SEARCH

SEARCH BY CITATION

Keywords:

  • Cantharidae;
  • Carabidae;
  • conservation biological control;
  • detrital subsidy;
  • intraguild predation;
  • multiplex PCR

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

1. Generalist predators such as carabid and cantharid beetles form an important component within natural enemy communities in arable land. Usually, predator–prey interactions are examined during the vegetation period, largely ignoring food web interactions occurring in the cold season. This is, however, when the larval stages of many polyphagous beetles develop whose survival critically depends on the availability of suitable prey.

2. In this study, we examined intra- and extraguild feeding links in larval Cantharis spp. (Coleoptera: Cantharidae) and Nebria brevicollis (Coleoptera: Carabidae), both abundant cold-adapted invertebrate predators in European arable land. As these immature beetles are fluid feeders, which impedes a microscopic analysis of their gut content, multiplex polymerase chain reaction assays were developed to examine trophic linkages in the field.

3. Collembolan DNA was detected in 49% and 54% of cantharid and carabid larvae, respectively. Earthworms were consumed by 34% of cantharid and 24% of carabid larvae. Significant differences in lumbricid and collembolan prey detection rates occurred between sampling dates in Cantharis spp. and N. brevicollis larvae, respectively, suggesting that these predators utilized different feeding strategies. In both predator taxa, however, only few individuals (0·2–1%) tested positive for DNA of intraguild prey, indicating that predator–predator trophic interactions were scarce in this community.

4.Synthesis and applications. We present conclusive evidence that cold-adapted predatory beetle larvae are strongly linked to the detrital food chain by feeding on collembolans and earthworms. By improving the habitat conditions for detritivores in arable land by mulching, compost applications, or provision of plant cover during winter, their densities can be increased easily. Applying these measures year round will retain and sustain predatory beetles during their whole life cycle in arable land. Ultimately, we expect that these measures will enhance the ability of polyphagous beetle predators to provide their fundamental ecosystem service as regulators of agricultural pests.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Throughout the world carnivorous invertebrates deliver the key ecosystem service of pest control by attacking a wide range of herbivorous insects which otherwise would inflict considerable damage in arable crops (Van Driesche, Hoddle & Center 2008). Aside from specialized predators and parasitoids, generalist predators are common in arable ecosystems where they consume both pest and non-pest foods. The latter sustain them when pest densities are low and allow generalist predators to impact pests immediately after colonization, making them effective biocontrol agents (Settle et al. 1996; Symondson, Sunderland & Greenstone 2002; Harwood et al. 2007). Besides consuming prey from lower trophic levels, polyphagous predators often feed on other predators and parasitoids, which also attack pest species (Symondson, Sunderland & Greenstone 2002). This intraguild predation can negatively affect the levels of pest control exerted by natural enemies and alter the trophic structuring of communities as well as their dynamics (Rosenheim 1998; Vance-Chalcraft et al. 2007).

In temperate agro-ecosystems, predator–prey interactions are typically examined during the growth period of the crop (i.e. spring and summer), whereas food web interactions occurring in the cold months of the year remain largely unexplored. Aside from diapausing species, there are, however, several invertebrate taxa that become active in autumn and winter (Tischler 1965). These include, for example, the larvae of predatory beetles such as carabids and cantharids, which develop between late summer and spring (Noordhuis, Thomas & Goulson 2001; Juen, Steinberger & Traugott 2003; Yamazaki, Sigiura & Kawamura 2003). The performance of these juvenile beetles determines the size of adult beetle populations in summer and hence their potential to control pests. As larval survival critically depends on the availability of high quality prey (Bilde, Axelsen & Toft 2000; Kyneb & Toft 2004), assessing their prey choices under natural conditions is of great importance. Examining the dietary choices of these cold-adapted predatory beetle larvae in the field, however, is fraught with difficulties as they are fluid feeders prohibiting microscopic analysis of their gut contents. Moreover, their small size and cryptic life style hinder direct observation and identification. Molecular analysis of predation can overcome these difficulties (King et al. 2008) and provides a powerful tool for determining trophic linking in fluid-feeding polyphagous predators (e.g. Agusti et al. 2003; Juen & Traugott 2007; Lundgren, Ellsbury & Prischmann 2009; Szendrei et al. 2010).

