1999. The Journal ofArachnology 27:249-254 THE EFFECTS OF DIFFERENT RATES OF THE HERBICIDE GLYPHOSATE ON SPIDERS IN ARABLE FIELD MARGINS Alison J. Haughton‘, James R. Bell-, Nigel D. Boatman^ and Andy Wilcox*: 'Crop & Environment Research Centre, Harper Adams University College, Newport, Shropshire TFIO 8NB UK; ^Department of Environmental & Geographical Sciences, Manchester Metropolitan University, Chester St., Manchester Ml 5GD. UK; ^Allerton & Educational Research Trust, Loddington House, Loddington, Leicestershire LE7 9XE. UK ABSTRACT. Field margins are susceptible to agro-chemical spray drift, and the effects ofherbicideon spiders in semi-natural habitats have been little studied. In this experiment, an arable field margin was & sprayed with three rates of glyphosate (90 g active ingredient/hectare (a.i/ha), 180 g a.i./ha 360 g a.i./ ha) and control plots left unsprayed. Spiders were sampled monthly (June-October) using a converted garden-vac and adult spiders were identified to species. A total of 23,393 spiders was sampled with the web-spinners representing more than 90% ofthe individuals. The effects ofglyphosate application on the abundance ofwandering and web-spinning prey-capture guilds, andthetwo mostabundant species {Gona- tium rubensandLepthyphantes tenuis) wereanalyzedusingANOVAFtests. Thehighestrateofglyphosate consistently reduced the total number of spiders, the numbers of web-spinners, G. rubens and L. tenuis, but not numbers of wandering spiders. Changes in vegetation structure and microclimate caused by the glyphosate are implicated in the reduction of numbers of spiders in plots receiving the highest rate of glyphosate. We conclude that glyphosate drift at rates of more than 360 g a.i./ha (active ingredients per hectare) into arable field margins could result in significant losses of important arthropod predators in farmland and a reduction in spider biodiversity in agroecosystems. In the United Kingdom, arable field mar- likely to increase (Mueller & Womac 1997). gins commonly comprise a boundary (hedge, Research into the optimum width of buffer fence, wall, or ditch) and a grass-dominated zones for reducing spray drift into sensitive boundary strip, and these constituent parts areas has recommended margins in the order m have been shown to be beneficial in enhancing of 6 wide for reduction of the most toxic flora, mammals, game birds and insects on ar- effects of various pesticides (Marrs et al. able farmland (e.g., Boatman 1994). Arable 1992; de Snoo 1997). field margins are important as overwintering Although the impact of insecticides on spi- sites (Bayram & Luff 1993), permanent hab- der behavior (Samu & Vollrath 1992) and itats (Alderweireldt 1994a) and refuges forre- mortality (Everts et al. 1989) has been studied, covery (Thomas et al. 1991) for spiders in the the effects of herbicide contamination on spi- agroecosystem. Not only do arable field mar- ders remain little-researched (Raatikainen & gins increase the opportunity for enhancing Huhta 1968; Asteraki et al. 1992). Spiders are spiders as predators (Alderweireldt 1994a), sensitive to changes in vegetation structure, but they are able to increase spider-biodiver- where a highly variable structure provides sity within biologically-impoverished arable web-spinners with increased web-site oppor- land (Duelli et al. 1990). tunities. Availability of structural support for Field margins are susceptible to direct her- webs and a suitable micro-climate (ameliorat- bicide applications (Boatman 1989) and also ed fluctuations in humidity and temperature) to spray drift by the virtue of their proximity are the most important factors in web site se- to high-input cropped areas. Glyphosate is a lection (Samu et al. 1996). The intrinsic action commonly used herbicide and with the devel- of herbicide on plants alters both the vegeta- opment ofherbicide resistant crops, the use of tion structure and therefore microclimate con- non-selective herbicides like glyphosate is ditions, and so it is likely that changes in spi- 249 — — 250 THE JOURNAL OF ARACHNOLOGY — der fauna would occur when a habitat is Statistical analysis. Total spider abun- exposed to a herbicide application. Here, we dance data and prey-capturing guild data were subjected an arable field margin to a herbicide log(x + 1) transformed while spider species application to