392 Jounuel oF THEAlannrcet Moseurro ConrRor,Assocrlrrox Vor,.5, No. 3

PREY-SZE SELECTION BY TRIOPS LONGICAUDATUS (NOTOSTRACA:TRIOPSIDAE) FEEDING ON IMMATURE STAGES OF CULEX QUINQUEFASCIATUS

NOOR S. TIETZE ,c,NoMIR S. MULLA

Department of Entomology, Uniuersity of California, Riuerside,CA g%21

^ ABSTRACT. The tadpole shrimp, Triops longicaudafus,was found to be a size-dependentpredator of Culex quinquefasciatuslawae in the laboratory. However, changesin tadpole shrimp size *-e.e a""o-- panied by cl'angesin prey-sizepreference: larger-sized predators consumedan increising proportion of ljrrger prey-items- Very large tadpole shrimp may be nonselectivepredators of this mosquitb species. Quantified behavioral observationsindicated that while secondinstai mosquito larvae were-encountered significantly less frequently than were fourth instar larvae or pupae, they were captured at significantly greaterrates and with shorter handling times. It is hypothesizedthat prey vulnerability has anlmportant influence on tadpole shrimp prey size "preferences."

INTRODUCTION oratory, this specieswas found to be a predator of mosquito larvae (Culex quinquefasciarr.mSay) In many predator-preyrelationships, the body (Tietze 1987)and is currently being considered sizes of the predator and the prey play an im- as a biological control agent against mosquitoes. portant role dictating the size of prey items These studies investigated the feeding rates consumed(Krebs 1978).Prey mlnerability may of two tadpole shrimp sizes on mosquito be a more important cause of size-dependent larvae; smaller tadpole shrimp (mean body length : predation than is active selection by inverte- 1.6 cm) consumedfirst and secondinstar mosquito brate predators. As prey size increases,so does Iarvae at higher rates than the third and fourth the efficiency of their escaperesponses. Inver- instars while Iarger tadpole shrimp (mean body tebrate predators, termed size-dependentpred- Iength : 2.4 cm) consumedall mosquito instars ators, have been found to prefer intermediate- at nearly equal rates (Tietze 1987).According sized prey items due to capture and handling to the classification scheme of Greene (1985), tradeoffs(Allan et al. 1987,Pastorok 1981). Triops longicaudatusare cruising raptorial pred- Pastorok (1981) found that Chaoborustriuit- ators. totus Saether larvae feed in an opportunistic Unlike floodwater mosquitoes(i.e., Aedes and. manner on all sizesof Daphniapulicari.a Forbes; P sorophora),stagnant water mosquitoes (Culex however,intermediate-sized prey were overrepre- and Anopheles) may oviposit during the entire sented in their diet. He found this disparity to flooding period, thus producing an heteroge- be due to probabilities of encountering prey and neous assemblageof larval instars and pupae. prey escapecapabilities and not active selection The ability of growing tadpole shrimp to feed on the part of the predator. Molles and Pie- upon various mosquito instars may determine truszka (1983)presented the stonefly,Hespero- their effectivenessas biological control agents perLapacifica (Banks) with two mayfly species, of stagnant water mosquitoes.The goal of this Ephemerellaaltana AIlen and Baetis tricaudatus study was to determine from laboratory experi- Dodds, and demonstratedthat "preference" for ments whether tadpole shrimp show sigrrificant E. altana was due to a better escaperesponse sizepreference for stagnant water mosquito lar- (i.e., lower vulnerability) bV B. tricaudatus. vae and pupae and if these preferenceschange Tadpole shrimp (Notostraca: Triopsidae) are as the tadpole shrimp grow. The issueof whether crustaceans inhabiting transient fresh waters active selection or vulnerability is involved in worldwide (Longhurst 1956). The diet and nu- this predation scenariois addressedthrough lim- trition of tadpole shrimp remain relatively un- ited behavioral observations. studied.Pont and Vaquer (1986)reported ?riops cancriformis (Bosc.) to progress from juveniles feedingon phytoplankton to mature adults feed- MATERIALS AND METHODS ing on plant material, and only the largest in- Laboratory selection experiments: Tadpole dividuals (>10 mm) are carnivorots. Triops Inn- shrimp (Triops longicaudatus)were reared from gicaudatusLeConte was called polyphagousand eggsin fiberglassmicrocosms at the UCR (Uni- found to feed upon invertebrates and plant ma- versity of California, Riverside) Aquatic and terial in both daylight and darkness (Scott and Vector Control Research Facility. The micro- Grigarick 1979). They also noted that, in the cosms were 0.84 m2 in area and equipped with laboratory, early instar mosquito larvae were float valves to maintain the proper water depth consumed.In a more detailed study in the lab- (30 cm). Five microcosmswere flooded at differ- Teopor-p SnRrvp PnsolrroN 393 ent times over a 5-day period to produce a range n1arrd 11,respectively, but are used to sum over of tadpole shrimp sizes.Larval and pupal stage the various prey