SrprnN4eun 1989 B.r.r. rN A MIcHIcAN RIvER 397

A BROAD EVALUATION OF B.T.I. FOR BLACK FLY (DIPTERA:SIMULIIDAE) CONTROL IN A MICHIGAN RIVER: EFFICACY, CARRY

AND NONTARGET EFFECTS ON INVERTEBRATES AND FISH

RICHARD W, MERRITT,I EDWARD D. WALKER,I MARGARET A. WILZBACH,,KENNETH W. CUMMINS2 eno WILLIAM T. MORGAN1

ABSTRACT. Efficacy for black fly control, carry and nontarget effects of B.f.i. (Teknar@ HP-D^-):applied in the Betsie River, Michigan, were studied in June 1988. Black fly mortality was high (-100%)f6i a2,200 m stretch downstream from the application site, declined to 30% at 3,200 m, and was_nil at4,500 m. Drift of black flies greatly increased after application at a downstream site, but did not changeai an upstream site. There *ere no detectable nontarget effects of B.t.i. application on: 1) invertebraternu".o--o. micro-driftt 2) numbers of invertebrates in benthic Surber samples; 3) mortality or feeding ofdrifting and nondrifting insects; 4) growth or mortality of caged Stenomena sp. Iarvae; 5) invertebratefunctional group composition; 6) mbrtality or weight change of caged rock bass; or 7) fish numbers,species composilion, length-weight (rock bass only) relationships or rock bass diet. Sampling of Rheo'tinytarsus sp. midges onnatural substrates indicated low (27%) mortality owing to B.t.i. at only 100 mdownstreapfrom ihe application site, with negligible mortality at all other downstream and upstreamsites. This information, iombined with no pronounced changes in numbers of midges in macro-driftafter application, indicated that midge populations were not adversely affected by B.t.i. in the study.

INTRODUCTION

Judged by the degree of annoyance and irri-tation they cause in North America, black fliesrank with mosquitoes as major pests of peoplein recreation areas (Newson 1977, Kim and Mer-ritt 1987). Annually, black flies account for largefiscal losses in Canada and the north centraland northeastern United States. They greatlydiscourage tourism and outdoor recreation, andinterfere with lumbering, mining and buildingactivities during early spring and summer(Jamnback 1969, Fredeen 1977, Merritt andNewson 1978). In addition to humans, otheranimals also are affected by black fly attacks(Steelman 1976). Fredeen (1985) estimated thatlosses to beef and dairy producers in Sasketche-wan in one year exceeded US $2.9 million owingto mass outbreaks of a single species. Besideseconomic effects, the bites of these insects cancreate a variety of pathological conditions inhumans requiring hospitalization of sensitizedindividuals (Newson 1977).

In recreation areas, spraying to control adultblack flies with insecticides has sometimes beeneffective, but the benefits are of short durationand this method presents risk of environmentalcontamination. Broad spectrum chemical insec-ticides applied to streams for larval control re-sult in death of many nontarget organisms (Fre-deen 1975, 1983; Mohsen and Mulla 1982). Inaddition, biomagnification of insecticides in thefood chain has undesirable consequences(Woodwell et al. 1967). These adverse effects

I Department of Entomology, Michigan State Uni-versity, East Lansing, MI 48824.

2 Pymatuning Laboratory of Ecology, Departmentof Biological Sciences, University of Pittsburgh, Pitts-burgh, PA 15260.

have prompted the development of more ecolog-ically sound management strategies employingnonpersistent agents presumed to have little orno toxicity to nontarget organisms.

The most successful biological control agentdeveloped to date for black flies is B.t.i. fBacillusthuringiensis var. israelensis de Barjac (serotypeH-14)1, first isolated by Goldberg and Margalit(1977) from samples taken in the Negev Desertof Israel. It has proven to be very effective in amultinational black fly control program con-ducted by the World Health Organization inWest Africa (Lacey et al. 1982), and at severallocations in North America (Molloy and Jamn-back 1981, Molloy 1989, Colbo and O'Brien1984, Back et al. 1985, Gibbs et al. 1986, Pistrangand Burger 1984). Lacey and Undeen (1986,1987), and most recently Molloy (1989) havereviewed uses of B.r.l. for black fly control.

