An Evaluation of Gambusia Aff/N/S and Bacillus Thuringiensis Var

470 JounNer, oF THE Arr.rrRtcer.t Moseurro CoNrnor, AssocrATIoN VoL.4,No.4

AN EVALUATION OF GAMBUSIA AFF/N/S AND BACILLUS THURINGIENSIS VAR. ISRAELENS/S AS MOSQUITO CONTROL AGENTS IN CALIFORNIA WILD RICE FIELbS

VICKI L, KRAMER,I RICHARD GARCIA, eNo ARTHUR E. coLwELL3

ABSTRACT. The mosquito potential of the mosquitofish and Bacillus thuringiensis var. (Bti) .control "fields israelensis were evaluatedin experimental wild rice fieldi in Lake County, California. w;;; assignedone of six treatme.nrs:^co1_t1o], 1.1 kg/ha Q. afflnis,3.4kg/ha c. iyitnis,'Gambusin bil ""ty io tyt" Vectobac" granules), 1.1 kglha G..affinis,plusBti and e.+'ue/na C. aiiinis plus'bd. iffiii!, "7 both releaserates, significantly reduced the mosquito population at densitiesexceeding fOO nsf/-i"tio* trap. Treatments with Bti significantly reducedlarval populations; however,the populition. i" itt" ftet,lt without fish reboundedto pretreatment levels within two weeks. In fields stodked*iitt C. "ffi"i";;e treated with Btr, populations remained low after Bti,treatment. Nontarget populations "f #h;;pft; were significantly lower in fields stocked with G. affinis than in fields witloot fish ott on" o. oro." sampling dates.

INTRODUCTION lbs/acre), and no significant differences were found among the mosquito populations in fields Wild rice (Zizania palustris Linn.) is grown with or without fish (Kramer et al. 1987).The in Lake County, California from May through authors postulated several reasonsfor this lack October, providing ca. 300 hectares of breeding of control, such as the short (90 day) growing habitat for Cul,ex tarsalis Coquillett, Anopheles season,the wild rice plant structure, and the freeborni Aitken and. An. Mc- franciscanus omnivorous feeding nature of G. affinis. They Cracken.nWild rice acreageis increasingin Cal- concludedthat higher releaserates of G. affinis ifornia and totaled more than 6,400hectares in may be necessaryfor significant mosquito con- 1986(M. D. Andres,unpublished data). trol in wild rice fields. The mosquitofrsh, Gambusia affinis (Baftd This study therefore evaluated G. affinis at a and Girard), has been shown by several re- higher releaserate of3.4 kglha (3 lbs/acre). This searchers (Hoy and Reed 1970, Craven and rate is 15 times greater than the usual release Steelman 1968)to be an effective mosquito con- rate (0.2 lbs/acre) in California white rice fields trol agent in white rice (Oryza satiua Linn.) (Combs 1986) and, although feasible for Lake fields. Consequently, mosquito abatement dis- County, is unrealistic for many mosquito abate- tricts (MADs) in the Central Valley of Califor- ment districts. Thus G. affiniswas also evaluated nia make seasonalreleases ofthe fish into their at a morepractical rate of 1.1kg/ha (1 lb/acre). rice fields. The use of G. affinis instead of chem- Although significant mosquito control was not icals can reducemosquito control costs in white achieved at 1.7 kg/ha during the 1986 season, rice fields substantially (Lichtenberg and Getz the 1.1 kg/ha rate was selectedfor 1987because 1985).Wild and white rice plants, although cul- many variables in a rice field systemcan change tivated in a similar manner, have several differ- from one seasonto the next, potentially affect- ences, such as plant height and structure and ing the control capabilities of G. affinis. Llso, length of the growing season,that could affect the Lake County experimental wild rice fields the mosquito control effectivenessof G. affinis were first year fields in 1986 and this may have (Kramer et al. 1987). affected mosquito production (Collins and Gambusiaaffinis was evaluatedin experimen- Washino 1980). The impact of G. affinis on tal wild rice fields in Lake County in 1986 at aquatic insect and zooplankton populations was releaserates of 0.6 and I.7 kg/ha (0.5 and 1.5 also evaluatedin this study. B acillus thuringiensis Berliner v ar. israe le nsis de Barjac (Bti) has been used effectively to control mosquito larvae in a wide range of hab- 'Division of BiologicalControl, Universitv of Cal- itats (Lacey and Undeen 1986),including white ifornia, Berkeley,CA g4?20.Current address(reprint rice fields, but had not previously been tested in requests):Contra CostaMosquito Abatement Disirict, wild rice. Studies (ibid) have shown that Bti is 1330Concord Ave., Concord. CAg452O. a highly selectivecontrol agent and that natural 2Division of Biological Control, Universitv of Cal- enemies of mosquito larvae are conserved.To ifo-rnia.Berkel ey. C A 94720. 'Lake evaluatewhether natural enemies,including in- County Mosquito AbatementDistrict. 410 Esplanade,Lakeport, CA 9b458. troduced fish, can maintain mosquito popula- a Tompkins, D. 198?. Lake County Agricultural tions at a low level following an initial reduction Crop Report. Depart. Food and Agriculiure,lakeport. by Bti, tests were conductedusing the pathogen 5 pp. alone and in combination with G. affinis. Ducn\,rarn1988 Gtuaush AFFrNrsAND B"r rlt Wrlo Rtcn Ftnr,os

