Introduction
The positive effect of biodiversity on ecological functions and processes is well documented (Johnson et al., Reference Johnson, Vogt, Clark, Schmitz and Vogt1996; Loreau Reference Loreau2000; Cardinale et al., Reference Cardinale, Harvey, Gross and Ives2003; Hooper et al., Reference Hooper, Chapin, Ewel, Hector, Inchausti, Lavorel, Lawton, Lodge, Loreau, Naeem, Schmid, Setälä, Symstad, Vandermeer and Wardle2005; Straub et al., Reference Straub, Finke and Snyder2008). The general opinion is that more diversified ecosystems sustain higher levels and higher number of ecological functions and services delivered to humans. However, for ecosystem services such as biological control, the effect of biodiversity is frequently ambiguous and highly dependent on the trophic level at which the biodiversity is considered (Montoya et al., Reference Montoya, Rodríguez and Hawkins2003). Indeed, while several studies showed a positive relationship between natural enemies biodiversity and pest biological control (Losey & Denno, Reference Losey and Denno1998; Snyder et al., Reference Snyder, Snyder, Finke and Straub2006; Letourneau et al., Reference Letourneau, Jedlicka, Bothwell and Moreno2009), others have also indicated that species-rich predator assemblages could be more sensitive to competitive interactions or intraguild predation, diverting them from pest predation (Rosenheim, Reference Rosenheim1998; Finke & Denno, Reference Finke and Denno2005; Davey et al., Reference Davey, Vaughan, Andrew King, Bell, Bohan, Bruford, Holland and Symondson2013). For instance, pest predation could be altered by diet shifting resulting from competitive interactions between generalist predators (Tixier et al., Reference Tixier, Dagneaux, Mollot, Vinatier and Duyck2013). Hence, our capacity to anticipate the effect of a change in predator diversity on biocontrol is directly dependent on our capacity to disentangle predator trophic interactions at the scale of the interacting community (Thompson et al., Reference Thompson, Brose, Dunne, Hall, Hladyz, Kitching, Martinez, Rantala, Romanuk, Stouffer and Tylianakis2012; Bohan et al., Reference Bohan, Raybould, Mulder, Woodward, Tamaddoni-Nezhad, Blüthgen, Pocock, Muggleton, Evans, Astegiano, Massol, Loeuille, Petit and Macfadyen2013). However, empirically characterizing complex trophic interactions within a pest–predator food web is challenging, even in presumably simplified ecosystems such as intensively managed farmland areas (Pocock et al., Reference Pocock, Evans and Memmott2012; Mollot et al., Reference Mollot, Duyck, Lefeuvre, Lescourret, Martin, Piry, Canard and Tixier2014). A typical community of service-providing predators frequently comprises dozens of small-sized arthropods with very wide diet spectrum (Agustí et al., Reference Agustí, Shayler, Harwood, Vaughan, Sunderland and Symondson2003; Eitzinger & Traugott, Reference Eitzinger and Traugott2011; Staudacher et al., Reference Staudacher, Jonsson and Traugott2016). Moreover, in response to frequent perturbations caused by management practices, predator–prey interactions could be very dynamic in space and time (Bommarco et al., Reference Bommarco, Firle and Ekbom2007; Bell et al., Reference Bell, Andrew King, Bohan and Symondson2010), requiring a significant capacity to track and integrate these spatio-temporal variations. This is particularly relevant with regard to carabid beetles – a major guild of generalist predators in agricultural landscapes (Kromp, Reference Kromp1999; Kulkarni et al., Reference Kulkarni, Dosdall and Willenborg2015) that could feed upon a wide variety of resources (Larochelle, Reference Larochelle1990) including numerous pests (Bohan et al., Reference Bohan, Boursault, Brooks and Petit2011; Boreau de Roincé et al., Reference Boreau de Roincé, Lavigne, Ricard, Franck, Bouvier, Garcin and Symondson2012) but also beneficial organisms (King et al., Reference King, Vaughan, Bell, Bohan and Symondson2010; Davey et al., Reference Davey, Vaughan, Andrew King, Bell, Bohan, Bruford, Holland and Symondson2013) or carrion (i.e., scavenging, Young, Reference Young1984, Reference Young2005). Additionally, carabids have been shown to dynamically change their feeding behavior in response to variations in prey abundance (Bohan et al., Reference Bohan, Bohan, Glen, Symondson, Wiltshire and Hughes2000; Bell et al., Reference Bell, Andrew