To be effective, B.t.i. must be eaten by blackfly larvae. The principal mortality agent of B.t.i.is a parasporal particle which contains a pro-teinaceous protoxin (Dubois and Lewis 1981,Aronson et al. 1986). The toxin becomes activeafter solubilization of protoxin in the presenceof proteases and alkaline pH in the black flyIarval midgut. The toxic polypeptides bind tosurface receptors on the midgut epithelial cellsand cause the cells to lyse and disintegrate, andthe larva dies.

To date, there have been no published studiesin the use or effect of B.t.i. against black flies inMichigan waterways. In June 1988, the Michi-gan Water Resources Commission and Depart-ment of Natural Resources issued a permit al-Iowing B.t.i. to be experimentally applied to asection ofthe Betsie River, a brown trout streamIocated in the lower peninsula. This paper re-ports results of this study, the major objectivesof which were to: 1) determine the extent of

Seprnrr,rsnn 1989 B.r.r. rN I Mrcnrcelt Rrvnn

spores in the river water were done to estimatethe carry of the B.t i. downstream from therelease site. Water samples were taken at sta-tions 100, 300, 600 and 1,800 m below the treat-ment site. For each station, samples were drawn5 min before B.t.i. application and at intervalsof 5, 10, 15, 30 and 60 min after application.Samples were taken by submerging and com-pletely filling 500 ml sterile tissue culture flasks.Samples were put on ice immediately after col-lection and kept on ice until they were returnedto the lab, where subsamples were drawn andpreserved in 10% formalin (final concentration,v/v). Dilutions (1:5 or greater) of the sampleswere prepared using filter-sterilized, deionizedwater and then stained with DAPI fluorochro-matic stain (Porter and Feig 1980, Walker et al.1988) at a concentration of -6 pg/mI for 25 minat 4'C in the dark. Slides were prepared fromeach sample according to the method of Hobbieet al. (1977), using lrgalan-black stained filtersfor a neutral, nonfluorescing background. Directcounts of total bacteria (including B.t i. spores)were done for 4 replicates of each sample, fromat Ieast 15 fields to a total count of at least 200bacteria for each filter, using a Leitz Laborlux11 microscope with the appropriate epifluores-cent light fittings and excitation and barrierfilters. Counts were converted to numbers ofbacteria per ml of river water sample, using astandard formula (APHA 1980).

Two attempts were made to obtain indirectcounts of bacterial spore numbers using agarplating procedures of Gibbs et al. (1986). Sub-samples (i0 ml) drawn from the original sampleswere heat-shocked in a water bath at 60'C for30 min to kill vegetative growth, serially diluted,and 1 ml of each dilution plated onto agar inpetri plates, which were then incubated for 24 hbefore inspection for the presence of bacterialcolonies. In the first trial, dilutions were inpowers of 10 from 1:1 to 1:10,000, and platingwas directly onto the top of solidified blood agar.In the second trial, dilutions were in powers of5 from 1:1 to 1:625, and the samples were mixedinto unsolidified tryptose agar during the platingprocedure.

Macro-drift samples: Drift sampling hasproved an effective method for evaluating theeffects of B.t.i. on black flies and nontargetinvertebrates (Gibbs et al. 1986, Molloy 1989),and was used in this study. Drift nets wereplaced in riffle areas above and below the site ofB.t.i. release. Six nets were set in a transectacross the width of the stream just above therelease site (ca. 20 m below World's Bridge), andsix more were set in a similar transect 100 mbelow the treatment site. The drift nets had amesh size of 350 pm and were anchored in thestream with iron rods passed through the long

axis of an aluminum frame (45 x 15 cm rectan-gular aperature) holding the net. AII drift sam-ples were preserved in 70% ethanol and returnedto the lab for later analysis.