MATERIALS AND METHODS sample G. affinis and other aquatic organisms. The trap stations, one per each side of the field, The eighteen 0.1 hectare (0.25 acre) Lake were randomly selectedon each trapping date. County rice plots used to evaluate G. affinis in On the final monitoring date (September8), four 1986(Kramer et al. 1987)were used for the 1987 traps were also set in the interior of each field study. The fields were flooded and seededon (a total of eight traps per field). All organisms June 9 using a seedbroadcaster attached to an were counted at the study site and returned to all-terrain vehicle. The wild rice plants matured their trapping location. On August 11, ten in about 90 days, and the fields were drained for trapped fish from each field (120 total) were harvest on September16. frozen for later gut content analysis. Maximum/minimum thermometers were The immature mosquito populations were placed in two of the plots, and the water tem- monitored weekly by taking 40 dips around the peratures were recorded weekly throughout the perimeter and 20 dips through the interior of season.Plant height and water depth were also each field. The perimeter density was sampled monitored weekly. A water sample from the by randomly selecting eight (two per field side) center of each field was collected on September ofthe flaggedstations eachweek. Five dips (400 13 and analyzedfor nitrate and phosphatecon- ml each) were taken at each station in a semi- tent, alkalinity, hardness,conductivity and pH. circularpattern, within 2 metersfrom the levee. There were three replicatesof each of the The interior sampleswere taken along a transect followingsix treatments:control, 1.1kg/ha (low beginning at a randomly selectedstation on one fish rate) G. affinis,3.4kg/ha (high fish rate) G. side of the field and ending at a randomly se- affinis, Bti only,1.1 kglha G. affinis plus Bri and lected station on the oppositelevee. A single dip 3.4 kg/ha G. affinis plus Bfi. All Bti treatments at 2 m along the transect. were 6 ke/ha of granular Vectobac'" (200 ITU/ was taken intervals placed in mg). Prior to flooding,treatments were assigned AII dip samples were concentrated, with rice water, and brought to to the fields using a randomizedblock design containers field (Fig.1). the Iaboratory to be immediately counted and were Gambusiaaffinis were seined from the Lake identified. Insects other than mosquitoes New Jersey light County MAD rearing ponds, weighed and re- also identified and recorded.A (Mulhern to leasedinto the 12 treatment plots on June 23, trap 1942) was operated adjacent mosquito two weeks post-flooding. Fish survival is be- the wild rice fields to monitor adult populations sample Iieved to be optimized by this two week post- in the area. The light trap flooding release schedule (Farley and Younce was collectedand countedweekly. populations were sampled 1977a).Fish releasedincluded adults and fry of The zooplankton both sexes. using a 202 micron net to concentratethe inte- Monitoring stations were flagged at 2 meter rior transect samples taken on August 3 and per intervals around the perimeter of each field. September 13. Thus 8 liters of water field (20 were filtered each date. After the Each flag was assigneda number so that stations dips) on could be randomly selectedas monitoring sites mosquito Iarvae and other aquatic organisms for the fish, mosquito and other aquatic insect were identified, the zooplankton were stained populations. with rose bengal to aid identification; they were preserved,with Every two weeks,four minnow traps (3.2mm then 5% formalin solution.Two mesh) per field were set overnight (ca. 24 hr.) to 5 ml subsampleswere drawn from eachconcen- trated 200 ml plot sample,and the zooplankton were identified and counted at 30x magnifica- tion. The Bti was applied at 6 ke/ha with a back- pack blower to the nine Bti fields on August 20 and again to the three Bti only fields on Septem- ber 8. Treatments were made when mosquito population densities were relatively high and near peak numbers, as based on expectedpop- ulation trends. Treated and control fields were sampled1, 3 and 5 dayspost-treatment after the first application, and 2 dayspost-treatment after the secondapplication. co.tro. r.r ffi ffi .or"^o^rcur,^ ffi r.. ror*^o^".ur,^ One-wayanalysis of varianceand Tukey'stest W;:"tl**" m*'xo/a^q"!', (for pairwiseconrparisons, P : 0.05)were used F.,J;:' to detect differences in the mosquito, aquatic Fig. 1. Experimental designof wild rice plots (water insect and zooplankton populations among the flow indicated by arrows), Lake County, CA, 1982. treatments. The mosquito population data were /1q JouRNer, oF THEArr,rnRrceN Mosqumo CoNrRor, Assocurror.r Vol.4, No.4

