Journal of the American Mosquito Control Association, 15(3):380-390, 1999Copyright O 1999 by the American Mosquito Control Association, lnc.

COMPARATIVE EFFICACY OF THE THREESPINE STICKLEBACK(GASTEROSTEU S ACULEATU,S) AND THE MOSQUITOFISH

(GAMBUSIA AFFINIS) FOR MOSQUITO CONTROL

YVONNE A. OFFILL aNo WILLIAM E. WALTON'

Department of Entomology, University of California, Riverside, CA 92521

ABSTRACT, The effectiveness of the threespine stickleback as a mosquito control agent was compared tothat of the mosquitofish in 28-m'� earthen ponds during 2 6-wk experiments where the 2 fish were stocked aloneand together. Relative to ponds without fish, the stickleback was not effective for controlling larval mosquitopopulations; however, sticklebacks reduced the abundance of Culex pupae. Mosquitofish provided significantlevels of control whether stocked alone or concurrently with the stickleback. As compared to mosquitofish alone,mosquito control was not significantly enhanced when both fish were stocked together. Mortality of adult stick-lebacks was related to a gradient of increasing water temperature across the ponds rather than the direct effectsof other abiotic factors such as low dissolved oxygen concentrations or biotic interactions with the mosquitofish.The stickleback exhibited a lower thermal tolerance and slower population recruitment as compared to themosquitofish populations, which reproduced successfully in water >33"C and grew rapidly. Stickleback bromasseither declined or increased slightly (-5OVo of initial stocking weighQ. Mosquitofish biomass increased 33- to38-fold at rates averaging between 0.079 and 0.095 g wet weight/g/day and total wet weight per pond at 6 wkafter stocking did not differ significantly between the 2 mosquitofish treatments.

KEY WORDS Culex, larvlorous fish, biological control, mosquitohsh, stickleback

INTRODUCTION

The mosquitofish (Gambusia affinis Baird andGirard) is the most widely distributed larvivorousfish used for mosquito control (Meisch 1985). Theeffectiveness of G. affinis as a mosquito controlagent depends on the abundance of vegetation, theabundance ofmosquitoes and relative abundance ofother prey, the abundance of mosquitofish preda-tors, and biotic factors affecting mosquitofish re-production such as fecundity and seasonal breedingcycles (Sawara 1974, Gratz et al. 1996). Gambusiahas a broad diet that includes phytoplankton, zoo-plankton, aquatic insects, gastropods, terrestrial in-sects trapped on the water surface, and eggs andyoung of fish (Washino and Hokama 1967, Laiu:d1977,Farley 1980, Hurlberr and Mulla 1981, Miuraet al. 1984, Bay 1985). Although mosquitofish re-production varies seasonally (Sawara 1974, Ceclnand Linden 1987). maximum recruitment often co-incides with peaks in host-seeking adult mosquitoactivity throughout much of California and thefish's geographic range. Not surprisingly, mosqui-tofish perform best as a control agent for mosqui-toes in situations where vegetation, alternative foodsources to mosquitoes, and predatory fish are lim-ited.

The use of mosquitofish has become controver-sial in recent years because Gambusia is purportedto affect biodiversity and the abundance of localfauna and, furthermore, the mosquitofish is an ef-fective biological control agent for mosquitoes in asubset of habitats to which the fish has been pur-posefully introduced (Gamradt and Kats 1996,Gratz et al. 1996, Rupp 1996). This debate has

come to the forefront in southern California and inother arid regions where multipurpose constructedwetlands are used for water reclamation, recreation,and wildlife habitat for the maintenance of localand regional biodiversity. Because treatment wet-lands can support populations of pestiferous anddisease-vectoring mosquitoes and are frequentlywithin proximity to human development, interven-tion to control mosquitoes is often necessary (Mor-tenson 1982, Carlson et al. 1986, Walton et al.1998). ln order to fulfill the objectives of maintain-ing local biodiversity and minimizing mosquitoproduction in constructed treatment wetlands, aneffective alternative larvivorous fish to the mosqui-tofish is needed.

The threespine stickleback (Gasterosteus aculea-tus L.) has been suggested as an alternative biolog-ical control agent for mosquitoes because of itsfood preferences, behavioral characteristics, andwide geographic distribution (Hubbs 1919, Waltonet al. 1996). The threespine stickleback is an en-demic species that is distributed throughout Cali-fornia, inhabiting streams, rivers, lakes, and brack-ish waters (Swift et al. 1993). The stickleback ispredaceous and feeds on small organisms through-out the water column (Schooley and Page 1984).Studies have shown that the threespine sticklebackhas a feeding preference for mosquito larvae andpupae over various other food items (Bay 1985). Ithas been speculated that simultaneously stockingthe top-feeding mosquitofish with a 2nd fish, suchas the stickleback, that feeds throughout the watercolumn will provide a level of mosquito controlgreater than will stocking either fish alone (Wood-ridge and Davidson 1996).

In this study, the efficacy of the threespine stick-

380

rTo whom correspondence should be addressed

SepreNrenn 1999 Mosquno Corurnol st G. ecutetrus vs. G. errr,vrs 3 8 1

A

Both

Mosguitofish

Stickleback

Conilol

Not used

Fig. l. The arrangement for treatments assigned toponds at the University of California-Riverside AquaticResearch Facility, Riverside, CA, during 2 studies: (A)1996 and (B) 1997. Water flow is represented bv arrows.

leback as a mosquito control agent was comparedto that of the mosquitofish and naturally occurringmacroinvertebrate predators in ponds without fish.The potential for the coexistence of the 2 speciesand the effectiveness of concurrent stocking ofbothfish as mosquito conffol agents were also examined.

MATERIALS AND METHODS

Study site: Two experiments were carried out in12 mesocosms (4 m X 7 m) at the University ofCalifornia-Riverside (UCR) Aquatic Research Fa-cility. The lst study was carried out from Jlune 23to August 12, 1996, and the 2nd study was carriedout from April 2 to May 31, 1997. The mesocosmswere :urzrnged into 2 rows (rows C and D) (Fig. l)and water was supplied through a single pipelinefrom a reservoir at the Agricultural Experiment Sta-tion, UCR. All ponds were equipped with floatvalves that maintained the water depth at approxi-mately 0.3 m. Before the 1996 study, all of theponds were excavated to approximately 0.16 m,lined with 0.015-mm-thick plastic sheeting, and soilwas then replaced over the sheeting.