establish whetherrelative spider abundance data were square-root (x + 0.5) population, prey-capture guild and number of transformed. Two-way univariate repeated individuals of abundant species were affected. measures ANOVAs were used to test for dif- ferences in mean number of spiders between METHODS — treatments because the samples of spiders A Site. well established arable field mar- through the season could not be considered to gin was selected on the Allerton Research and be independent of each other (Von Ende Educational Trust’s Loddington Estate in Lei- 1993). Where an interaction between treat- cestershire, UK. The field margin was domi- ment and date existed, indicating that the ef- nated by couch grass (Elymus repens (L.)) and fect of treatment differed between dates, a false oat grass (Arrhenatherum elatius (L.)) one-way univariate ANOVA was used to test and lay adjacent to a dense uncut hawthorn for differences in mean number of spiders be- {Crataegus monogyna Jacq.) and blackthorn tween treatments in each month. Planned (Prunus spinosa L.) hedge. The field margin comparisons were used to test for differences was east-south-east facing on slightly stoney implicit in the experimental design: we used clay soils from the Hanslope Series and the Least SignificantDifference (LSD) tests to de- field was sown to winter barley (cultivar: termine differences between means (Sokal & Fighter). Rohlf 1995). Treatments. Eight replicates of four RESULTS treatments (90 g active ingredient/hectare (a.i./ ha), 180 g a.i./ha & 360 g a.i./ha glyphosate A total of 23,393 spiders from 11 families and control) were randomly applied to adja- and 67 species was recorded and the dominant m cent field margin plots, which measured 12 family was the Linyphiidae. Specimens have X 1 m. The glyphosate (Roundup Biactive®, been deposited at the Liverpool Museum, UK. Monsanto) was applied to the plots at a vol- Pre-treatment. Spider abundance did not ume rate of 200 liters/ha and a pressure of 25 differ between plots prior to treatment appli- bars using an Oxford Precision sprayer on 30 cation (ANOVA F( = 1-31; P = 0.2901). May 1997. We therefore consid3e,r28e)d the plots to be similar Sampling.—“Spiders were sampled using a in spider fauna composition and proceeded modifiedgarden-vac (g-vac) (RyobiRSV3100E). with analysis. As a relatively new arthropod sampling de- Total spider abundance.-Two-way re- ANOVA vice, the g-vac has received critical attention. peated measures indicated that there Its sampling efficiency has been reviewed and was a significant date X treatment interaction the machine used in this experiment has been (^(15. 140) ^ 2.69; P < 0.0013), so we analyzed considered to be an effective method of sam- data from individual months. Total abundance pling spiders (Samu et al. 1997). The g-vac of spiders was only significantly different be- samples comprised 10 sub-samples of 30 sec- tween treatments in September (one-way AN- ond ‘sucks’ at 1 m intervals along each ex- OVA F(3,28) = 4.01; P < 0.0171), where sig- perimental plot. This approximated to a total nificantly fewer spiders were found in the 360 sampling area per plot of 0.13 m^. The inver- g a.i./ha treatment than in all other treatments tebrate samples were cooled immediately and (Table 1). — then extracted with an aspirator into 70% al- Prey-capture guilds. Web-spinning adult cohol before being identified. All adult spiders spiders from the Tetragnathidae, Theridiidae were identified to species, whilst immatures and Linyphiidae and wandering adult spiders were included in total number of spiders. from the Thomisidae, Clubionidae, Pisauridae, Spiders were sampled prior to the herbicide Zoridae, Oonopidae and Lycosidae were application to confirm that plots did not sup- grouped to investigate the treatment effects on port different abundances of spiders. Spiders these two prey-capture guilds. Table 2 shows were then sampled two weeks post-herbicide the mean number ofindividuals from the fam- application and monthly thereafter. Spiders ilies in the two guilds in each of the treat- were sampled from June to October inclusive. ments. Web-spinning spiders were dominated HAUGHTON, ET AL.