1. Introduction
Ground beetles (Coleoptera: Carabidae) are one of the most common arthropods in agroecosystems and their ability to aid in biological pest control has been demonstrated for several groups of pests including slugs [
1,
2,
3,
4,
5,
6] and caterpillars [
7,
8]. However, most carabids are generalist predators and their diet can comprise a variety of invertebrates including insects, slugs or earthworms, as well as seeds [
9]. Feeding experiments conducted under limited or no-choice conditions in the laboratory can hence be problematic, as the carabids might feed on select pests that they would not or only sporadically eat in the field, where there is a greater availability of alternative prey. A more accurate way to assess actual pest predation is molecular gut content analysis of field-collected carabids using either standard polymerase chain reaction (PCR) or quantitative PCR (qPCR), which has been shown to be even more effective in detecting DNA traces of prey items in predator guts [
10]. This approach has been used successfully in many studies investigating predation by carabids [
11,
12,
13] and also has the added benefit of being able to determine the timing of feeding incidences. This is of particular interest as pest control by generalist predators is usually most effective when the pest in question is just becoming active and the predator to prey ratio is highest [
14,
15]. The natural enemies are thus able to delay, dampen or even prevent exponential pest population growth in the crop [
16,
17,
18] while they are unlikely to significantly impact pest populations and prevent crop damage once the pests have become established [
19]. Despite evidence of the opportunistic feeding behavior of certain species of carabids [
20,
21,
22], it has been shown that some pests are consumed even if their numbers are low [
23,
24,
25,
26]. A prerequisite for pest control potential is the sufficient abundance of generalist predators in the fields at critical times, and to increase their effectiveness, habitat favorability should also be considered [
17].
Carabid assemblages in agricultural fields are not related to crop type but rather to the timing of cultivation measures and changes in phenology and resulting microclimate specific to the crop [
27]. In general, soil cultivation, such as tilling, disturbs the vertical and horizontal distribution of soil biota and, depending on the type of cultivation, buries all or some of the crop residue [
28] which can serve as a habitat for pests and beneficials alike [
29,
30]. Different studies show different impacts of soil cultivation on carabid abundance [
31,
32,
33,
34,
35], depending, for example, on depth or timing of cultivation as well as on carabid species. Carabid beetle assemblages are also affected by size and isolation of agricultural patches and the presence of permanent landscape elements such as natural woodlands or field margins [
36,
37,
38,
39,
40,
41,
42]. These can serve as important refuge habitats and overwintering sites for a variety of beneficial arthropods [
43,
44] and may contribute to enhanced pest regulation in neighboring fields [
45,
46].
While much research has been carried out on the role of carabid beetles as predators in agroecosystems, this is mostly restricted to Europe and there is less knowledge of carabid assemblages and their pest control potential from cropping systems in the United States (US). Annual ryegrass
Lolium multiflorum L. (AR) is a major cash crop in Western Oregon, where it is grown for seed. It is seeded in late summer or autumn around the first rain of the wet season and grows well in poorly drained soils, which are not suited for perennial grasses. The crop is most susceptible to damage in autumn and early winter when it is in the seedling stage (i.e., before tillering occurs), during which time a number of common pest species including slugs (Gastropoda), cutworms (Lepidoptera: Noctuidae) and cranefly larvae (Diptera: Tipulidae) start to become active [
47]. Slugs thrive in the mild and wet climate of Western Oregon and account for ~
$60 million damage to the state’s
$500 million grass seed industry annually [
48]. Here, slugs are most active between September and June, when humidity and soil moisture are high and they aestivate in the ground or beneath residue during the hot dry summer [
49]. Mating and egg laying by the most common species
Deroceras reticulatum (Müller) can occur in autumn and spring [
50], leading to a large proportion of neonate and juvenile slugs in spring and early summer. Slugs are usually nocturnal but can be found feeding by day during fog or rain and damage to crops can be especially extensive around weedy and grassy field margins which serve as critical habitat [
50]. Several cutworm species can pose a threat to grass seed crops, including glassy cutworm
Cymodes devastator (Bruce), black cutworm
Agrotis ipsilon (Hufnagel) and winter cutworm
Noctua pronuba Linnaeus [
47]. The latter species, in particular, is of concern as it is cold tolerant and can feed on crops throughout the winter when temperatures are above freezing [
47]; its largely night-active larvae are present in the soil between September and March [
51]. As this species was only discovered in Oregon in 2001, little is known about the impact of natural enemies. The larvae of the European cranefly
Tipula paludosa Meigen and Common cranefly
Tipula oleracea L. can be minor pests in non-irrigated autumn sown grasses such as AR and their activity increases usually after heavy rainfall, with stand loss occurring generally when plants are already weakened or stressed [
47]. The present study takes a multi-step approach to assess, for the first time, the biological control potential of carabid beetles for these pests in AR in Western Oregon.