Upon flooding, each pond was enriched with 3.5kg of rabbit pellets (Forti-diet, Kaytee Products

Inc., Chilton, WI) to promote oviposition by mos-quitoes. During the 2nd study, a 2nd enrichmentwas carried out at 36 days after flooding on May8, 1997. Five days postflooding (April 7, 1997), allponds were treated with bifentMn (Capture@ FMCCorporation, Philidelphia, PA) 2EC (20Vo active in-gredient [AI]) at a rate of 0.5 g AI/ha to eliminatepopulations of predaceous tadpole shrimp (Triopslongicaudatus Le Conte). This short-lived chemicaltreatment has no effect on tJle mosquitoes (Mullaet al. 1992).

Because the mesocosms were devoid of naturalvegetation, artificial vegetation was constructed andintroduced to each pond before stocking flsh. Ar-tificial vegetation consisted of 0.015-mm-thickblack plastic sheeting cut into 61 X 5.1-cm strips(Offill 1998'�). Strips were folded and stapled toform a frond. Five fronds were positioned equidis-tantly around a l5-cm-diameter circle of 22-gaugegalvanized steel wire and weighted with 7+-in. hexnuts. Vegetation was placed in each corner and inthe center of each pond to provide visual oviposi-tion cues for mosquitoes and to provide shelter forfish.

Physicochemical measurements: Water temper-atures were measured using maximum-minimumrecording thermometers (Markson Scientific Inc.,Del Mar, CA). The northern ponds in the rows werepartially shaded, whereas the southern ponds re-ceived full sun. Temperafures were measured inponds Dl, C4, and D6 to provide representativetemperature data at various positions within the ar-ray of ponds. Thermometers were positioned at ap-proximately 25 cm from the water surface and wereread every 48-72lr for the duration of the studies.

The pH, water temperature, and dissolved oxy-gen concentration were recorded at 30-min inter-vals over 24 h using electronic sensors (ICM WaterAnalyzers, Perstorp Analytical, Wilsonville, OR)during the final week of the 2nd study. Measure-ments were taken at a depth of 18 cm in ponds D1and D6 on May 27-28, 1997.ln order to determinethe effect of organic enrichment on pH and the dis-solved oxygen concentration, ponds Dl and D6were again effiched after the 2nd experiment (June6-7, 1997). The pH, water temperature and dis-solved oxygen concentrations were measured underenrichment conditions at 3o-min intervals for a 24-h period.

Mosquitoes and nontarget organisms: Mosqui-toes and nontarget organisms were sampled by dip-per (350 ml) and by tow net (mesh aperture size =153 pm). Four dips were collected twice each weekin the corners of each pond and combined using aconcentrator cup (mesh opening : 200 U,m). Du-plicate tow net hauls were taken weekly along the

'� Offill, Y. A. 1998. Comparison of the effectiveness ofthe three-spined stickleback (Gas te rosteus aculeatus) andthe mosquitofish (Gambusia affinis) for mosquito control.M.S. thesis. University of California, Riverside, CA.

tlI

ffimN

JounNel or rne AMEnTcAN MoseulTo CoNtnol AssocIATIoN VoL. 15, No. 3

long axis in each pond. All samples were collectedbetween 0700 and 1100 h. Specimens were pre-served in alcohol (final concentration approximate-ly SOVo). Insects were counted under a stereodis-secting microscope at 25-50x and nonculicine tiD(awere identified to genera using Merritt and Cum-mins (1984). Mosquito immature stages were sep-arated into early (stages I and II) and late larvalinstars (stages III and IV) and pupae. All late-stagemosquitoes were categorized to species using Bo-hart and Washino (1978).

Treatments were assigned to ponds based on ini-tial mosquito larval densities in pretreatment dipsamples. Four treatments were assigned so that thevariation in mosquito abundance among treatmentswas equivalent. The treatments were control (nofish stocked), stickleback only, mosquitofish only,and both fish stocked together. Tfeatments werereplicated 3 times.

Fisft.' Mosquitofish were supplied by NorthwestMosquito and Vector Control District (Corona, CA)and sticklebacks (Gasterosteus aculeatus micro-cephalus Girard) were collected with D-nets fromthe Mohave River near Apple Valley, CA. Fishwere transported in 32-quarf ice chests containingwater fiom the collection site which was continu-ously aerated by portable air pumps. Upon arrivalat the UCR facility, fish were acclimated for ap-proximately 24 h in a pond similar to those usedfor the study. Fish were held in 3 1.25-m3 cages(1.22 m long x 1.13 m high X 0.91 m wide) con-structed of polyvinyl chloride (PVC) frames cov-ered by fiberglass window screening. The top ofeach cage was covered with plastic netting (BirdBlock@, Easy Gardner Co., Waco, TX) to deter pi-scivorous birds. Minimal losses (<lVo\ occurredduring transport and no losses occurred during ac-climation.

Each mesocosm was stocked at a rate of 2.3 kg/ha (approximately 10-14 fish per pond) for the1996 experiment and 3.4 (each species alone) or4.5 kglha (both fish: approximately 17-22 fish perpond) for the 1997 study. Stocking occurred at 10and 15 days postflooding for each of the 2 experi-ments, respectively. Fish were weighed individuallyand reproductive individuals were distributedequally among treatments. Any fish succumbingduring the 1997 experiment were collected using adip net and measured for standard length. Standardlength was converted to wet weight using weight-length regressions (G. aculeatus: Walton et al.1996; dead G. afinis were not observed).

Fish yield was estimated at the end of each ex-periment by collecting fish from each pond using aseine (0.64-cm mesh opening). Three hauls weretaken from each pond and a gross wet weight ofeach species was measured for each haul. Any fishremaining at 2 days after seining and turning offthe water supply were collected by dip net andweighed. The distribution of fish wet weight withineach treatment was determined by weighing a rep-

resentative sample of 100 G. ffinis from eachpond. All G. aculeatus individuals were weighed.The number of mosquitofish in each pond was es-timated by converting the total wet weight to thenumber of individuals using weight-length regres-sions for a representative sample of fish (n : 168)from the 1996 study. The number of sticklebacksin each pond was determined by direct count.

