Bulletin of Entomological Research (2006) 96, 637–645 DOI: 10.1079/BER2006467

Biodiversity vs. biocontrol: positive and negative effects of alternative prey on control of by carabid

W.O.C. Symondson1 *, S. Cesarini1, P.W. Dodd1, G.L. Harper1y, M.W. Bruford1, D.M. Glen2·, C.W. Wiltshire3 and J.D. Harwood1z 1Cardiff School of Biosciences, Cardiff University, PO Box 915, Cardiff CF10 3TL, UK: 2IACR-Long Ashton Research Station, Department of Agricultural Sciences, University of Bristol, Long Ashton, Bristol, BS41 9AF, UK: 3Arion Ecology, The Brambles, Stinchcombe Hill, Dursley, Gloucestershire, GL11 6AQ, UK

Abstract

Environment-friendly farming techniques seek to increase invertebrate biodi- versity in part with the intention of encouraging greater numbers of predators that will help to control crop pests. However, in theory, this effect may be negated if the availability of a greater abundance and diversity of alternative prey diverts predators away from feeding on pests. The hypothesis that access to alternative prey can lead to reduced pest suppression under semi-field conditions was tested. Alternative prey type and diversity were manipulated in 70 mesocosms over 7+ weeks in the presence of the carabid melanarius (Illiger), a known predator of slugs, and reproducing populations of the reticulatum (Mu¨ ller). Significantly fewer slugs survived where no alternative prey were provided. Maximum slug numbers and biomass were found in treatments containing either carabids plus a high diversity of alternative prey (many of earthworm and three of Diptera larvae) or a single additional prey (blowfly larvae, vomitoria Linnaeus). In these treatments slug numbers and biomass were as high as in plots lacking predators. The effects of alternative prey were taxon-specific. Alternative prey strongly affected carabid fitness in terms of biomass and egg load. The fittest predators (those with access to high alternative prey diversity or C. vomitoria larvae) reduced slug numbers the least. The mean individual slug weights were greater in treatments with alternative prey than where no alternative prey were provided to the carabids. These results suggest that pests may survive and reproduce more rapidly in patches where predators have access to alternative prey.

*Fax: +44 (0)29 20 874 305 E-mail: [email protected] ·Honorary Professor of Cardiff University and independent consultant at Styloma Research and Consulting, Phoebe, The Lippiatt, Cheddar, BS27 3QP, UK yPresent address: School of Applied Sciences, University of Glamorgan, Pontypridd, Mid-Glamorgan, CF37 1DL, UK zPresent address: Department of Entomology, University of Kentucky, S-225 Agricultural Science Center North, Lexington, KY 40546-0091, USA 638 W.O.C. Symondson et al.