categories.Chesson's alpha can mosquitoes of the speciesCx. quinquefasciatus be usedfor calculatingprey selectionin constant were provided from a laboratory colony main- or changingprey densities(Chesson 1978). AI- tained at UCR. pha was calculatedfor each prey size class in Predator and prey sizeswere measuredusing each arena (replicate) and each predator size a dissection microscope fitted with calibrated group. A multivariate analysis of variance (SAS ocular micrometer. These measurementswere Institute Inc. 1982) was performed to test the subsequentlyconverted to millimeters. Due to null hypothesis:there were no significant differ- telescopingof the abdomen,tadpole shrimp size encesin selectionbetween predator size groups. was measuredas the length ofthe carinal suture SeparateANOVAs andDuncan'smultiple range that divides the carapacealong the longitudinal tests (SAS Institute Inc. 1982) were performed bodyaxis (Linder 1959).Longhurst (1956) found to determine whether significant differencesex- the carapace lengfh of T. can'criformls to be ist for prey sizepreferences within eachpredator closely conelated (growth coefficient k = 1.02) size group. to total body length. Immature mosquito size Quantified predator-prey interoctions: Behav- was measuredfrom the apex ofthe head capsule ioral observations were made to quantify en- to the baseof the air siphon for larvae and from counters,captures and handling times of a Iarge the anterior ofthe headto the paddlesfor pupae. tadpole shrimp (carapacelength = 7.0 mm). Preferencetests were performed in a Gibson Attacks could not be quantified due to the cruis- environmental chamber. Temperature was ing predation strategy used by tadpole shrimp. maintained at 26 + 2'C and measuredusing a An encounter was defined as the prey's coming continuous hygrothermograph. Illumination within contact distance of the predator or the within the chamberwas producedby one 60 watt length of the shrimp's first thoracic appendage cool white fluorescentbulb. Arenas were plastic which is held anteriad, like an antenna. Hand- containers filled with 200 ml of well water and ling time was the duration between a capture a water depth of 4 cm. Tadpole shrimp were and subsequentsearching. Due to difficulty in acclimated to the arena and temperature by visually distinguishing between consecutivelar- placing individuals into arenas inside the envi- val instars, only three of the original five prey ronmental chamber for t h pretreatment. The groups could be analyzed:three of each second shrimp were starved during the acclimation and fourth instar Iarvae and pupae were simul- period. Immature mosquitoes were acclimated taneouslyadded, and the ensuingbehaviors were to the same test temperature by placing them recorded for 30 min. This was done four times into the environmental chamber for t h pre- using different shrimp each time. treatment. Ten representatives of each larval Behavior was recordedusing a Sony Beta I instar and pupae were preservedin 70% EbOH SLO-340video recorder, an RCA CC 0-1-0cam- and were later measuredin body length. Tadpole era and a Tamron 90 mm telemacro lens. The shrimp size was measured immediately before recordings were reviewed by playing back on a placing them into the arena for acclimation. Sony Beta I SLO 323 and viewed on a Sony Preferencetests consisted of exposingindividual SVM-1010 Motion Analyzer. Number of en- tadpole shrimp in the arena to five of each counters,captures and handling time was deter- mosquitolarval instar and pupae(25 total). The mined by careful replay of each predation se- prey items were simultaneously added to the quence.Capture successrate was calculated by arena and the time of addition noted. After t h dividing number of captures by number of en- of predation each arena was inspectedfor num- counters.These values were arcsine transformed bers of prey consumed,Six replicateswere made and the three prey groups were comparedusing for each predator size class. Due to the short ANOVA and Duncan's multiple rangetest (SAS period of exposure (1 h), replacement of prey Institute Inc. 1982). Handling times and en- consumedwas not implemented. counter rates were similarly compared between Prey selectionwas calculatedusing an electiv- the three prey groups. ity indice, Chesson'salpha (A) (Manly et al. 1972,Chesson 1983): RESULTS (P . ln [(nr - rr)/nr] Prey selection: Significant differences < "' = prey preference(6) were found for >, ln - .ivrrJ 0.05)in size [(nr first, third and fourth instar larvae and pupae where ni is the number of prey items of type i when comparing six predator (tadpole shrimp) present at the beginning of a foraging bout, 11is size classes;preference for second instar larvae the number of prey items of type i consumed,n; was not significantly different (P > 0.05) be- and r.;are symbols having the samedefinition as tween predator size classes(Table 1). 394 Jounulr, oF THEAunRrcl,r.r Mosqurro CoNrnol Assocrnrron VoL.5,No.3