In this study, we examine the feeding behaviour of predatory beetle larvae during autumn and winter in arable land. Our analysis focuses on Cantharis spp. and Nebria brevicollis larvae, two abundant generalist predator taxa in European arable land during the cold period of the year (Nelemans 1988; Noordhuis, Thomas & Goulson 2001; Juen, Steinberger & Traugott 2003; Traugott 2006). Based on field survey data on the above- and below-ground invertebrate community active between autumn and spring, we developed two multiplex polymerase chain reaction (PCR) assays allowing us to screen cantharid and carabid larvae, collected in a grassland and an arable field, for both extra- and intraguild prey. Our objectives were twofold: (i) to identify molecularly the frequency of trophic links between larval beetle predators and detritivorous, herbivorous and carnivorous prey; and (ii) to assess how trophic linking changes during the cold season. We tested the hypotheses that cold-adapted predatory beetle larvae show strong trophic links to both extra- and intraguild prey and that prey detection is unrelated to season.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Field sampling

Larvae of Cantharis spp. and Nebria brevicollis were caught near Innsbruck, Austria, by dry pitfall trapping within a small field (0·1 ha, 12 traps) planted with spelt Triticum spelta, a grassland (0·2 ha, 12 traps) and the borderline of these two habitats (10 traps). To reduce predation among caught animals, traps were filled with clay granules (Lecaton®) and arranged in 0·5 × 0·5 m trap grids (three rows and five columns per trap grid; trap grids in field and grassland only). Trapping was conducted six times during the investigation period (27–29 October 2006, 17–19 November 2006, 9–11 December 2006, 7–8 January 2007, 15–17 February 2007 and 11–13 March 2007) and traps were emptied daily, both in the morning and late afternoon (morning only at fourth and fifth sampling date due to low temperatures). Caught larvae were transferred individually into 1·5-mL reaction tubes and stored immediately upon collection at −28 °C. Since catches of Cantharis spp. and N. brevicollis larvae were low in January and March, only catches from October, November, December and February were considered for molecular gut-content analysis.

To record the species spectrum and activity patterns of autumn- and winter-active epigaeic invertebrates, pitfall traps [each trap filled with c. 50 mL ethylene glycol–water (1 : 2)] were installed between 26 October 2006 and 13 March 2007 and emptied every 7–12 days. The endogaeic invertebrate fauna was sampled by taking soil cores (Ø 30 cm, 5 cm deep) on 30 October 2006 (n = 30), 21 November 2006 (n = 30), 7 February 2007 (n = 40) and 12 March 2007 (n = 40). Half of the soil samples each were taken in the field and the grassland along a diagonal transect between the habitat border and centre. From these soil cores, animals were extracted using a Kempson extractor (Kempson, Lloyd & Ghelardij 1963). Soil temperature (at c. 5 cm depth) and air temperature (at c. 15 cm above the ground) were recorded by HOBO Pro data loggers (Onset, Bourne, MA, USA) installed within the grassland and the spelt field.

DNA extraction and sequencing

Each taxon targeted by the molecular assay (see below), as well as the most abundant ‘other’ taxa obtained by pitfall trapping and soil sampling (‘non-target’ taxa), was used for DNA extraction. Non-target taxa included Enchytraeidae, Opiliones (Phalangium opilio), Araneae (Linyphiidae; Centromerus sylvaticus, Meioneta rurestris, Oedothorax apicatus, Tiso vagans; Lycosidae: Alopecosa sp., Pardosa agrestis, Pardosa sp.; Tetragnathidae: Pachygnatha degeeri), Coleoptera (Carabidae: Amara consularis, Dyschirius globosus, Poecilus versicolor, Pterostichus melanarius; Staphylinidae: Ocypus fuscatus, Quedius nitipennis, Xantholinus linearis, Atheta sp., unidentified Hydrophilidae; Scarabaeidae: Aphodius sp.), Dermaptera (Forficula auricularia), Homoptera (Anaceratagallia ribauti), and a variety of Diptera (Bibionidae, Cecidomyiidae, Chironomidae, Limoniidae, Muscidae, Scatopsidae, Stratiomyidae). DNA extracts of these non-target taxa were used in cross-reactivity tests to evaluate the specificity of the molecular assays (see below). The target species included the abundant cold-adapted predators Cantharis fusca, Cantharis livida and Cantharis rustica (Coleoptera: Cantharidae), Nebria brevicollis and Calathus fuscipes (Coleoptera: Carabidae), Centromerita bicolor and Erigone dentipalpis (Araneae: Linyphiidae), the widespread granivorous larvae of the carabid Harpalus rufipes as well as collembolans and earthworms (species listed below) which are ubiquitous inhabitants of arable systems. Small animals such as linyphiid spiders were extracted as a whole, while from larger species muscle tissue of legs was taken, avoiding inclusion of DNA from their gut contents.