analyzedin two groups:1) Control, l.L kg/ha G. abundant prey population, a larger proportion affinis and 3.4 kg/ha G. affinis (to evaluate the of pregnant femalesreleased into the fields and impact of G. affinis) and 2) Control, Btl only, an earlier (by 10 days)fish releasedate. 1.1 kglha G. affinis plus Btl and 3.4 kg/ha G. Although the 3.4 kg/ha releaserate is three affinis plus Bti (to evaluate the impact of Btj times greater than the 1.1 kg/ha rate, fish trap alone and in combination with G. ofinrs). Since counts throughout the season seldom ap- the effect of Bti on rice field nontarget orga- proacheda three-folddifference (Fig. 2). Thus, nisms appearsto be negligible (except for mor- the stocking of G. affinis at increasingly higher tality among some chironomids and dixids) rates does not appear to yield proportionally (Garcia et al. 1980,Garcia 1986,Miura et al. greater fish populations. This result may be 1980),the six treatment goups were combined related to factors that determine the carrying into three groups-l) control plus Bti only, capacity of the rice field habitat. As Norland 2) all 1.1 kg/ha G. affinis f:'eldsand 3) all 3.4 and Bowland (1976) stated, fish stocking rates ke/ha G. affinis fie\ds-to analyze the aquatic do not necessarily determine ultimate popula- insect and zooplankton populations. tion density, and food supply is likely the lim- iting factor to final population numbers.In their RESULTS AND DISCUSSION study, two white rice fields stocked at the same rate (1.1 kg/ha mature female G. affinis) had The seasonalaverage minimum water tem- vastly different final (September) trap counts perature was 17"C and the averagemaximum (30 vs. 130 fish/trap). Another field stockedat 30"C. Water quality was similar among the 3.4 kg/ha of mature femalesyielded a frnal catch treatments. The nitrate concentration on Sep- of 150fish/trap, substantially lessthan the Lake tember 13 averaged0.34 ppm (range0.31-0.40), County wild rice catch at the same releaserate. phosphate0.76 ppm (0.47-0.86),alkalinity 207 Their fish populations (in the l.t, 2.2 and 3.4 ppm (200-220),hardness 153 ppm (110-190), kg/ha fields) Ieveled off after ca. 4 weeks sug- conductivity 392 micromhos/cm(380-410) and gesting the carrying capacity of the fields had pH 7.6 (7.5-7.9).Water depth averaged17 cm beenreached. and maximum plant height was 2.6 m. Based on the regressionequation developed The G. affinis population increased steadily by Stewart and Miura (1985) to estimate abso- throughout the growing season (Fig. 2). The lute densities of G. affinis in white rice fields, averagetrap count at the end of the seasonwas the 1987 Lake County trap counts of 182 and 182 fish/trap in the Iow rate fish fields and 230 230 in the low and high fish fields respectively fish/trap at the high releaserate. The interior are equivalent to densitiesof ca. 586,000and trap counts were similar to the perimeter 746,000fish/ha (ca. 244 and 311 kg/ha since catches.These counts were well above the 1986 about 2,400fish, including fry, malesand mature peak trap counts of 20 and 76 fish/trap in the females,captured from the wild rice fields at the 0.6 and 1.7 kglha fields, respectively(Kramer et end of the season,equaled one kilogram). The al. 1987).The much Iarger G. affinis population growth curvesof Stewart and Miura (1985)in in 1987than 1986may havebeen due to a more white rice estimate a maximum carrying capacity of ca. 120,000fish/ha. Our studiesindicate that wild rice fields may support higher popu- - coNnoL lation densities of G. affinis lhan white rice __ 1.1 KA/V W (e.g., mark -...--- t.1 KC/A W fields; however, additional studies i r., *o/u * and recapture)would haveto be applied to verify Z ".. "o/o * this hypothesis. Studies in the Central Valley : 200 haveindicated that the nutrient content ofwild is greater that ofwhite rice E rice field water than '-- - water (Kramer and Garcia 1988). populations in G. affinis C Immature mosquito treated fields were significantly (P < 0.01) lower 5 3 t.s than in control fields from August 11 until harvest (Fig. 2). The late instar (3rd and 4th) populations were separately evaluated and also found to be significantly lower. The G. affinis trap countson August 11 were 101and lillttap !o 7 t4 21 2A 1 tt t! 25 2 9 t6 JUf,E JULY AUOUST AEPT. in the low and high fish fields, respectively.The Fig.2. Gambusiaaffinis population in wild rice fields data imply that a count of more than 100 fish/ (right axis-bars) and larval mosquito populations in trap will effectively control mosquitoes. G. affinis treated and control wild rice fields (left The perimeter mosquito counts averaged(all axis-lines), Lake County,CA, 1987. fields combined throughout the entire season) DECEMBER1988 GAntaush AFFrNrsAND Bfl IN WrLD RIcu FIor,os +lJ

