATTTCCTC Hymenoptera GAGGGTTTGGAAATTGATTAATTCCTTTAATATTAGGAGCTCCTGACATGGCTTTTCCTC Ephemeroptera GTGGATTTGGGAATTGGCTTGTACCCCTTATGTTAGGTGCCCCAGACATGGCTTTCCCTC ZBJ-ArtR2c GAGGATTTGGWAATTGATTAGTW

1

accession number, species, and family of all the reported sequences are the following ones:

Accesion # Species Family

Collembola 1 JN298134 Parisotoma notabilis Isotomidae Collembola 2 HM398042 Tomocerus minor Tomoceridae Araneae 1 HQ924416 Erigone atra Linyphiidae Araneae 2 HQ979349 Philodromus rufus vibrans Philodromidae Hemiptera EU701907 Sitobion avenae Aphididae Diptera HQ979110 Drosophila simulans Drosophilidae Coleoptera JF296222 Coccinella septempunctata Coccinellidae Lepidoptera JF859814 Cydia pomonella Tortricidae Psocoptera HQ978921 Psocoptera sp

Hymenoptera JN300351 Aphidius sp. Braconidae Ephemeroptera JN297844 Baetis tricaudatus Baetidae