Ecc CeNNrsnlrsu rN ?oxoRnyNcHITEs 359

EGG CANNIBALISM AND CARNIVORYAMONG THREE SPECIESOF TOXORHYNCHITES

J. R. LINLEY exo D. DUZAK

Florida Medical Entomolngy Laboratory, Institute of Food and.Agricultural Sciences,Uniuersity of Flnrida, Vero Beach, FL 32962

ABSTRACT. In laboratory experiments, third and early fourth instar larvae of Toxorhynchites arnboinensis,Tr. splendensand Tx. breuipalpis,previouslystarved 24 h, rapidly cannibalizedeggs of their own species, or ate the eggs of other species present on the water surface in small containers. Toxorhynchitesamboinensis and, Tx. splendenslarvae of either instar cannibalizedeggs somewhat more rapidly than Tx. breuipalpis,probably becatse brevipalpiseggs distributed themselvesaround the edgeof the container and were less accessible.When offered heterospecificeggs, fourth instar larvae of all three speciesate them as efficiently as they cannibalized their own except that eggsof Tr. breuipalpiswerc eaten very slowly. Toxorhyrchites atnboinensrslarvae were offered conspecific hatched eggs,and these also were consumed.

INTRODUCTION 12:12.Lawae were reared throughout develop- ment in the presence of an excess of Aedes (Linley In a recent study 1988) of oviposition aegypti (Linn.) prey of the same instar. Adults (Doles- behavior in Toxorhynchitesamboinensis emerged in cages 38 x 50 x 65 cm in which chall), it was observedthat when females laid mating took place and eggs were obtained as their eggson water containing 20 third or fourth needed by allowing the females to oviposit in instar Iarvaepreviously starved 24 h, large num- black, plastic pots 9 cm diam x 7 cm deep bers of eggs were cannibalized. Although this containing tap water. was apparently the first published report of egg All experiments were done in a laboratory cannibalism in Toxorhynchites, egg carnivory maintainedat27.0 + 2.0'C under approximately (Kazana had been reported previously et al. 500 ft-candles illumination. Each experiment (see 1983). It has been known for many years consisted of 10 replicates, each of one third or Steffan and Evenhuis 1981) that the larvae are fourth instar , starved 24 h, then allowed mutually cannibalistic, probably in all instars accessto 20 unhatchedeggs (with one exception, (Furumizo and Rudnick 1978, Steffan et al. seebelow) of either their own or one of the other 1980).The extent to which cannibalism among species,for 6 hours. Only fourth instar larvae larvae may regulatepopulations in nature is not weretested with heterospecificeggs. One control known, but the behavior certainly occursin nat- pot, containing 20 eggs,but no larva, was set up (Trpis ural habitats 1973, Kazana et al. 1983). for each experiment. Secondinstar larvae were Similarly, if egg cannibalism is widespreadand not tested because,in earlier experiments with occurs naturally, then it also may play a role in Tx. amboinensrs(Linley 1988),they showed no regulating larval populations. This is not only significant ability to cannibalize eggs. Fourth of intrinsic interest, but it is also of practical instar larvae were taken in the first 2 days of importance in view of the biological control ap- the instar to ensure that none had yet reached plications (Gerberg of Toxorhynchites 1985). the period of compulsive killing (prey killed but In experiments reported here, we have tested not eaten) that precedespupation (Trpis 1972, whether egg cannibalism is exhibited also by Furumizo and Rudnick 1978,Padgett and Focks (Wied.) Toxorhynchitessplendens and Tx. breu- 1980). lpolpis Theobald. Experiments were also done During the 24-h period of starvation, larvae with Toxorhynchites amboinensisfor purposes were confined individually in tap water in wells of comparison,as the methods in earlier exper- (3.5 cm diam, I cm deep). Experiments were (Linley iments 1988) differed from those used done in small, black plastic pots 4 cm diam X 3 here.Additional experimentstested whether lar- cm deep, filled with tap water to a depth of vae of the three specieswould eat each other's about 2.5 cm. At the beginning of each experi- eggsand, in the caseof. Tx. arnboincnsis,whether ment, a larva was placed in each of the 10 pots, fourth instar larvae would eat conspecific which already contained 20 eggsof the selected hatched eggs. specieson the water surface. The numbers of eggs remaining (uneaten) were then counted MATERIALS AND METHODS every 0.5 h for 6 hours. In one experiment,fourth instar 7.r. amboinensislarvae were tested with Larvae and eggswere obtained from labora- hatched eggs.These consisted mainly of entire tory colonies maintained at 27'C under LD individual eggs,but also of somelarge fragments, JounNlr, oF THE ArraeRrceNMoseurro Coxrnol Assocretron Vol.5, No.3