King, Bohan and Symondson2010). Hence, carabid contribution to biological control is yet hardly predictable, rendering the implementation of concrete management actions difficult. In this paper, we tackle the long-standing problem about the contribution individual species make to biological control within the carabid community. Previous studies have showed that carabid species could be lumped into broad trophic categories according to their feeding preferences (Larochelle, Reference Larochelle1990) or according to functional traits, such as the mandible morphology or body size (Forsythe, Reference Forsythe1983; Rouabah et al., Reference Rouabah, Lasserre-Joulin, Amiaud and Plantureux2014; Rusch et al., Reference Rusch, Birkhofer, Bommarco, Smith and Ekbom2015). However, morphological trait-based categories do not provide a quantitative estimate of diet, also the link between functional attributes and feeding behavior is not always consistent, especially in such flexible feeders as carabids. Recently, molecular techniques were very successful in allowing the direct quantification of trophic links from predators’ gut contents (Symondson, Reference Symondson2002; Clare, Reference Clare2014). However, the main limitation with the molecular diet analysis is that it only provides a snapshot of a species diet – usually the most recent feeding event, thus making the incorporation of spatio-temporal variations in feeding choice of multiple species resource-demanding. Finally, both functional and molecular approaches fail to distinguish active predation from scavenging (Foltan et al., Reference Foltan, Sheppard, Konvicka and Symondson2005; von Berg et al., Reference Von Berg, Traugott and Scheu2012). The analysis of naturally occurring carbon and nitrogen stable isotope ratios (expressed as δ, the ratio of heavy to light isotope, and reported in parts per thousand as per mil) could help overcome some of the above-mentioned methodological challenges.
Stable isotopes provide quantitative and time-integrative estimate about the trophic space occupied by an organism based on the dependency of isotopic signatures observed in consumers on their dietary resources (DeNiro & Epstein (Reference DeNiro and Epstein1978, Reference DeNiro and Epstein1981)). For instance, measuring the stable isotope ratios of carbon (δ13C) allows pinpointing the actual food source among a pool of potential ones, provided that their ratios differ (Gannes et al., Reference Gannes, Martínez del Rio and Koch1998). In contrast, for elements such as nitrogen (δ15N), a predictable enrichment of the heavier isotope from diet to consumer is observed (Martínez del Rio et al., Reference Martínez del Rio, Wolf, Carleton and Gannes2009), leading to identify the trophic level of an organism relative to a reference baseline (Eggers & Jones, Reference Eggers and Jones2000). Considering these advantages, we applied δ13C and δ15N stable isotope analysis in order to estimate the degree of involvement in biological control of the most common carabid species in two dominant crop types. We measured the mean and the variance of the δ13C and δ15N trophic space occupied by each species and compared it with the baseline crop plant. The crop plant is an important agronomical factor structuring arthropod assemblages through the frequency and the intensity of the different management practices (Marrec et al., Reference Marrec, Badenhausser, Bretagnolle, Börger, Roncoroni, Guillon and Gauffre2015; Puech et al., Reference Puech, Poggi, Baudry and Aviron2015) but is also an important resource at the very basis of farmland food webs (Tixier et al., Reference Tixier, Dagneaux, Mollot, Vinatier and Duyck2013; Mollot et al., Reference Mollot, Duyck, Lefeuvre, Lescourret, Martin, Piry, Canard and Tixier2014). Our objectives were twofold: (i) identify variations in the trophic position across species and between crop types for the same carabid species; (ii) assess whether variations in the crop plant baseline influence variations in carabid isotopic signatures. Our expectations were that (i) we would be able to identify distinct trophic positions – through strict herbivores to strict carnivores – within the carabid community; (ii) that variations in isotope ratios of the crop baseline will cascade at upper trophic levels, thus allowing the distinction between crop-derived and non-crop-derived trophic groups.