DNA was extracted following the CTAB-protocol provided in Juen & Traugott (2005) with the following modifications: Five μL of RNase (20 mg mL−1; Sigma, St. Louis, MO, USA) were added to the TES-buffer and samples were incubated overnight for lysis. DNA-pellets were dried at 30 °C for c. 45 min, pellets were dissolved in 150 μL TE buffer (10 mm TRIS, 1 mm EDTA, pH 8·0; Sigma) for 30 min at 30 °C using a sample shaker and samples finally stored at −28 °C.

For the design of primers, part of the mitochondrial cytochrome oxidase subunit I (COI) gene was sequenced from a minimum of two individuals per target species using the universal invertebrate primers LCO1490 and HCO2198 (Folmer et al. 1994). Each 10 μL PCR contained 0·2 mm dNTPs (Genecraft, Cologne, Germany), 1× PCR buffer (Genecraft), 3 mm MgCl2 (Genecraft), 0·5 μg bovine serum albumin (AppliChem, Darmstadt, Germany), 0·375 U Taq polymerase (Genecraft), 1 μm of each primer, 1·5 μL of DNA extract and PCR water to fill the reaction up to 10 μL. PCR cycling conditions were 94 °C for 15 s followed by 35 cycles of 94 °C for 15 s, 48 °C for 20 s, 72 °C for 30 s and a final elongation of 2 min at 72 °C. PCR products were purified and sequenced in both forward and reverse directions. Sequences were corrected manually and checked for similarity with published COI sequences in GenBank using the BLAST algorithm (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi). Sequences were submitted to GenBank (accession numbers GU323020GU323032).

Primer design and multiplex PCR

Primers were designed using PrimerPremier (PREMIER Biosoft International, Palo Alto, CA, USA) following the guidelines of King et al. (2008). Several primer pairs were designed for targeting intraguild prey including cantharid beetles (one group-specific primer pair covering all three Cantharis species), carabid beetles (species-specific primer pairs for C. fuscipes, H. rufipes and N. brevicollis) and linyphiid spiders (group-specific primer pair covering C. bicolor and E. dentipalpis). Predation on extraguild prey was assessed by employing group-specific primers for collembolans (Col4F/Col5R; Kuusk & Agusti 2007) and earthworms (185F/14233R; Harper et al. 2005) as well as a primer pair specific to H. rufipes. To verify that the latter primers target the most abundant collembolan and earthworm species present at the study site, they were tested with DNA extracts of the collembolans Desoria hiemalis, Hypogastrura cf. vernalis, Isotoma anglicana, Isotoma viridis, Lepidocyrtus cyaneus and Sminthurus viridis as well as the earthworms Aporrectodea caliginosa, Aporrectodea rosea, Lumbricus castaneus, Lumbricus rubellus, Lumbricus terrestris and Octolasion lacteum.

To screen the gut contents of Cantharis spp. and N. brevicollis larvae for DNA of extraguild and intraguild prey, two multiplex PCR assays were developed. The same primer mix was used in both multiplex systems except that the assay used to screen Cantharis spp. larvae contained a primer pair targeting N. brevicollis, whereas the assay used to screen N. brevicollis larvae included a primer pair targeting Cantharis DNA. Each 10 μL multiplex PCR reaction contained 1 μL PCR water, 5 μL multiplex PCR reaction mix (Qiagen, Hilden, Germany), 1 μL primer mix (for primer concentrations see Table 3) and 3 μL of DNA extract. PCR cycling conditions were 94 °C for 15 min followed by 35 cycles of 94 °C for 30 s, 61 °C for 90 s, 72 °C for 90 s and a final elongation at 72 °C for 10 min. PCR products were electrophoresed in 3% ethidium bromide-stained agarose gels at 90 V and visualized by ultraviolet transillumination. Both multiplex PCR assays were tested for their cross-reactivity using DNA from all non-target taxa and all target species, including all earthworm and collembolan species listed above.

Table 3.   Primer names and sequences, expected product sizes (Size) and primer concentrations (Con.) for primers used in multiplex PCRs to detect cantharid, carabid, linyphiid, collembolan and earthworm DNA within the gut contents of predatory beetle larvae
Taxon targetedPrimer namePrimer sequence (5′–3′)Size (bp)Con. (μm)
  1. *The multiplex PCR used for screening Cantharis spp. larvae included the primer pair for N. brevicollis, whereas the multiplex PCR used for screening N. brevicollis larvae included the primer pair targeting all Cantharis species.