Keywords: Carabidae, , diet, generalist predators, predator fitness, , slugs

Introduction Prey diversity has been shown to have a powerful effect on predator nutrition, reproduction and survival. Green- Alternative prey help to sustain and retain generalist stone (1979) suggested that predators seek to diversify their predators within crops when target pests are absent or at low diets in order to balance their amino acid requirements and, density. This dietary flexibility of generalists can, in theory, more recently, Mayntz et al. (2005) demonstrated that give them a significant advantage over specialist natural invertebrate predators are capable of selective foraging in enemies, allowing them to be present within a crop early order to balance their intake of proteins and lipids. Other in the year before the pests arrive in any significant numbers studies have shown the benefits of a mixed diet, either to (Murdoch et al., 1985; Chiverton, 1987; Chang & Karieva, balance dietary needs or to avoid prey toxins that affect 1999; Harwood et al., 2004). Certain agricultural practices can predator fitness (e.g. Toft & Wise, 1999; Oelberman & Scheu, encourage greater numbers of alternative prey during this 2002; Fisker & Toft, 2004). Importantly, however, beneficial/ early period, allowing predators to reach high densities detrimental effects of a mixed diet are taxon-specific. J.D. by the time the crop needs to be protected from immigrating Harwood, S.W. Phillips, K.D. Sunderland, D.M. Glen, M.W. or rapidly reproducing pests (Settle et al., 1996). Such Bruford, G.L. Harper, and W.O.C. Symondson (unpublished systems depend upon temporal separation between periods data) have demonstrated that both linyphiid spiders and of maximum alternative prey availability and periods of carabid beetles benefit from a diverse diet; for example pest abundance. In many circumstances this is probably not carabids fed a more diverse diet weighed more, laid more achievable and practices such as conservation tillage eggs, continued to lay eggs for longer and the eggs (minimal tillage systems including direct drilling that are developed and hatched more rapidly. Carabids and spiders designed to conserve the soil (Ko¨ller, 2003)) are now fed on pests only (slugs or aphids) were the least fit. There is employed, in part, to maximize biodiversity (Kladivko, some evidence that these effects translate to the field where 2001), in the hope that this will foster a larger and more carabids in more complex habitats, with more diverse prey, diverse predator community that will go on to control the were larger and contained more eggs (Bommarco, 1998, pests. Empirical studies have shown that this is often an 1999). Clearly, prey diversity, if exploitable by the predator, effective strategy for increasing predator numbers (Stinner & improves predator fitness and between years this may lead House, 1990), but the effects of mechanical operations to increased predator numbers, even of univoltine species (cultivation, harvesting) on predators and their prey can be such as carabid beetles. The predators may go on to limit complex and taxon-specific (El Titi, 2003; Holland & growth in pest numbers, although where the pests form a Reynolds, 2003; Holland, 2004; Thorbek & Bilde, 2004). The substantial proportion of the total available prey, suppres- additional predators in minimal or no-tillage systems may sion may lead to subsequent limitation of predator densities sometimes go on to increase predation pressure on pests, but through loosely-coupled feedback mechanisms (Symondson in other instances predator numbers, pest numbers and crop et al., 2002b). However, in the medium term (within a year) damage may all increase (Stinner & House, 1990), a problem univoltine carabids do not have time to respond numerically that has been particularly noted with slugs (Kendall et al., to prey abundance and diversity. For flightless species such 1995; Symondson et al., 1996; Glen & Symondson, 2003). as the ground Pterostichus melanarius (Illiger) field A theoretical problem with the conservation tillage scenario boundaries have been shown to be effective barriers to is that alternative prey may divert predators away from movements (Thomas et al., 1998), preventing significant feeding on target pests (Halaj & Wise, 2002; Symondson aggregative responses to prey, although within-field aggre- et al., 2002a; Koss & Snyder, 2005; Rypstra & Marshall, 2005), gation to pest density can take place (Bohan et al., 2000; and the more numerous and diverse the community of Winder et al., 2005a,b). Thus, P. melanarius responses to alternative prey, the more likely it would be that prey dietary components and prey diversity will be restricted to choice would favour one or more of these alternative prey the prey available in the fields in which they are sampled, species rather than the target pests. This may happen and the fitness of those predators will depend upon the through simple substitution (non-pest for pest) or through abundance and diversity of available food resources. Within switching behaviour as relative densities of prey species year, therefore, reproductive numerical responses will not change (Holt & Lawton, 1994). Food web theory suggests have time to operate, and spatial numerical responses will be that, within a more diverse system, many of those alternative limited, restricting responses by the predators to alternative prey will be other predators, and that predation on each prey to functional responses. If this is correct, then the other potentially decreases the ability of the predator presence of alternative prey is likely to lead to reduced community as a whole to limit herbivore (pest) numbers predation on the target prey (slugs) (Harmon & Andow, (Polis et al., 1989; McCann et al., 1998; Finke & Denno, 2004). 2004). And so the question remains, what happens when The carabid P. melanarius is a highly generalist predator farming systems encourage a numerous and diverse fauna eating a wide variety of prey (Sunderland, 1975, 2002; within a crop coincident with periods when pest density is Symondson et al., 2000). However, Symondson et al. (1996) likely to be high? May the goals of encouraging alternative found, using anti-slug antibodies and enzyme linked prey and biodiversity, and seeking biocontrol of pests, immunosorbent assays, that over 80% of P. melanarius sometimes conflict? May not the fittest predators, replete captured in the field in one year contained slug proteins in with alternative prey, be the least effective at controlling their guts, clearly demonstrating that slugs are a signifi- pests? cant prey item. These carabids have now been shown, in Alternative prey reduce predation on pests 639 semi-field and field experiments, to be capable of affecting Table 1. Invertebrates added to each of 10 replicate mesocosms the temporal and spatial dynamics of slugs (Symondson for each of seven treatments. The target slug prey, Deroceras et al., 1996, 2002b; Bohan et al., 2000; McKemey et al., 2003, reticulatum, were added to all plots (28 per plot), and carabid reviewed in Symondson, 2004). We therefore tested the beetle predators (two female Pterostichus melanarius per plot) hypothesis that, despite these high rates of predation by were added to the appropriate treatments. P. melanarius on slugs reported in the field, when numerical Treatment Prey species responses by the predators was prevented slug survival/ reproduction would be greater where alternative prey were A Slugs B Slugs+beetles available to the predators. At the same time we tested the C Slugs+beetles+earthworms (E. hortensis) (50) hypothesis that alternative prey, and prey diversity, would D Slugs+beetles+Diptera larvae (C. vomitoria)(9a) have a positive effect upon carabid fitness. We further tested E Slugs+beetles+earthworms (mixed speciesb) the hypothesis that the strength of this effect would be prey (50)+Diptera larvae (three speciesc)(9a) taxon-specific. In addition, we tested the hypothesis that a F Slugs+earthworms (E. hortensis) (50) higher diversity of alternative prey would reduce still G Slugs+Diptera (C. vomitoria)(9a) further the ability of the beetles to limit slug numbers. a Overall, we wished to challenge the assumption that the This number of larvae was added every week. Where three fittest beetles would be most effective at controlling pest species of Diptera were provided, three individuals of each species were added each week. In every other instance the (slug) populations. Where reproductive numerical responses invertebrates were added just once, at the start of the are not possible, and immigration severely limited, as we experiment. argue may often be the case in the field for this predator– b In this high diversity treatment mixed species of earthworms prey system, we would expect that functional response by (Allolobophora chlorotica, Aporrectodea caliginosa, Aporrectodea longa the beetles to the total prey available to lead to increased and Octolasion cyaneum) were collected directly from the field. survival of the pests in the presence of alternative prey. Forty five of these mixed species were added, at random, to each We also tested the hypothesis that predation by the plot plus five Eisenia hortensis. c beetles would affect the size structure of the slug population. The three species were Calliphora vomitoria, and McKemey et al. (2001) demonstrated a preference for the Fannia canicularis. Numbers of alternative prey introduced to each plot are given in smallest slugs in laboratory trials, but could find no size parentheses. preferences where trials were performed, as here, in semi- field miniplots (McKemey et al., 2003). If the beetles showed a preference for smaller slugs, then treatments with the fewest slugs at the end of the experiment (i.e. those subject to this order are reported from arable fields (Glen et al., 1988; the highest predation pressure) would be expected to have Symondson et al., 1996). Each of the 70 mesocosms contained the largest slugs. 20 larger slugs, which had been collected directly from the field, and eight neonate slugs, hatched in the laboratory. A subset of 20 of the slugs from the field were weighed and Materials and methods ranged from 125 to 415 mg (mean 214 mg, SE 24.9), while Experiments were conducted in outdoor mesocosms neonate slugs were all between 2.2 and 3.3 mg. There were (miniplots). These were circular plastic tubs (35 cm diam- three control treatments without predators: slugs only (A), eterr18 cm deep), with drainage holes around the base slugs plus the earthworms Eisenia hortensis (Michaelsen) (F) covered with fine mesh to prevent invertebrates escaping or and slugs plus Diptera larvae, Calliphora vomitoria Linnaeus entering. The surface area of soil in each mesocosm was (G). These treatments were designed to measure background 0.08 m2. The tubs were half filled with a mixture of steam- rates of slug population increase and any possible effects of sterilized sandy loam (50%), peat (30%) and grit (20%). Over other invertebrates (earthworms and Diptera) on slug the summer a crop of spring wheat was grown in the tubs, numbers. There were four treatments with beetles (B, C, D then harvested and cut to within a few centimetres of the soil and E, table 1) where two female P. melanarius were added to and experiments conducted on the stubble. The inner rim of each plot. Two beetles is a higher density than might usually each plot was regularly painted with FLUON (polytetra- be encountered in the field but were used to compensate for fluoroethylene – Whitford Plastics, Runcorn, UK) to prevent the expectation that at least one beetle might die during the slugs (Symondson, 1993) and other invertebrates from course of this prolonged experiment (in practice a mean of climbing out (McKemey et al., 2003). Each plot contained a 1.45 beetles per plot survived, see Results). Treatment B refuge for the beetles consisting of a piece of polystyrene tile contained beetles and slugs only. Treatment C contained (10 cmr10 cm), weighted down with a stone to prevent it beetles, slugs and the earthworms E. hortensis, and treatment from blowing away. The plots were protected by netting to D contained beetles, slugs and the Diptera larvae prevent birds from eating the predators or prey. The area C. vomitoria. Treatment E contained the highest prey surrounding the plots was regularly treated with slug pellets abundance and diversity with slugs plus mixed earth- containing metaldehyde to destroy any other slugs in the worm species, together with three species of Diptera larvae, vicinity. Plots were watered as and when necessary to C. vomitoria, Lucilia caesar Linnaeus and Fannia canicularis maintain moist surface conditions and encourage slug and Linnaeus. Although none of these species would be earthworm activity. commonly found as larvae in the soil they acted as a readily There were seven treatments (table 1) with 10 replicate available surrogate for the many species that are common in plots for each treatment. Numbers of alternative prey per arable fields, such as crane flies (Tipulidae), cluster flies mesocosm are shown in table 1. All treatments included () and many species of Phoridae. The mean slugs, Deroceras reticulatum (Mu¨ ller), at a density of 28 per weights of each species of Diptera larvae were found to be plot, equivalent to 350 m2, a high number but densities of (n = 10): C. vomitoria 99.4 mg (SE 3.10), L. caesar 39.7 mg (SE 640 W.O.C. Symondson et al.