0.31 larvae/dip higher than the interior dip these two groupson all previous sampling dates. counts. However, differences were not signifi- Applications of Bti were applied to all Bti cant. fieldson August20 (Fig.4).The mosquitolarvae The larval Cx. tarsalis population was much were reducedby 72%,70% and 38Voin the Bti lower in 1987than in 1986,when it reacheda only, low fish plus Bti and high fish plus Btl maximumof 1.6larvae/dip (Kramer et al. 1987). fields, respectively,one day post-treatment. The In 1987, Cx. tarsalis reacheda maximum of 0.7 percent mortality in the interior of the fields larvae/dip (l6Vo ofthe total larval count) in the was slightly less than the perimeter reduction. control plots in early September.The population These reduction rates are low compared to re- never exceeded0.2 larvae/dip (7%) in the fish- duction rates achieved in other Lake County treatedfields or 0.08(6%) inthe Bti only fields. wild rice fields (>95% mortality) where the Anophelesfranciscanus larvae comprised ca. plants were of similar height and the application 5% ofthe total anophelinecount (basedon 4th procedure and dosagerate were the same but instar identification); whereas,in 1986, 40% of the field size many times larger (Garcia and the anophelines were An. franciscanus. Thus Colwell 1987, personal communication). The Anophelesfreeborni clearly dominated in 1987 relatively high rate of water flow through the with a peakof 4.7lawae/dip in the control fields, comparatively smaller experimental rice plots which exceededthe peak larval count for all might have diluted the Bti and lessenedits ef- speciescombined (3.6/dip) in 1986. fectiveness. Culex tarsalis was the most abundant species All post-treatment larval densities were sig- in the adult Iight trap collections (Fig. 3). The nificantly (P < 0.01) less than densities in the peak Cr. farsolispopulation was 444 femalesper control fields (pretreatmentpopulation numbers trap night in mid-July. Anophelesfreeborni and were not significantly different). Mosquito den- An. franciscanuspeaked at 110 and 70 females sities in the G. affinis plus Btl fields remained per trap night, respectively,in early September. significantly lower than in the controls for the Nearby breeding sources, including sloughs, duration of the season.On September1 and 8, ditches and adjacent commercialwild rice fields the number of larvae in the Bti only fields did (planted one month after the experimental not differ significantly from the control density, fields), probably contributed substantially to the and by September8, the larval number in these Iight trap collections.Differential speciesattrac- plots was significantly greater than in the high tion to the light trap may also have influenced G. affinis plus Bti fields. Apparently, the mos- the proportion of speciescollected. quito populations in the Bti only fields re- Just prior to Bti treatment on August 20, bounded while the populations in the G. affinis Iarval mosquito densitiesin the low fish plus Bti fields were kept at a low level by the fish. Stew- fields were substantially greater than in the low art et al. (1983),working with white rice fields, fish only fields (2.6 versus 0.9 larvae/dip). This obtained the greatestmosquito control in a field was probably due to habitat and sampling vari- treated with Bti and stockedwith G. affinis (0.2 ation as fish populations were similar (97 versus kglha). 104 fish/trap on Augrrst 12 in the Iow fish plus The Bti only fields were treated a secondtime Btl and low fish only fields, respectively).Larval on September 8, and water flow through the populations were relatively similar between

_ comoL

.-- 4 oNLY

__ 1.1 RO/M 6,@44

.,-.--- 5.4 KO,/U W & 4 550 3.5

H ".o - 6 - 250 L 2.s

zoo I 2 z.o J

100

o o 30 7 11 21 2a 4 1r ra 25 2 a UAY JUXE JULY UOUSI SEPI- JUTE JULY AUOUST SEPT. Fig. 3. Light trap counts of Culex tarsalis, Anophelcs Fig. 4. Larval mosquito densities in Btj treated and freeborni and.An. franciscanus females adjacent to wild control wild rice fields, Lake County, CA, 1987. Bti rice fields, Lake County, CA, 1987. applied on August 20 (arrow). 474 ,J0URNALoF THE AMERTCANMosquno CoNrRol Assocrerton VoL.4,No.4