Nebria brevicollis*Neb-bre-S200TCCGAGCAGAATCAGGAAAC1562
Neb-bre-A200CTCCTAGTATTAAGGGCACAAGC2
Cantharis group*Can-gen-S202TCCCACGAATAAATAATATAAGATTC1662
Can-gen-A203CTAAAAATAGCTARATCAACAGATGG2
Collembola groupCol4FGCTACAGCCTGAACAWTWG1774
Col5RTCTTGGCAAATGCTTTCGCAGTA4
Calathus fuscipesCal-fus-S203AGCTCATGCATTTGTAATAATTTTC1972
Cal-fus-A206CTTGTCCCAGCTCCTCTTTCG2
Earthworm group185FTGTGTACTGCCGTCGTAAGCA225–2362
14233RAAGAGCGACGGGCGATGTGT2
Linyphiidae groupLin-gen-S208GGrAGAWTRTTAGGKGATGATCAAT2452
Lin-gen-A211CATCCMGCCCCTACWCCTA2
Harpalus rufipesHar-ruf-S206CTGAATTGGGGACTCCAGGT2782
Har-ruf-A209AGGAGGGTAAATTGTTCATCCG2

Primer sensitivity was determined for all primer pairs used in the multiplex PCRs by testing dilutions of target DNA for amplification success (for the group-specific collembolan, Cantharis, linyphiid and earthworm primer pairs DNA from I. viridis, C. fusca, E. dentipalpis and L. rubellus was used, respectively, as these species were the most abundant ones in the respective group at our sampling site). DNA concentration in the original extracts was determined using NanoDrop ( NanoDrop Technologies, Wilmington, DE, USA), adjusted to 0·5 ng μL−1 and twofold serially diluted. The serial diluted target DNA was then used as template in the multiplex assays at concentrations of 300, 150, 75, 37·5, 18·75, 9·38, 4·69, 2·34, 1·17, 0·59, 0·29, 0·15, 0·07, 0·04 and 0·02 pg of target DNA per μL PCR.

Molecular gut-content analysis

Whole beetle larvae were homogenized using a pestle and a ball mill (Retsch MM301, Haan, Germany) and subjected to the CTAB-protocol described above. Extracted DNA was cleaned using Geneclean Turbo Nucleic Acid Purification Kit (QBiogene, Heidelberg, Germany). To check for DNA carry-over contamination during DNA extraction, one extraction blank was included within each batch of 47 predators and tested with universal primers (Folmer et al. 1994; PCR conditions see above). Each PCR (96-well plate) included 86 field-collected predator samples, four negative controls (PCR water) and six positive controls (DNA of C. fuscipes, H. rufipes, E. dentipalpis, I. anglicana, L. rubellus, and N. brevicollis or C. fusca).

Statistical analyses

To compare prey detection rates between predator taxa and time points at the < 0·05 level (Payton, Greenstone & Schenker 2003), 84% tilting confidence intervals (CI; Hesterberg et al. 2003) were calculated by 9999 bootstrap resamples using s-plus 8.0 (Insightful Corporations, Seattle, WA, USA). Mean activity trappability densities (Sunderland et al. 1995) and mean temperatures were compared between sampling sites and the four sampling periods by nonparametric tests [Kruskal–Wallis (K–W) anova, Mann–Whitney (M–U) U-test]. Spearman correlation coefficient was employed to calculate correlations between activity densities and ambient temperature. Nonparametric tests and correlation analysis were done in spss 15.0 (SPSS Inc., Chicago, IL, USA).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The autumn- and winter-active invertebrate community

Pitfall catches of predatory macroinvertebrates were dominated by soldier beetle larvae (Cantharis spp.), carabid larvae (N. brevicollis, C. fuscipes and H. rufipes), linyphiid spiders (E. dentipalpis, Erigone atra) and lycosid spiders (Alopecosa sp., Pardosa sp.) (Table 1). Overall, invertebrate epigaeic activity was positively correlated with air temperature (Spearman correlation coefficient 0·629; = 0·009) and was low between the end of December and the beginning of February when frost and snow prevailed (Fig. 1). Cantharis spp. and N. brevicollis larvae as well as E. dentipalpis and E. atra were active during the whole sampling period, while the surface activity of larval C. fuscipes and H. rufipes ceased at the end of November. With rising soil and air temperature, catches of E. dentipalpis, E. atra and lycosid spiders increased (Spearman correlation coefficient 0·753; = 0·001) and peaked between mid February and mid March. Pitfall catches of E. dentipalpis (U = 2·0; < 0·001), E. atra (U = 3·5; < 0·001) and N. brevicollis (U = 0·0; < 0·001) were significantly higher in the grassland than in the field (M–W U-test), whereas no significant differences in pitfall catches occurred in larval Cantharis spp., C. fuscipes and H. rufipes. Soil samples revealed a rich endogaeic invertebrate fauna, dominated by collembolans, mites, nematodes and enchytraeids (Table 2). Aside from these taxa, dipteran larvae, adult beetles (mostly Aphodius sp.), beetle larvae and earthworms were abundant.