Material and methods
Collection and sample preparation
Carabid beetles were intensively sampled between 03 and 23 May 2012, period corresponding to the peak of carabid activity in our study area this year (Supplementary fig. S1). Sampling took place in two arable fields (winter wheat and oilseed rape, size 1–5 ha each), situated approximately 10 km apart within the long-term ecological research area ‘Armorique’ (http://osur.univ-rennes1.fr/za-armorique/, 48°36′N, 1°32′W), Brittany, France. For maximizing the number of species collected, each field was checkered with a high number (>50) of dry pitfall traps. All traps were opened and closed in four sampling sessions of 24 h each. Carabid beetles were collected alive. Living individuals were freeze-killed at −20 °C as soon as possible (and in all cases less than 5 h after collection), sorted out at the laboratory and identified to the species level (Roger et al., Reference Roger, Jambon and Bouger2012). For assessing the crop baseline, fresh plant tissues (stems, leaves, pods and ears) were sampled from randomly selected plants for each crop type: oilseed rape (Brassica napus, Brassicaceæ) and wheat (Triticum aestivum, Poaceæ). Plant tissues per crop were mixed together, frozen at −20 °C within 5 h after collection and stored at −20 °C prior analyses (N = 5 for wheat and N = 8 for oilseed rape). Both carabid and plant samples were freeze-dried for 24 h. To avoid bias induced by the presence of prospective prey within carabid gut contents, all species sizing >4.5 mm (90% of species) were dissected and gut contents were removed. Because of their small size, dissection was impractical for species smaller than 4.5 mm. These specimens were analyzed in their entirety. All freeze-dried samples were manually ground into fine powder. Tin capsules containing between 1 and 2 mg of tissue of each individual were processed with isotope-ratio mass spectrometer (Delta Plus, Thermo Quest, Waltham, MA, USA), coupled to an elementary analyzer (Flash EA, Thermo Quest, Waltham, MA, USA) at the Roscoff Biological Station, Brittany, France. The stable isotopic composition of carbon and nitrogen (δ13C and δ15N) was expressed as a relative ratio, in parts per thousand, to an international standard:
$$\delta X{\rm} = {\rm} \left[ {(R_{{\rm sample}} /R_{{\rm standard}} ){\rm} - {\rm} 1} \right]1000,$$
where R sample is the absolute isotopic ratio (heavy/light) of the sample and R standard is the correspondent ratio in the standard (Peterson & Fry, Reference Peterson and Fry1987; Ehleringer & Rundel, Reference Ehleringer, Rundel, Rundel, Ehleringer and Nagy1988). The international standards used were Vienna Pee Dee Belemnite for δ13C and atmospheric nitrogen for δ15N. Measurement uncertainty was ±0.2‰ for δ15N and ±0.1‰ for δ13C.
Statistical analyses
Statistical analyses were run with the R software version 3.1.0 (R Core Team, 2013). Variations in the δ13C and δ15N trophic position among carabid species and between crop types as well as their interaction were compared by fitting general linear models (GLM, Gaussian distribution family, link = identity) using the R function glm. The number of individuals per species and per field ranged between N = 1 and N = 28 (mode N = 2 and N = 4). Carabid species represented by less than two individuals in the whole dataset (seven species) were not taken into account in statistical analyses. Their individual δ13C and δ15N values are provided in Supplementary table S1.
The effect of variations in the crop plant baseline on differences in the carabid δ13C between crop types were also investigated using a GLM model (Gaussian distribution family, link = identity), including as variables the carabid species and the interaction carabid species × crop field. Hence, the interaction carabid species × crop field was an estimate of δ13C variation between the two crop fields within the considered carabid species. The crop was also treated like a carabid species in this model. Thus, the interaction crop × field was an estimate of the δ13C variation between the two crop species as crop baselines consist in wheat in one field and in oilseed rape in the other. Then, for each pair of carabid species, t-tests were performed to assess whether the interaction carabid species × field differed significantly. Similarly, t-tests were also performed to assess whether the δ13C variation between the two fields for each carabid species differed significantly from the δ13C variation between the two crop types (interaction crop × field). Only the most abundant carabid species occurring in the both crop types were selected for these analyses (11 species, mode N = 10).