fields was stopped for 24 hours following the tured by the minnow traps, which only retained treatment. The mortality rate of 8b% exceeded organisms greater than 4 mm in width. Five of that of the first treatment. The two day post- the six insect groups monitored by the minnow treatment population was significantly (P < traps and four ofthe sevengroups monitoredby 0.01) smaller than the control population.The dipping had population densities significantly fields were drained shortly after this monitoring lower in the fish-treated fields than in the con- date. trol fields on one or more sampling dates (Figs. Odonata, Corixidae, Belostomatidae, Noto- 5 and 6). Speciescollected in 1987were similar nectidae, Hydrophilidae and Dytiscidae popu- to those in the 1986 wild rice frelds and are lations were monitored with minnow traps (Fig. reportedby Kramer et al. (1987). 5) and by dipping (Fig. 6). The Ephemeroptera Dragonfly, primarily An ax junius (Drury) and were sampled only by dipping as few were cap- PantaLahymenaea (Say), and damselfly,primar-

A E o= ,.o z

=.! = E.5 trJ q. G ut o =.2 zf ."1

I 15 2t 12 26 t I 15 29 12 26 I JULY AUGUST SEPT. JULY AUGUST SEPT. Fig. 5. Population densities of (A) Odonata, (B) Corixidae, (C) Belostomatidae, (D) Notonectidae, (E) Hydrophilidae and (F) Dytiscidae (number per minnow trap) in wild rice fields with and without Gambusta affinis,Lake County, 1987 (no G. affinis -, 1.1 kelha G. affinis - - -,3.4 kg/ha G. affinis . . . .). DncnNrspn1988 GAMBUSTA AFFrNrs AND B?I rN Wrr-n RIcr FInr,os

ily Ennllagm,acarunculatum Morse, numbers in Numbers of the water boatmen, Corisellade- the minnow traps were low and there were no color Uhler and HesperocorixaIneuigata (Uhler) significant differences among the treatment (Corixidae),were significantly higher in minnow groups.Dip counts of Odonata naiads were sig- traps in the control fields than in the fields with nificantly (P < 0,01) greater in the control fields the high rate of G. affinis on July 15 (P < 0.05) than in the G. affinis-treatedfields at the end of and greater than all fish-treated fields on July the season (September 8 and 15). The mayfly 29 (P < 0.01). However,maximum population population, primarily Calliboetissp. (Ephemer- density was only ca. 2 corixids/minnow trap. No optera), peaked in early July and no differences significant differenceswere found among popu- were detected among populations in fields with lations monitored by dipping. A giant water bug, or without fish. Belostorna flumineum Say (Belostomatidae),

o o. 6 o 25 co

G u,l O. rs E lrJ Ero = f Zr

25 30 7 t4 2t 2a 4 !! !t 25 I A 13 JUNE JULY AUGUST SEPT. JUNE JULY AUGUST SEPT. Fig. 6. Population densities of (A) Odonata, (B) Corixidae, (C) Belostomatidae, (D) Notonectidae, (E) Hydrophilidae and (F) Dytiscidae (number per 180 dips) in wild rice fields with and without Gambusiaaffinis, Lake County, 1987(no G. affinis-, 1.1ke/ha G. affinis- - -,3.4 kg/ha G. affinis . . . .). 476 JouRNlr, oF THE AMERTcANMosquno CoNrRol AssocrATroN VoL.4,No.4