1.34) and F. canicularis 24.2 mg (SE 2.15). Mixed earthworms in subsequent analyses, regardless of how many beetles were were collected from an arable field near Cardiff while it was found. As the error caused by loss of beetles from some plots being ploughed. A subset were identified to species and through mortality would have reduced our chances of comprised Allolobophora chlorotica (Savigny) 27.8%, Aporrec- finding differences between treatments, any remaining todea caliginosa (Savigny) 30.6%, Aporrectodea longa (Ude) significant differences would have had to have been strong 19.4%, Octolasion cyaneum (Savigny) 2.8% plus unidentified to be detectable. The large number of replicates per treat- juveniles 19.4%. Earthworms were not weighed individually, ment (10) was designed to minimize this predictable but a sample batch of 50 E. hortensis had a mean individual problem. weight of 0.93 g while a batch of 50 mixed earthworms Data were analysed by ANOVA, with data transformed collected from the field had a mean individual weight of when necessary to stabilize variances. Where ANOVAs were 0.331 g. The P. melanarius beetles were obtained by pitfall significant, means were compared using Least Significant trapping in a field of winter wheat. Differences (LSD for P £ 0.05). Slugs and earthworms were introduced to the plots, directly from the field, over nine days. This was designed to Results avoid potential problems caused by disease transmission between invertebrates in the laboratory, especially the slugs The numbers, biomass and mean weight of slugs which are highly susceptible to the slug parasitic nematode extracted from the mesocosms at the end of the experiment Phasmarhabditis hermaphrodita (Schneider) (Wilson et al., 1993) are shown in fig. 1. Overall the treatments had a significant and the pathogen Microsporidium novacastriensis (Jones & effect on slug numbers (F6, 63 = 4.59, P < 0.001) and total slug Selman) which debilitates D. reticulatum in laboratory biomass per plot (F6, 63 = 5.83, P < 0.001). There was no cultures (Jones & Selman, 1985). The beetles were then significant difference between the three treatments without weighed and introduced to the plots on 12 September 2003, beetles, showing that neither earthworms nor Diptera were along with (in appropriate treatments) the first batch of directly affecting slug numbers. Fewest slugs were (and Diptera larvae (table 1). Diptera larvae, unlike the other prey, lowest slug biomass was) recovered from plots with slugs were added weekly because otherwise they would have and beetles only. Plots with slugs and beetles contained pupated, hatched and escaped. After 12 days, when significantly fewer slugs (and lower slug biomass) than plots temperatures began to fall (which would have limited beetle with slugs only, showing that the beetles were reducing activity), the plots were moved to a glasshouse with slug numbers. However, where there were additional prey temperature control. During the 12 days outdoors the mean species in treatments D (slugs, beetles and C. vomitoria minimum temperature was 11.6C (SE 1.15, lowest 2.7C) larvae) and E (the highly diverse diet treatment with slugs, and the mean maximum 20.8 (SE 0.91, highest 24.9C). beetles, mixed earthworms and mixed Diptera larvae), the During the 40 days under glass the minimum temperature plots contained significantly greater numbers and biomass of (on every day) was 16C, controlled by a thermostat slugs than plots with slugs and beetles alone, demonstrating and heater. The mean daily maximum was 25.2C (SE 0.51, that alternative prey were diverting the beetles away from highest 32C). Weeds growing in the plots were regularly feeding on slugs. However, treatment C (slug, beetles and removed mainly to prevent them from providing escape E. hortensis), did not contain a significantly greater number routes for predators and prey, but also because they were or biomass of slugs than plots with slugs and beetles only. found to attract aphids. The presence of beetles in treatment D (slugs, beetles and After the experiment had been running for five weeks, C. vomitoria) did not reduce slug numbers and biomass any beetles found under refuges were collected and frozen. significantly compared with treatment G (slugs and Where one or no beetles were found we did not know at that C. vomitoria but no beetles), suggesting that these Diptera stage whether the beetles were still present in the plots, were diverting the beetles away from feeding on slugs. buried in the soil, or whether they had died. After three days However, beetles in treatment C (slugs, beetles and of checking under the refuges, 35 of the plots were gradually E. hortensis) did reduce slug numbers and biomass compared flooded over 10 days, forcing the slugs and any remaining with treatment F (slugs and E. hortensis but no beetles), again beetles to the surface. This technique is considered the most suggesting that this species of earthworm was not signifi- accurate method for assessing slug populations (Glen et al., cantly diverting the beetles from preying on the slugs. 1989; Symondson et al., 1996). Each slug was identified and In addition to slug numbers and biomass, we analysed weighed. The process was repeated with the remaining 35 mean slug weights to see whether these varied between plots 14 days later. The plots were processed in two batches treatments at the end of the experiment. Overall, there were because the flooding unit could not accommodate all 70 at significant differences between treatments (F6, 63 = 3.28, once. Each batch contained five replicates of each treatment. P = 0.007). Treatment B (slugs and beetles only) contained Initial analyses of the data treated these two batches as significantly smaller slugs than in any other treatment separate blocks in analysis of variance (ANOVA) but, as (fig. 1c). Few, if any, of the adult slugs had survived to there were no significant differences found for slug numbers, the end of the experiment and the mean size of slugs biomass and mean size data (P > 0.2), ‘block’ was excluded extracted from all treatments was much smaller than the from subsequent analyses. mean size introduced to plots. In treatment B (slugs and All beetles were weighed and frozen as soon as they were beetles only) the mean size was that of very small neonates collected from the plots. They were then dissected and their (back-transformed mean 1.5 mg). There was no significant egg loads recorded. At the end of the experiment 21 plots difference between the remaining treatments, with a pooled still contained two beetles, 16 contained one and just three back-transformed mean of 7.2 mg, still very small but r4.8 had no live beetles. We cannot rule out the possibility that larger than in treatment B. some beetles may have drowned, and thus failed to come to Two measures of predator fitness, beetle biomass and egg the surface, during gradual flooding. All plots were included load, were analysed. At the beginning of the experiment Alternative prey reduce predation on pests 641 a 3 a 0.25 2.5 LSD LSD 2 0.2 1.5 0.15 slug number 1 e 0.5 0.1 0 Beetle weight (g) 0.05 b 6 LSD 5 0 4 b 25 3 LSD 2 20