Prey size preferences of tadpole shrimp Behauioral obseruations:The capture success changeddramatically with their growth (Table rate of secondinstar Culex quinquefascinrluswas 2 and Fig. la-f). The smallest tadpole shrimp significantly greater than that of fourth instars size group (2.2 mm) preferred first instar larvae or pupae (Table 3). Tadpole shrimp (7.0 mm) significantly more than secondinstar larvae and capture successrate on secondinstar larvae was did 19t consume larger instars or pupae (Fig. five times greater in magnitude than on fourth 1a). The secondpredator size group (8.0 mm) instar larvae and six times greater than that of fed upon first and secondinstar larvae without pupae.Similarly, handling time of secondinstar a significant difference (P > 0.0b) in preference mosquito larvae was significantly less than that (Fig. 1b). Membersof all prey sizeclasses were of fourths and pupae (Table 3). Comparing the taken when the tadpole shrimp had carapace numbers of encounters between the tadpole lengths of 4.0 mm or greater. Medium-sized shrimp and the three prey items, fourth instar tadpole shrimp (4.1 mm) showed significant larvae and pupaehad significantly more encoun- preferencefor earlier instar larvae but did feed ters than did second instar mosquito larvae upon third and fourth instar larvae and pupae (Table3). (Fig. 1c). Aside from between fourth instar lar- vae and pupae, larger-sizedtadpole (4.7 shrimp DISCUSSION mm) showeddistinct preferencesbetween prey size classes (Table 2 and Fig. 1d). Tadpole Tadpole shrimp were found to be size-depend- shrimp with 5.75 mm carapacelength exhibited ent predators of mosquito larvae and generally similar preferencesfor third and fourth instar displayed decreasingpreferences for increasing larvae and pupae but significantly greater pref- larval size. As tadpole shrimp become larger, erence for third instar larvae over pupae (Fig. their relative preference for Iarger prey sizes 1e). Finally, while the largest tadpole shrimp increases.Distinct prey-sizepreferences are ev- (9.9 mm) had no significant difference in pref- ident if tadpole shrimp are separatedinto arbi- erence between first, third and fourth instar trary size groups (i.e., small, medium and large). larvae and pupae, there was significant differ- Small tadpole shrimp consumed early instar ences between these and second instar larvae (first and second) mosquito larvae. Medium- (Table 2 and Figure 1f). sized tadpole shrimp consumedall available in- stars but preferred early instar larvae. As the Table 1. Analysis of preference for immature Culer predators grew larger, the relative preferencefor quinqu,efasciatus between six tadpole shrimp (?riops larger prey items increased.The largest tadpole longicaudatus) size groups. shrimp size group appearedto feed upon most prey sizesin a nonselectivemanner. MANOVA results Instar/stage Nonselectivepredation occurs when the rela- (mean size + SD)' df tive abundanceof prey size classesin the diet is L1 (1.2+ 0.10) 18 5 0.0001 not different from that ofthe environment. Five L2 (2.3+ 0.13) 2 5 0.0818 prey groups of equal abundancehave a theoret- L3 (4.1+ 0.26) 7 5 0.0003 ical nonselective value of 0.2. When the prey L4 (5.6+ 0.19) 8 5 0.0001 size preferences of the largest tadpole shrimp Pupae(3.3 + 0.62) t) 5 0.0011 sizeclass (9.9 mm) are comparedto 0.2 (t-test), u Mean size (mm) of mosquito larvae (n : 10) only secondinstar Iarvae are significantly (P < measured from apex of head capsule to base of siphon; 0.05) different. size of pupae was measured from apex of head to base Tadpole shrimp had an overall preferencefor of paddles. second instar mosquito larvae. Preference for