Drift was sampled from 2000 h to 0800 h (for12 h) during each of 6 successive days or sam-pling periods, as natural drift of stream orga-nisms is greatest during this time (Waters 1972).The first two sampling periods started 48 and24 h before B.t.i. application. The beginning ofthe third period was concurrent with the appli-cation, while the remaining three periods began24, 48 and 72 h after application. The drift netpositions were the same for each of the 6 sam-pling periods. Stream depth and current flow infront of each net were measured just prior toremoval of the sample to allow for correctionsin calculating actual drift densities. Drift collec-tions per net per day were then expressed uni-formly as the number of individuals of eachtaxon per total volume of water that passedthrough each net during the sampling period(Waters 1972).

The sampling design (6 replications at eachsite with 6 sampling periods, 2 before the treat-ment and 4 after the treatment) provided infor-mation on pretreatment and posttreatment driftabove and below the B.t i. release point. Use ofan experimental design incorporating both spa-tial (above-and-below) and temporal (before-and-after) sampling (cf., Green 1979) providedcontrols to the treatment effect in space andtime.

Large numbers of chironomids were capturedin drift nets: therefore. collections of these or-ganisms were subsampled with a device modifiedfrom Waters (1969). An evaluation of the sub-sampler with drift samples revealed no signifi-cant differences in the number of invertebratesdistributed among subunits after subsampling(12 tests, P > 0.05). Total counts were made forother invertebrates, including black fly larvae.

Micro-drift sarnples: ln order to determine theeffects of the B.t.i. application on smaller-sizedinsects and to determine the suitability of usingthe coarser mesh-sized, Iarger volume nets forsampling drift, micro-drift sampling was con-ducted as follows. The nets used to collect micro-drift were of the "windsock" type, 0.8 metersIong with a mesh size of 250 pm and a collectioncup, with a removable screw cap, at the terminusof the net (Wilzbach et al. 1986). Four sets offive replicate drift samples were taken. Two setsof the five replicates were taken the night beforeB.t.i. release (one set above and one below therelease point), and similar sets were taken im-mediately aft,er B.t.i. treatment. The sets takenbefore release indicated the natural variabilitybetween the upper (above the release point) andlower (below the release point) sampling loca-

400 JouRnlr, oF THE Arvrnnrclr Mosqurro Coxrnor, Assocra,norl V o L . 5 . N o . 3

tions. All sampling commenced within 20 minof 2200 h and was of a 2 h duration. Comparisonofthese collections with those taken at the samelocations after treatment with B.t i. permit eval-uation of the immediate effects of the pesticide.

The invertebrates from the micro-drift sam-ples were sorted to taxa, enumerated by sizeclasses (nearest mm) and converted to estimatedbiomasses with a special computer programwhich uses an array of 32 length-weight regres-sion equations that are suited for the variousgroups of organisms. The coefficients and for-mulations for the regressions were from theIiterature or independently determined. Driftinginvertebrates were also categorized according tofunctional feeding group (Merritt and Cummins1984, Cummins and Wilzbach 1985) and habit(i.e., mode of attachment or locomotion; Menittand Cummins 1984).

D rif t I benthos mortality experiments : In orderto compare mortality rates between drifting andnondrifting (benthic) invertebrates both beforeand after release of B.t.i., a specifically designedsampler was used to partition these drifting andnondrifting population cornponents. The plexi-glas partitioner (Wilzbach and Cummins 1989)encloses a 0.1 m2 area of stream bottom and has250 pm mesh panels at the fronts and sides thatpermit circulation of stream water. The samplercan be seated several cm into a gravel streambottom, with heavy cobbles on the lid to hold itin place. Water and invertebrate drift can exitthe enclosure through two downstream ports.These ports can be fitted with 250 pm meshmicro-drift nets to collect the drift. The designof the partitioning sampler permits collection ofdrift and nondrifting benthos from a definedarea of stream bottom and prevents colonizationby drift from outside ofthe area.