Results
A total of 295 individuals belonging to 45 species and 26 different genera were analyzed (table 1). Of these 28 species were common in both cultures. Overall, the amplitude of δ13C variation was higher in wheat (fig. 1a), whereas higher amplitude of δ15N variation was observed in oilseed rape (fig. 1b). Average δ13C values ranged from −14.3‰ (Harpalus rubripes) to −29.3‰ (Amara familiaris) in wheat (fig. 1a), and from −24.4‰ (Pterostichus melanarius) to −29.8‰ (Leistus fulvibarbis) in oilseed rape (fig. 1b). Average δ15N values ranged from 4.4‰ (Syntomus obscuroguttatus) to 9.6‰ (Pterostichus vernalis) in wheat (fig. 1a), and from 2.2‰ (Amara plebeja) to 12‰ (P. melanarius) in oilseed rape (fig. 1b). Based on GLM analysis, the carabid species, the crop type and their interaction explained a significant part of the variation in δ13C and in δ15N (table 2). The same results were observed when δ13C and δ15N for only the most abundant species, present in the both crop types, were considered (Supplementary table S2). GLM analysis of the interaction carabid species × field carried out on the species caught in both fields revealed two groups of species (fig. 2) depending on whether their δ13C variation between fields differed significantly from the δ13C variation between the two crops or not. For each species, isotopic signature was considered significantly different from plant isotopic signature for a probability threshold of 0.05. The first group included Nebria salina (P = 0.83), Anchomenus dorsalis (P = 0.54), Poecilus cupreus (P = 0.43) and Anisodactylus binotatus (P = 0.50). The second group included Amara aenea (P = 0.018), Brachinus sclopeta (P = 0.042), Loricera pilicornis (P = 0.021) and Ocydromus tetracolus (P = 0.0018). Asaphidion flavipes (P = 0.079) was also included in the second group because its isotopic signature was almost significantly different from the isotopic signature of the crop plant as well as almost significantly different from the isotopic signatures of all the species belonging to the first group. All species belonging to group 1 did not differ significantly from each other in their interaction with the field (fig. 2), neither from the interaction crop × field (fig. 2). In the group 2, species did not differ significantly from each other in their interaction with the field (with an interaction carabid species × field not significantly different from 0), but the interaction carabid species × field was significantly different from the interaction crop × field (fig. 2, Supplementary table S3). Finally, the interaction with the field of species belonging to the group 1 was frequently significantly different from the interaction with the field of species belonging to the group 2, and when it was not, the difference was always close to the significance (Supplementary table S3). Two species exhibited intermediate behavior (Agonum muelleri and Amara similata), probably due to the relatively low sampling size (respectively, only four and one individuals in either of the two crop types). However, A. muelleri tended to belong to group 1 and A. similata to group 2 (Supplementary table S3).
Comparison of δ13C and δ15N stable isotope ratios of carabid beetles in (a) wheat; (b) oilseed rape. Means ± SD are shown. Species codes are indicated in table 1.
Boxplot comparing the δ13C values of the crop baseline and the 11 most common carabid species in wheat (black) and oilseed rape (gray). Carabid species has been arranged according to whether their δ13C variation between the fields differed significantly (group 2) or not (group 1) from the δ13C variation between the two crop types. The species Agonum muëlleri and Amara similata exhibited intermediate behavior. For P values see Supplementary table S3. Species codes are indicated in table 1.
Carabid species (and their taxonomic affiliation) analyzed for δ13C and δ15N isotope ratios. Codes as presented in fig. 1.
Underlined species were present in both crop types.
1 Indicates species present only in oilseed rape (n = 11).
2 Indicates species present only in wheat (n = 6).
Results of GLM analyses testing the effect of the carabid species, the crop type, and their interaction on δ13C and in δ15N variation.