tot. slug biomass (mg)1 Log

e 0 15 c 3 10 Egg number 2.5 LSD 5 2 1.5 0 1 B C D E ind. slug weight (mg) Log 0.5 Beetles Beetles Beetles Beetles e Worms Worms+ 0 Log Diptera Diptera+ A B C D E F G Beetles Beetles Beetles Beetles Treatments Worms Worms+ Worms Diptera Diptera+ Diptera Fig. 2. Graphs of measures of beetle (Pterostichus melanarius) Treatments fitness at the end of the experiment in terms of mean numbers of eggs within the beetles (a) and beetle weights (b). Slugs, Fig. 1. Mean numbers (a), total biomass per plot (b), and mean Deroceras reticulatum, were added to all plots, while additional individual weights (c) of slugs (Deroceras reticulatum) in seven prey added are indicated. ‘+’ indicates multiple species. For full £ mesocosm treatments at the end of the experiment (n = 10 per details of treatments A–G see table 1. LSD bars are for P 0.05. treatment). Slugs, D. reticulatum, were added to all plots, while additional prey added are indicated. ‘+’ indicates multiple species. For full details of treatments A–G see table 1. LSD bars are for P £ 0.05. (table 2). Overall, the fitter the beetles (in terms of weight and egg load at the end of the experiment) the greater the number and biomass (per plot) of slugs surviving. Relation- ships between beetle fitness parameters and mean individual there was no significant difference between the weights of slugs weights at the end of the experiment proved to be non- beetles in the different treatments (F3, 76 = 0.23,P= 0.879). significant once missing data (plots with zero slugs) were However, by the end of the experiment there were highly removed from the analyses. significant differences (F3, 33 = 13.44, P < 0.001) (fig. 2a), with significantly greater beetle biomass in treatments D (beetles, Discussion slugs and Diptera) and E (beetles, slugs and diverse prey) than in treatments B (beetles and slugs only) and C (beetles, This carabid, P. melanarius, has been shown in many slugs and the worm E. hortensis). Similarly, analysis of the previous studies to feed on slugs in the laboratory egg loads within these beetles showed highly significant (Symondson, 1997; McKemey et al., 2001; Oberholzer & differences (F3, 33 = 16.19, P < 0.001) (fig. 2b) with beetles in Frank, 2003; Oberholzer et al., 2003), microcosms and treatments D (beetles, slugs and Diptera) and E (beetles, miniplots (Buckland & Grime, 2000, Thomas, 2002; McKe- slugs and diverse prey) containing significantly more mey et al., 2003) and field (Cornic, 1973; Tod, 1973; eggs than beetles in treatments B (beetles and slugs only) Symondson et al., 1996, 2002b; Bohan et al., 2000; Dodd and C (beetles, slugs and the worm E. hortensis). Beetle et al., 2003; Harper et al., 2005), and together these results weights at the end of the experiment were highly correlated provide clear evidence that they are killing, consuming and with egg load (beetle weight = 0.167+0.0025 egg load, r2 in many cases reducing the number of these pests. This 58.3%, P < 0.001). experiment was designed to measure the effects of the A number of regression analyses were performed to presence of alternative prey, and prey diversity, on the examine the relationship between slug numbers and biomass ability of P. melanarius to limit slug numbers when numerical per plot at the end of the experiment and beetle fitness responses were not possible. As might be expected, lowest 642 W.O.C. Symondson et al.