The experiment was conducted the night be-fore and the night after B.t.i. release at a point100 m below the release point. Each night, thepartitioner was seated in place for 10 h (from1000 to 2000 h) with ports open, then the portswere closed with drift nets for 2 h e000-2200h). The ports were opened until the next morn-ing (0900 h) at which time the drift nets wereagain attached and the bottom sediments gentlydisturbed to wash individuals from the benthosinto them. These individuals represented ben-thos that had not drifted out of the enclosedarea the previous night.

Drift and benthos collections were carefullywashed into enamel trays, to which ice wasadded to maintain temperature, sorted under amicroscope, and insects were transferred withfine mesh scoops to holding chambers for as-sessment of mortality. These were S0-cell plasticculture chambers (each cell 12 x 12 X 8 mm),with 250 rrm mesh on the bottom to allow water

circulation into each cell. Animals were placedin the floating culture chambers with one indi-vidual per cell. The taxon and length of eachindividual (nearest mm) and the fullness (fuII,75%,50%,25% or empty) of the fore-, mid- andhindgut were recorded afber visual inspection at12x or greater magnification. The chamberswere placed in a closed cooler with stream waterfor 12 h. Styrofoam blocks attached to the cham-bers allowed them to float so that each cell was757o full of water. Water temperature was main-tained at L8-22'C by adding ice as needed.

After being held without food in the dark forthe 12 h period, the animals were censused formortality under a dissecting microscope withoutremoving the individuals from their cells. Anyindividual that exhibited damage from handlingwas excluded from the mortality census. Theprocedure yielded 12-h mortality estimates frompartitioned samples of drift and benthos priorto and after release of B.t.i. In the latter case,the partitioner was in place when the B.t.i. wasintroduced.

Substrate sarnples for black fly and mi.dge mor-tality estirnates: The effect of B.t.i. on black fliesand on the chironomid midge, Rheotanytarsussp. (Chironominae, Tanytarsini), was deter-mined by counts of living and dead larvaepresent on natural and artificial substrates fol-lowing B.t i. application. Rheotanytors.rs waschosen as a test nontarget organism because: 1)it is a nematocerous dipteran, closely related tothe Simuliidae, that was common at the studysite, 2) it attached to substrates identical orsimilar to those used by black flies, and 3) it isa filter-feeding insect that removes fine particlesfrom the passing water that are in the same sizerange used by black flies (Wallace and Merritt1980). Thus, Rheotanytarsrzs would be the kindof non-target organism most likely to be affectedby B.t.i. (Ali et al. 1981, Back et al. 1985, DeMoor and Car 1986).

Two types of artificial substrates were placedin the stream (in April and May 1988) just abovethe treatment site, and at 100, 300, 600, 1,800and 4,500 m below the treatment site. The firsttype of substrate consisted of 20-cm2 clay tilesplaced on cobble or sand in the stream bed. Thesecond tlpe consisted of a combination of 4 x35 cm clear acetate strips and 5 x 60 cm coloredvinyl flagging tape anchored in the stream withwire pegs. All substrates were placed in thestream at least 48-72 h prior to the pesticideapplication to allow time for colonization tooccur. In addition to artifrcial substrates, natu-rally occurring, trailing leaves of the grass Vol-lisncria sp. were used as a substrate when arti-ficial substrates were missing or had been van-dalized.

Mortality counts were made 2 h after treat-

SEPTEMBER 1989 B.r.r. rN l MrcxrceN Rrvnn

ment at the site above the release of B.t.i.. andat the 100, 300, 600, 1,800 and 4,500 m sites.Counts were also made at 12 h after the B.t.i.application at all 6 sites, and also for additionalsites at 2,200 and 3,200 m, where there were noartificial substrates and counts for both orga-nisms were fuom Vallisnerio. Counts were ac-complished by taking note of the total numberof dead and live black fly or Rheotarrytarsuslarvae on 3-6 substrates at each site and record-ing the number and percent dead per substrate.Because these larvae respond to touch with wrig-gling movements, it was easy to discern liveand dead individuals. Qualitative pretreatmentcounts showed no apparent mortality of blackflies or Rheotanytarsus prior to B.t.i. release.