Discussion
Overall, the species identity explained a significant part of the variation in the carabid δ13C and δ15N ratios. Despite these differences, we were not able to identify clear trophic groups within the carabid community based on the species δ13C and δ15N values. This is mainly due to the high intraspecific variation, especially for δ15N. This suggests that carabid beetles are indeed generalist, plastic foragers, particularly in agricultural areas (Lövei & Sunderland, Reference Lövei and Sunderland1996; Bennett & Hobson, Reference Bennett and Hobson2009; Okuzaki et al., Reference Okuzaki, Tayasu, Okuda and Sota2010; Kamenova et al., Reference Kamenova, Tougéron, Cateine, Marie and Plantegenest2015), where such trophic generalism could be easily explained by important variations in resource availability (Bohan et al., Reference Bohan, Bohan, Glen, Symondson, Wiltshire and Hughes2000; Bell et al., Reference Bell, Andrew King, Bohan and Symondson2010). However, similar levels of intraspecific variability in isotopic values have also been reported within carabid communities residing in more stable habitats (Zalewski et al., Reference Zalewski, Dudek, Tiunov, Godeau, Okuzaki, Ikeda, Sienkiewicz and Ulrich2014), suggesting that other mechanisms could be in play. For instance, numerous non-trophic sources of variation could also explain the high intraspecific variability in δ13C and δ15N signatures of a consumer. Factors such as individual differences in metabolic rates, fasting time, or diet quality have all been shown to influence isotopic values without any direct link to differences in trophic choice (reviewed by Martínez del Rio et al., Reference Martínez del Rio, Wolf, Carleton and Gannes2009). Carabid beetles typically could experience extended fasting periods (Bilde & Toft, Reference Bilde and Toft1998; Laparie et al., Reference Laparie, Larvor, Frenot and Renault2012) and the proportion of individuals displaying an empty gut within a population could be exceptionally high (Sunderland, Reference Sunderland1975; Hengeveld, Reference Hengeveld1980), which in turn may significantly impact the isotopic fractionation. However, little is known about the specific factors affecting isotopic fractionation in insects in general, particularly in carabid beetles. Hence, identifying the most important sources of non-dietary variation and the magnitude of their effect is an important requirement for taking full advantage of the stable isotope analysis for this group of organisms. Another methodological constraint hampering the interpretation of stable isotope data in our case comes from the lack of information about the time lag in isotopic turnover associated with ontogenetic niche shifts (i.e., changes in the carabid isotopic signature between larval and adult stages). Compared with the adult stages, carabid larvae usually exhibit distinct or more specialized trophic habits (Löveï & Sunderland, Reference Lövei and Sunderland1996), suggesting that a delayed response in isotopic turnover during the dietary shift after metamorphosis will result in an isotopic signature that does not match the actual adult's diet. Moreover, for some carabid species, two or more generations that are not distinguishable morphologically could co-occur at the same season (Thiele, Reference Thiele1977). This suggests that intraspecific variations in isotopic signature would have more to do with time since metamorphosis than with feeding habits. This point requires further consideration within the dynamic agricultural landscapes where spatio-temporal turnover of crops is high (Holland et al., Reference Holland, Thomas, Birkett, Southway and Oaten2005; Fahrig et al., Reference Fahrig, Baudry, Brotons, Burel, Crist, Fuller, Sirami, Siriwardena and Martin2010). For instance, an individual emerging in the late season on a previous year may have experienced a very different basal resource, and still exhibit the isotopic signature of its previous diet at the moment of the field sampling. The same reasoning holds true for dispersal. One plausible scenario is that overwintering carabid beetles may migrate to other fields after emergence, while keeping the isotopic imprint of their native location (Girard et al., Reference Girard, Baril, Mineau and Fahrig2011). Indeed, we cannot guarantee that no migrants from the nearby fields were present at the time of sampling, but we believe that such scenarios are highly unlikely in our case. First, it takes about 17 days for a mid-sized adult carabid to reach the isotopic level of its prey after overwintering (Makarov et al., Reference Makarov, Matalin, Goncharov and Tiunov2013). Second, beetles were collected in the middle of the reproductive season, when carabids usually tend to stay within the crop they select after emergence (Marrec et al., Reference Marrec, Badenhausser, Bretagnolle, Börger, Roncoroni, Guillon and Gauffre2015).