since hatchedeggs arejoined only by a thin strip RESULTS of chorion, which breaks easily when manipu- lated. Controls:Twenty eggswere invariably present To facilitate statistical comparison of the on the controls after 6 h; consequently,no fur- rates at which eggs were consumed, the data ther referenceis madeto control pots. (e.g.for Tx. splendens,Fig. 2) were rectified to Egg cannibalisrn:Starved third and fourth in- Iinear by log (n + 1) transformationof the time star larvae of Tx. arnboinensis,Tx. splendens axis (Fig. 1) and tests run for differences be- and ?x. breuipalpisrapidly cannibalizedtheir tween the regressioncoefficients. eggs(Fig. 2). There wasno significantdifference in consumption rate between instars in Tx. am- boinensis,but fourth instars ate the eggs more o g) rapidly in the caseof Tx. splendens(P < 0.001) ctt q) and ?r. breuipalpis(P < 0.001).Both instars of c o Tx. breuipalpisconsumed conspecific eggs more G slowly than either of the other two species(P < o 0.001in all instars). Consumptionof heterospecificeggs: Fourth in- o o star Tx. breuipalpis larvae ate eggs of Tx. arn- boinensissignificantly more (P 5 rapidly < 0.001) than eggsof their own species(Fig. 3 cf. Fig.2), c a! but at about the same rate as Tx. splendens q) = larvaeate Tx. amboinensrseggs (Fig. 3i. Fourth instar larvae of Tx. am.boinensrsconsumed Tx.

Logro (n + 1) time (hr) Tx.brevipalpis Fig. 1. Transformed data (from Fig. 2) for egg t cannibalism by third (open symbols) and fourth lt... ' (closedsymbols) instar Toxorhynchitessplendens lat- 1 vae. ?.,r't".j.., rt?.+lItI fx. brqlpalpls rllrit, ul t.. o +l zof--r o Tx. splendens UJ 0) a ctt 'i. c, 'uF o 20A- c ot "sp/endens o '.F CD FiS-d---i'... | (E o 16l- il ir-{. | , o ,F |..I 1 I o J t-+--+.i 'f-t'I-I l CI ':Fti+11 o r o ,lt .F ,r11.. t E 'l- 5 L tl+4{+ '-t o oI .ct t tr t G E :l ++ f f+"i o tr l E c (E o =

u123456 Time (hr) Time (hr) Fig. 3. Consumption of heterospecific eggs by single fourth instar Toxorhynchites breuipalpis (eating am- Fig. 2. Egg cannibalism by single third (open sym- boinensis eggs), Tx. splendens (eating amboinensis bols) and fourth (closed symbols) instar Toxorhyn- eggs) and Tx. amboinensrs (eating breuipalpis eggs chites breuipalpis,Tx. splcndensand ?r. amboinensis (dotted line) and splendens eggs (continuous line)) Iarvae previously starved 24 h. larvae previously starved 24 h. Ecc ClNNralr,rsM rN ?oxoaHYNCHITES 361