Substrate samples for benthos estimates: Sam-ples of benthic invertebrates were taken with aSurber sampler (mesh size, 1 mm) at transectsacross the stream located 105 and 110 metersdownstream from the release site. Five sampleswere taken before and 5 after B.t.l. application,and samples were evenly distributed across thelength ofthe transects, after allowing for a 3-mbuffer zone from each bank to minimize edgeeffects. Samples taken before release were fromthe 105 m transect at 0700 h, June 15; andsamples taken after release were from the tran-sect at 110 m on June 16. All samples were takenat 0700 h and were preserved in 70Vo ethanoland returned to the lab for sorting and identifi-cation.

Short-term inuertebrate growth experiment:Live larvae of the scraper mayfly, Stenonemasp., were collected from the study site prior tothe release of B.t.i. and held for 6 days (138degree-days) in plastic field growth chambers(30 x 16 x 9 cm). These enclosures were fittedwith 500 pm mesh panels on all sides and in theremovable lid to allow circulation of streamwater. The growth box chambers were posi-tioned in the current 12 h before release of B.t.i..with 3 above and 3 below the release site. Cham-bers were held in position with sections of rein-forcement bar driven into the stream bottom.The boxes were provisioned with natural foodfrom the location, which consisted of small cob-bles of uniform size coated with periphyton.Care was taken to remove all macroinverte-brates from the cobbles before placing them inthe boxes. Each box was stocked with 25 larvaeand before stocking the boxes, a random sub-sample of 51 larvae was removed for determi-nation of initial weights. These larvae were air-dried and then placed in individual Beam@ cap-sules with perforated lids and placed in a dessi-cator. The capsules were oven-dried (50'C for24 h) desiccated for 24 h, and then each larvawas weighed to the nearest 0.1 pg. After the 6-day incubation period, the surviving larvae re-

moved from the growth chambers were treatedin the same fashion as the others for determi-nation of final weight.

Fish studi.es: Species composition, relativeabundance and length-weight relationships ofresident fish were studied to examine effects ofB.t.i. application. A boat equipped with an alter-nating current, electrofishing shocker with gas-oline generator was used to collect fish along100 meter stretches situated above (ca. 1 km)and below (300 m) the release point. Two passeswere made for each collection along each stretchboth before and after application of B.t.i., givinga total of four collections (above-before. above-after, below-before, below-after). A crew of fourpeople shocked and netted the fish, and all fishfrom both passes collected at a site were retainedIive for species identification and length (nearestmm) and weight (nearest 0.1 g) determinations.Fish were anesthetized with club soda (COz)prior to measuring and weighing.

One day prior to release of B.t.i., rock bass(Ambloplites rupestris) representing a range ofthe sizes in the population were taken from theabove- (10 fish) and the below-release (8 fish)stretches and placed in 1-m3 cages constructedon l-cm mesh wire. Prior to caging, gut contentsof the fish were removed to evaluate diet com-position and to insure that all fish had the samegut fullness initially. Gut contents were flushedfrom the digestive tract with water injected froma syringe, and preserved in alcohol. Cages wereplaced along the bank under cover of overhang-ing vegetation, and fish were held there for 3days. At the end of the 3-day period, the cageswere checked for mortality, and all individualswere measured and weighed again, and their gutcontents were removed.

RESULTS

Carry of B.t.i.: Bacterial numbers in the riverranged from 2,027,834 to 4,424,365 per ml andwere within the range observed in other rivers(e.g., Kondratieff and Simmons 1985). One-wayANOVA and Duncan's multiple range test (Steeland Torrie 1980) revealed distinct peaks in bac-terial numbers, appearing above the natural,background bacterial numbers in the river,which we attributed to the presence of B.t.i.spores moving in the water column down river(Fig. 2). At the site 100 m downstream from therelease point, there was a peak in direct countsat 5 min posttreatment; afterwards, directcounts dropped to preapplication levels. At the300-m downstream site, there was a peak indirect counts at 10 min postapplication, and atthe 600 m downstream site there were peaks atthe 10 to 15 min sampling times. At the 1,800 m

402 JounNnl oF THE AunRrc.tN Moseurro Colrrnol Assocrauor Vol. 5, No. 3

4.2

3.7

3.2

2.7

2.2

4.2

3.7

3.2

2.2

55351 5

+ .2

3.7

3.2

2.7

2.2

4.2

a 7

3.2

(.oo

xEooo(Elr

oz

1 0 0 m

600 ma

ab

2.2

downstream site, there were no peaks in directcounts, which were equal among sampling times,

Attempts to culture spores on media for col-ony counts were unsuccessful.