Yet, according to our dataset, some species occupied well-differentiated δ13C and δ15N isotopic niches. Interestingly, differences were more marked for carbon compared with nitrogen, possibly indicating that resource partitioning is mostly related to the identity of the basal resource rather than to a differentiation in the trophic position. This confirms results from at least three independent studies, suggesting that the primary resource uptake could be the prevalent axis of trophic differentiation between carabid species (as opposite to the δ15N trophic position) (Ikeda et al., Reference Ikeda, Kubota, Kagawa and Sota2010; Okuzaki et al., Reference Okuzaki, Tayasu, Okuda and Sota2010; Zalewski et al., Reference Zalewski, Dudek, Tiunov, Godeau, Okuzaki, Ikeda, Sienkiewicz and Ulrich2014). Here, Harpalus species were more carbon-enriched compared with all other species, even if enrichment was less pronounced in oilseed rape compared with wheat (fig. 1). In laboratory conditions, Harpalus species usually show marked granivorous preferences (Johnson & Cameron, Reference Johnson and Cameron1969; Forsythe, Reference Forsythe1982; Goldschmidt & Toft, Reference Goldschmidt and Toft1997) as well as a capacity to consume a large variety of seed species (Honek et al., Reference Honek, Martinkova and Jarosik2003, Reference Honek, Martinkova, Saska and Pekar2007; Wallinger et al., Reference Wallinger, Sint, Baier, Schmid, Mayer and Traugot2015). Thus, the carbon enrichment (compared with the crop baseline) as well as the relatively low δ15N trophic position might suggest that in arable fields, Harpalus species not only consume plant material but also that this plant material probably originates from weed plants rather than the crop. Another genus of supposedly granivorous species, Amara, showed a rather intriguing pattern in δ15N signature. Most species occupied intermediate-to-high δ15N tropic positions in both wheat and oilseed rape, while only two species in oilseed rape showed a trophic position compatible with phytophagous behavior (A. plebeja and Amara lunicollis). The explanation about this pattern is not evident, but it will be interesting to investigate at what extend observed differences in isotopic signatures among Amara species could be linked to various degrees of specialization in seed consumption within the genus (Saska, Reference Saska2005; Honek et al., Reference Honek, Martinkova, Saska and Pekar2007).
From applied perspective, our results show that solely based on the mean and the variance of δ13C and δ15N signatures, it seems challenging to consistently infer carabids’ potential to contribute to biological control. Nevertheless, when comparing variations in carbon signatures between carabid species and the crop baseline, we identified two distinct groups. The first group varied in a similar pattern to the variations in δ13C observed between the two crop plants. The carabid species forming this group – N. salina, A. dorsalis, P. cupreus and A. binotatus – are all supposedly mid- to large-sized carnivorous or omnivorous beetles, usually indicated as feeding upon a large variety of insects, comprising aphids and other pests (Sunderland, Reference Sunderland1975). The second group of species had their δ13C ratios varying independently from the variations in the crop. These species include A. aenea, L. pilicornis, O. tetracolus, and A. flavipes, which are all small- to mid-sized species, mainly phytophagous or carnivores, specialized in the consumption of small detritivores like Collembola (Sunderland, Reference Sunderland1975; Hintzpeter & Bauer, Reference Hintzpeter and Bauer1986). The carabid B. sclopeta, which also belongs to this group of species, is most likely an ectoparasitoid of Amara species (Saska & Honek, Reference Saska and Honek2004, Reference Saska and Honek2008). This apparent trophic partitioning mediated by the baseline resource allows distinguishing between crop-derived and non-crop-derived trophic groups of species, and thus confirms our expectations. Consequently, based on our results and on previous information about carabid trophic behavior, we were able to indirectly assess the likely contribution of these generalist predators to biological control. For example, large carnivorous or omnivorous carabids most seemingly rely on the crop plant for their carbon uptake suggesting the consumption of animal species directly associated with the crop (i.e., herbivorous pests). On the other hand, small granivorous or detritivore feeders probably rely on alternative carbon resources with no direct link to the crop. The latter might still be contributing to biocontrol of weeds (Girard et al., Reference Girard, Baril, Mineau and Fahrig2011), as weed plants constitute another important and diversified source of carbon at the basis of agricultural food webs (Marshall et al., Reference Marshall, Brown, Boatman, Lutman, Squire and Ward2003; Tixier et al., Reference Tixier, Dagneaux, Mollot, Vinatier and Duyck2013; Mollot et al., Reference Mollot, Duyck, Lefeuvre, Lescourret, Martin, Piry, Canard and Tixier2014).