Table 1.   Autumn- and winter-active predatory macroinvertebrates caught in pitfall traps in a spelt field (12 traps), a grassland (12 traps) and the border between the two habitats (10 traps) between 26 October 2006 and 13 March 2007. For each taxon, the total number of animals caught (n), the arithmetic mean per trap (mean) and the standard error (SE) are provided
TaxonFieldGrasslandBorder
nMean ± SEnMean ± SEnMean ± SE
Coleoptera: Cantharidae
 Cantharis spp. larvae76253·4 ± 12·4193355·8 ± 14·68135964·1 ± 17·36
Coleoptera: Carabidae
 Nebria brevicollis larvae494·1 ± 0·9434828·4 ± 4·57836·9 ± 1·32
 Harpalus rufipes larvae432·5 ± 1·21302·3 ± 1·02853·8 ± 1·42
 Calathus fuscipes larvae242·5 ± 0·83201·9 ± 0·48373·9 ± 0·96
 Other carabid larvae161·1 ± 0·4360·4 ± 0·24271·2 ± 0·5
 Carabid adults673·7 ± 1·351707·3 ± 3·222907·9 ± 2·80
Coleoptera: Staphylinidae
 Staphylinid larvae373·1 ± 1·39131·1 ± 0·51332·1 ± 0·83
 Staphylinid adults876·1 ± 1·46543·3 ± 1·081417·4 ± 1·89
Araneae: Linyphiidae
 Erigone dentipalpis13510·8 ± 4·0851340·3 ± 14·6138433·5 ± 12·20
 Erigone atra393·3 ± 1·4916213·4 ± 5·36878·7 ± 3·23
 Centromerita bicolor312·5 ± 0·83373·1 ± 1·05212·1 ± 0·87
 Oedothorax sp.816·2 ± 1·99956·3 ± 1·86261·9 ± 0·59
Araneae: Lycosidae
 Lycosid spiders1449·2 ± 3741488·5 ± 2·542168·5 ± 2·59
Araneae: other families
 Other spiders715·3 ± 1·1218812·8 ± 2·741147·6 ± 2·03
image

Figure 1.  Pitfall trap catches of autumn- and winter-active macroinvertebrates and air temperature (c. 15 cm above the ground) in arable land between 26 October 2006 and 13 March 2007.

Download figure to PowerPoint

Table 2.   Soil-dwelling invertebrates recorded in soil samples taken in a field (= 66) and a grassland (= 66) between 30 October 2006 and 12 March 2007. For each taxon, the number of extracted individuals (n), the arithmetic mean per sample (mean) and the standard error (SE) are provided
TaxonFieldGrassland
nMean ± SEnMean ± SE
Collembolans11 469184·9 ± 23·6914 110227·6 ± 33·16
Mites411466·4 ± 13·456320101·9 ± 21·09
Nematodes247839·7 ± 9·51313550·6 ± 11·21
Enchytraeids8178131·9 ± 16·18340754·9 ± 9·71
Dipteran larvae149924·2 ± 2·96337754·5 ± 8·26
Beetle adults4477·2 ± 1·473876·2 ± 0·83
Beetle larvae711·1 ± 0·25198·4 ± 1·01
Earthworms1842·9 ± 0·533746·0 ± 0·93
Other invertebrates911·5 ± 0·2763110·2 ± 1·46

Assay specificity and sensitivity

The multiplex PCR assays, including the newly designed primers, were successful in detecting the six prey taxa –Cantharis spp., N. brevicollis, H. rufipes, C. fuscipes, earthworms and collembolans – by amplifying taxon-specific DNA fragments ranging from 156 to 278 bp (Table 3). Within the two multiplex PCR assays the primers amplified only DNA from the target organisms. Only small differences in the sensitivity of the different primers used within the multiplex PCR assays were found: the detection limit for the primer pairs targeting H. rufipes, collembolans, earthworms was 0·07 pg of target DNA per μL PCR and for the detection of Cantharis, linyphiid and C. fuscipes DNA the minimum amount of DNA was 0·15 pg DNA per μL PCR. DNA of N. brevicollis was detectable down to 0·29 pg DNA per μL PCR.