Macro-drift sampling: Thirty-eight inverte-brate taxa were identified from the macro-driftsamples (Table 1). The drift was dominated bygammarid amphipods and Iarvae of hydropsy-chid caddisflies, black flies, baetid mayflies andchironomid midges. Chironomid pupal exuviae,indicating adult emergence, were abundant inthe collections but were not counted. Interpre-tation of the effect of B.t.i. treatment on inver-tebrate drift must take into account both theresults of ANOVA (Table 1) and a graphicaldepiction of the temporal and spatial pattern ofdrift (Figs. 3 and 4). Drift of black fly larvaegenerally decreased over time at the site abovethe treatmentpoint (Fig. 3, A and B), suggestingthe occurrence of natural population processesat the study site. At the site below the treatmentpoint, black fly drift was similar to that abovefor the 2 days of pretreatment sampling, butgreatly increased immediately following appli-

300 m

1800 m

2.7

cation of B.f.i. For the frnal 3 days of sampling,black fly larvae were rare in the below-treatmentsite drift samples. ANOVA showed a highly sig-nificant effect (P < 0.001) of both site locationand sampling day on black fly drift. Drift ofchironomid midge larvae showed slightly differ-ent patterns at the above-treatment and below-treatment sites (P : 0.045) and was generallyhigher at the above treatment site among the 6sampling days (Fig. 3, C and D). There wassignificant variation of chironomid drift amongsampling days nested within treatment sites (P< 0.05). In general, there was no change inchironomid midge drift at the down-stream site(on posttreatment sampling days 3-6) that couldbe attributed to B.t.i. application (Fig. 3D). Bae-tid mayfly larvae (Fig. 3, E and F) and gammar-ids (see Fig. 5, G and H) varied in drift samplesamong sampling days (P < 0.001 and P < 0.01,respectively), but did not differ in numbers ofindividuals in drift at above-treatment and be-Iow-treatment sites (P > 0.05). Hydropsychidcaddisflies varied significantly in drift both be-tween treatment sites (P < 0.05) and among

2.7

' r5 35 55 75 -5 15 35 55

Time of sampling (min) from first sampleFig. 2. Direct counts of bacteria (including viable B.t.i. spores), using DAPI stain and epifluorescence

microscopy, in the Betsie River at four sites downstream from site of B.t.i. release. Data are means (+ SEM)of four replicates per site. Means with the same letter are not significantly different (one-way ANOVA andDuncan's rnultiple range test, P > 0.05).

75- q

SEPTEMBER 1989 B.r.r. rN A MrcHrcAN RrvER

Table 1. Relative abundance (as total in 72 collections) of individuals of the invertebrate taxa (order andfhmily) in macro-drift net collections in the Betsie River, and results of nested ANOVA on drifting numbers

""tt""t"d rb""" ""d

ANOVA F-tests

403

TaxonRelative

abundance SitesDays

within sites

AmphipodaGammaridae

DecapodaPalaemonidae

IsopodaAsellidae

OdonataAeshnidaeLibellulidaeCalopterygidaeCoenagrionidae

EphemeropteraBaetidaeHeptageniidaeEphemerellidaeLeptophlebiidaeSiphlonuridaeTricorythidaeNon-Baetid, Non-Heptageniid

PlecopteraPerlidae

HemipteraPleidaeSaldidae

MegalopteraCorydalidae

ColeopteraElmidae (larvae)Elmidae (adults)Dytiscidae (larvae)Dystiscidae (adults)Gyrinidae (Iarvae)Haliplidae (adults)Hydrophilidae (larvae)Hydrophilidae (adults)

