International Association for Ecology

Dragonfly Predators Influence Biomass and Density of Pond Snails Author(s): Andrew M. Turner and Michael F. Chislock Reviewed work(s): Source: Oecologia, Vol. 153, No. 2 (Aug., 2007), pp. 407-415 Published by: Springer in cooperation with International Association for Ecology Stable URL: http://www.jstor.org/stable/40210876 . Accessed: 07/03/2012 13:17

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Springer and International Association for Ecology are collaborating with JSTOR to digitize, preserve and extend access to Oecologia.

http://www.jstor.org Oecologia (2007) 153:407-415 DOI 10.1007/s00442-007-0736-9

COMMUNITYECOLOGY

Dragonfly predators influence biomass and density of pond snails

Andrew M. Turner • Michael F. Chislock

Received: 7 October 2006 /Accepted: 20 March 2007 /Published online: 25 April 2007 © Springer-Verlag 2007

Abstract Studies in lakes show that fish and crayfish not have a strong effect on species composition. A model predatorsplay an importantrole in determiningthe abun- suggested that A. junius and P. hymenaeahave the largest dance of freshwatersnails. In contrast,there are few studies effects on snail biomass in the field. Given that both pul- of snails and their predatorsin shallow ponds and marshes. monate snails and dragonfly nymphs are widespread and Ponds often lack fish and crayfishbut have abundantinsect abundantin marshesand ponds, snail assemblagesin these populations. Here we present the results of field surveys, water bodies are likely regulated in large part by odonate laboratory foraging trials, and an outdoor mesocosm predation. experiment, testing the hypothesis that are impor- • • • tant predatorsof pulmonate snails. In laboratoryforaging Keywords Food webs Predation Gastropoda trials, conducted with ten species of insects, most • Snail taxa consumed snails, and larval were espe- cially effective predators.The field surveys showed that dragonfliesconstitute the majorityof the insect biomass in Introduction fishless ponds. More focused foraging trials evaluated the ability of the dragonfliesAnax junius and hyme- Ecologists seek to identify mechanisms that regulate the naea to prey upon different sizes and species of pulmonate distributionand abundanceof species along environmental snails (Helisoma trivolvis, Physa acuta, and Stagnicola gradients.A key gradientin freshwateris the environmental elodes). Anax junius consumed all three species up to the continuumfrom small, temporaryponds and marshes largely maximum size tested. consumed snails lacking predatorsto deep, permanentlakes containing fish with a shell height of 3 mm and smaller, but did not kill (Wellborn et al. 1996). Freshwater gastropods are key larger snails. P. acuta were more vulnerable to predators players in many freshwaterfood webs, and given suitable than were H. trivolvis or S. elodes. In the mesocosm abiotic conditions are ubiquitousinhabitants of lentic sys- experiment, conducted with predatortreatments of A. ju- tems spanningthe habitatgradient (Pip 1986; Jokinen 1987; nius, P. hymenaea, and the hemipteranBelostoma flumi- Dillon 2000). Lodge et al. (1987) argue that abiotic condi- neum, insect predatorshad a pronouncednegative effect on tions (Ca2+availability) limit gastropodsat a regional scale, snail biomass and density. A. junius and B. flumineumre- but biotic interactions determine snail distribution and duced biomass and density to a similar degree, and both abundanceat a local scale. A numberof experiments,con- reducedbiomass more than did P. hymenaea.Predators did ducted in deep permanentlakes, show that food limitation (Osenberg 1989) and predationby fish and crayfish (Bron- market al. 1992; Martinet al. 1992; Lodge et al. 1994) can Communicated by Jay Rosenheim. regulatesnail abundanceand determine the structureof snail assemblages. In contrast, there are few studies of snail Turner • M. F. Chislock A. M. (El) regulation and community structurein shallow of Clarion Clarion, population Department Biology, University, and marshes. Such habitats lack fish and PA 16214, USA ponds typically e-mail: [email protected] crayfish but have large insect populations,many of which