—HERBICIDE AND FIELD MARGIN SPIDERS 251 — Table 1. Mean total number ofspiders in treatments and LSD Pvalues fordifferences betweenmeans in all treatments and 360 g active ingredient/hectare (a.i./ha). Control 90 g a.i./ha 180 g a.i./ha 360 g a.i./ha mean 220.50 205.13 209.38 15—2.13 P <0.0037 <0.0174 <0.0136 by the Linyphiidae and were more abundant significantly more individuals (LSD P < than wandering spiders, where they represent- 0.0043; LSD P < 0.0072 for control and 90 ed more than 90% of individuals in these two g a.i./ha respectively) than the 360 g a.i./ha guilds. Wandering spiders were not found to treatment. differ between treatments (repeated measures Lepthyphantes tenuis (Blackwall 1852) ANOVA F(3,28) ^ 0-67; P < 0.5779). showed different abundances in different Two-way repeated measures ANOVA in- treatments in September and October (repeat- dicated that there was a significant treatment ed measures ANOVA “ 7.63; P < by date interaction (F(i2 i^) “ 2.61; P < 0.0007), where each of the28o)ther treatments 0.0042) for web-spinners,,so we analyzeddata had significantly more individuals than the from individual months. The number of web- 360 g a.i./ha treatment (Table 4). spinners was significantly different among DISCUSSION treatments in August, September and October, — where more spiders were found in the control General effect of treatment. Applica- plots than in the 360 g a.i./ha in September tions of glyphosate at 360 g a.i./ha signifi- and October only (Table 3). cantly reduced the abundance of total spiders, Species data.—Only species which oc- web-spinners, Gonatium rubens and Lepthy- curred in sufficientnumbers (mean numberin- phantes tenuis, but not of wandering spiders. dividuals > 1.5 in each month) were analyzed The lower rates of herbicide had little or no individually. Only two linyphiid species ful- effect on the abundance of spiders per se; filled this criterion and showed significantly however, this study does not take into effect different mean abundances among treatments. possible changes in wandering and mating. Gonatium rubens (Blackwall 1833) showed The initial effects of the herbicide on the different abundances in different treatments total number of spiders and prey-capture (repeated measures ANOVA 4.41; P guilds were insignificant, but became more < 0,0116) in months August to October, profound as the season progressed. Therefore, where the control and 90 g a.i./ha plots had it is assumed that spiders are not affected di- rectly by glyphosate (which is generally non- — toxic to animals), but indirectly by modifica- Table 2. Mean number of individuals in each tions of other factors, such as habitat, prey family from treatments (June to October), a.i./ha = availability and microclimatic conditions. The active ingredient/hectare. time taken for the herbicide to act on vege- tation and change the habitat sufficiently for Con- 90 g 180 g 360 g spiders to exert preferences clearly takes trol a.i./ha a.i./ha a.i./ha months rather than weeks. Where such effects Wandering spiders — — are widespread, numbers of spiders may be Thomisidae 0.4 0.3 low in the following spring, which is a time Clubionidae —0.4 0.4 —0.3 —0.4 when spiders are a determining factorin aphid Pisauridae — 0.3 population dynamics in wheat crops (Coc- Zoridae — —0.1 0.1 —0.1 quempot & Chambon 1990). Thus, our single Oonopidae 0.1 season study indicates that the longer term ef- Lycosidae 2.6 2.4 1.8 4.3 fects of herbicide on spiders as biocontrol Web spinners agents and spider species diversity in agroe- Tetragnathidae 0.8 0.6 0.8 1.3 cosystems are of concer—n. Theridiidae 6.8 6.3 10.9 7.3 Wandering spiders. The highest rate of Linyphiidae 29.2 27.3 32.7 24.1 herbicide did not significantly reduce the 252 THE JOURNAL OF ARACHNOLOGY — Table 3. Comparison of mean number ofweb-spinners between different treatments & LSD P values for differences between means in all treatments and 360 g active ingredient/hectare (a.i./ha). August September October Treatment mean P mean P mean P control 10.00 ns 28.0 <0.0007 85.25 <0.0061 90 g a.i./ha 8.25 ns 24.75 <0.0059 82.13 <0.0159 180 g ai./ha 19.25 <0.—0226 29.86 <0.—0004 82.50 <0.