Overall, the stable isotope analysis we applied in this study shows to be practical, rapid, and relatively cheap method for directly assigning carabid species into loose trophic categories. However, in order to take more advantage of such stable isotope data, it would be beneficial to diversify the tools and the methods. For instance, more quantitative tools such as mixing models or functional metrics (Layman et al., Reference Layman, Arrington, Montaña and Post2007; Parnell et al., Reference Parnell, Phillips, Bearhop, Semmens, Ward, Moore, Jackson, Grey, Kelly and Inger2013; Cucherousset & Villéger, Reference Cucherousset and Villéger2015), or methods for more direct assessment of diet (e.g., DNA metabarcoding, Pompanon et al., Reference Pompanon, Deagle, Symondson, Brown, Jarman and Taberlet2012; Vacher et al., Reference Vacher, Tamaddoni-Nezhad, Kamenova, Peyrard, Moalic, Sabbadin, Schwaller, Chiquet, Smith, Vallance, Fievet, Jakuschkin, Bohan, Woodward and Bohan2016) could help characterizing with more precision the carabid contribution to biological control (Boreau de Roincé et al., Reference Boreau de Roincé, Lavigne, Ricard, Franck, Bouvier, Garcin and Symondson2012).
Yet, carabid beetles are also notorious scavengers, but DNA techniques cannot tease apart active predation from scavenging (Foltan et al., Reference Foltan, Sheppard, Konvicka and Symondson2005; Juen & Traugott, Reference Juen and Traugott2005; Heidemann et al., Reference Heidemann, Scheu, Ruess and Maraun2011; von Berg et al., Reference Von Berg, Traugott and Scheu2012). Thus, the question about the importance of carrion in carabid diet remains open. Stable isotope analysis could possibly inform about scavenging, but it is not known yet how the consumption of carrion impacts the δ13C and δ15N signatures as well as the trophic groups we observe here. In order to elucidate if scavengers could be clearly delimitated based on their isotopic signature, it would be interesting to experimentally feed carabid beetles with fresh or decayed prey (Wallinger et al., Reference Wallinger, Staudacher, Schallhart, Peter, Dresch, Juen and Traugott2013), and compare their δ13C and δ15N signatures.
In conclusion, stable isotope analysis appears as a straightforward, relatively cheap, and complementary tool that could be used to assess species trophic behavior and possibly provide a mean for the functional categorization of carabid beetles based on their diet. Although our study does not provide direct evidence about carabid contribution to biological control, we show that the isotopic signal can inform us about the juxtaposition in arable fields of two independent and functionally complementary carabid trophic groups, comprising each several functionally redundant species. Based on these results, we argue that maintaining high levels of species richness within farmland carabid assemblages appears as an important prerequisite for preserving the integrity of ecological functions that could be important to humans (Jonason et al., Reference Jonason, Smith, Bengtsson and Birkhofer2013; Trichard et al., Reference Trichard, Alignier, Biju-Duval and Petit2013; Peralta et al., Reference Peralta, Frost, Rand, Didham and Tylianakis2014).
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0007485317000542.
Acknowledgements
This study was a part of a wider project and the authors greatly thank all persons that voluntereed for the fieldwork: Frédéric Hamelin, Nolwenn Génuit, Maël Dugué, Kévin Tougeron, Théo Vantsteenkeste, Frédérique Mahéo, Lucie Mieuzet, Nathalie Leterme, Jean-François Le Gallic, Bernard Chaubet and Sarah Polin. The authors also thank the LTER area ‘Armorique’ for providing infrastructures as well as the network of farmers that kindly allowed them to sample in their fields. Finally, the authors thank the two anonymous referees for their insightful suggestions on the first version of the manuscript that the study was also funded by the ‘Peerless’ project ANR-12-AGRO-0006. This study was funded by the French National Research Agency through the ‘Landscaphid’ project ANR-09-STRA-05. Stefaniya Kamenova was partially funded by the Région Poitou-Charentes (France) during her PhD thesis.
Author contribution
S.K. and M.P. designed the study, collected and analyzed the data. C.L. ran the stable isotope analysis. S.E.P. helped with data collection and statistical analyses. S.K. wrote the paper and M.P. provided comments.