£l Springer 408 Oecologia(2007) 153:407^15 function as top predatorsin pond food webs (Zaret 1980; species of carnivorousaquatic insects in orderto determine McPeek 1990; Wellbornet al. 1996). if they were capable of feeding on snails. We chose to Larval dragonflies are particularlyabundant in fishless screen insects that were relatively large and commonly ponds and can play an importantrole in structuringpond encounteredin the ponds of northwestPennsylvania, USA. communities (McPeek 1990, 1998; Wissinger and McGr- Insect species screened were the dragonfliesAnax junius, ady 1993; Skelly 1996). Although a wide variety of ani- Pantala hymenaea, Pachydiplax longipennis, Tramealac- mals are known to prey on freshwatergastropods (reviewed erata, and Erythemis simplicicollis, the damselfly Lestes in Lodge et al. 1987; Brown 1991; Dillon 2000), there are vigilax, the water bugs Belostomaflumineum, Ranatra ni- no studies examining whether dragonflies prey on snails. gra and Pelecoris fermoratus, and the larvae of the beetle Here we present the results of laboratory experiments, Dytiscus verticalis. Prey in the screening study were Physa mesocosm experiments, and field surveys, testing the acuta between 3.0 and 6.0 mm shell height. Due to their hypothesis that dragonflies are important predators of small size and thin shell, this size class and species is pulmonate snails. Our larger goal is to evaluate the po- among the most vulnerableof the snails commonly found tential importance of predation in structuring the snail in ponds. Trials were conducted indoors in polypropylene assemblages of fishless ponds. shoeboxes (20 x 35 x 14 cm deep) filled to a depth of Our study focuses on three species of pulmonatesnails; 4 cm (2.8 1) with aged well water. We chose to use a rel- Physa acuta, Stagnicola elodes, and Helisoma trivolvis. P. atively small foraging arena so as to maximize predator- acuta and H. trivolvis are both widespreadand abundantin prey encounter rates and minimize the role of any behav- the ponds of northwest Pennsylvania (Mower and Turner ioral avoidance by snails. Fifteen snails and one insect 2004). S. elodes has a narrow habitat distribution,being predatorwere stocked into each containerand we recorded restrictedto shallow ponds with forest cover, but within snail survivorshipafter 48 h, along with the method of kill such ponds they are also quite abundant (Mower and (shell crushedor body extractedfrom shell). Four replicate Turner2004). All three snail species commonly co-occur trails were conducted for each predatortaxa. In addition, with the insect predators screened in this study. Of the for each set of four predatortrials we ran a no-predator three species, S. elodes is the largest and has a relatively control trial in orderto assurethat there was no background thick shell, whereas P. acuta is the smallest and has a mortalityin the absence of predators.The environmentwas relatively thin shell (Mower and Turner2004; A. Turner, controlled with a temperatureof 22°C and 14:10 h day:- unpublisheddata). Because these three snail species differ night lighting. in body size, degree of shell development, and anti-pred- ator behavior (Mower and Turner2004), we expected that Field survey their vulnerabilityto dragonflieswould also differ. We first screened ten species of carnivorousinsects in The impactof a particularspecies of insect predatoron snail order to determine which might be the most important assemblages will depend both on the feeding rates of indi- predatorsof pulmonate snails. Based on these results, and vidual predatorsand on the abundanceof the predator.We informedby the results of a field survey, we then chose for surveyed the biomass of predatoryinsects in 12 ponds in furtherstudy three species that likely are the most impor- orderto providean assessmentof the relative abundancesof tant snail predators:Pantala hymenaea, a libellulid drag- different sorts of insect predators.Sampled ponds were lo- onfly, Anax junius, an aeshnid dragonfly, and the cated in CrawfordCounty, Pennsylvania, USA. All 12 ponds hemipteranBelostoma flumineum. A. junius and B. flumi- were more or less permanentbut lacked well-developedfish neum are abundantin permanent,fishless ponds (Dunkle communities, though a couple contained scatteredfathead 1989; Kesler and Munns 1989; McPeek 1990; Chase 1999). minnows(Pimephales promelas). Insects were sampledwith P. hymenaea are largely restricted to and abundant in a stove-pipe-typesampler, 25 cm diameter,which has been ephemeralponds, but we view them as broadly represen- shown to perform well in sampling vegetated habitats tative of the many species of libellulid dragonfliesthat are (Turner and Trexler 1997). Eight replicate samples were especially abundantin fishless ponds. taken from randomly selected locations in each pond, and samples were pooled for furtheranalyses. While in the field we identified insects to the level of family and measured Materials and methods their wet mass. We used biomass as an index of abundance, and not number,because when comparingacross size clas- Insect screening ses or taxa, the biomass of a predatorspecies is a better predictorof its feeding rate (Peters 1983; Yodzis and Ennis In order to determine which insects may pose the highest 1992). Each pond was sampledtwice duringthe summer,in degree of risk to pulmonate snails, we first screened ten late May and again in late July, so as to providea seasonally