Molecular gut-content analysis of field-caught predators

A total of 429 Cantharis spp. (sixth instar) and 98 N. brevicollis larvae (second and third instar), collected at four sampling dates, were screened for DNA of the four intraguild (C. fuscipes, Cantharis spp., N. brevicollis, Linyphiidae) and the three extraguild (Collembola, Lumbricidae, H. rufipes) prey taxa. In October, November and December 76·2% (89·7%) and 5·8% (6·1%) of the Cantharis spp. larvae were caught during the night and day, respectively, while 18·0% (4·2%) of the total catch was obtained in February when traps were emptied every 24 h in the morning only (numbers in parentheses are percentages of predators testing positive for prey DNA for the respective catches). Similarly, 70·4% (55·6%), 4·1% (0·0%) and 25·5% (44·4%) of the N. brevicollis larvae were caught during the night, day and within the 24-h catches in February, respectively. Figure 2 shows an example for the detection of prey DNA in field-collected Cantharis spp. larvae. The nightly mean air temperatures (±SD) were significantly higher during the catching periods in October (10·9 °C ± 1·8) and November (9·9 °C ± 3·7) compared to December (2·8 °C ± 4·8) and February (−0·3 °C ± 4·4) (K–W anova: χ2 = 95·9, d.f. = 3, < 0·001; M–W U-tests: October vs. December U = 87·0, < 0·001; October vs. February U = 4·5, < 0·001; November vs. December U = 275·5, < 0·001; November vs. February U = 116·5, < 0·001).

image

Figure 2.  Detection of prey DNA in field-collected Cantharis spp. larvae using multiplex PCR. PCR products were electrophoresed on 3% agarose gels. Each lane represents a single predator showing PCR products of (1) collembolans, (2) earthworms, (3) collembolans and Harpalus rufipes, (4) collembolans and earthworms, (5) collembolans and H. rufipes, (6) collembolans, earthworms and H. rufipes, (7) collembolans, Calathus fuscipes and earthworms; L, 100-bp DNA ladder.

Download figure to PowerPoint

Extraguild prey constituted the majority of prey detected in cantharid larvae. Significantly more larvae tested positive for collembolan DNA (49·0%; CI: 45·7–52·7%) than for earthworm DNA (34·0%; CI: 30·8–37·5%). In 22·4% of the specimens screened, both collembolan and earthworm prey were detected. Only in five Cantharis spp. larvae was DNA of the carabid H. rufipes detected along with DNA of collembolans (one larva) or earthworms (two larvae), or both springtail and lumbricid DNA (two larvae). One Cantharis spp. larva tested positive for the carabid C. fuscipes and for collembolan and earthworm DNA, but none tested positive for N. brevicollis.

Similar to Cantharis spp. larvae, collembolans were the most frequently detected prey in larval N. brevicollis (collembolan prey: 54·1%; CI: 45·5–60·5%; earthworm DNA 23·5%; CI: 17·4–30·5%). Both collembolan and earthworm DNA were detected in 19·4% of these carabid larvae. Within one N. brevicollis larva besides collembolan DNA also DNA of linyphiids was amplified. In none of the larvae, however, could DNA from Cantharis spp. be detected.

In both carabid and cantharid larvae, extraguild prey detection rates changed between sampling dates (Fig. 3): compared to October (55·1%) and November (51·7%), collembolan prey detection in Cantharis spp. larvae dropped significantly in December (37·7%) but increased again in February to 53·3% (Fig. 2a). Earthworm DNA detection rates decreased monotonically with sampling date from 52·4% in October to 10·4% in February (Fig. 3a). In N. brevicollis significantly fewer larvae contained collembolan DNA in October (26·7%) than in December (69·2%), whereas earthworm DNA detection rates were not significantly different among the four sampling dates (Fig. 3b).