DipteraChironomidaeSimuliidaeTipulidaeAthericidae

TrichopteraHydropsychidaeBrachycentridaeHelicopsychidaeHydroptilidaeLimnephilidaePhilopotamidaePolycentropodidaePsychomyiidaeNon-Hydropsychids

LepidopteraPyralidae

10,000

24

78

3,81766882I J

13I

109

439

12

D

0.00NS

ND

ND

NDNDNDND

2.44NS8.87r*

NDNDNDND

2.04NS

40.73***

NDND

ND

6.32*NDNDNDNDNDNDND

4.18+29.40+**

NDND

5.69*NDNDNDNDNDNDND

0.11NS

ND

2.35*

ND

ND

NDNDNDND

3.59***0.61NS

NDNDNDND

1.13NS

2.98**

NDND

ND

0.59NSNDNDNDNDNDNDND

2.23+16.00***

NDND

2.07*NDNDNDNDNDNDND

3.39**

ND

8It7

48344

t)

7I13

I347

3,9016,970

/1

7,7912I

I J

o

422313

103

2Degrees offreedom were: sites, 1; days, 10; and error, 60.* P < 0.05.** P < 0.01.*t* P < 0.001.NS, not significant (P > 0.05); ND, ANOVA not done.

404 Jounuer, oF THE ArrapRIclx Moseurro Covrnor, Assocrenorl Vor,. 5, No. 3

1 500

t 500

| 200

I 500

1200

1 s00

I 200

900

600

300

0

I 500

' -o

xa)

trIIJ@+lEe

.:

oF

F(EoE

.=oco

E

=L

o 1 500

1 200

0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

Dav- - ,

Fig. 3. Drift density [expressed as mean number of individuals (+ SEM) per drift net per ms x 1,0001 ofselected invertebrate taxa in macro-drifb collections at above-(left column) and below-(right column) treatmentsites; collections were made for 2 days before and 4 days after application, at night. Arrows indicate when B.t.i.was applied above the below-treatment site.

t 2001 2 A O

'1s00 1 500Baetidae

1 200

sampling days (P < 0.01). However, the patternsof drift at the above- and below-treatment sitesappeared rather similar, except that drift washigher at the above-treatment site on the firstsampling day (pretreatment) and somewhatlower on the Iast sampling day (posttreatment).Hydropsychid drift on days 3-5 (the first 3 daysafter treatment) was similar between above-and-below treatment sites.

--Jr---*--0 1 2 3 4 5 6 7

0 1 2 3 4 5 6 7

Results of macro-drift samples for less abun-dant organisms are shown in Fig. 4. There wasIarge variation in numbers of heptageniid may-flies in the drift, with significant variation be-tween treatment sites (P < 0.01), but no signif-icant variation among sampling days (P > 0.05).Early larval instar mayflies (mainly Heptageni-idae, but also including Ephemerellidae, Lepto-phlebiidae, Siphlonuridae and Tricorythidae)

1 200

0

1 500

1 200

900

600

300

0

1 500

1 200

900

600

300

0

0

0

D Cl-',ronomidae I

0

0

SEPTEMBER 1989 B.r.r. IN A MIoHIGAN RrvER 405

-oF

xe)EIJJ@+l

a

!t.sI

o?

?(goE

.=ocoE+t=L

o

3 4 3 4 5 6

DayIig. 4. Drift density (as in Fig. 3) of selected invertebrate taxa in macro-drift collections at above-(leFt

column) and below-(right column) treatment sites. collection times are as in Fig. B.

were more abundant in the drift at the below-treatment site than at the above-treatment siteboth before and after B.t.i. application (Fig. 4,A and B). When numbers from these Iast fourfamilies were combined, there were no signifi-cant differences among sampling days or be-tween treatment sites (P > 0.05). Caddisflies inthe families Brachycentridae, Helicopsychidae,Hydroptilidae, Limnephilidae, Philopotamidae,Polycentropidae and Psychomyiidae were alsouncommon in collections, and numbers of indi-