© Springer Oecologia (2007) 153:407-415 409 integrated view of insect abundance. Thus, the relative in springand ran into fall. In May we filled each mesocosm biomass of each insect family, calculatedfor each of the 24 with well water, fertilized the tanks at a rate of 50 jugP I"1 sampling events and then averaged across the two sample (added as K2HPO4, N was added as NaNO3 so as to dates and across the 12 ponds, was used as an index of their maintain a N:P molar ratio of 16:1), added 10 g dry red potentialimpact on snail populations. maple {Acer rubruni)leaves to provide physical structure and additional nutrients, and inoculated the tanks with Foragingtrials zooplanktonand phytoplanktoncollected from local ponds. Mesocosms were then covered with 50% shade cloth to Because dragonfliescapture prey by seizing it between the preventunwanted insects from colonizing the tanks. In late lobes of their prehensilelabium, we hypothesizedthat snail June, after algae and zooplanktonhad become established, body size would influence their vulnerabilityto odonates, an assemblage comprising three species of pulmonate with largersnails being less vulnerablethan smaller snails. snails (P. acuta, H. trivolvis, and 5. elodes) was stocked Therefore, we conducted foraging trials in shoebox con- into the tanks as adults (ten of each species for a total of 30 tainers in order to establish, for two dragonfly species snails) and allowed to oviposit. Adults deposited eggs for identified in the screening process and field survey as 15 days, at which time they were removed and the preda- potentially important snail predators (A. junius and P. tors (A. junius, P. Hymenaea, or B. flumineum) were hymenaea), which species and size classes of snails the stocked. Hatching time for all three species is approxi- dragonflieswere capable of killing. B. flumineumwas not mately 2 weeks, so initiation of hatching coincided with included in these trials because its ability to forage on predatorstocking. Zooplanktonwere also abundantin the snails has already been studied (Kesler and Munns 1989; mesocosms and provided an alternatefood source for the Chase 1999; Hovermanet al. 2005). Each trial included a odonates. In additionto no-predatorcontrols, we includeda single species (P. acuta, H. trivolvis, or S. elodes) and size treatmentof the giant water bug Belostomaflumineum, a class of snail. As with the predatorscreening trials, for- known predator of snails (Kesler and Munns 1989), in aging trials were conducted indoors in plastic shoeboxes order to establish a benchmarkagainst which to evaluate filled with 2.8 1 of aged well water. Temperature was the ferocity of the dragonfly predators.The three species maintainedat 22°C, and lighting was set at 14:10 h day:- were stocked at a density of two A. junius, four B. flumi- night. Snails were sorted by shell size into 1-mm size neum, or five P. hymenaea per tank, which was aimed at classes. For S. elodes and P. acuta we measured shell achieving a stocking rate of 2.5 g wet biomass per meso- height, for H. trivolvis we measured shell width. In each cosm. This stocking rate (1.25 g m"2)is about one-quarter case this dimension representsthe longest axis of the shell. of the average overall insect biomass measured in local Fifteen snails were stocked into each container, and for ponds. There were eight replicate mesocosms for each of each predatorx snail-species x size class combination, the four predatortreatments, yielding a total of 32 meso- five replicate trials were conducted. Four replicates re- cosms. ceived predators,and the fifth remained as a no-predator Mesocosms were periodically inspected, and dead or control. Survivorshipwas measuredafter 48 h, and method emerging predatorswere replaced. After 10 weeks, tanks of kill (shell crushedor body extracted)was also recorded. were drainedand all survivingsnails were recovered.None An individualdragonfly was not used more than once in a of the snails had begun reproduction at the time the trial for a particularsnail species, thoughthey were in some experimentwas terminated.We tallied the numberand dry cases used in predationtrials for other species. The effects mass of each species (dry mass includes shell material). of predatoridentity, prey identity, and prey body size on Upon draining the tanks we found that one no-predator prey mortality were assessed with a 2x3x7 factorial tank contained several P. hymenaea, and data from this ANOVA. Analyses were performedwith SPSS 11.0. tank were deleted from subsequentanalyses. The effects of predatortreatment on snail biomass, density and body size Mesocosm study were analyzed with one-way ANOVA, and pairwise dif- ferences among treatments were assessed with Tukey's We hypothesized that insect predationwould most affect test. Treatmenteffects on species composition were eval- survivorshipof juvenile snails and thus set out to evaluate uated by performingmultivariate ANOVA (MANOVA) on the effects of insects on snail recruitmentfrom egg to adult. the relative abundance matrix and the relative biomass The effect of insect predation on snail recruitment was matrix. Because the relative measuresof composition sum assessed in large outdoor mesocosms (1.6 m diameter, to 1, there are just two independentvariables and we per- 0.6 m deep, 1,200 1). Pulmonate snails generally have an formed MANOVA on vectors of two rather than three annual life history with peak reproductiveeffort in early species. We also evaluated treatmenteffects on evenness, summer (Dillon 2000), so our experiment was established using Simpson's formulation(Krebs 1989), by performing