Statistical analyses: To test for differences in theabundance of mosquitoes among the treatments,abundance data for dip samples were ln-trans-formed and analyzed using a repeated measuresanalysis of variance (ANOVA) (Wilkinson andCoward 1997). Immature mosquitoes were catego-rized into 3 groups: 2 latval subpopulations (lst-and 2nd-stage larvae, 3rd- and 4th-stage larvae)and pupae. Because the variance for each of 2 sub-populations (older larval instars and pupae) was nothomogeneous among the treatments in the 1997study (e.g., no older larval instars were present insome treatments on particular sampling dates), anonparametric repeated-measures ANOVA wasused to test for differences among the 4 treatments.Pairwise comparisons between treatment means foreach of the 3 subpopulations were made using theStudent-Newman-Keuls method (Fox et al. 1995).Similar analyses were carried out for nontarget or-ganisms that were sufficiently abundant to permitstatistical comparisons among the treatments.

In order to compare fish production among thetreatments containing fish, treatment means werecalculated for biomass (wet weight) of each speciesand compared using a /-test. The estimated numberof mosquitofish per pond was compared betweentreatments (mosquitofish only vs. both fish stockedtogether) using a t-test.

RESULTS

Physicochemical factors

A north-south gradient in temperature occurredacross the ponds. Maximum water temperatures ofponds at the south end of the rows were 2-5oCwarmer than were water temperatures in partiallyshaded ponds at the north end of the rows duringthe 1996 study (Fig. 2A). In the warmest pond, Dl,the average maximum water temperature was35.3'C and the minimum was 21.1'C. In C4 andD6, the maximum water temperature averagedacross sample dates was 33.5'C and the averageminimum temperatures were 21.9 and 22.loC, re-spectively.

Maximum water temperatures during the 2O daysafter stocking fish in 1997 (Fig. 2B) were 5-7"Ccooler than those recorded during the 1996 study(Fig. 2A) and the 2nd half of the 1997 study. PondDl was the warmest, with an average maximumwater temperature of 32.5"C and an average mini-mum temperature of 2O.6"C throughout tlle 199'7study. These averages were only slightly higher

Moseutro CoNrnol By G. ecutztrus vs. G. li'rnrs 383

5U

2 5 d3

!t

2 0 $c

" " d

I't0

b

ZJ

r 2 0o

c

9, tsxo!9 r o6.9,

^ 3 5

9 a nl

68. 25E0)F 2 0

+Dl :mi l imum--o-01: minimum+-C4: meximum{FC4: minimum-a-D6: milimum--i-D6: minimum

10 15 20 25

Days after stocking

0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 0

Days after stocking

Fig. 2. Maximum and minimum water temperaturesin 3 ponds at the University of Califomia-RiversideAquatic Research Facility, Riverside, CA, (A) from June23 to August 12, 1996 and (B) from April 2 to May 31,199'7.

than those for ponds C4 and D6, which had averagemaximum temperatures of 31.8 and 3 1.6'C, respec-tively, and average minimum temperatures of 18.7and 19.5"C, respectively.

Dissolved oxygen concentrations at 3 wk afterenrichment fluctuated between 11-13 mgfliter dur-ing a 24-h period (Fig. 3). The dissolved oxygenconcentration in pond Dl dropped to 2 mg/liter dur-ing the night and subsequently increased rapidlybeginning at O9OO h. The amplitude of the diel fluc-tuation of dissolved oxygen concentrations in thecooler pond, D6, was similar to that observed inpond Dl; however, the minimum dissolved oxygenconcentration was approximately 10 mg/liter duringlate May (Fig. 3). Because the northern ponds wereshaded during the morning, the daily increase ofdissolved oxygen concentration in pond D6 beganapproximately 2 h later than in pond D1.

Nighttime dissolved oxygen concentrations de-creased appreciably shortly after enrichment com-pared with dissolved oxygen measurements taken20 or more days following enrichment. Dissolvedoxygen concentration declined abruptly during the

15:00 21:00 3:@ 9:00 15:00 21:00

Time of day

09:00 15:00 21:00 3:@ 9:00 15:00 21:00

Time of day

Fig. 3. Dissolved oxygen concentration (A) and watertemperature (-) measured over a 24-h period (May 27-28, 199'7) in 2 ponds at the University of California-Riv-erside Aquatic Research Facility, Riverside, CA, at 3 wkpostenrichment.

late afternoon and early evening (after 163O h) andwas 0 mg/liter throughout the night (Fig. 4). Dis-solved oxygen concentration remained at 0 rng/literfor approximately 13 h in pond D6 and for t h inpond Dl. After enrichment, the maximum dis-solved oxygen concentrations in the ponds were 9-10 mg/liter.

Mosquitoes and nontarget organisms

The predominant mosquito species collected dur-ing both studies were Culex quinquefasciatzs Say,Culex stigmatosome Dyar, and Culex tarsali,s Coq.During the 1996 study, C.r. tarsalis and Cx. stig-matosoma were more common than was Cx. quin-quefasciatus (average relative abundance : 47,48,and 5Vo, respectively). Culex stigmatosoma and Cr.tarsalis were also prevalent during spring 1997; av-erage relative abundance was 37 and 47Vo, respec-

, t 4

40

;: 35

! *ob o co - vEo' - 2 0

25 -to3

2 0 Eg

1 5 d

I1 0

d ) i

o - -E

coCD ra

o

o) , ^o l v0

.oo1 5

384 JounNnr- oF THE AMERTCAN Moseurro Contnol AssocnrroN VoL. 15, No

35

3 0 i

^ - J' " 9 e2 o E s1 5 ; 6" ^ d P

6

IJoEcoo

o

o_zo

.t2o

30l

25 E'a

E2 0 9

o, . (

1 0 35

0

J

oE

coo

o

0_z

.2

1 0

9

8

7

6

54

1

9:00 15:00 21:OO 3:00

Time of day

1 09I7o

54

'1

09:00 15:00 21:OO 3:00 9:00 15:00

Time of day

Fig. 4. Dissolved oxygen concentration (A) and watertemperature ( ) measured over a 24-h period (June 6-7,1997) in 2 ponds at the University ofCalifornia-RiversideAquatic Research Facility, Riverside, CA, at 2 days afterenrichment.

tively. Only 2 Culiseta sp. larvae were collectedduring the 1997 experiment.