was significantly more abundant in the control in G. affinis fields (Farley and Younce 19?2, frelds than in the fish-treated fields toward the Miura et al. 1984). In Lake County wild rice end of the growing season(August 26 and Sep- fields in 1986, when the G. affinis population tember 9) as monitored by minnow traps (P < was much lower than in 1987, there were no 0.01)and by dipping (P < 0.05). significant differencesamong insect populations Backswimmer populations, primarily Noto- in fields with or without fish (Kramer et al. necta unifasciofo Guerin and N. undulata Sav 1987).The impact of G. affinis on aquatic insects (Notonectidae), peakedin mid-July and minnow probably varies dependingon fish numbers,the trap counts were significantly (P < 0.01) greater availability of refugia and alternative prey den- in the control fields than in the G. affinistreated sities as well as other factors. fields from August 12 to Septemberg. The no- Two cladocerans(Ceriodaphnin sp. and Clry- tonectids monitored by dipping showed signifi- dorus sp.), a copepod(Cyclops sp.), ostracods cant differences between the control and fish- and chironomids were commonly found in the treated fields earlier in the season.The control zooplankton samples (Fig. 7). Total body field density was significantly greater than that Iengthsof Ceriodaphmroranged from 0.27 to0.74 in the high G. affinis fields on July 21 and 28 (P mm and of Chydorusfrom 0.17to 0.39 mm (40 < 0.05) and than in all fish-treated fields on individuals of each measured).In August, Cer- August 11 (P < 0.01).The bias of the minnow iodaphnia were significantly (P < 0.01) more traps for the larger (later instar) notonectids abundant in the control fields than in all G. and of the dipping for the smaller (early instar) affinis tueatedfields. No significant differences notonectids may explain why the two sampling were detected among the copepod,ostracod or regimes detected population differences on dif- chironomid populations in fields with or without ferent dates. fish. Adult beetles were monitored primarily with All zooplanktonpopulations increased in Sep- minnow traps while beetle larvae were sampled tember, and the Ceriodaphnia,Cyclops and os- via dip counts.Water scavengerbeetles, primar- tracod populationswere significantly(P < 0.01, rIy Tropisternus lateralis (Fabricius) and Hydro- philus triangularis Say (Hydrophilidae), were abundant throughout the season, reaching a 160 maximum of 23/ftap in the controls in mid-July. %- A* The hydrophilid density was significantly (P < i50 U "** 0.01)greater in the control fields than in the G. N o"*.oo^ affinis f,teatedfields on the final trapping date, E aa,.o*o*o* Septernber 9. No significant differences were detected among populations monitored by dipping. Predaceousdiving water beetles,primarily ao Therrnonectesbassilarus insi.gnis (McWilliams), Rhantus gutticollis (Say) and Laccophilus sp. E lrJ (Dytiscidae), were less abundant than hydro- F philids. There were significantly more dytiscid = Eto beetlestrapped in the control fields than in all lr,l (P 4o fish-treated fields on August 12 < 0.01) and E llJ soo significantly more than in the low G. affinis @ fields on August 26 (P < 0.05). Dytiscids col- = J 45o lected by dip sampling were more abundant in z the control fields than in either or both the fish- treated fields on August 4 and August 18 to September8. In general, significant differences among aquatic insect groups in fields with and without G. affinis were detected during the later part of the growing seasonwhen the G. affinis population was relatively high. Other studies have shown that populations of notonectids and dam- selflies(Farley and Younce 1977b,Miura et al. 1984), dragonflies (Farley and Younce 1977b), and mayflies and chironomids (Miura et al. NO FISH 1.1 KG,/HA3.4 KG,/HA 1984) were significantly lower in G. affinis- Fig.7. Zooplankton populations on (A) August 3 treated fields. Aquatic beetle,corixid, and belos- and (B) September13 in wild rice fields, Lake County, tomatid populations were not significantly lower 1987. DECEMBER1988 Gtwausu, AFFrNrsaNo Brr rN Wrr,o RIcn Ftnr,os