—0230 360 g a.i./ha 7.25 16.38 66.50 number of wandering spiders. Wandering spi- to have preferences for specific prey type (Al- ders generally contain few examples of sten- derweireldt 1994b). Many web-spinning spi- ophages (Nentwig 1986) and they may be ders, therefore, may not utilize sub-standard more adept at finding suitable food items in habitat with a poor prey availability, since disturbed habitats due to their prey-capture they invest energy in web-building (Uetz strategy (Young & Edwards 1990). Thus, a 1991). The web-spinners in this study repre- combination of feeding strategy and an avail- sented the dominant prey-capture guild and able diverse prey source may not have suffi- indicated that higher levels of herbicide re- ciently deterred the wandering spiders from sulted in unfavorable habitat. Such losses of using the herbicide treated plots. important farmland spiders from herbicide Vegetation structure can influence not only misapplications could be significant in terms wandering spider prey recognition (Rovner of conservation of spiders in agroecosystems & 1980) but also mate detection (Uetz Strat- and in enhancing spiders as predators. ton 1982). The indirect effects ofherbicide on Gonatium rubens: This linyphiid is a litter the ability of spiders to detect mates was not species (McFerran et al. 1994) and it showed recorded, and we suggest that long-term ex- a preference away from heavily sprayed plots. periments should concentrate on mating suc- Although autecological literature about G. cess and feeding ability to investigate any cor- rubens is sparse, as a web-spinning spider it relations with herb—icide use. has similar habitat requirements as those out- Web-spinners. The action of herbicide lined above. It must be concluded that all, or on vegetation results in sparse cover and re- a combination of, abundance of web-building duced vegetation height (Raatikainen & Huhta sites, availability of prey and level of humid- 1968) as plants lose their vigor. Web-spinning ity were sub-standard. spiders rely on vegetation structure to provide Lepthyphantes tenuis: The most abundant both web-attachment sites and appropriate hu- spider in the British agroecosystem is L. ten- midity (Greenstone 1984; Young & Edwards uis (Topping & Lovei 1997). This linyphiid & 1990; White Hassall 1994). Unlike wan- builds webs at 10 cm above the ground and dering spiders, web-spinning linyphiids tend is completely dependent upon web-building for prey (Alderweireldt 1994b). As vegetation — height is reduced under exposure to herbicide LthSyTDpahbaFlnetve4as.luetesCnoufmiopsradbrieiftfsweoreneennocfesdmibfefeaetrnweennteunmtrbmeeearatnmoesfntiLsneap&l-l b(uRialadtiinkgaihneeingh&t fHoruhLt.ate1n9u6i8s)m,atyhebeidceoamlewedibs-- placed to a height with reduced humidity. treatments and 360 g active ingredient/hectare (a.i./ Aphids form a large part of the diet ofL. ten- ha). uis (Alderweireldt 1994b) and the spider can September & October reduce the aphid {Rhopalosiphum pad! L.) population on wheat plants by 34% (Maesour Treatment mean P & Heimbach 1993). Thus, Lepthyphantes ten- control 34.38 <0.0002 uis is an important predator in farmland and 90 g a.i./ha 34.19 <0.0005 reductions caused by herbicide applications 180 g a.i./ha 31.75 <0.—0072 should be considered against the benefits of 360 g a.i./ha 24.38 biocontrol. HAUGHTON, ET AL.—HERBICIDE AND FIELD MARGIN SPIDERS 253 — Conclusions. Herbicide applications at Everts, J.W., B. Aukema, R. Hengeveld & J.H. higher rates reduce the abundance of impor- Koeman. 1989. Side-effects of pesticides on tant predators. Field margins, which are val- ground-dwelling predatory arthropods in arable ued as refuges for farmland spiders during ecosystems. Environ. Poll., 59:203-225. winter and periods of disturbance, are suscep- Greenstone, M.H. 1984. Determinants ofweb spi- tible to herbicide spray drift and may suffer der species diversity: vegetation structural diver- sity vs. prey availability. Oecologia,62:299-304. losses in spider fauna. Reduced herbicide use McFerran, D.M., W.I. Montgomery & J.H. Mc- ianndanedlsenwehaerrefie(lYdoumnagrg&insEdiwsarsudgsge1s9t9e0d)haesrea mAudnaimt.ies1o9f94g.rouTnhde-diwmeplalcitngosfpigdrearzsi(nAgraonneaceo)mi-n way ofenhancing spider populations in agroe- upland vegetationtypes. Proc. Royal IrishAcad., cosystems not only for biocontrol but also for 94B:119-126. conservation of spider biodiversity. Mansour, F. & U. Heimbach. 1993. Evaluation of lycosid, micryphantid and linyphiid spiders as ACKNOWLEDGMENTS predators ofRhopalosiphumpadi (Horn: Aphid- We wish to thank the farm manager at the idae) and their functional response to prey den- Loddington Estate, Phil Jarvis for access to sity - Laboratory experiments. Entomophaga, 38: 79-87. tsihteyfioefld Osixtfeoarndd fPoarulsJtaothinstsiocanlatadtvhieceU.nivAe.rJ-. Marrs, R.H., A.J. Frost, R.A. Plant & P. Lunnis. 