In 1996, all sticklebacks died in both treatments.Because of this mortality, the 1996 data were ana-lyzed for the presence versus absence of mosqui-tofish. Mosquitofish provided significant levels ofcontrol for larval (lst and 2nd instars: F,.,,, : 6.17,P < O.O32; 3rd and 4th instars: F,.,u : 5.96, P <0.035) and pupal (F,., : l4.Ol, P < 0.006) mos-quito subpopulations as compared to ponds withoutfish. Late-stage (3rd and 4th instars) mosquitoabundance declined from approximately 250 larvaeper dip during the lst week after stocking mosqui-tofish to less than 3 larvae per dip from 17 to 3ldays after stocking fish (Fig. 5). In contrast to the100-fold reduction in larval abundance observedbetween days l0 and 17 in ponds containing themosquitofish, the abundance of late instars in pondswithout mosquitofish declined approximately 3O-fold between the lst and 3rd week of the experi-ment. During the last half of the 1996 study, thelate-stage subpopulations in ponds without Gam-busia were 5- to l0-fold larger than in ponds con-taining mosquitofish and averaged between 7 and2O larvae per dip.

Gambusia did not significantly affect the abun-dance of nontarget organisms (repeated measures

'1000

0 5 ' 1 0 1 5 2 0 2 5 3 0 3 5

Days post-stocking

Fig. 5. Abundance (mean + SE) of Culex spp. 3rd-and 4th-stage larvae in dip samples from treatments withand without mosquitofish during the period from hne 23to August 12,1996.

ANOVAS, P > 0.05). Cyclopoid copepods were themost abundant potential predator of mosquito lar-vae during the 1996 experiment (Table 1). Preda-tory insects were rare.

In 1997, sticklebacks failed to provide signifi-cantly better mosquito control than did the naturallyoccurring invertebrate predators in the control treat-ment (Table 2). This was true for both larval mos-quito subpopulations (Table 2). Many of the stick-lebacks survived in the 1997 experiment and,therefore, comparisons of the effects of the 4 treat-ments on mosquitoes and nontarget organisms werepossible. The abundance of both larval subpopula-

Table l. Nontarget taxa collected from experimentalponds at the University of Califomia-Riverside Aquatic

Research Facility in Riverside, CA, from lune 23through August 12, 1996, and from April 2 through

May 31, 1997.

Abundancer

Nontarget group r997

Pond Do

Anisoptera: AeshnidaeAnisoptera: LibellulidaeCeratopogonidaeChironomidaeCladoceraCopepodaCorixidaeDytiscid larvaeEphemeropteraEphydrid larvaeHydrophilid lrvaeLaccophilus spp.NotonectidaeOstracodaVeliidaeZy gopter a: Coenagrionidae

rA, abundant (>10,000 individuals collected); C, common(1,000 < C < 10,000 individuals collected); U, uncommon (100< U < 1,000 individuals collected); R, rare (<100 individuals

collected).

RRUCAARRUURRRCRR

RRRAAAURRRRRARRR

Spsrerraenn 1999 MoseulTo CoNTRoL sy G. tcurcnrus vs. G. arrr,rns 385

Table 2. Pairwise comparisons of larvivorous fish treatment means for the abundance of immature Culex spp.collected in dip samples during 1997.

Stage'Difference of

means or ranks n2Comparison q' P < 0.05LI-LII Stickleback vs. mosquitofi sh

Stickleback vs. bothStickleback vs. controlControl vs. mosquitofishControl vs. bothBoth vs. mosquitofish

Stickleback vs. mosquitofi shStickleback vs. bothStickleback vs. controlControl vs. mosquitofishControl vs. bothBoth vs. mosquitofish

Stickleback vs. mosquitofi shStickleback vs. bothStickleback vs. controlControl vs. mosquitofishControl vs. bothBoth vs. mosquitofish

r.991.550 . 1 8I . 8 1t . 3 7o.44

26.O20.o4.O

22.O16.06.0

22.019.01 . 0

21.018.03 .0

6.705.220.616.094.611.48

5.595.551.576 . 1 06.282.35

4.735.270.395.827.06I . 1 8

4.,2J

22A

J

2-1

2z

4J

25

22

YesYesNoYesYesNo

YesYesNoYesYesNo

YesYesNoYesYesNo

LI[_LIv

Pupae

:- 10000lepP 1000ooo

I 100g

I6 r o

E

l

{ 1

:- 10000.-o

I 1o0oooE9 loogooC a ^o t uufo< 1

rL, larval instan'� n, number of treatments in the compaison.r 4, Student-Newman-Keuls statistic.

A

0 1 0 2 0 3 0 4 0

Days post-stocking

Fig. 6. Abundance (mean + SE) of Culex spp. lst-and 2nd-stage larvae in dip samples from treatments (A)with stickleback and no fish (control) and (B) with mos-quitofish and both mosquitofish and sticklebacks duringthe period from April 2 to lNf.ay 31, 1997. The arrow in-dicates the 2nd enrichment. Points are offset horizontallvto facilitate illustration.

tions (1st and 2nd instars: Fr., : 5.02, P < 0.030;3rd and 4th instars: X2, : 21.89, P < 0.001) andpupae (12. : 18.69, P < 0.01) differed significantlyamong the 4 treatments. Although larval abundancein the stickleback treatment increased lO-fold afterthe 2nd enrichment (Figs. 64. and 7A), the declinein both larval subpopulations across the experimentwas negligible compared to that observed in themosquitofish treatments (Figs. 68 and 7B). Late in-stars in ponds without fish declined approximatelylO-fold during the lst week of the experiment andthen increased to an average of 200 larvae per dipafter the 2nd enrichment of the ponds. Populationsof older larvae in the control ponds declined by 2orders of magnitude between day 2O and 30 (Fig.7A).

The abundance of Culex pupae was greatly re-duced in treatments with mosquitofish and reducedto a lesser extent in ponds containing only the stick-leback (Table 2). Although the number of pupaepresent in the stickleback and control ponds fluc-tuated throughout the study (Fig. 8A), pupal abun-dance in ponds containing sticklebacks alone de-clined until day 2O after stocking and resurgedthereafter to approximately 10 pupae per dip. Thenumber of pupae present in ponds with mosquito-fish or both fish together decreased after the lst 1Odays and remained suppressed throughout the ex-periment (Fig. 8B).