0.01, and 0.05, respectively)lower in the G. tected in the fish guts. Although the mosquito o/fnis versus control fields. Chydorus and chi- population was not significantly reducedby G. ronomid populations were not significantly affinis in 1986 (Kramer et al. 1987),the per- lower in the fish-treatedfields. In contrast to centageof fish (9%) containing mosquito Iarvae these data, Miura et al. (1984) found that G. was similar to the 1987finding. Therefore,the affinis did not reduce populations of copepods percentageof fish containing mosquitoes does or ostracods. Bay and Anderson (1966) found not provide an index of the control effectiveness that G. affinis did not reduce populations of of G. affinis. The percent of fish guts containing chironomids even at a fish density of more than larvae seemsto be, in part, a function of the 250lbs/acre; whereas, Miura et al. (1984)found relative abundanceof mosquito larvae and fish. that the fish significantly reduced chironomid Other factors, such as availability ofalternative larvae. Different sampling regimes (benthic vs. prey and prey accessibility, undoubtedly influ- free-swimming)or the presenceof different chi- encethe number of mosquito larvae consumed. ronomid species may account for the varying Gambwia affinis had an extensive array of results. alternative prey available in the wild rice fields. Apparently G. affinis prefer to feed on the Cladocerans(primarily Chydorusand someCer- larger Ceriodaphniathan the smaller Chydorus. iodaphnia) were the most abundant organisms Gambusiaaffinis has been shown to essentially in the fish guts. Copepods,ostracods and rotifers eliminate Ceriodaphnia in experimental ponds were also found. Forty chironomid larvae were (Hurlbert and Mulla 1981).Bence and Murdoch found in 20 (16.7%)of the fish, and there were (1983) found that G. affinis rcduced the abun- 20 hydrophilid larvae in 8 (6.7Vo)of the fish dance and mean size of zooplankton in white dissected.Other insects found in the guts of G. rice fields (in fields without fish, zooplankton affinis included the following: 2 Anisoptera, 1 were(1.6 mm and with fish. <0.8 mm). Zygoptera, 1 Cercopidae,1 Corixidae, 2 Hom- Since Bti potentially reducespopulations of optera, S Thysanoptera, 1 Cecidomyidae,1 Stra- some chironomids,the data were analyzedto tiomyidae, 4 other Diptera larvae and 1 Diptera detect differences among chironomid popula- adult. Eighty-six snails (Physa) were found in 9 tions in fields treated with Bti versusuntreated (7.5%) fish (one fish had consumed38 snails fields. No significant differenceswere found (av- and a second,20). One fish contained} G. affinis erage of 12 and 12.7 chironomids/liter in the fry. Five fish (4.2%) had tapeworms (Bothrioce- untreated and Bti treated fields, respectively). phalus)in their gut. Algae was also found in the However, only free swimming larvae were sam- guts of many fish. pled (not epiphyticor benthic larvae). Eight of the 39 fish containing insects or Ofthe 120G. affinis dissectedfor gut analyses, snails had more than 7 organisms of the same 427ocontained zooplankton only (fish sizerange type (mosquitoes,chironomids, hydrophilids or 16-50 mm standardlength (S.L.), | : 27 mm), snails) in their guts. These numbers are more 28% had zooplankton and insects or snails (17- than would be expected by random encounter 50 mm S.L., i = 29),17% had insectsonly (19- and support the notion of a prey searchprefer- 52 mm S.L., i : 31) and 39% had empty guts ence. The omnivorous feeding behavior of G. (16-50 mm S.L., x : 31). Basedon these data, affinis is evident by these gut dissections and there does not appear to be a correlation be- has been reported by many researchers(Hess tween fish size and generalprey preference. and Tarzwell 1942,Washino and Hokama 1967, Nine female frsh (7.5% of those dissected), Miura et al. 1979,Farley 1980). ranging in sizefrom 25to 47 mm S.L., contained In conclusion, G. affinis significantly reduced 58 mosquito larvae (all anophelines:6 first in- the mosquito population in Lake County wild star,27 second,12 third, 12 fourth and 1 pupa). rice fields at fish densities exceeding100/trap. One fish (female,44 mm S.L.) had consumed38 In 1986, when mosquito populations were not larvae (2nd-4th instars) and a secondfish (fe- significantly different in fields with and without maIe,25mm S.L.) contained10 larvae (lst-3rd fish, densities of G. affinis did not reach this instars).The presenceof largenumbers of mos- level(Kramer et al. 1987).Since one release rate quito larvae in only 2 of the 120 fish examined in 1987 (1.1 kg/ha) was Iess than one 1986 suggeststhat some individuals form a search releaserate (1.7 kg/ha), the assumption that G. preferenceand consistently seek the same prey affinis will be an effective control agent of mos- item. The remaining seven G. affinis had, l-B quitoes at a given release rate is not always Iarvae in their guts. The fish containing mos- reliable.These studies indicate that the G. affinis quito larvae were only found in half of the fields population must be monitored post-releaseto sampled.These fields had a higher averagemos- assessits control potential. quito population (0.68/dip vs. 0.27ldip) and Granular Bti effectively reduced mosquito lower G. affinis population (112ltrap vs. 150/ populations in wild rice. When G. affinis were trap) than the fields where no larvae were de- present, the larval populations did not rebound JouRNu, oF THE AMERTcANMosqurro Cot.rRor, ASsocrATroN VoL.4,No.4 after a Bti treatment. Thus effectivemosquito Hu, pp.14b-b0. 