£) Springer 410 Oecologia(2007) 153:407-415

ANOVA on the evenness of biomass and evenness of factors (Table 2). Most of the variationin mortality (86% abundances.All analyses were performedwith SPSS 11.0. of treatmentsums-of-squares) was attributable,however, to just two factors: the additive and non-additive effects of predatoridentity and prey body size (Table 2). Averaged Results across prey species and size classes, A. junius consumed 72% of the snails presentedto them (Fig. la), whereas P. Insect screening hymenaea consumed 17% of their snails (Fig. lb), a 4.2- fold difference. Prey size also played a key role in deter- Of the ten taxa of insects assayed, nine consumed P. acuta, mining vulnerability, as small snails were, on average, with only larvalD. verticalis (Dytiscidae) failing to kill any much more vulnerableto predatorsthan were larger snails snails. Feeding rates varied a great deal, however, among (Fig. 1). The snail size effect was largely dependent on the taxa that consumed snails (Table 1). The most lethal predatorspecies. Pantala hymenaeahad high feeding rates predatorswere A. junius, B. flumineum,E. simplicicollis, P. on the smallest size classes, but was largely unable to hymenaea, and T. lacerata. The remaining taxa killed consume snails >3 mm body length (Fig. lb). In contrast, fewer than 10% of the snails presentedto them (Table 1). A. junius readily consumed all size classes tested (Fig. la). Prey identity and the two-way interactions involving Field survey prey identity had small but significanteffects on mortality rates (Table 2). P. acuta was most vulnerable and H. tri- Survey data showed that large insects were quite abundant volvis the least vulnerableto predators(Fig. 2). Examina- in local ponds, with an overall average biomass of tion of prey identity x body size interactionplots suggests 4.91 g m~2.Libellulid dragonflieswere the most prevalent that prey vulnerabilityrankings were consistent for all size of the insect predators, with a biomass that comprised classes >3 mm shell height, but that prey identity did not 46.3% of the overall large-insect biomass (Table 1). influence mortality of snails in the 2-mm size class: all Aeshnid dragonflies were the next most abundantgroup were highly vulnerable, regardless of identity. The mor- (32% of total biomass), followed by belostomatids and tality imposed by A. junius depended on prey identity to a naucorids, with 8 and 7% of the total insect biomass, greater extent than did the mortality imposed by P. respectively (Table 1). hymenaea (Fig. 2). Dragonflies consumed a snail by extracting its body Foragingtrials from the intact shell or by crushing the shell. Averaged across predatorspecies, prey species, and size classes, 69% Dragonflies readily preyed on snails, but the feeding rate of all consumed snails were killed by extraction.Dragon- dependedon predatoridentity, prey identity, and prey body flies extracted snails from their shells by first seizing size, as well as all two-way interactions of these three the snail's foot with their labium and then using their

Table 1 Laboratoryassay of feeding rates3and abundanceof large carnivorousinsects in local ponds.SD in parentheses b Order Family LaboratoryAssay Field survey Taxa screened Body size (g) Consumed(%) Wet biomass(g m"2) Relativebiomass (%)

Coleoptera Dytiscidae Dytiscusverticalis 1.36 0 0.03(0.10) 0.6 Hemiptera Belostomatidae Belostomaflumineum 0.37 68(3) 0.41(0.46) 8.4 Hemiptera Nepidae Ranatranigra 0.19 5(3) 0.14(0.34) 2.9 Hemiptera Naucoridae Pelecorisfemoratus 0.06 3 (4) 0.35 (0.58) 7.2 Odonata Lestidae Lestes vigilax 0.13 3(4) 0.14(0.25) 2.8 Odonata Aeshnidae Anaxjunius 1.27 98(3) 1.56(1.67) 31.8 Odonata Tramealacerata 0.74 13 (5) 2.27 (2.83) 46.3 Odonata Libellulidae Pachydiplaxlongipennis 0.22 3 (4) Odonata Libellulidae Erythemissimplicicollis 0.29 55 (6) Odonata Libellulidae Pantala hymenaea 0.48 17 (7) a Consumptionrates are for late-instarinsects feeding on juvenile Physa acuta (3.0-6.0 mm shell height) as measuredin foragingtrails conductedin aquariaand run for 48 h b Field surveyresults are based on a surveyof 12 ponds,each sampledtwice. Surveydata are organized at the level of family,thus the biomass datafor the Aeshnidae,Lestidae, and Libellulidaerepresent all species in the family pooled together

£} Springer Oecologia (2007) 153:407-415 411

Table 2 Effects of predator identity (Dragonfly species), prey identity (Snail species), and prey body size (Shell height) on snail mortality rate in foraging trials, rj2Effect size

Factor n2a df F P

Dragonfly species 0.446 1 377.9 <0.001 Snail species 0.033 2 13.9 <0.001 Shell height 0.167 6 23.6 <0.001 Dragonfly species x Snail species 0.028 2 11.9 <0.001 Dragonfly species x Shell height 0.119 6 16.8 <0.001 Snail species x Shell height 0.035 12 2.5 0.006 Dragonfly species x Snail species x Shell height 0.023 12 1.7 0.084 a For each factor, rj2is the proportion of total variance attributable to the factor, and is calculated as the ratio of the effect variance to the total variance (rj2= SSfactOr/SStota, ). Model R2 = 0.85