Table 2. Analysesof preferencefor immature Cul,exquinquefasciotrzs within six tadpole shrimp (?riops Iongicaudatus)size groups (ANOVA; Duncan's multiple-range test: unshared letters indicate significant differences). ANOVAresults Prey size class TPS size" (mm) df L3 L4 2.20 833 0.0001 0.684a 0.316b 3.00 37 A 0.0001 u.br6a 0.482a 4.10 26 0.0001 0.411a 0.373a o.rs?u 0.047b 0.030b 4.70 qt 0.0001 0.424a 0.31lb 0.193c o.o72d 5.75 1n 0.000r 0.332a 0.323a 0.196b 0.110bc 0.042c 9.90 A 0.01 0.207b 0.289a 0.175b 0.168b 0.141b 'Mean (n : 6) tadpole shrimp size measured as carapace length. SEPTEMBER 1989 TADPoLE SsRtlr,tp PnPolttott

a< T 0.8 I 0.8 o. C.L. = 2.2 mm o. C.L.=3.0 mm 0.7 J 0.7 0.6 t!) 0.5 z 0.5 9'' o z 0.5 Ir a 0.4 (to 0.4 a 0.3 U' uJ ul 0.3 T I o o.2 o 0.2 z 0.1 z 0.1 ul 0.0 lrl 0.0 L3 L4 = Ll L2 = L1 L2 L3 L4 INSTAR INSTAR rr< r 0.8 0.8 o- = 4.75 o- C.L. = 4.1 mm J C.L. J 0.7 o.7 0.5 t!) 0.5 91 z 0.5 z 0.5 o o U' 0.4 a o.4 a U' { UJ 0.3 uJ 0.3 T 0.2 o 0.2 o z 0.1 z 0.1 Lr.l 0.0 u,t 0.0 = L1 L2 L3 L4 = L1 L2 L3 L4 INSTAR INSTAR f< o.s I 0.8 E = 9.9 mm o. o.t C.L. J 0.7 C.L. = 5.75 mm i !, 0.6 t!) 0.6 z 0.5 6 o.u o (tl 0.4 3 o.o at LlJ ul 0.3 :tr 0.3 T 0.2 o o z o'2 z 0.1 fi o't lrJ 0.0 = L1 L2 L3 L4 P = L1 L2 L3 L4 P INSTAR INSTAR -r$5 Fig. 1a-f. Mean preference(Chesson's alpha) for four Cul,exquinoquefasciatus lawal instars and pupae : by piedatory tadpoie shrimp (?riops longicaudatus)at six size classes(C.L. carapacelength).

secondinstar larvae did not changesignificantly Ieastvulnerable to predation. Behavioral analy- with growth of the tadpole shrimp (Table 1) and sis indicated slightly lower capture successand the largestpredator group did show greaterpref- greater handling times for mosquito pupae than erencefor these than for first, third and fourth fourth instar larvae. instar larvae and pupae (Table 2). Behavioral Although this study is not a defrnitive test of analysis gave similar results: high capture suc- whether apparent prey size preferenceis due to cessand short handling times for secondinstar active predatory selectionor vulnerability ofthe larvae. Second instar larvae may be the most prey, behavioral observations tend to support vulnerable mosquito larval instar to tadpole the latter hypothesis. The behavioral analysis shrimp predation. showedthat although the number of encounters Altering of prey body shapeis known to influ- betweentadpole shrimp and secondinstar larvae ence predator capture and handling success were fewer in frequency than with fourths and (Kerfoot 1975, L977). Mosquito pupation pro- pupae,capture successof secondinstars was far duces drastic changes in body shape; from a greater. Since second instar larvae were ob- Iinear fourth instar larva to a smaller, round served (but not quantified) on the videotaped pupa. The latter stage was generally the least predation sequences to consistently display "preferred" by tadpole shrimp, or similarly, the weaker escaperesponses (in velociW and dis- 396 JouRNlr, oF THE AMERTCANMoseurro CoNrRor, AssocrATroN VoL.5,No.3

Table 3. Capture successrates, handling times (sec)and encounter frequencies(no./30 min) of Cubx quinquefasciatussecond and fourth instars and pupae by Triops longicaudittu (carinal iength : T mm) from 30-min video recordings (n : 4) in the laboratory (ANOVA; Duncans multiple-range test: unshared letters indicate significant differences).