Table 2. Relationships between beetle weights and egg loads, at the end of the experiment, with slug numbers and biomass.

Response (Y) Predictors (X) Equation r2 % Probability

Beetle weight Loge slug number Y = 0.170+0.0171 X 25.2 P = 0.002 Loge slug biomass Y = 0.239+0.0104 X 24.6 P = 0.002

Beetle egg load Loge slug number Y = 3.06+5.85 X 31.5 P < 0.001 Loge slug biomass Y = 25.5+3.22 X 25.3 P = 0.002

residual slug numbers and biomass were found in the interaction between dietary quality and diversity, as shown treatment where there were no alternative food resources previously by Toft & Wise (1999), Oelberman & Scheu (2002) (B, beetles and slugs only). In all other treatments containing and Fisker & Toft (2004). P. melanarius, residual slug number and biomass were Our experimental design was not, in retrospect, ideal greater (fig. 1) in most cases significantly so, providing with respect to the high/low diversity comparison. We did direct evidence that the presence of alternative prey reduced not realize beforehand that E. hortensis would prove to be a the rate of predation on slugs. The one exception was non-preferred prey item and therefore the comparison treatment C (beetles, slugs and the worm E. hortensis), where between high and low diversity of worms was confounded the earthworms E. hortensis were provided as alternative by this factor. Molecular analyses using both monoclonal prey. Eisenia hortensis is closely related to Eisenia fetida antibodies and DNA primers have shown that earthworms (Savigny), a species known to generate defensive chemicals are frequently the largest component of the diet of within coelomic fluid exuded by dorsal pores when attacked P. melanarius, but these techniques did not distinguish (Sims & Gerard, 1985). It is possible that such deterrent between predation on different species (Symondson et al., chemicals exist in E. hortensis and are effective against 2000; Harper et al., 2005 and unpublished data). P. melanarius, especially, as here, where the carabids have the In a previous miniplot trial, under similar conditions, option of eating other prey, the slugs, instead. At the end McKemey et al. (2003) could find no significant difference in of the experiment, beetles in treatments with slugs plus predation rates on different size classes of slugs. It was E. hortensis were no fitter (in terms of their biomass and concluded that the preference shown in the laboratory for egg loads) than beetles in plots with slugs only, possibly smaller slugs (McKemey et al., 2001) was counteracted by suggesting that these two prey species were of approxi- environmental heterogeneity, which provided a greater mately the same quality to the predator. The only significant number and diversity of refugia for the smaller slugs. effect of the presence of E. hortensis was on slug size; slugs However, in the current experiment this seems to have gone with beetles only were significantly smaller at the end of the one step further, with the smallest slugs found in the experiment than in any other treatment, including treatment treatment subject to the greatest predation pressure (treat- C ((beetles, slugs and the worm E. hortensis)), suggesting that ment B, slugs and beetles only). The mean size of the slugs predation pressure on the slugs may have been marginally was, at 1.5 mg, smaller than the smallest slugs used in lower where these worms were present. McKemey et al. (2003) and it is probable that these neonate In the two other treatments that included predators plus slugs were either overlooked by the predator or found alternative prey, treatment D (beetles, slugs and Diptera) numerous refugia in the soil and within the stems of the and treatment E (the high prey diversity treatment), slug wheat stubble. Many other studies have shown that there is a numbers, biomass and mean weight were all significantly lower size threshold below which predators ignore prey (e.g. greater than where slugs alone were present with beetles. In Greene, 1975; Finch, 1996). addition, beetles in treatments D and E with alternative prey The length of the experiment was designed to allow time were the fittest, with significantly greater predator biomass for the slugs to reproduce. The size of the slugs recorded at and egg loads. The latter two are likely to be correlated as the end of the experiment suggests that most hatched from beetle weight is affected by egg load, and this proved to be eggs laid by adults during the experiments and that, the case. There was no evidence that the high diversity following egg laying, most adults died. Predation on slug treatment had any different effect on predation on slugs or eggs, as well as neonates, was possible (Oberholzer & Frank, predator fitness than the one incorporating just slugs and fly 2003; Symondson, 2004) but could not be tracked. larvae. larvae and pupae are used commercially (e.g. The use of LSD to define post-ANOVA differences GAB Biotechnologie GmbH) as a complete diet for rearing between treatment means is widely used, but often restric- carabid larvae such as (Linnaeus) and have tions are placed on the number of comparisons that are been used to maintain P. melanarius adults in the laboratory allowable, usually x-1 where x is the number of treatments for over a year (W.O.C. Symondson, unpublished data). It (e.g. Fry, 1993). However, often more a priori comparisons is likely that a diet of slugs plus C. vomitoria contained most are ecologically justified, especially where more than one of the nutritional requirements of these beetles and a more control is used (as here). It has been argued that where more diverse diet provided no significant additional benefit. than x-1 comparisons are made, Bonferroni adjustments are Theory suggests that predators presented with an abundant necessary. However, this has been strongly criticized in and diverse diet should be more selective, taking the most recent papers by Perneger (1998) and Moran (2003), who profitable prey (Pyke, 1984; Stephens & Krebs, 1986), which argue that Bonferroni is too conservative, and that the more in this case may have been the Diptera larvae. By contrast, treatments that are included in the experiment, and the more a diet of slugs plus E. hortensis was significantly worse than a controls there are in place (normally considered a good highly diverse diet, showing that there is a taxon-specific thing), the greater the chances that type II errors will occur Alternative prey reduce predation on pests 643