The objectives of this study were: (i) to identify the most commonly occurring carabid species in this system, and to investigate their spatial and temporal overlap with those of the key pests in AR; (ii) to quantify the predatory capacity of the selected carabid species on the key pests using qPCR-based prey identification of their gut contents; and (iii) to investigate the influence of field margin structure and disk tillage on the activity abundance of the common carabid species.
4. Discussion
The three pest groups that were monitored for this study were selected as they are active between autumn and spring and are known to cause serious damage to the emerging AR crop [
47,
49]. Thus, the identification of natural enemies that are active during the same time period is an important consideration for biological control. Of the common carabid species in AR, only
N. brevicollis was found in large numbers during the early parts of this critical period, when it emerges from summer diapause to mate and lay eggs [
66]. Abundance of this nocturnal carabid, which overwinters mainly in its larval stage [
67], decreased in November but activity of
N. brevicollis briefly overlapped with that of slugs and
N. pronuba in late October/early November (
Figure 1A,E,F).
Nebria brevicollis larvae were found between December and April, with the highest activity between March and April followed by the emergence of the first tenerals around the same time, starting the spring activity phase of this species. Other common carabids in the fields also started to become active in late March/early April (
A. muelleri,
Figure 1D) or late April/early May (
P. laetulus,
Figure 2C), while
C. cancellatum emerged in May (
Figure 1B). Even if these three carabid species are not active during the time critical for crop damage in AR, slugs and their eggs and, to a lesser degree, caterpillars and cranefly larvae were still present in the fields and predation on these pests might still occur. This can lessen pest pressure the following autumn either indirectly by reducing the number of caterpillars and cranefly larvae metamorphosing into egg laying adults or directly by feeding on slug eggs and juvenile slugs that would otherwise have fed on emerging AR later in the year. The only distinct spatial overlap of carabid and pest abundances was between
N. brevicollis and slugs, which were most abundant in the sites with vegetation margins.
Unlike with quadrats or soil sampling, the refuge traps we used in our study do not provide data on the actual density but rather the activity of the monitored organisms. This is dependent on conditions such as temperature, humidity or soil moisture (e.g., [
68,
69]), which can have different effects on activity depending on vegetation cover in the habitat [
70,
71]. While refuge traps were also reported to overestimate the abundance of surface-active species such as
D. reticulatum [
72] and underestimate juveniles [
73] in comparison to soil sampling, Clements and Murray [
74] found that ‘wadding square traps’, which are most similar to the refuge traps used in this study, caught more juveniles than saucer traps. The primary advantage of refuge traps over quadrats or soil samples is the comparably low sampling effort, which is crucial for large-scale experiments. Hence, they are frequently used to monitor slugs (e.g., [
75,
76,
77]), but to our knowledge have not been used previously for cranefly larvae and caterpillars. As we were interested mainly in the activity periods of the pests and the differences in relative pest numbers between the sites, refuge traps were the ideal choice for this study.
Nebria brevicollis was the only common carabid in this study whose distribution was significantly associated with a vegetation margin. This species has traditionally been considered a woodland species associated with litter of deciduous trees and high soil moisture [
67]. However, more recent publications report it to be rather eurytopic and abundant in a range of different habitats including agricultural grasslands and even urban areas [
78,
79]. While it occurred in all sites during this study, it was most abundant in those with a vegetation margin. It is likely that the vegetation field margins provide suitable habitat and refuge during harvest and post-harvest operations, which, in AR, also correspond with the timing of summer diapause and subsequent mating in
N. brevicollis. However, its high mobility both as adult and larva [
80] make it an effective colonizer of nearby fields that might offer suboptimal conditions for breeding or summer diapause. While the model containing the effect ‘field margin’ was also a significantly better fit than the null model for
P. laetulus and
A. muelleri, their activity densities in the different fields indicate an association with a gravel rather than a vegetation margin (
Figure 2C,D).