ANOVA results Instar/stage Quantified behaviors df L2 L4 Capture success 492 0.0001 0.26a 0.05b 0.04b Handling time 162 0.0000 31a 73b 175b Encounters 520.0356 15a 48b 58b tance)when encounteringa predator,they may reviewing this paper. These studies were sup- be more vulnerable to being captured. These ported by funds from UNDP/World Bank/ observations suggest vulnerability to have an WHO SpecialProgram for Researchand Train- important influence on size-dependentpreda- ing in Tropical Diseasesand from University of tion, However, lower encounter rates of second California Special Funds for Mosquito Re- instar larvae may be in part due to their smaller search. size, swimming behavior and position in the REFERENCES CITED water column. Allan,D. Prey-size selection is an important factor to J.,A. S.Flecker and N. L. McClintock.1987. Prey size selection considerwhen evaluating new by carnivorous stoneflies. Lim- biological control nol. Oceanogr.32:864-872. agents. The proper degreeof control may only Chesson,J. 1978. Measuring preference in selective be realized when there are appropriate relative predation. Ecology 59:217-215. sizesof predatorand prey. For example,in this Chesson,J. 1983.The estimation and analysisofpref- study we found that smaller tadpole shrimp erenceand its relationship to foraging models.Ecol- (carapacelength <4.0 mm) do not consumethird ogy 64:1297-1304. or fourth instat Cr. quinquefasciarfuslarvae. In Greene,C. H. 1985.Planktivore functional groupsand the field, some staglant water mosquito larvae patterns of prey selection in pelagic communities. producedfrom ovipositions during the first J. Plankton Res.7:35-40. days Kerfoot, after flooding may escapepredation W. C. 1975.The divergenceof adjacent pop- by becoming ulations.Ecology 56: 1298-1313. third and fourth instars before the tadpole Kerfoot, W. C. 1977. Implications of copepodpreda- shrimp have grown to 4.0 mm in carapace tion. Limnol. Oceanog.22:316-325. length. Similar situations may occur where prey Krebs, J. R. 1978.Optimal foraging: decision rules for growth rates are very high, as in the floodwater predators, pp. 23-63. In: J. R. Krebs and N. B. mosquito, Psorophora columbiae Dyar and Davies (eds.) Behavioral ecology: an evolutionary Knab, whose eggshatch immediately upon hy- approach,Sinauer Associates,Sunderland, MA. dration. These mosquitoes may completely es- Linder, F. 1959. Notostraca, pp. 572-576. In: W. T. (ed.), capepredation by rapid developmentand emer- Edmondson Freshwater biology, John Wiley gence. and Sons,New York. "preference" Longhurst, A. R. 1956. A review of the Notostraca. Whether similar prey-size rela- Bull. Br. Mus. Nat. Hist. Zool.3:1-57. tionships exist for other prey speciesof Triops Manly, B. F., P. Miller and L. M. Cook.1972. Analysis longicaudatusremains to be investigated. What of a selective predation experiment. Am. Nat. happens if tadpole shrimp are offered a variety 106:719-736. of plants and invertebrates as a food source? Molles.M. C.. Jr. and R. K. Pietruszka.1983. Mech- The presence of an alternative food source anisms of prey selection by predaceousstoneflies: prey predator (TetraMin' fish flakes) was found to signifi- roles of morphology, behavior and hunger.Oecologia (Berlin) 57:25-3I. cantly (P < 0.05) reduce the consumption of Pastorok, R. A. 1981.Prey vulnerability and size se- second instar Cr. quinquefasciarlusIarvae by lection by Chaoboruslarvae. Ecology 62:.1311-1324. Triops longicaudatusin the laboratory (unpub- Pont, D. and A. Vaquer. 1986.Influence du phyllopode Iisheddata). Studiesconcerning tadpole shrimp Triops cancriforrnis (Bosc.) sur la biocenose des relative preference for various prey taxa and rizieresde Camargue.Acta Ecologica7:75-88. possiblyprey-switching between these taxa de- SAS Institute Inc. 1982.SAS UsersGuide: Statistics, serve future investigation. 1982Edition. Cary, NC. Scott, S. R. and A. A. Grigarick. 1979.Observations on the biology and rearing of the tadpole shrimp ACKNOWLEDGMENTS Triops longicaudatus(LeConte) (Notostraca:Triop- sidae).Wasmann J. BioI. 36:116-126. We thank T. M. Perring,J. Bethke and T. C. Tietze, N. S. 1987.Tadpole shrimp (?riops lnngirau- Baker for assistance and for providing video datus) new candidatesas biological control agents equipment. We also thank W. E. Walton for for mosquitoes.Bio Briefs 13(2):1.