(i.e. real significant differences will be missed). Moran (2003) This simple plot experiment must, however, be inter- argues that this is particularly relevant to ecological studies, preted with caution. It does not, for example, tell us anything where variances are usually high. Both authors argue for about how much alternative prey will be available in the simply presenting the data, pointing out where significant field or how temporal change in alternative prey availability differences lie (as here, with an LSD bar), and leaving might affect predator–target prey interactions (e.g. Settle interpretation to the reader. We have followed this policy. et al., 1996). Although the plots were designed to emulate All of the comparisons we made were, a priori, predictably field conditions, they inevitably restrict the movement meaningful. patterns of predators and prey as well as the choice of prey We accept that the definition of ‘fitness’ used in this available. Nevertheless, we believe the implications for paper is limited, given that the experiment ran for only a few interpretation of field data are profound. Our results suggest weeks and we do not know what longer-term consequences that future field studies of interactions between generalist greater beetle weight or egg load would have on survival or predators and particular prey species should ideally monitor reproduction. Nevertheless, there were significant associa- consumption of alternative, competing prey (Symondson tions between these measures of fitness and slug number et al., 2000; Agustı´ et al., 2003), which is now possible using and biomass in the plots. These analyses show that one of the molecular approaches now being applied in the the beetles that were fittest had the least impact on slug field (Symondson, 2002; Sheppard & Harwood, 2005; populations and were found in association with the highest Sunderland et al., 2005), especially multiplex polymerase residual numbers of slugs. Although predictable from our chain reactions (PCRs) (Harper et al., 2005). Examination of results, and an effect of treatment rather than fitness per se, just one small part of a food web, especially one involving this may appear counter intuitive and could easily lead to generalist predators, cannot be relied upon to reveal the misinterpretation of data from the field. Spatial associations ecological processes leading to observed dynamics. between generalist predators and a target prey species in the field may be interpreted in very different ways. If predator and target prey numbers are positively spatially associated it Acknowledgements might be concluded that the predators are aggregating to the prey in order to feed on them (Symondson et al., 1996; Bohan The authors wish to thank Ms Jenna Willis and Ms Clare et al., 2000; Winder et al., 2005a,b). However, if high predator Armsworth for supplying the neonate slugs, Mr Simon density is associated with low prey density this may also be Shayler for identifying the fly larvae and, with Dr Samuel seen as evidence of high predation pressure (prey density Sheppard, helping to collect slugs, and Mr Lyndon Tuck for has been reduced by the predators) (Bohan et al., 2000). recording temperatures and watering the mesocosms. The In practice one would expect transient dynamics with lag authors also thank the Biotechnology and Biological Sciences phases, such that highest predator density is associated with Research Council of the United Kingdom for funding this previous, rather than current, prey densities, as found by work as part of a larger study of the effects of biodiversity on Winder et al. (2005a). Both interpretations could be true biological control of crop pests. under different circumstances, but availability of alternative prey is rarely, if ever, considered as the primary factor References driving the observed dynamics. Our results suggest that an equally valid interpretation of the same data might be that Agustı´, N., Shayler, S., Harwood, J.D., Vaughan, I.P., Sunder- pest species may survive and reproduce more rapidly in land, K.D. & Symondson, W.O.C. (2003) Collembola as patches where the predators are feeding on alternative prey. alternative prey sustaining spiders in arable ecosystems: This would be predictable from theory (Harmon & Andow, prey detection within predators using molecular markers. 2004) in that the predators are reproducing on an annual Molecular Ecology 12, 3467–3475. time scale, but responding to total prey on much shorter Bohan, D.A., Bohan, A.C., Glen, D.M., Symondson, W.O.C., timescales, primarily in terms of functional responses. Wiltshire, C.W. & Hughes, L. (2000) Spatial dynamics of Symondson et al. (1996) found greater numbers of slugs and predation by carabid beetles on slugs. Journal of P. melanarius in long-term no-tillage plots compared with Ecology 69, 367–379. plots that had been subjected to various forms of tillage over Bommarco, R. (1998) Reproduction and energy reserves of many years. Analysis of the beetles’ gut contents using a predatory carabid beetle relative to agroecosystems antibodies showed that where there was greater slug complexity. Ecological Applications 8, 846–853. biomass in the soil each beetle was eating greater quantities Bommarco, R. (1999) Feeding, reproduction and community of slug. However, ratio-dependent predation pressure on the impact of a predatory carabid in two agricultural habitats. slugs in the no-tillage plots must have been lower, because Oikos 87, 89–96. by September (just before cultivation and therefore an effect Buckland, S.M. & Grime, J.P. (2000) The effects of trophic of treatments applied one year previously) there were structure and soil fertility on the assembly of plant 1.8 times as many beetles in the no-tillage plots compared communities. Oikos 91, 336–352. with those that were tilled, but 91 times more slug in terms Chang, G.C. & Kareiva, P. (1999) The case for indigenous of biomass. The beetles in the no-tillage plots were the best generalists in biological control. pp. 103–115 in Hawkins, fed, with the greatest fore-gut biomass, but apparently the B.A. & Cornell, H.V. (Eds) Theoretical approaches to biological least capable of limiting slug numbers. It is probable that control. Cambridge, Cambridge University Press. prey in general were more numerous and diverse in the Cornic, J.F. (1973) Etude du re´gime alimentaire de trois espe`ces no-tillage treatment, as has been found in many other studies de Carabiques et de ses variations en verger de pommiers. (Stinner & House, 1990; Kladivko, 2001; Holland, 2004), and Annales Socie´te´ Entomologique de France 9, 69–87. that as a result the slugs in this treatment were under the Chiverton, P.A. (1987) Predation of Rhopalosiphum padi (Homo- least predation pressure. ptera: Aphididae) by polyphagous predatory 644 W.O.C. Symondson et al.