Agonum muelleri is described by some authors as a species that prefers open and dry habitats [
81], while other authors classify this species as ‘hygrophilous’ [
82] and list damp grasslands or even open woodlands as habitats [
79]. No other studies have investigated the habitat preferences of
P. laetulus to date and further work is needed to determine the ecological requirements of this species. ‘Field margin’ or ‘soil moisture’ did not explain the site preference of
C. cancellatum, which is a species that is associated with open and dry country [
54]. However, other soil characteristics such as grain size, pH and organic matter content (not measured in this study) could be important for this species as it was completely absent from site C3 (
Figure 2B), the only site with slightly gravelly soil. In general, none of the associations with ‘field margin’ were strong (R
2 < 0.21;
Table 3) for any of the common carabid species, indicating that other factors not measured in this study are likely of greater importance.
Tilling can influence carabid beetles either directly by killing adults, larvae or pupae or indirectly, by altering biotic and abiotic habitat characteristics making the environment more or less suitable for different species [
27,
28]. None of the common carabid species in this study showed a significant response to disk tilling in August following the BACI analysis (
Table 4), which is comparable with the findings of other studies. Holland and Reynolds [
83] determined that tilling in October or February had no effect on the abundance of emerging
N. brevicollis in fields of wheat stubble and undersown grass, and other studies even indicate that tilling might have a favorable impact on the species (summarized in [
84]). Hatten et al. [
85] state that, while
C. cancellatum shows no consistent response to tillage, trap catches of this species were higher in conventional as opposed to no-till fields of peas, wheat and winter wheat.
Agonum muelleri activity has been shown to be reduced by moldboard ploughing in July in a previous timothy/red clover field [
86], but in Gareau et al. [
87] it was significantly more abundant in full October tillage compared to reduced October tillage in cereal rye. While no studies have been carried out involving
P. laetulus, related species
P. scitulus was more abundant in conventional tillage systems compared to no till in spring wheat [
88] and peas, wheat and winter wheat [
85].
The large decreases in
N. brevicollis numbers in sites E4 and E5 (
Figure 3A) might be due to other effects: soil cultivation in this study took place at a time when
N. brevicollis is in summer diapause, which is likely taking place in the field margin—Kromp [
27] refers to an unpublished manuscript, that mentions
N. brevicollis moving to the permanent boundary strip and woodland edge of the field during soil cultivation and seedbed preparation. If tilling had a direct mortality effect, it should have impacted the number of specimens found in autumn and subsequently also the number of larvae. However, activity density of the autumn 2018 population of
N. brevicollis was high in site E4 and the number of larvae trapped the following spring was similar to those of the previous year. It is rather more likely that the larvae, all of which were caught in the edge traps in site E4, were pupating in or close to the field margins and died when the adjacent river burst its banks in early April 2019. Only one larva was caught at site E5 in spring 2018, followed by many adults. The autumn captures for
N. brevicollis at this site were low, only two specimens were trapped in late October followed again by only a few larvae caught in winter 2018/spring 2019. This indicates that the species primarily inhabits the field margin at this site and only sporadically ventures into the field. This is supported by the finding that, unlike in site E4 where
N. brevicollis was also found in large numbers in the field traps, nearly 90% of the captures at site E5 were from the edge traps. Why significantly fewer adults were captured at this site in the second year is unclear, but conditions might have been more suitable in the adjacent field margin and garden than within the field.
The majority of feeding incidences on gastropods as established by qPCR was accounted for by
N. brevicollis, which is plausible given the temporal and spatial overlap between this beetle species and slugs. However, only one specimen tested positive for gastropod DNA during pest emergence in the autumn when slugs would pose the greatest problem. The remainder of the screened beetles that tested positive for Gastropoda were collected in spring/summer, possibly due to the higher abundance of smaller slugs at this time which are much easier to consume than larger adults [
3] or due to a lack of more palatable prey. Feeding incidences on slugs were also highest in the two fields where these gastropods were most abundant and in the year in which they were present in especially high numbers, indicating opportunistic feeding events including scavenging rather than targeted predation. The overall feeding incidence was just under 10%, compared with 37% of
N. brevicollis testing positive for slug DNA from a study in the United Kingdom [
89], but according to the farmers of our study fields, pest pressure by slugs was much lower in the two years of the study compared to previous years, likely owing to the unusually warm and dry autumn weather in 2017 and 2018. Thus, slug numbers in the fields and associated feeding incidences by carabids might generally be higher than those recorded during the current study. Lower feeding rates for gastropods were observed for
P. laetulus and
A. muelleri, which is likely due to these beetles having lower abundances in the sites with high slug numbers (E4 and E5), and thus fewer encounters.