The mosquitofish provided significantly bettercontrol of immature mosquitoes as compared toboth the naturally occurring macroinvertebrates inthe fishless ponds and the stickleback (Table 2).The treatment with both fish present also showed

JounNru- or tne AwsnrclN Mosqurro CoNTRoL AssocrerroN VoL . 15 . No .3

A

.sE

co ' 1 0 0oooo

8 t ococf

I

oc t6n

eE 1 0

^ 1000--eEoco 100o

Eog

8 t ocoE

< 1

.g

;c 100ooEG

o

P r oo

Days post-stocking

Fig. 7. Abundance (mean t SE) of Culex spp. 3rd-and 4th-stage larvae in dip samples from treatments (A)with stickleback and no fish (control) and (B) with mos-quitofish and both mosquitofish and sticklebacks duringthe period from April 2 to May 31, 1997. The arrow in-dicates the 2nd enrichment. Points are offset horizontallvto facilitate illustration.

significantly improved control for all stages of im-mature mosquitoes relative to the control and stick-leback treatments (Table 2). The abundance of lateinstars declined from approximately 5O larvae perdip at the time of stocking to nearly undetectablelevels after 17 days (Fig. 7B). Population trends forthe 2 larval subpopulations were similar in bothGambusia treatments. Larval mosquito abundancedid not differ significantly between the mosquito-fish alone and the mosquitofish * stickleback treat-ments (Thble 2).

The mosquitoflsh also quickly reduced larvalmosquito populations after a 2nd enrichment. Late-stage mosquito abundance was only about 3 and 9larvae per dip in the mosquitofish and the combinedfish treatments, respectively, after the 2nd enrich-ment of the ponds (Fig. 7B). In ponds containingGambusia, the abundance of late-stage mosquitoesdeclined to preenrichment levels by the next sam-pling date. After the 2nd enrichment, late-stagepopulations in the ponds containing mosquitofishwere l5-45Vo of those in the stickleback treatment(approximately 50 larvae per dip: Figs. 7A, 78) and2-5Vo those in the control ffeaftnent (200 larvae perdip: Fig. 7A).

1 000

20 30

Days post-stocking

Days post-stocking

Fig. 8. Abundance (mean + SE) of Culex spp. pupaein dip samples from treatments (A) with stickleback andno fish (control) and (B) with mosquitofish and both mos-quitofish and sticklebacks during the period from April 2to May 31, 1997. The arrow indicates the 2nd enrichment.Points are offset horizontally to facilitate illustration.

No significant differences were found in nontar-get species abundance among treatments (repeated

measures ANOVAs; P > 0.05). Species abundancevaried slightly between seasons (Table 1). Cope-pods and cladocerans were the most abundant zoo-plankton throughout the experiment. Abundance ofnotonectids during the L997 experiment was greaterthan during 1996.

Fish populations

Although sticklebacks did not survive to the endof the 1996 study, the mosquitofish populations in-creased substantially. At 6 wk after stocking, mos-quitofish biomass (wet weight) per pond increasedan average of 21-fold (from 5.80 to 118.6 g) whenstocked alone and 44-fold (from 3.5 to 155.5 g)when stocked concurrently with the stickleback(Fig. 9A). Except for 1 pond in the mosquitofishalone treatment where the mosquitofish populationincreased only about lO-fold, the wet weight of theresultant fish populations in the 2 mosquitofishtreatments was similar, approximately 150 g perpond (53.6 kg/ha). The average increase for mos-quitofish biomass was approximately 33-fold. The

B

Ssprel.rsen 1999 Mosqurro Conrnol By G. ACzTEATUS vs. G. er'r.ri.vrs

^ 180'b 160

cg 140I 12o

E 100

.E ro! A no - 'P a o

e . ,0

59 ann

9 2 f i

ft zooE.9 150

I tooE

Treatment

Troetmeni

Fig. 9. Fish biomass at stocking (In) and after 6 wk(Out) for 2 studies at the University ofCalifornia-River-side Aquatic Research Facility, Riverside, CA, (A) sum-mer 1996, (B) spring 199'7. M, mosquitofish alone; MB,mosquitofish when both fish species were present: S,stickleback alonel SB, stickleback when both fish soecieswere present.

final biomass of mosquitof,sh in the 2 treatmentsdid not differ significantly (to: 1.028; P : 0.36).

During the 1997 study, the average sticklebackbiomass per pond decreased from 9.71 to 9.56 gwhen stocked alone and from 6.56 to 0.330 g whenstocked with the mosquitofish (Fig. 9B). Stickle-back mortality occurred lst (May 1) and was great-est in the comparatively less shaded southern ponds(ponds Cl and D1). On May 11, dead adult stick-lebacks were also found in ponds D3, D5, and D6.For the entire study, the number of dead adult stick-lebacks collected by dip net declined along thenorth-south gradient of ponds (pond [number ofdead adults collectedl): cl (8), D1 (9), D3 (4), D5(6), D6 (1), and C7 (1). Despite these losses, thesurviving sticklebacks in the 2 northernmost pondssuccessfully reproduced and between l0 and 20 ju-veniles per pond were recovered by seining. Stick-leback biomass (wet weight) in ponds C7 and D6increased approximately 5OVo from that stocked.

Seining removed >98Vo of the mosquitoflsh bio-mass from the earthen ponds, with the exception ofpond D5, where only 737o of the fish biomass wasremoved by seining. The bottom of pond D5 wasmore irregularly shaped than were the bottoms ofthe other ponds; a subset of the fish in D5 was ableto avoid the seine.

Weight (9)

0 1 0 2 0 3 0 4 0 5 0

Length (mm SL)

Fig. 10. Relative abundance of Gambusia affinisweight (wet weight) classes in the (A) mosquitofish aloneand (B) both fish treatments and (C) the relationship be-tween mosquitofish wet weight and standard length inponds at the University of California-Riverside AquaticResearch Facility, Riverside, CA. The mean 1 SD is il-lustrated for each weight category.

The biomass of the mosquitofish population in-creased almost 38-fold during the 2nd study and,after 6 wk, was approximately 2-fold larger thanduring summer 1996. The average mosquitoflshbiomass per pond increased from 9.23 to 299.96 gin the mosquitofish-only treatment and from 6.08to 281.94 g in ponds with both flsh species (Fig.9B). The final biomass of mosquitofish in the 2treatments (mosquitofish alone: 107.13 kg/ha; bothfish: 100.7 kg/ha) did not differ significantly (r, :O.9@, P : 0.39). The distribution of G. affinisamong the weight size categories was similar for 2mosquitofish treatments (Figs. 1OA, 1OB). The ma-jority of individuals (approximately 6OVo) were inthe 0.1 and 0.2 g weight classes. Wet weight in-creased directly as a cubic power of length (Fig.10c).