during the aphids’ pre-peak period in spring barley. Annals ment on cereal pests and their natural enemies. pp. 83–102 of Applied Biology 111, 257–269. in Glen, D.M., Greaves, M.P. & Anderson, H.M. (Eds) Dodd, C.S., Bruford, M.W., Symondson, W.O.C. & Glen, D.M. Ecology and integrated farming systems. Chichester, John (2003) Detection of slug DNA within carabid predators Wiley & Sons. using prey-specific PCR primers. pp. 13–20 in Slug and snail Kladivko, E.J. (2001) Tillage systems and soil ecology. Soil and pests: agricultural, veterinary and environmental perspectives. Tillage Research 61, 61–76. Alton, British Crop Protection Council. Ko¨ller, K. (2003) Techniques of soil tillage. pp. 1–25 in El Titi, A. El Titi, A. (2003) Effects of tillage on invertebrates in (Ed.) Soil tillage in agroecosystems. London, CRC Press. agroecosystems. pp. 261–296 in El Titi, A. (Ed.) Soil tillage Koss, A.M. & Snyder, W.E. (2005) Alternative prey disrupt in agroecosystems. London, CRC Press. biocontrol by a guild of generalist predators. Biological Finch, S. (1996) Effect of beetle size on predation of cabbage root Control 32, 243–251. fly eggs by ground beetles. Entomologia Experimentalis et Mayntz, D., Raubenheimer, D., Salomon, M., Toft, S. & Applicata 81, 199–206. Simpson, S.J. (2005) Nutrient-specific foraging in inverte- Finke, D.L. & Denno, R.F. (2004) Predator diversity dampens brate predators. Science 307, 111–113. trophic cascades. Nature 429, 407–410. McCann, K.S., Hastings, A. & Huxel, G.R. (1998) Weak Fisker, E.N. & Toft, S. (2004) Effects of chronic exposure to a trophic interactions and the balance of nature. Nature 395, toxic prey in a generalist predator. Physiological Entomology 794–798. 29, 129–138. McKemey, A., Symondson, W.O.C., Glen, D.M. & Brain, P. Fry, J.C. (1993) Biological data analysis: a practical approach. Oxford, (2001) Effects of slug size on predation by Pterostichus IRL Press. melanarius (Coleoptera: Carabidae). Biocontrol Science and Glen, D.M. & Symondson, W.O.C. (2003) Influence of soil Technology 11, 83–93. tillage on slugs and their natural enemies. pp. 207–227 in El McKemey, A., Symondson, W.O.C. & Glen, D.M. (2003) Titi, A. (Ed.) Soil tillage in agroecosystems. London, CRC Predation and prey size choice by the carabid beetle Press. Pterostichus melanarius (Coleoptera: Carabidae): the dangers Glen, D.M., Wiltshire, C.W. & Milsom, N.F. (1988) Effects of of extrapolating from laboratory to field. Bulletin of straw disposal on slug problems in cereals. Aspects of Entomological Research 93, 227–234. Applied Biology 17, 173–179. Moran, M.D. (2003) Arguments for rejecting the sequential Glen, D.M., Milsom, N.F. & Wiltshire, C.W. (1989) Effects of Bonferroni in ecological studies. Oikos 100, 403–405. seed-bed conditions on slug numbers and damage to Murdoch, W.W., Chesson, J. & Chesson, P.L. (1985) Biological winter wheat in a clay soil. Annals of Applied Biology 115, control in theory and practice. American Naturalist 125, 177–190. 344–366. Greene, A. (1975) Biology of five species of Cychrini Oberholzer, F. & Frank, T. (2003) Predation by the carabid (Coleoptera: Carabidae) in the steppe region of south- beetles Pterostichus melanarius and Poecilus cupreus on slugs eastern Washington. Melanderia 19, 1–43. and slug eggs. Biocontrol Science and Technology 13, 99–110. Greenstone, M.H. (1979) Spider behaviour optimises dietary Oberholzer, F., Escher, N. & Frank, T. (2003) The potential of essential amino acid composition. Nature 282, 501–503. carabid beetles (Coleoptera) to reduce slug damage to Halaj, J. & Wise, D.H. (2002) Impact of a detrital subsidy on oilseed rape in the laboratory. European Journal of Entomol- trophic cascades in a terrestrial grazing food web. Ecology ogy 100, 81–85. 83, 3141–3151. Oelbermann, K. & Scheu, S. (2002) Effects of prey type and Harmon, J.P. & Andow, D.A. (2004) Indirect effects between mixed diets on survival, growth and development of a shared prey: predictions for biological control. BioControl generalist predator, Pardosa lugubris. Basic and Applied 49, 605–626. Ecology 3, 285–291. Harper, G.L., King, R.A., Dodd, C.S., Harwood, J.D., Glen, Perneger, T.V. (1998) What’s wrong with Bonferroni adjust- D.M., Bruford, M.W. & Symondson, W.O.C. (2005) Rapid ments? British Medical Journal 316, 1236–1238. screening of invertebrate predators for multiple prey DNA Polis, G.A., Myers, C.A. & Holt, R.D. (1989) The ecology and targets. Molecular Ecology 14, 819–827. evolution of intraguild predation: potential competitors Harwood, J.D., Sunderland, K.D. & Symondson, W.O.C. (2004) that eat each other. Annual Review of Ecology and Systematics Prey selection by linyphiid spiders: molecular tracking of 20, 297–330. the effects of alternative prey on rates of aphid consump- Pyke, G.H. (1984) Optimal foraging theory: a critical review. tion in the field. Molecular Ecology 13, 3549–3560. Annual Review of Ecology and Systematics 15, 523–575. Holland, J.M. (2004) The environmental consequences of Rypstra, A.L. & Marshall, S.D. (2005) Augmentation of soil adopting conservation tillage in Europe: reviewing the detritus affects the spider community and herbivory in a evidence. Agriculture, Ecosystems and Environment 103, 1–25. soybean agroecosystem. Entomologia Experimentalis et Appli- Holland, J.M. & Reynolds, C.J.M. (2003) The impact of soil cata 116, 149–157. cultivation on (Coleoptera and Araneae) emer- Settle, W.H., Ariawan, H., Astuti, E.T., Cahyana, W., Hakim, gence on arable land. Pedobiologia 47, 181–191. A.L., Hindayana, D., Lestari, A.S. & Sartanto, P. (1996) Holt, R.D. & Lawton, J.H. (1994) The ecological consequences of Managing tropical rice pests through conservation of shared natural enemies. Annual Review of Ecology and generalist natural enemies and alternative prey. Ecology Systematics 25, 495–520. 77, 1975–1988. Jones, A.A. & Selman, B.J. (1985) Microsporidium novacastriensis Sheppard, S.K. & Harwood, J.D. (2005) Advances in molecular n. sp., a mircosporidian parasite of the grey field slug, ecology: tracking trophic links through predator–prey food Deroceras reticulatum. Journal of Protozoology 34, 581–586. webs. Functional Ecology 19, 751–762. Kendall, D.A., Chinn, N.E., Glen, D.M., Wiltshire, C.W., Sims, R.W. & Gerard, B.M. (1985) Earthworms. London, E.J. Winstone, L. & Tidbold, C. (1995) Effects of soil manage- Brill/Dr W. Backhuys. Alternative prey reduce predation on pests 645