1992. The effects ofherbicide drift on semi-nat- Haughton was funded by the Higher Educa- ural vegetation: the use of buffer zones to mini- tion Funding Council for England. mize risks. Aspects of Appl. Biol., 29:57-64. LITERATURE CITED Mueller, T.C. & A.R. Womac. 1997. Effect offor- mulation and nozzle type on droplet size with Alderweireldt, M. 1994a. Habitat manipulations isopropylamine and trimesium salts of glyphos- increasing spider densities in agroecosystems: ate. Weed TechnoL, 11:639-643. possibilities for biological control? J. Appl. En- Nentwig, W. 1986. Non-webbuilding spiders: prey tomol., 118:10-16. specialists or generalists? Oecologia, 69:571- Alderweireldt, M. 1994b. Prey selection and prey 576. capture strategies oflinyphiid spiders in high in- Raatikainen, M. & V. Huhta. 1968. On the spider put agricultural fields. Bull. British Arachnol. fauna of Finnish oat fields. Ann. Zool. Fennica, Soc., 9:300-308. 5:254-261. Asteraki, E.J., C.B. Hanks & R.O. Clements. 1992. Rovner, J.S. 1980. Vibration in Heteropoda vena- The impact of the chemical removal of the toria (Sparassidae): athird method ofsoundpro- hedge-base flora on the community structure of duction in spiders. J. Arachnol., 8:193-200. carabid beetles (Col., Carabidae) and spiders Samu, F. & E Vollrath. 1992. Spider orb web as (Araneae) ofthe field andhedgebottom. J. Appl. bioassay forpesticide sideeffects. Entomol. Exp. Entomol., 113:398-406. Bayram, A. & M.L. Luff. 1993. Winter abundance SamAupp,.,E,62K:.1D1.7-1S2u4n.derland, C.J. Topping & J.S. and diversity of Lycosids (Lycosidae, Araneae) A Fenlon. 1996. spider population in flux: se- and otherspiders in grass tussocks in afieldmar- lection and abandonment of artificial web-sites gin. Pedobiologia, 37:357-364. Boatman, N.D. 1989. Selective weed control in and the importance of intraspecific interactions in Lepthyphantes tenuis (Araneae: Linyphiidae) field margins. Brighton Crop Protection Confer- Boaetnmcaen-,WNe.eDd.s,(e1d9.)8.9:7189594-.794F.ield Margins: Inte- Saminu,whEe,atJ.. NOeecmoeltohgi&a,B1.06K:is2s2.8-2193997.. Assessment grating Agriculture and Conservation. BCPC of the efficiency of a hand-held suction device Famham, UK. for sampling spiders: improved density estima- Cocquempot, C. & J-P. Chambon. 1990. L’activite tion or oversampling? Ann. Appl. Biol., 130: des Araignees et son incidence sur la limitation 371-378. & des populations de pucerons des biocenoses cer- Sokal, R.R. EJ. Rohlf. 1995. Biometry. 3rd ed. ealieres. Rev. Ecol. Sol., 27:205-219. W.H. Freeman & Co., New York. De Snoo, G.R. 1997. Arable flora in sprayed and Thomas, M.B., S.D. Wratten & N.W Sotherton. unsprayed crop edges. Agric. Ecosystems and 1991. Creation of ‘island’ habitats in farmland Environ., 66:223-230. to manipulate populations of beneficial arthro- Duelli, R, M. Studer, I. Marchand & S. Jacob. pods: predator densities and emigration. J. Appl. 1990. Population movements of arthropods be- EcoL, 28:906-917. tween natural and cultivated areas. Biol. Con- Topping, C.J. & G.L. Lovei. 1997. Spider density serv., 54:193-207. and diversity in relation to disturbance in agroe- 254 THE JOURNAL OF ARACHNOLOGY cosystems in New Zealand with a comparison to VonEnde, C.N. 1993. Repeatedmeasuresanalysis: England. New Zealand J. EcoL, 21:121-128. growth and other time-dependent measures. Pp. Uetz, G.W. 1991. Habitat stracture and spider for- 113-137. In Design and Analysis of Ecological aging. Pp. 325-348. In Habitat Stracture: The Experiments. (S.M. Scheiner & J. Gurevitch, Physical Arrangement ofObjects in Space. (S.S. eds.). Chapman & Hall, London. Bell, E.D. McCoy & H.R. Muschinsky, eds.). White, P.C.L. & M. HassalL 1994. Effects ofman- Chapman & Hall, London. agement on spider communities of headlands in Uetmzu,niGc.aWt.i,on&aGn.dE.reSptrroadtutcont.ive19i8s2o.latAicoonusintiscpicdeorms-. Youcnerge,alO.fiPe.lds&. PGe.dBo.bioEldowgairad,s.38:1196990-.184S.piders in Pp. 123-159. In Spider Communication: Mech- UnitedStates fieldcropsandtheirpotentialeffect anisms and Ecological Significance. (P.N. Witt& on crop pests, J. ArachnoL, 18:1-27. J.S. Rovner, eds,). Princeton Univ. Press, Prince- Manuscript received29April 1998, revised 11 No- ton, New Jersey. vember 1998.