The estimated number of G. ffinis per pond wassimilar in the mosquitofish-only ponds (mean +SD: 896 + 96) than in the ponds stocked with borhfish species (84: -r 99). The estimated number of

c?

6 Z . A

; ^E Zo

I ' 5e 1

* o i0

40

a 3 0c

9-'oEr .10

0

B40

d 3 0

3 r opL 1 0

0

388 JounNrL oF THE AMERICIN Mosqutro CoNrnol AssoclATIoN Vor. 15, No. 3

G. affinis per pond did not differ significantly be-tween the 2 mosquitofish treatments (t" : 0.67, P< 0.55).

DISCUSSION

The stickleback was less effective than was themosquitofish for reducing populations of immaturemosquitoes. Even though sticklebacks were not aneffective control agent for Culex larvae, G. aculea-trs provided significantly better control of mosquitopupae than did the naturally occurring macroinver-tebrate predators in the ponds without fish at 2-3wk after stocking fish in the 1997 study. Althoughsticklebacks persisted during the 1997 experiment,the abundance of mosquito larvae in sticklebackand control treatments did not differ significantly.This lack of a significant difference for mosquitoabundance between the 2 treatments was in part dueto the variability in mosquito larval abundanceamong ponds assigned to the stickleback treatmentand was caused by differential mortality of stick-lebacks among the ponds. Stickleback mortalitywas lst observed at day 2O after stocking and there-after pupal abundance increased in ponds stockedwith only G. aculeatus. The signiflcant reduction ofthe pupal subpopulation by G. aculeatus mighthave been caused by enhanced detection of thedarkly pigmented pupae as compared to the morelightly pigmented larvae.

The mortality of the stickleback was caused bythermal stress rather than the direct effects of otherabiotic factors such as low dissolved oxygen con-centrations or biotic interactions with the mosqui-tofish. The timing of stickleback mortality duringthe 1996 experiment was unknown. Maximum wa-ter temperatures were =31"C throughout the 1996experiment and, in the warmest ponds at the southends of the rows, maximum water temperatureswere >35'C for much of the study. During the sum-mer months, water temperatures reached a maxi-mum of 36.7"C. Because thermometers were situ-ated ^t approximately O.25 m depth, thetemperature at the water surface might have beenslightly higher than tJrose recorded by the thermom-eters.

An increase in the maximum water temperatureto )33"C was associated with death of reproductivesticklebacks during the 2nd study and mortality wasdirectly related to the gradient of water temperatureobserved in the ponds. Stickleback mortality wasobserved at around 20 days after stocking in 199'7.Before day 2O, maximum water temperatures were<31"C. After day 20 poststocking, maximum watertemperatures were between 33 and 35"C. The dailymaximum temperature in the warmest pond of the1997 study averaged 35.3'C. Water temperaturesprobably exceeded the thermal tolerance of thestickleback subspecies (G. aculeatus microcepha-Ius) used in our study. The thermal tolerance for arelated subspecies, G. aculeatus williamsoni Girard,

collected from the Santa Clara River in Los An-geles County, was approximately 34'C (Feldmeth

and Baskin 1976). The geographic ranges of G. a.microcephalu.r and G. a. williamsoni overlap in theLos Angeles Basin (Swift et al. 1993). Upper lethaltemperatures were even lower (28.8-21.6'C) intests of thermal and osmotic acclimation for the eu-ryhaline G. aculeatus aculeatus (L.) collected fromthe middle of its eastern geographic range in Hal-ifax, Nova Scotia, Canada (Jordan and Garsider972).

Even though the stickleback is primarily a mid-water fish (Schooley and Page 1984), it can with-stand low dissolved oxygen levels by altering be-havior and swimming at, or just below, the watersurface (Whoriskey et al. 1985, Walton et al. 1998).Under normal conditions in the Riverside ponds,the dissolved oxygen concentration was lowest be-tween approximately 0200 and 0700 h. The night-time dissolved oxygen concentration decreased toapproximately 2 mglliter in pond Dl and was 5times greater in a cooler pond, D6. These levels aresufficient for normal swimming behavior of thestickleback (Feldmeth and Baskin 1976). Under en-richment conditions, dissolved oxygen levels in theRiverside ponds were sustained at 0 mg/liter forperiods of 9-13 h during the late afternoon throughmidmorning. Hypoxia caused by the 2nd enrich-ment might have forced the sticklebacks to residein thermally stressful conditions near the water sur-face. Other studies indicated that the sticklebackwas able to survive periods of oxygen deprivationwhen water temperature was <29oC (.Whoriskey etal. 1985, Walton et al. 1997).

Mortality of sticklebacks stocked concurrentlywith mosquitofish was no greater than for stickle-backs stocked alone. Even though the mean finalbiomass of sticklebacks in the 2 treatments differedappreciably in 1997, sticklebacks were eliminatedfrom the warmest ponds in both treatments and,consequently, the variation around mean final bio-mass for each treatment was large. Although mos-quitofish biomass increased appreciably (averageincrease 38 times initial wet weight) over thecourse of the experiment, the stickleback biomassincreased a comparatively small amount (-SOVoinitial stocking weight) or declined. In this study,mosquitofish and sticklebacks were able to coexistand successfully reproduce in relatively cool ponds.Long-term studies may be more appropriate toevaluate the persistence of G. aculeatus because thesticklebacks exhibit complex mating behaviors andrequire sufficient time for nesting, mating, hatchingof eggs, and paternal care of the young before theiropen-water dispersal. Because of the short durationof this study, an evaluation of the long-term persis-tence of both species when reared together was notpossible. Coykendall (1980) cited observational ev-idence that resident stickleback populations exclud-ed mosquitofish from several types of aquatic hab-itats in Oregon.