image

Figure 3.  Collembolan and earthworm DNA detection rates in Cantharis spp. (a) and Nebria brevicollis (b) larvae collected in arable land at four time points between 27 October 2006 and 17 February 2007. Sampling period O 27–29 October 2006 (n = 118 and 15 for a and b, respectively), N 17–19 November 2006 (n = 120 and 32 for a and b, respectively), D 9–11 December 2006 (n = 114 and 26 for a and b, respectively), F 15–17 February 2007 (n = 77 and 25 for a and b, respectively). Error bars indicated 84% confidence intervals and letters denote significant differences in DNA detection rates at < 0·05 (Payton, Greenstone & Schenker 2003).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This study presents a trophic network in an autumn- and winter-active invertebrate community where trophic linking has been identified empirically using molecular methods, including the detection frequency of specific prey taxa. Our results show that DNA of collembolans and earthworms was regularly detected in larval Cantharis spp. and N. brevicollis, the most abundant macroinvertebrate predators in the study sites. Trophic interactions to predatory arthropods (intraguild prey) and to the granivorous larvae of H. rufipes, however, were negligible. This suggests that detritivorous organisms such as collembolans and earthworms primarily sustained the autumn- and winter-active predators, whereas intraguild and phytophagous prey were less important. The importance of detritivores as food for generalist predators in arable systems has already been acknowledged in previous work. However, these studies did not focus on the predator community present in autumn and winter and/or the resolution of trophic links was limited (Symondson et al. 2000; Agusti et al. 2003; Traugott 2003; Oelbermann, Langel & Scheu 2008). According to our observations, springtails and earthworms were abundant throughout autumn and winter, providing constant food sources for cantharid and carabid beetle larvae in the cold season. Moreover, collembolan prey has been shown to be of high nutritional value for invertebrate generalist predators (Marcussen, Axelsen & Toft 1999; Bilde, Axelsen & Toft 2000). This might also be true for earthworms, which are a soft-tissued and protein-rich prey (Lawrence & Millar 1945; Sun et al. 1997). While it was already known that Cantharis spp. larvae feed on earthworms (Lukasiewicz 1996; Traugott 2003), it came as a surprise that c. 50% of the screened specimens contained collembolan DNA, albeit they are slow moving predators. Similarly, a considerable proportion of N. brevicollis larvae, which are agile hunters and purportedly specialize in collembolan prey (Penney 1966; Nelemans 1988), were feeding on earthworms, too. Considering, however, that gut-content analysis does not allow discriminating between active predation and scavenging (Foltan et al. 2005; Juen & Traugott 2005) and both cantharid and carabid beetle larvae readily accept carrion food (Penney 1966; Bilde, Axelsen & Toft 2000; Traugott 2003; Juen & Traugott 2005), scavenged prey may have contributed to the diet of these generalist predators.

The identity of both predator and prey can affect prey DNA detection success by PCR (King et al. 2008), complicating comparisons of prey detection rates between different predator and prey taxa. Although recent experiments indicated that prey DNA detection rates do not differ significantly between carabid and cantharid larvae (Juen & Traugott 2006; T. Waldner & M. Traugott, unpublished data), prey detection rates were not statistically compared between cantharid and carabid predators in this study to avoid misleading conclusions. Comparisons between earthworm and collembolan prey DNA detection rates within a predator taxon should be interpreted against the background of post-feeding detection intervals, which might differ for these two prey types. However, effects of factors such as meal size, feeding frequency, consumption of alternative prey, or ambient temperature will affect post-feeding prey DNA detection intervals, making the adjustment of the current field data a formidable task. Moreover, it has to be kept in mind that the current approach does not allow quantifying the number of prey taken or the mass of prey consumed (King et al. 2008). Considering that collembolans represent a much smaller prey than earthworms, it is likely that the latter constituted a larger proportion in the diet of both predators than the current detection rates suggest, a hypothesis which needs to be tested in future work. Differences in primer sensitivity, another factor potentially affecting prey DNA detection success, are unlikely to have significantly biased the current screening results as all fragments of prey DNA amplified were shorter than 300 bp and the detection limits ranged from 0·07 to 0·29 pg target DNA per μL PCR.

In the present data, prey detection rates varied during autumn and winter: while lumbricid DNA was detected in c. 50% of Cantharis spp. larvae collected in October and rates continually decreased thereafter, no temporally significant differences occurred in earthworm prey detection rates in larval N. brevicollis. By contrast, springtails were found to be frequently consumed by cantharid larvae throughout the investigation period (40–50% of larvae testing positive), whereas detection rates of collembolan DNA in N. brevicollis were 30% in October and constantly increased thereafter to c. 70% in December. The latter suggests that these carabid predators were mainly feeding on cold-resistant springtails such as I. anglicana, I. viridis and D. hiemalis, which were abundant in December (data not presented). Note, however, that ambient temperatures were significantly lower for larvae caught in December and February which might also have contributed to the increased collembolan detection rates. The cantharid larvae, on the other hand, were probably less selective in their feeding behaviour and consumed both live and carrion collembolan prey, leading to the observed constant rates of collembolan prey detection throughout the season. The varying feeding periods of these predators also might have affected the temporal patterns in prey detection: for example, sixth instar Cantharis spp. larvae show their main feeding period in October and November (Traugott 2003), probably explaining why detection rates of lumbricid prey – an important food source for larval cantharids (Lukasiewicz 1996; Traugott 2003) – steadily decreased from November onwards.