100 100 -, % 1 a) Anaxjunius 80

* 60 J \J ^ 60 ±

40 o 40 20 20 $ $ § °^ ' ' ' o-l , . . | P. acuta S. elodes H. trivolvis 1 10°1 ux 2 -i- D) Pantala hymenaea Prey identity 80 - t Fig. 2 Effect of prey identity (Physa acuta, Stagnicola elodes and Helisoma trivolvis) on predation by A. junius (filled circle) and P. 60 \ hymenaea (open circle) as measured in laboratory feeding trials. Each point represents mean mortality (±1 SE, n = 28) suffered by each 40- V species over 48 h, and is averaged across size classes

20 K and significantlylower in all three predatortreatments than in no-predatorcontrols (ANOVA, FX21= 35.8, P < 0.001; 0 -I . . T^» • • • 2 3 4 5 6 7 8 all pairwise comparisons of predatortreatments with no- control at P < The effects of P. Snail shell height predator significant 0.05). hymenaea, A. junius and B. flumineum on snail biomass Fig. 1 Effect of snail body size (shell height, mm) on predation by a were not statistically distinguishable from one another as measured in Anax junius and b Pantala hymenaea laboratory 3a). Insect also depressed total snail abun- mean n = (Fig. predators feeding trials. Each point represents mortality (±1 SE, 12) = P = P. A. suffered each 1-mm size class over 48 h, and is averaged across dance (ANOVA, F3,27 4.60, 0.01). hymenaea, by effects three snail species junius, and B. flumineumeach had similar negative on snail abundance,as all three reduced snail density rel- not mouthpartsto pull the snail's body from its shell. The ative to no-predator controls but were significantly had a foraging mode employed by dragonfliesdepended on snail different from each other (Fig. 3b). Predators sig- body size, however, as 76% of the consumed snails in the nificant effect on mean body size of P. acuta (ANOVA, 2- and 3-mm size classes were killed by shell crushing,but ^3,27= 4.75, P < 0.01), with mean body mass in the more than 50% rel- only 2% of the consumed snails >3 mm were crushed. presence of B. flumineumreduced by ative to the no-predatorcontrol (Fig. 3c). Mean body size Mesocosm study of P. acuta in other predatortreatments was not signifi- cantly different from controls (P > 0.10), and predatoref- Insect predators had a strong negative effect on snail fects on body size of H. trivolvis and S. elodes were not recruitment,with biomass of recovered snails substantially significant (P > 0.10).

£} Springer 412 Oecologia(2007) 153:407-415

Discussion

Lodge et al. (1987) hypothesized that given adequateCa, snail assemblages in lakes will be regulated by fish and crayfish predators, whereas snails in shallow, fishless ponds will be structuredlargely by competition, and snail assemblages in temporary ponds will be limited by the frequency and severity of dessication. The hypothesis that predation by fish and crayfish limits snails residing in lakes has been tested repeatedly(e.g., Brown and DeVries 1985; Osenberg 1989; Weber and Lodge 1990; Bronmark et al. 1992; Martin et al. 1992; Osenberg et al. 1992; Lodge et al. 1994; Nystrom and Perez 1998; Lewis 2001), but the mechanisms structuringthe snail assemblages of fishless ponds remains little studied (but see Eisenberg 1966; Kesler and Munns 1989; Chase 2003). While admitting that there are very few data regarding the abundance and feeding rates of invertebrate predators, Lodge et al. (1987) argued that competition will pre- dominate in fishless ponds, as the available data suggest that invertebrate predators have low feeding rates on snails and thus are ineffective predators relative to fish and crayfish. Here we show that insects common in fishless ponds are effective predators of snails and can suppress the abun- dance of species with which they co-occur. Thus, these results do not supportthe hypothesis that predationwill be unimportantin governing the snail assemblages of fishless ponds. A number of insects fed on snails in laboratory feeding trials, with the anisopteranA. junius, the hemipt- eran B. flumineum, and the anisopteranE. simplicicollis the rates. Our mesocosm study Fig. 3 Mesocosm study showing effects of three insect predators on scoring highest predation snail a biomass, b abundance, and c body size. Bars represent showed, in a semi-natural setting and using a realistic treatment means (+1 SE), n = 8 replicate mesocosms per treatment, predator biomass, that odonates and B. flumineum can = except for no-predator, for which n 7. Lower case letters above bars suppress the recruitmentof snails with which they com- denote treatment means not significantly different from each other co-occur. Our data show that insects are both (P < 0.05, Tukey's test) monly survey widespread and abundant in the marshes and ponds of northwestern Pennsylvania. Thus, our data, especially when taken together with the results of Kesler and Munns H. trivolvis dominated the snail biomass of the meso- (1989), provide evidence that snail abundancesin fishless cosms, comprising on average 66% of the total biomass, ponds are likely governed in part by insect predation, and P. acuta was the most abundantspecies, comprising especially by dragonflies. 51% of the total snail abundance.There was, surprisingly, Other inhabitantsof fishless ponds known to be snail no significant effect of predator treatment on species predatorsinclude tiger salamanders(Ambystoma tigrinum; composition, measured either as patterns of relative bio- Brophy 1980; Benoy et al. 2002), red-spottednewts (Not- mass or relative abundance(MANOVA; relative biomass, ophthalmus viridescens; Burton 1976), turtles (Mahmoud Wilk's X = 0.686, F6,52= 1.80, P > 0.10; relative abun- 1968), sciomyzid fly larvae (Eckblad 1976), leeches dance, Wilk's X = 0.795, F6>52= 1.05, P > 0.10). An (Bronmark and Malmqvist 1986; Bronmark 1992), and analysis of Simpson's evenness, as calculated from the waterfowl (Krapuand Reinecke 1992). In the ponds of our relative biomass and relative abundance matrices, also region, these taxa are much less abundantthan insects and found no treatmenteffects (ANOVA; evenness of relative thus less importantas predators,but these studies do further biomasses, FX2i = 1.73, P > Q.10; evenness of relative support the general idea that predators can limit snail abundances,FXT1 = 130, P > 0.10). populations in fishless ponds.