SerrBrr.rsen 1999 Mosqulro Coxrnol ey G. tcutetrus vs. G. .armvls

The mosquitoflsh, whether stocked alone or withthe stickleback, provided significantly better mos-quito control than did the stickleback alone or thepredaceous aquatic macroinvertebrates in the pondswithout fish. The mosquitofish reproduced rapidly.Mosquitofish production is directly related to pri-mary production (Goodyear et aL 1972) and exhib-its seasonality (Sawara t974, Cech and Linden1987). During 1997, mosquitofish populations grewat rates between 0.651 and 0.671 individuals/indi-viduaVwk. Population growth rates were somewhatgreater than those observed in longer studies (0.288individuals/individuaVwk 13- to l4-wk study ofReed and Bryant tl974h 0.401-0.414 individuals/individuaVwk: l0-week study of Hoy and O'Gradytl97ll) where resource limitation and seasonal re-duction in reproduction might have occurred. Theability to reproduce rapidly following introductionand wide environmental tolerances (Brett 1956) arefavorable characteristics of the mosquitofish, par-ticularly in the environmentally stressful conditionsfound in constructed treatment wetlands (Walton etal. 1997).

No significant difference was found in mosquitocontrol between the treatment with both fish andthe mosquitofish alone. The reduction of larvalmosquito populations in the treatment with bothfish can therefore be attributed to the presence ofthe mosquitofish. Our data do not support an en-hancement of mosquito control by the concunentstocking of mosquitofish with a fish, such as thestickleback, that feeds throughout the water col-umn. We conclude that mosquito control is not ben-efited by simultaneously stocking both fish species.

The lack of a significant difference in mosquitocontrol between the ffeatment with both fish, whichwas stocked with one half of the amount of mos-quitofish in the treatment with only mosquitofish,and the mosquitofish alone treatment indicates thatthe lower stocking rate may provide adequate con-trol of larval mosquito populations. A previousstudy showed that mosquito abundance did not dif-fer significantly when mosquitofish were stocked inthe spring at rates of 4 kglha and I kg/ha (Waltonand Mulla 1991). Another study of 2 stocking rates,1.1 kg mosquitofish./ha and 3.4 kg mosquitofish,/ha,concluded that the lower rate was effective for con-trolling mosquito larvae in rice fields (Kramer et al.1988). Mosquitofish biomass increased approxi-mately lVolday in fish ponds (Coykendall 1977).lnour studies, mosquitofish biomass after 6 wk duringthe summer was, on average, approximately onehalf of that for a similar period during the spring,54 vs. 104 kg/ha. Within each study, flnal mosqui-tofish biomass in the 2 treatments was roughlyequivalent. Even though stocking rates differed byalmost 2-fold for the 2 studies, the instantaneousrate of increase of mosquitofish biomass (wetweight) in the mosquitofish alone and the combinedfish treatments averaged 0.081 and O.095 glglday,respectively, during the summer and 0.079 and

O.O87 glglday, respectively, during the spring. Ourresults support previous findings that a lower stock-ing rate of mosquitoflsh of approximately 1.5 kg/ha may be sufficient to achieve adequate larval con-trol in a habitat relatively devoid of vegetativecover.

The stickleback was not effective in controllinglarval mosquito populations in the Riverside ponds,whereas the mosquitofish provided signiflcant lev-els of control whether stocked alone or concurrent-ly with sticklebacks. As compared to mosquitofishalone, mosquito control was not significantly en-hanced when both fish were stocked together. Wefurther found that neither species of fish affectednontarget fauna, but caution that these experimentalponds are unlike many vegetated habitats intowhich fish may be introduced as larvivorous agents.Because the mesocosms are not representative ofall natural field conditions, further investigation ofthe stickleback as a biological control agent for lar-val mosquito populations is warranted.

ACKNOWLEDGMENTS

This research was supported in part by SpecialFunds for Mosquito Research from the Division ofAgriculture and Natural Resources of the Univer-sity of California and ca:ried out as partial fulfill-ment of an M.S. degree from the Depanment ofEntomology, University of California-Riverside toY.O. We thank J. Chaney for assistance applyingbifenthrin and P Chawla, T. Pach6n, L. Randall, M.Wfuth, P Workman, and M. Zia for assisting withvarious aspects of the study. M. Gates, M. Holman,S. Hudson, D. Johnson, M. Metzger, and A. Work-man assisted with lining the ponds. The NorthwestMosquito and Vector Control District provided themosquitofish used in our studies. We benefited fromdiscussions with J. Malcolm, R. Rodriguez, and C.Swift. J. Keiper, J. Malcolm, B. Mullens, and M.Mulla kindly provided comments on an earlier draftof the manuscript.

REFERENCES CITED

Bay, E. C. 1985. Other larvivorous fishes, pp. 18--24. In:H. Chapman (ed.). Biological control of mosquitoes.American Mosquito Control Association, Fresno, CA.

Bohart, R. M. and R. K. Washino. 1978. Mosquitoes ofCalifornia. Division of Agricultural Sciences Publica-tions, Publ. 4084. Univ. Calif., Berkeley, CA.

Brett, J. R. 1956. Some principles in the thermal require-ments of fishes. Q. Rev. Biol. 31:75-87.

Carlson, D. B., R. R. Vigliano and G. L. Wolfe. 1986.Distribution of mosquitoes in different wastewater stag-es of secondarily treated domestic effluent and untreat-ed citrus washwater. J. Am. Mosq. Control Assoc. 2:526-531.

Cech, J. J., Jr. and A. L. Linden. 1987. Comparative lar-vivorous performances of mosquitofish, Gambusia af-finis, and juvenile Sacramento blackfish, Orthodon mi-crolepidtus, in experimental paddies. J. Am. Mosq.Control Assoc. 3:35-41.

390 JouRNAL oF THE AMERIcIN MosQutro Conrnol AssoclertoN Vol . 15, No. 3

Coykendall, R. L. 1977. Aquacultural studies of mosqui-tofish, Gambusia ffinis, in earthen impoundments:stocking rate optimization for yield, protection of over-wintering fish stocks. Proc. Mosq. Vector Control As-soc. Calif. 45:80-82.

Coykendall, R. L. (editor). 1980. Fishes in California mos-quito control. California Mosquito Vector Control As-sociation, CMVCA Press, Sacramento, CA.

Farley, D. G. 1980. Prey selection by the mosquitofishGambusia aflinis rn Fresno county rice fields. Proc.Mosq. Vector Control Assoc. Calif. 48:51-55.

Feldmeth, C. R. and J. N. Baskin. 1976. Thermal andrespiratory studies with reference to temperature andoxygen tolerance fbr the unarmored stickleback Gas-terosteus aculeatus williamsoni Hubbs. Bull. South.Calif. Acad. Sci. 7 5:127 -731.