The multiplex PCR assay used to discern the trophic linkages proved to be specific when tested against the most abundant animals which occurred in the study sites. Food chain errors such as secondary predation (Sheppard et al. 2005) can potentially corrupt gut-content analysis. However, this type of error would have been revealed in our study as the molecular assay included primers for the most abundant intraguild predators. The present results show that secondary predation is unlikely to bias the outcomes as predator DNA, along with lumbricid and collembolan DNA, was detected in only two individuals. In general, the number of predators which contained intraguild prey was very small. Predation events within the dry pitfall traps are also unlikely to have biased the current data: only few invertebrates were caught in these traps which were emptied at a regular basis and never were any remains of prey found in the traps, an indication of predation within beakers.

Although the low ambient temperatures in autumn and winter probably had a positive effect on prey DNA detection rates (von Berg et al. 2008), the weak intraguild trophic links observed here indicate that intraguild predation played only a minor role in shaping this autumn- and winter-active predator community. By contrast, intraguild predation can be extensive within arthropod predator communities in spring and summer in comparable habitats (e.g. Snyder & Ives 2001; Lang 2003; Prasad & Snyder 2004; Rosenheim 2005). Note, however, that intraguild predation does not necessarily interfere with biological control (Colfer et al. 2003; Snyder, Clevenger & Eigenbrode 2004) and that even in summer the levels of intraguild predation can be negligible in some predatory arthropod communities (e.g. Harwood et al. 2007). Perhaps the long-term organic management of the study site, which provided a rich detritivorous invertebrate fauna, dampened intraguild interactions, a phenomenon that has been observed previously (Halaj & Wise 2002).

Currently, knowledge of the ecology of predatory beetle species is largely restricted to adults, although larval mortality is assumed to be the key factor in overall mortality (Kleinwächter & Bürkel 2008). Thus, overwintering field populations of predatory beetle larvae should be taken into account when trying to enhance adult beetle numbers in summer to improve the levels of natural pest control (Noordhuis, Thomas & Goulson 2001; Purvis & Fadl 2002). The strong trophic connection of carabid and cantharid larvae to the detrital food chain found here offers a new strategy for implementing such conservation biological control measures. By improving the habitat conditions for detritivores in arable land (e.g. by mulching, compost applications, or provision of plant cover during winter) their densities can be increased easily (Bell et al. 2008). Applying these measures not just in spring and summer, the main focus of previous studies (e.g. Landis, Wratten & Gurr 2000; Rypstra & Marshall 2005; Bell et al. 2008), but also during autumn and winter will provide the cold-adapted juvenile stages of predatory beetles with an important food source. This should enhance larval survival (Bilde, Axelsen & Toft 2000; Kyneb & Toft 2004) and subsequently lead to larger adult beetle populations. Moreover, as the larval nutritional status affects adult longevity and fecundity (Ernsting, Isaaks & Berg 1992), the species’ fitness might increase as well. Ultimately, we expect that these measures will enhance the ability of polyphagous beetle predators to provide their fundamental ecosystem service as regulators of agricultural pests.

The present results provide conclusive evidence that autumn- and winter-active predatory beetle larvae are strongly linked to the detrital food chain by feeding on earthworm and collembolan prey. By contrast, intraguild trophic interactions between predatory beetle larvae and with linyphiid spiders were weak or absent, indicating that these cold-adapted predators coexist without major predator–predator trophic interactions. Future work should examine how the availability of detritivorous prey affects trophic linkages to both intra- and extraguild prey within food webs of cold-adapted generalist predators. This knowledge will help to assess whether improved larval feeding conditions translate to enhanced levels of pest control.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank Erwin Meyer for discussing this work with us and Erhard Christian, Alexander Rief and Johannes Schied for their help with identification of collembolans, spiders and carabids, respectively. Rudolf and Martha Nagiller kindly allowed us to collect invertebrates within their fields. Thomas Waldner supported DNA-extractions. Corinna Wallinger provided valuable comments which improved the manuscript.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References