& Springer Oecologia (2007) 153:407-415 413

In the mesocosm study, B. flumineum had a stronger results support the notion that anisopteran odonates effect on snail biomass thandid A. junius and P. hymenaea. (dragonflies,including aeshnids and libellulids), relative to Recall, however, that predatortreatments in the mesocosm other insect taxa, are the most importantpredators of snails study were matched for biomass. The degree to which in fishless ponds. different predatortaxa in a pond will each govern snail Insect predatorsin the mesocosm study had substantial biomass depends not just on differences in mass-specific effects on snail abundanceand biomass, but there was little feeding rates, but also on the biomass density of each evidence that predatorscaused a shift in species composi- predatortaxa (Peters 1983; Yodzis and Ennis 1992). We tion. Although high variability in snail abundanceamong can generatean index of predatorimpact on snail biomass, tanks may have limited our power to detect subtle effects, it and compare the importanceof the three taxa tested here, is clear that insect predatorsdid not induce a large shift in by constructing a simple model incorporatingour field species composition. Thus, in the relatively naturalenvi- survey data. Let the mass specific interaction strength of ronment of outdoor mesocosms the three species are of predator(/) be defined as the additive reduction in snail similar vulnerabilityto predators.This result contrastswith biomass as observed in the mesocosm study, standardized the laboratory foraging trials where P. acuta was more by predatorbiomass (2.5 g in all cases): vulnerableto predatorsthat were H. trivolvis and S. elodes. The small spatial and temporal scale of the laboratory - a( = (snail biomass, no predatortrt.(g) snail biomass, foraging trials are designed to maximize predator-prey predatortrt i(g)) / predatorbiomass(g) encounters and preclude the use of inducible defenses, so the trials measure the of to (i) predation ability predators capture and consume prey. In contrast, the larger spatial at is a dimensionless coefficient representingthe reduction scale, physical complexity, and temporal scale of the in snail biomass per unit predatorbiomass added. Predator mesocosm study allow for the employment of behavioral impact,the predictedreduction of snail biomass in the field defenses which lower encounter rates with predatorsand (g m~2)attributable to predatortaxa (/), is then: morphological trait shifts which lower vulnerability to predators(e.g., DeWitt et al. 2000; Turnerand Montgom- • Hovermanet al. Turneret al. rt = {Biomassdensity predator taxa i(gm~2)) at. (2) ery 2003; 2005; 2006). Insect predators are patchy at the landscape scale, Because the mesocosms differ from ponds in a number reachingtheir greatestabundance in shallow, fishless ponds of ways, including initial snail biomass, r, is an index of and marshes (McPeek 1990; Skelly 1996). Their role in predatorimpact and not a quantitativeestimate of predator structuringsnail assemblagesat large spatialscales remains impact. This analysis suggests that aeshnid dragonfliesare an open question as the snail species used in our study are the most importantof the three predatorsconsidered, with those that most typically coexist with insect predatorsin r, = 3.94 g snails m"2(Table 3). Libellulids are the second small ponds and marshes. However, most of the 30-odd most importantpredator taxa, with a predicted predator gastropodsspecies found in northwesternPennsylvania do impact 85% as large as that of the aeshnids (Table 3). not occur in these ponds and their vulnerabilityto insect Belostomatids, despite having the highest mass specific predatorsis not yet known. Studies of predationin lentic interactionstrength (ah) have a predicted predatorimpact systems generally find that regardlessof a predator'sshort- just 29% that of the aeshnid rh due to the lower abundance term effects on coexisting prey, predators can shape of belostomatidsin the field survey (Table 3). Thus, these among-lakedistribution of prey throughlocal exclusion of vulnerable prey species, thereby restricting community membership(Wellborn et al. 1996). Thus, it is possible that odonates play an important role in determining species Table 3 Relative impact of three predator taxa on snail biomass. composition of snails in ponds by excluding more vulner- Mass interaction of i as measured in the at specific strength predator able species, but this hypothesis has not been tested. mesocosm e.g., the additive reduction in snail biomass experiment, et al. out that (g) per gram of predator biomass added to mesocoms; r, predator Twenty years ago, Lodge (1987) pointed impact- the predicted reduction of snail biomass in the field (g m"2), we have little understanding of the factors governing taking into account predator biomass measured in field surveys, among-lake variation in snail species composition and attributable to i predator abundance,and made the valuable suggestion that it would Taxa a, n be profitableto examine how regulatingmechanisms shift across the lake size work with other Aeshnidae 2.52 3.94 gradient. Subsequent taxa and Libellulidae 1.48 3.36 aquatic (e.g., amphipods, amphibians, odonates) such an and showed that there is a Belostomatidae 2.82 1.16 adopted approach predictablepattern of species replacementalong the lake-