Nebria brevicollis also tested positive for Tipulidae and Lepidoptera DNA, and the beetles were observed feeding on live mature
N. pronuba caterpillars as well as
Tipula spp. larvae in the laboratory. Mair and Port [
4] reported that
N. brevicollis readily fed on Diptera larvae and Seric Jelaska et al. [
90] showed that 20% of specimens tested positive when screened for Lepidoptera DNA with PCR, compared to 7.4% in this study. However, when looking solely at the presence of positive feeding events in September and October when early season predation would be at its most beneficial as the predator to pest ratio is largest, it was found that 18.6% of the 59 screened
N. brevicollis, tested positive for Lepidoptera. This indicates that caterpillars might be an important food source during the second activity phase of this carabid species. While no species-specific primers were used,
N. pronuba was the only abundant caterpillar in the AR fields during the autumn and winter season. This study provides the first feeding record of carabid beetles on this relatively new introduced pest species. Both caterpillars and cranefly larvae were consumed during pest emergence in the autumn, even when they were not found under refuge traps. This indicates that consumption of these two prey groups is probably not just the result of opportunistic feeding, as simple chance encounters with the prey would be less likely.
Lepidoptera were also consumed by
C. cancellatum, which is described as ‘purely predaceous’ [
91] and one pair of beetles in captivity was reported to have killed between five and twelve
Noctua clandestine larvae a day [
92]. In addition, Gidaspow [
93] stated that
Calosoma spp. could kill approximately 300 caterpillars per season with even their newly hatched larvae being able to overcome caterpillars of considerable size. It is hence surprising that only a few individuals (5.9%) tested positive for Lepidoptera DNA. However, only a few caterpillars were detected in our sites during peak
C. cancellatum activity. The same applies for
P. laetulus—while this latter species as well as
A. muelleri showed some potential for contributing to pest control in AR, this study is the first to investigate their predation on these pest groups; still more studies are needed to elucidate their feeding preferences. A limitation of molecular gut content analysis is that it cannot distinguish between predation, scavenging [
94,
95] or secondary feeding [
96,
97]. Additionally, not every positive qPCR result necessarily indicates that the prey was killed. While the latter issue is difficult to resolve, as even an exact quantification of the amount of prey being consumed cannot prove the death of that prey, the ability and/or likelihood of a generalist predator actually feeding on live rather than dead prey can be determined through feeding experiments. We carried out two different laboratory trials (at 21 and 16 °C) using 20 adult
N. brevicollis that were starved for seven days, which showed the species neither consuming eggs nor live juvenile
D. reticulatum (~0.05 g;
Appendix A) while dead slugs were readily eaten on other occasions, even when beetles were not starved (Inga Reich, personal observation). Other laboratory studies on the predation of slugs by
N. brevicollis also showed very limited feeding on
D. reticulatum, with most being restricted to very small, injured or dead individuals or eggs [
3,
4,
98,
99]. However, Ayre [
2] found that up to 80% of one-day-old slugs presented to
N. brevicollis at an ambient temperature of 8 °C were killed but mortality rate of slugs dropped at higher temperatures.
Nebria brevicollis larvae were also found to feed on slugs, with ambient temperature being less important [
2] but only one of the 36 larvae tested in the latter study were positive. It is hence questionable whether the positive results herein show actual predation events, even though they mainly occurred at a time where many neonate slugs were present. On the other hand,
N. brevicollis was observed to feed on both, live mature cranefly larvae and final stage
N. pronuba caterpillars in the laboratory, demonstrating that the positive qPCR results for these pests might constitute actual predation events rather than scavenging or secondary predation. Another limitation of molecular gut content analysis is that digestion will break down the prey DNA in the predator gut as time progresses. Using qPCR technique, slug DNA could still be detected in 42 out of 47
N. brevicollis even after 24 h after feeding on ~0.001 g tissue (
Appendix B).
The next step in investigating the impact of commonly occurring carabids in AR fields should be to quantify the effect of predation by manipulating pest and predator densities in field cage experiments [
100,
101,
102]. This is essential as Firlej et al. [
13] found that despite a high incidence of feeding on soybean aphid
Aphis glycines by
P. melanarius (up to 33.7% of field-collected carabids screened positive for aphid DNA), a field cage experiment showed very little impact of carabid feeding on
A. glycines population growth at moderate infestation levels.