Fox, E., K. Shotton and C. Ulrich. 1995. Sigmastat: sta-tistical software. User's manual. Jandel Cotp., San Ra-fael, CA.

Gamradt, S. C. and L. B. Kats. 1996. Effect of introducedcrayfish and mosquitofish on California newts. Conserv.Biol. 10: 1 155-1162.

Goodyear, C. P., C. E. Boyd and R. J. Beyers. 1972. Re-lationships between primary productivity and mosqui-tofish (Gambusia ffinis) production in large meso-cosms. Limnol. Oceanogr. 17 :445-45O.

Gratz, N. S., E. E Legner, G. K. Meffe, E. C. Bay, M. W.Service, C. Swanson. J. J. Cech. Jr. and M. Laird. 1996.Comments on "Adverse Assessments of Gambusia af-fn is." J . Am. Mosq. Control Assoc. 12:160-166.

Hoy, J. B. and J. J. O'Grady. 1971. Populations of mos-quitofish in rice fields. Proc. Mosq. Vector Control As-soc. Calif. 39:1O7.

Hubbs, C. L. 1919. The stickleback: a fish eminently fittedby nature as a mosquito destroyer. Calif. Fish Game 5:1 1 a A

Hurlbert, S. H. and M. S. Mulla. 1981. Impacts of mos-quitofish (Gambusia affinis) predation on planktoncommunities. Hydrobiologia 83: 125-l 5 l.

Jordan, C. M. and E. T. Garside. 1972. Upper lethal tem-peratures of threespine stickleback, Gasterosteus acu-leatus (L.), in relation to thermal and osmotic accli-mation, ambient salinity, and size. Can. J. Zool. 50:1405-141 1.

Kramer, V. L., R. Garcia and A. E. Colwell. 1988. Anevaluation of Gambusia affinis and Bacillus thuringren-sis var. israelensis as mosquito control agents in Cali-fornia wild rice fields. J. Am. Mosq. Control Assoc. 4:470-478.

Laird, M. 1977. Enemies and diseases of mosquitoes.Their natural regulatory significance in relation to pes-ticide use, and their future as marketable componentsof integrated control. Mosq. News 37:331-339.

Meisch, M. V. 1985. Gambusia ffinis affinis, pp. 3-17.1n.' H. Chapman (ed.). Biological control of mosquitoes.American Mosquito Control Association, Fresno, CA.

Merritt, R. W and K. W. Cummins. 1984. An introductionto the aquatic insects of North America, 2nd ed. Ken-dall/flunt Publishing Company, Dubuque, IA.

Miura. T.. R. M. Takahashi and W. H. Wilder. 1984. Im-pact of the mosquitofish (Gambusia affnis) on a ice

field ecosystem when used as a mosquito control agent.Mosq. News 44:510-5t'7.

Mortenson, E. W 1982. Mosquito occurrence in waste-water marshes: a potential new community problem.Proc. Mosq. Vector Control Assoc. Calif. 50:65-67.

Mulla, M. S., M. Zgomba, H. A. Darwazeh and J. D.Chaney. 1992. Efflcacy and selectivity of two pyre-throid insecticides against the predator Triops longicau-/atus (Notostraca: Triopsidae) and Culex tarsalis lar-vae. Bull. Soc. Vector Ecol. 17:51-56.

Reed, D. E. and T, Bryant. I974. The use of minnow trapsto monitor population trends of Gambusia affinis in ricefields. Proc. Mosq. Control Assoc. Calif. 42:49-51.

Rupp, H. R. 1996. Adverse assessment of Gambusia af-

finis: an alternate view for mosquito control practition-ers. J. Am. Mosq. Control Assoc. 12:155-159.

Sawara, Y. 7974. Reproduction in the mosquitofish (Gam-busia affnis), a freshwater fish introduced into Japan.Jpn. J. Ecol. 24:14O-146-

Schooley, J. K. and L. M. Page. 1984. Distribution andabundance of two marsh fish: the mosquitofish (Gare-busia afrinis) and the threespine stickleback (Gaster-osteus aculeatu.s). Proc. Mosq. Vector Control Assoc.Calif . 52:134-139.

Swift, C., T. R. Haglund, M. Ruiz and R. N. Fisher. 1993.The status and distribution of freshwater fishes of south-ern California. Bull. South. Calif. Acad. Sci. 92:101-16'7.

Walton, W. E. and M. S. Mulla. 1991. Integrated controlof Culex tarsalis larvae using Bacillus sphaericus andGambusia affinis: effects on mosquitoes and nontargetorganisms in field mesocosms. Bull. Soc. Vector Ecol.16:2O3-221.

Walton. W. E., P D. Workman and S. A. Pucko. 1996.Efficacy of larvivorous fish against Culex spp. in ex-perimental wetlands. Proc. Mosq. Vector Control As-soc. Calif. 64:96-10l.

Walton, W. E., M. C. Wirth, P D. Workman and L. A.Randall. 1997. Survival of two larvivorous fishes in amultipurpose constructed wetland in southern Califor-nia. Proc. Mosq. Vector Control Assoc. Calif. 65:51-) / _

Walton. W. E.. P. D. Workman. L. A. Randall. J. A. Jian-nino and Y. A. Offill. 1998. Effectiveness of controlmeasures against mosquitoes at a constructed wetlandin southern California. J. Vector. Ecol. 23: 149-160.

Washino, R. K. and Y. Hokama. 1967. Preliminary reporton the feeding pattem of two species of fish in a rice-field habitat. Proc. Mosq. Control Assoc. Calif. 35:84-87 .

Whoriskey, E G., A. Gaudreault, N. Martel, S. Campeauand G. J. FitzGerald. 1985. The activity budget andbehavior patterns of female threespine sticklebacks,Gasterosteus aculeatus _(L.), in a Qudbec tidal saltmarsh. Nat. Can. (Rev. Ecol . Syst . ) 112:113- l 18.

Wilkinson, L. and M. Coward. 1997. Analysis of variance.Chapter 6, pp. 119-223. Iru; Systat 7.0 user's manual.SPSS Inc., Chicago, IL.

Woodridge, J. and E. W. Davidson. 1996. Biological con-trol of mosquitoes, pp. 530-563. In: B. J. Beaty and W.C. Marquardt (eds.). The biology of disease vectors.Univ. Press of Colorado. Niwot. CO.