^ Springer 414 Oecologia (2007) 153:407^15 size continuum,driven in large partby shifts in the identity Dunkle SW (1989) Dragonflies of the Florida Peninsula, Bermuda, Fla. of the top predatorand tradeoffsinvolving traitsconferring and the Bahamas. Scientific Publishers, Gainsville, Eckblad J (1976) Biomass and energy transfer by a specialized (Wellborn 1994; 1997; safety against predators Skelly predator of aquatic snails. Freshwater Biol 6:19-21 McPeek 1998; McPeek et al. 2001). Behavioral, physio- Eisenberg RM (1966) The regulation of density in a natural logical, or morphological traits that confer protection population of the pond snail, Lymnaea elodes. Ecology against the predatorsof larger ponds and lakes often incur 47:889-906 Hoverman Auld RA back costs that the bearerof those traits at a in JT, JR, Relyea (2005) Putting prey together put disadvantage again: integrating predator-induced behavior, morphology, and ponds lacking predatorsor with differentsorts of predators life history. Oecologia 144:481-491 (e.g., Wissinger et al. 1999; Wellborn 2002; McPeek Jokinen E (1987) Structure of freshwater snail communities: species- Am Malacol Bull 5: 2004). Unfortunately,less is known about the role of such area relationships and incidence categories. 9-19 mechanisms in Our structuring gastropod assemblages. Kesler DH, Munns WR Jr (1989) Predation by Belastoma flumineum results suggest that predation is an important limiting (Hemiptera): an important cause of mortality in freshwater factor even in ponds lacking fish and crayfish, but field snaik I N Am Renthoi Soc 8:342-350 and nutrition. In: experiments conducted across the lake size gradient are Krapu GL, Reinecke KJ (1992) Foraging ecology Batt Afton Anderson Johnson in order to more understand the BDJ, AD, MG, Ankney CD, DH, necessary completely Kadlec JA, Krapu GL (eds) Ecology and management of factors governing the landscape-scale distribution of breeding waterfowl. University of Minnesota Press, Minneapo- freshwatersnails. lis, Minn., pp 1-29 Krebs CJ (1989) Ecological methodology. Harper and Row, New York Acknowledgements We thank Scott Ray, Rebekah Turner, Amanda Lewis DB (2001) Trade-offs between and survival: Harold, Hannah Turner, Mike Pettyjohn, Martin Heben, and Sharon growth responses of freshwater snails to 82:758-765 Montgomery for assistance with field work in all weather conditions. predacious crayfish. Ecology DM, Brown KM, Klosiewski SP, Stein RA, Covich AP, Chuck Williams, Sharon Montgomery, Jason Hoverman, Rebekah Lodge Leathers BK, Bronmark C (1987) Distribution of freshwater Turner, and two anonymous reviewers provided helpful comments on snails: scale and the relative of the manuscript. This study was supported by the National Science spatial importance physicochem- ical and biotic factors. Am Malacol Bull 5:73-84 Foundation (grant No. 0444939) and is a contribution of the Pyma- DM, Kershner MW, Aloi JE, Covich AP (1994) Effects of an tuning Laboratory of Ecology. Lodge omnivorous crayfish (Orconectes rusticus) on a freshwater littoral food web. Ecology 75:1265-1281 Mahmoud I (1968) Feeding behavior in kinosternid turtles. Herpe- toloeica 24:300-305 References Martin TH, Crowder LB, Dumas CF, Burkholder JM (1992) Indirect effects of fish on macrophytes in Bays Mountain Lake: evidence Benoy GA, Nudds TD, Dunlop E (2002) Patterns of habitat and for a littoral trophic cascade. Oecologia 89:476-481 invertebratediet overlap between tiger salamanders and ducks in McPeek MA (1990) Determination of species composition in the prairie potholes. Hydrobiologia 48 1:47-59 Enallagma damselfly assemblages of permanent lakes. Ecology Bronmark C (1992) Leech predation on juvenile freshwater snails: 71:83-98 effects of size, species, and substrate. Oecologia 91:526-529 McPeek MA (1998) The consequences of changing the top predatorin Bronmark C, Malmqvist B (1986) Interactions between the leech a food web: a comparative experimental approach. Ecol Monogr Glossiphonia complanata and its gastropod prey. Oecologia 68:1-23 69:268-276 McPeek MA (2004) The growth/predation risk tradeoff: so what is the Bronmark C, Klosiewski SP, Stein RA (1992) Indirect effects of mechanism? Am Nat 163:88-1 1 1 predation in a freshwater benthic food chain. Ecology 73:1662- McPeek MA, Grace M, Richardson JML (2001) Physiological and 1674 behavioral responses to predators shape the growth/predation Brophy T (1980) Food habits of sympatric larval Ambystoma tigrinum risk trade-off in damselflies. Ecology 82:1535-1545 and Notophthalmus viridescens. J Herpetol 14:1-6 Mower C, Turner AM (2004) Behavior, morphology and the Brown KM (1991) Mollusca: Gastropoda. In: Thorp JH, Covich AP coexistence of two pulmonate snails with molluscivorous fish: (eds) Ecology and classification of North American freshwater a comparative approach. Am Malacol Bull 19:39-46 invertebrates. Academic Press, San Diego, Calif., pp 285-314 Nystrom P, Perez JR (1998) Crayfish predation on the common pond Brown K, DeVries D (1985) Predation and the distribution and snail {Lymnaea stagnalis): the effect of habitat complexity and abundance of a pond snail. Oecologia 66:93-99 snail size on foraging efficiency. Hydrobiologia 368:201-208 Burton T (1976) An analysis of the feeding ecology of the Osenberg CW (1989) Resource limitation, competition, and the salamanders (Amphibia: Urodela) of the Hubbard Brook Exper- influence of life history in a freshwater snail community. imental Forest, New Hampshire. J Herpetol 10:187-204 Oecologia 79:512-519 Chase JM (1999) To grow or to reproduce? The role of life-history Osenberg CW, Mittelbach GG, Wainwright PC (1992) Two-stage life plasticity in food web dynamics. Am Nat 154:571-586 histories in fish: the interaction between juvenile competition Chase JM (2003) Strong and weak trophic cascades along a and adult performance. Ecology 73:255-267 productivity gradient. Oikos 101:187-195 Peters RH (1983) The ecological implications of body size. DeWitt TJ, Robinson BW, Wilson DS (2000) Functional diversity Cambridge University Press, Cambridge among predators of a freshwater snail imposes an adaptive Pip E (1986) The ecology of freshwater gastropods in the central tradeoff for shell morphology. Evol Ecol Res 2:129-148 Canada region. Nautilus 100:56-66 Dillon RT Jr (2000) The ecology of freshwater mollusks. Cambridge Skelly DK (1996) Pond drying, predators, and the distribution of University Press, New York Pseudacris tadpoles. Copeia 1996:509-605