1n: T. D. st. George,B. H. Kayand control can be achievedin wild rice when fields J. Blok (eds.),proc. 4th Symp.,{.6ovirus Res. in are stockedat 1.1kg/ha of G. affinis,monitored Aust.,Q.LM.R. Brisbane. Hess, and treated with Bti when the mosquito popu-' A. D. and C. M. Tarzwell.1942. The feeding lation increasesbeyond an acceptablelevei. habitsof Gambusiaaffinis affinis, with specialref- erence to the malaria mosquito Anophclcs quadri- morulatus. Am. J. Hyg. 35:142-751. ACKNOWLEDGMENTS Hoy, J. B. and D. E. Reed.1970. Biological control of Culex tarsalis in a California rice field. Mosq. News This study was supported in part by special 30:222-230. state funds for mosquito researchin California Hurlbert, S. H. and M. S. Mulla. 1981.Impacts of mosquitofish (Gambusiaaffinis) predation plank- and conductedin partial fulfillment of the ph.D. on ton communities.Hydrobiologia 83:125-151. degree,Department of Entomological Sciences, Kramer, V. L., R. Garcia and A. E. Colwell. 198?.An University of California at Berkeley. We thank evaluation of the mosquitofish, Gambusia affinis, David Woodward, Norman Anderson. Bill Dav- and the inland silverside, Menidia beryllino, as mos- idsonand William Voigt for their assistanceand quito control agentsin California wild rice fields. J. cooperation, and Bruce Eldridge for his com- Am. Mosq. Control Assoc.3:626-632. ments on this manuscript. Kramer, V. L. and R. Garcia.1988. A comparisonof mosquito population density, developmental rate and ovipositional preference in wild versus whi e REFERENCES CITED rice fields in the Central Valley. Proc. Calif. Mosq. Vector Control Assoc.In press. Bay,E. C. andL. D. Anderson.1966. Studies with the Lacey,L. A. and A. H. Undeen. 1986.Microbial control mosquitofish,Garnbusia affinis, as a chironomid of black flies and mosquitoes.Annu. Rev. Entomol. control.Ann. Entomol. Soc. Am. 59:150-153. 3l:265-296. Bence,J. R. and W. W. Murdoch.1983. Ecological Lichtenberg, E. R. and W. Getz. 1985.Economics of studiesof insectpredators and, Gambusio in rice rice-field mosquitocontrol in California. BioScience fields:a preliminaryreport. Proc. Calif. Mosq. Vec- 35:292-297. tor ControlAssoc. 50:48-50. Miura, T., R. M. Takahashiand R. J. Stewart.1979. Combs, J. C. (ed.). 1986. California Mosopito and Habitat and food selection by the mosquitofish Vector Control Association Yearbook. pp. 32-33. Gambusiaaffinis. Proc. Calif. Mosq. Vector Control CMVCA Press. Assoc.47:46-50. Collins, F. H. and R. K. Washino. 1980. The effects Miura, T., R. M. Takahashi and F. S. Mulligan, III. of irrigation water sourceand crop rotation on the 1980. Effects of the bacterial mosquito larvacide, abundanceof Culex tarsalisin California rice fields. Bacillus thuringiensis serotype H-14 on selected Proc. Calif. Mosq. Vector Control Assoc. 48: aquatic organisms.Mosq. News 40:619-22. 103-108. Miura. T.. R. M. Takahashiand W. H. Wilder. 1984. Craven,B. R. and C. D. Steelman.1968. Studies on a Impact of the mosquitofish (Gambtnia affinis) on a biological and chemical method of controlling the rice field ecosystemwhen usedas a mosquito control dark rice field mosquito in Louisiana. J. Econ. En- agent.Mosq. News44:510-51?. tomol. 61:1333-36. Mulhern, T. D. 1942.New Jerseymechanical trap for Farley, D. G. 1980.Prey selectionby the mosquitofish mosquito surveys. New Jers. Agr. Exp. Stn. Circ. Gambtrciaaffinis in Fresno County rice fields. Proc. 421:l-8. Calif. Mosq.Vector Control Assoc.48:51-54. Norland, R. L. and J. R. Bowland. 1976.Population Farley,D. G. and L. C. Younce.1977a. Stocking date studies of Gambusiaaffinis in rice fields: sampling versusefficacy of Gambusiaaffinis in Fresno County design,fish movement and distribution. Proc. Calif. rice fields. Proc. Calif. Mosq. Vector Control Assoc. Mosq.Control Assoc.44:53-56. 45:83-86. Stewart, R. J. and T. Miura. 1985.Density estimation Farley, D. G. and L. C. Younce. 1977b. Effect of and population gxowth of mosquitofish (Gambusia Gambusia affinis (Baird, and Girard) on selected affinis) in rice fields. J. Am. Mosq. Control Assoc. non-target organisms in Fresno County rice fields. 1:8-13. Proc. Calif. Mosq. and Vector Control Assoc. 45: Stewart. R. L.. C. H. Schaefer and T. Miura. 1983. 87-94. Sarnpling Culex tarsalis (Diptera: Culicidae) imma- Garcia.R.. B. DesRochersand W. Tozer. 1980.Studies tures on rice fields treated with combinations of on the toxicity of Bacillus thuringiensis var. israe- mosquitofish and.Bacillus thuringiensis H-14 toxin. lensls against organisms found in association with J. Econ.Entomol. 76:91-95. mosquito larvae. Proc. Calif. Mosq. Vector Control Washino, R. K. and Y. Hokama. 1967. Preliminary Assoc.48:33-36. report on the feeding pattern of two speciesof frsh Garcia, R. 1986. Strategies for the management of in a rice habitat. Proc. Calif. Mosq. Control Assoc. mosquito populations with Bacillw thuringi.ensis 35:84-87.