ut quickly accumulatedin the meniscusaround the o +l edgeof the dish, whereasamboinensis and splen- o dens eggs invariably floated away trom the h ct) !P Ct) edges,usually in one or two adherent clumps. :o Eo Larvae of all three species tended to remain ?4, E-C away from the edges.Thus, in the caseof breu- F() $aE ipalpis,larvae simply may not have had compa- o.E =c rable accessto the eggs.It is noteworthy that o when fourth instat breuipalpislarvae had access 6 o 123456to amboinensiseggs, they consumed them far (Fig. mentioned Time (hr) moreefficiently 3). It shouldbe also, that every 0.5 h the clumped amboinensis Fig.4. Consumptionof hatchedconspecific eggs by and.splendens eggs were separatedto facilitate singlefourth instar Toxorhynchites amboinensis larvae counting, but they regroupedinto clumps almost previously starved24 h. immediately. The behavior that enableslarvae to find float- splendensand Tx. breuipalpiseggs more slowly ing eggsmerits further study. Larvae were often (P < 0.001in both instances)than thesespecies seenwith their bodies parallel to the water sur- had eaten Tx. amboinensiseggs. When Tx. breu- face, with heads and mouthparts touching or ipalpis eggswere the prey, Tx. amboinensisIar- very close to the interface. We noticed on a vae found and ate them much more slowly (P < number of occasionsthat contact by groups of 0.001)than Tx. splendenseggs (Fig. 3). eggson a larva's siphon would often causethe Cannibalismof hatchedeggs: When 20 hatched to investigate the contact with its head. Tx. arnboinenslseggs were placed on watet con- Since the pot interiors were black, the behavior taining conspecific fourth instar larvae, they at detection and seizure of an egg could not be were eaten somewhat more slowly (P < 0.02) seen clearly, but disappearanceof an egg was than unhatched eggs (Fig. 4), but quite rapidly instantaneous,suggesting that larvae seizeeggs nonetheless.There may have been some error with the same extremely rapid head and man- in these counts inasmuch as it was not always dibular action that accompaniescapture ofother possible to determine whether larvae ate the prey (Breland 1949, Furumizo and Rudnick entire hatched egg, or whether it sometimes 1978).Eggs of amboinensisand splenderu,being separated into one or more additional pieces. extremely hydrofuge,float very high on the sur- Small fragments intermittently appearedon the face, and presumably can be seizedonly if the surface, suggesting that some breakage had mouthparts are used to strike sharply upwards, taken place. Effectively, this addedegg material with the headinclined. to the surface and slowed the apparent rate of The experiments described here were done consumption.To minimize error, however,we under highly artificial conditions that are un- did not count very smallpieces obviously outside doubtedly at variance with conditions likely to the size range ofthose initially introduced. occur in nature. Single larvae might often be found in a natural container and, in the caseof some species,for example Tx. haemoffhoidalis DISCUSSION (Fabricius) occurring in bracts of Heliconia car- ibaea(Lounrbos et al. 1987),in one of dimen- Our main objective in these experiments was sions similar to those used here. Even though to determinewhether egg cannibalism,as pre- the presenceof 20 eggs simultaneously on the viously noted in Tx. amboinensrs(Linley 1988), surfacein such a limited habitat seemsimprob- also occurred in other species.The results indi- able, the rate at which eggs are eaten experi- cate that hungry third and fourth instar larvae mentally suggeststhat eggcannibalism may oc- of both Tx. splendens and Tx. breuipalpis also cur in nature. According to Trpis (1972) and will eat eggslaid by femalesof their own species. Furumizo and Rudnick (1978),respectively, the All three speciesshow a propensity for eggcan- egg stage of Tx. breuipolplslasts about 44 h (at nibalism and can detect, seize and ingest eggs 25.0"C) and that of Tx. splendens48-60 h (at floating on the water surface. 26.5'C). Single fourth instat splendensandbreu- Among the three species,Tx. breuipalpis ap- ipalpislawae, respectively,ate about 17 and 11 peared to be least efficiently cannibalistic. In of the initial 20 eggsin 6 h, and Tx. arnboinensis both instars it disposed of its own eggs at a had eaten about 16 in this time. It is probable, significantly slower rate than the other two spe- therefore, that someeggs would be eaten in over cies.This differencewas, however,almost cer- 40 h, even if eggand larval densities were much tainly causedmore by the distribution of breui- Iower. The rate of cannibalism would presum- palpis eggson the water surface than by differ- ably be affected by egg density and also by the ences in behavior. Most eggs of breuipalpis amount of other prey present. 362 Jounlrar, oF THE ArranRrcl,NMosquno CoN,rRor,AssocrlTroN VoL. 5. No. g

ACKNOWLEDGMENTS thcobal.di(Diptera: Acta Cient. Venez. 34:151-158. This paper is Institute of Food and Agricul- Linley, J. R. 1988.Laboratory experimentson factors tural Sciences, University of Florida Experiment affecting oviposition site selection in Toxorhyn- Stations Journal Series No. 8985. chttesamboinensis (Diptera: Culicidae),with a re- port on the occurrence of egg cannibalism. Med. Vet.Entomol. 2:27 l-277 Padgett, P. D. and D. A. Focks. 1980. Laboratorv REFERENCES CITED observations on the predal ion of Toxorhynchitis rutilus rutilus on Aedesaegpti (Dipteta: Culicidae). Breland, O. P. 1949. The biology and the immature J. Med. Entomol. L7:466-472. stagesof the Megarhinus septentrionalis Steffan,W. A. and N. L. Evenhuis.1981. Biology of Dyar and Knab. Ann. Entomol. Soc.Am. 42:Sg-47. Toxorhynchites.Annu. Rev.Entomol. 26:159-181. Furumizo, R. T. and A. Rudnick. 19Tg. Laboratorv Steffan,W. A., R. D. Stoaksand N. L. Evenhuis.1980. studies of (Diptera: Culi_ Biological observations of Toxorhynchites atnboi- cidae):biological observations. Ann. Entomol. Soc. nensrs (Diptera: Culicidae) in the laboratorv. J. Am. 71:670-673. Med. Entomol. l7:515-518. Gerberg, E. J. 1985.Sequential biocontrol application Trpis, M. 1972.Development and predatory behavior in the use of Toxorhynchlresspp. pp. 33-46. Ia M. of Toxorhynchitesbreuipalpis (Diptera: Culicidae) in Laird and J. W. Miles (eds.), Integrated mosquito relation to temperature. Environ. Entomol. 1:53?- control methodologies,Vol. 2, New york: Academic 546. Press. Trpis, M. 1973.Interaction betweenthe predator ?or- Kazana, M., C. E. Machado-Allison and L. A. Bulla. orhynchites breuipalpisand its prey Aedesaegypti. 1983. Preferenciasalimentarias de Toxorhynchites Bull. W. H. O. 49:359-365.