^ Springer Oecologia (2007) 153:407-415 415

Skelly DK (1997) Tadpole communitities: pond permanence and Wellborn GA (2002) Trade-off between competitive ability and predation are powerful forces shaping the structure of tadpole antipredator adaptation in a freshwater amphipod species communities. Am Sci 85:36-45 complex. Ecology 83:129-136 Turner AM, Montgomery SL (2003) Spatial and temporal scales of Wellborn GA, Skelly DK, Werner EE (1996) Mechanisms creating predator avoidance: experiments with fish and snails. Ecology community structure across a freshwater habitat gradient. Annu 84:616-622 Rev Ecol Syst 27:337-363 Turner AM, Trexler JC (1997) Sampling aquatic invertebrates from Wissinger SA, McGrady J (1993) Intraguild predation and compe- marshes: evaluating the options. J N Am Benthol Soc 16:694- tition between larval dragonflies: direct and indirect effects on 709 shared prey. Ecology 74:207-218 Turner AM, Turner SL, Lappi HM (2006) Learning, memory, and Wissinger SA, Whiteman HH, Sparks GB, Rouse GL, Brown WS predatoravoidance by freshwater snails: effects of experience on (1999) Foraging trade-offs along a predator-permanencegradient predator recognition and defensive strategy. Anim Behav in subalpine wetlands. Ecology 80:2012-2116 72:1443-1450 Yodzis P, Innes S (1992) Body size and consumer-resource Weber LM, Lodge DM (1990) Periphytic food and predatory crayfish: dynamics. Am Nat 139:1151-1175 relative roles in determining snail distribution. Oecologia 82:33- Zaret TM (1980) Predation and Freshwater Communities. Yale 39 University Press, New Haven, Conn. Wellborn GA (1994) Size based predation and the evolution of prey life histories: a comparative study of freshwater amphipod populations. Ecology 75:2104-2017

£) Springer