Stephens, D.W. & Krebs, J.R. (1986) Foraging theory. Princeton, Symondson, W.O.C., Sunderland, K.D. & Greenstone, M.H. Princeton University Press. (2002a) Can generalist predators be effective biocontrol Stinner, B.R. & House, G.J. (1990) Arthropods and other agents? Annual Review of Entomology 47, 561–594. invertebrates in conservation tillage agriculture. Annual Symondson, W.O.C., Glen, D.M., Ives, A.R., Langdon, C.J. & Review of Entomology 35, 299–318. Wiltshire, C.W. (2002b) Dynamics of the relationship Sunderland, K.D. (1975) The diet of some predatory arthropods between a generalist predator and slugs over five years. in cereal crops. Journal of Applied Ecology 17, 389–396. Ecology 83, 137–147. Sunderland, K.D. (2002) Invertebrate pest control by carabids. Thomas, C.F.G., Parkinson, L. & Marshall, E.J.P. (1998) pp. 165–214 in Holland, J.M. (Ed.) The agroecology of carabid Isolating the components of activity-density for the carabid beetles. Andover, Intercept. beetle Pterostichus melanarius in farmland. Oecologia 116, Sunderland, K.D., Powell, W. & Symondson, W.O.C. (2005) 103–112. Populations and communities. pp. 299–434 in Jervis, M.A. Thomas, R.S. (2002) An immunological and behavioural study of the (Ed.) as natural enemies: a practical perspective. Berlin, role of carabid beetle larvae as pest control agents in cereal crops. Springer. PhD thesis, Cardiff University, Cardiff, UK. Symondson, W.O.C. (1993) Chemical confinement of slugs: an Thorbeck, P. & Bilde, T. (2004) Reduced numbers of generalist alternative to electric fences. Journal of Molluscan Studies 59, arthropod predators after crop management. Journal of 259–261. Applied Ecology 41, 526–538. Symondson, W.O.C. (1997) Does Tandonia budapestensis Tod, M.E. (1973) Notes on beetle predators of molluscs. (: ) contain toxins? Evidence from Entomologist 106, 196–201. feeding trials with the slug predator Pterostichus melanarius Toft, S. & Wise, D.H. (1999) Growth, development and survival (Coleoptera: Carabidae). Journal of Molluscan Studies 7, of a generalist predators fed single- and mixed-species diets 457–465. of different quality. Oecologia 119, 191–197. Symondson, W.O.C. (2002) Molecular identification of prey in Wilson, M.J., Glen, D.M. & George, S.K. (1993) The rhabditid predator diets. Molecular Ecology 11, 627–641. nematode Phasmarhabditis hermaphrodita as a potential Symondson, W.O.C. (2004) Coleoptera (Carabidae, Drilidae, biological control agent for slugs. Biocontrol Science and Lampyridae and Staphylinidae) as predators of terrestrial Technology 3, 503–511. gastropods. pp. 37–84 in Barker, G.M. (Ed.) Natural enemies Winder, L., Alexander, C.J., Holland, J.M., Woolley, C. & Perry, of terrestrial molluscs. Wallingford, Oxon, CAB International. J. (2005a) Modelling the dynamic spatio-temporal response Symondson, W.O.C., Glen, D.M., Wiltshire, C.W, Langdon, of predators to transient prey patches in the field. Ecology C.J. & Liddell, J.E. (1996) Effects of cultivation techniques Letters 4, 568–576. and methods of straw disposal on predation by Pterostichus Winder, L., Alexander, C.J., Holland, J.M., Symondson, melanarius (Coleoptera: Carabidae) upon slugs (Gastro- W.O.C., Perry, J. & Woolley, C. (2005b) Predatory activity poda: Pulmonata) in an arable field. Journal of Applied and spatial pattern: the response of generalist carabids to Ecology 33, 741–753. their aphid prey. Journal of Animal Ecology 74, 443–454. Symondson, W.O.C., Glen, D.M., Erickson, M.L., Liddell, J.E. & Langdon, C.J. (2000) Do earthworms help to sustain the slug predator Pterostichus melanarius (Coleoptera: Carabidae) within crops? Investigations using a monoclonal antibody-based detection system. Molecular (Accepted 10 August 2006) Ecology 9, 1279–1292. Ó CAB International, 2006