D Other Trichoptera

*

F Perlidae

viduals from these families were combined foranalysis as well (Table 1; Fig. 4, C and D). Therewas no significant difference between treatmentsites for this group (P > 0.05), however, thenumbers of these caddisflies did vary sigrrifi-cantly among sampling days (P < 0.01). In gen-eral, collections for this group decreased in bothabove- and below-treatment sites over time. Per-lid stoneflies (Table 1; Fig. 4, E and F) weresignificantly more abundant in below-treatmentdrift compared to above treatment drift (P <

90

60

30

90

30

90

60

on

60

A otherEphemeroptera

C Other Trichoptera

E Perl idae

60

30

H Elmidae

0

406 JouRNar, oF rHE ArrannrclN Moseurro Comrnol AssocrlrroN V o L . 5 , N o . 3

0.001) and also varied significantly among treat-ment days (P < 0.01). Beetles were generallyuncommon in the drift; however, there weresufficient numbers of elmid larvae for analysis(Table 1; Fig. 4, G and H). Elmid larvae weresignificantly more abundant in below-treatmentdrift compared to aborre-treatment drift (P <0.01). Drift collections of elmid larvae did notvary significantly amorrg sampling days nestedwithin treatment sites lP > 0.05).

Micro-drift sampling: The micro-drift samplesindicated that only an insignificant amount ofdrift was represented by size classes of animalsin the size range of 1 mm or less (Fig. 5). Thus,the macro-drift samples adequately representedthe total drift composition in the stream duringthe sampling period. This was true for the sitesabove and below the poi.nt of release of B.t.i. andfor the same sites before and after the pesticidewas applied.

In addition, no significant general effect ofthe B.t.i. application on size class compositionof the drift could be d,emonstrated when com-paring paired observations (n : 9) either at theabove-treatment site berfore and after treatment(Wilcoxon signed-rank statistic : 17.5, P >0.05) or at the below-treatment site before andafber treatment (Wilcoxon signed-rank statistic: 16.5, P > 0.05). Comparisons of total driftnumbers at the site below treatment (Mann-

A = l p reabove

El Post above

B .Er I Pre belowtr Post below

0 1 2 3 4 5 6 7 8

Size in mmFig. 5. Comparison of size class distributions of the

micro-drift before and after the application of B.t.i.both above (A) and below (B) the application site.

A I preabove

tr Postabove

Shred Col-ga Col-fi l Scr Pred

B t Pre belowEl Post below

80

60

40

20

80

60

40

20

CL=oo),=oo(Il

E.9lr

o)ooo.

0

oatgoo.Nan

q)o

0)o-

40

30

20

1 0

40

30

20

1 0

" Shred Col-ga Col-fil Scr Pred

Funct ional grouPFig. 6. Comparison of the relative percent biomass

in each functional group in the micro-drift before andafter treatment above (A) and below (B) the applica-tion site. Abbreviations: SHRED, shredders; COL-GA,gathering collectors; COL-FIL, filtering collectors;SCR, scrapers; PRED, predators including frsh fry(see Merritt and Cummins 1984, Cummins and Wilz-bach 1985).

Whitney U : 8.0; rIr,II2 = 5,5; P > 0.05) or totalbiomass (Mann-Whitney U : 8.0; nr,nz :5,5; P> 0.05) showed no significant differences be-tween pre- and posttreatment samples. Further,no discernible shifts were observed in the taxo-nomic or functional group composition of thesamples that could be associated with the appli-cation of B.r.i. (Fig. 6). The major source ofvariation was due to the sporadic presence inthe samples of a few fish fry, included along withinvertebrates in the predator category, whichhad high biomass relative to drifting inverte-brates,

Drift / benthos mortality experiments: Drift andbenthos collections taken using the drift/ben-thos partitioner showed no significant effects ofthe B.t.i. release on mortality of the partitionedpopulation components (Table 2), with the ex-ception of an increase (from 1 to 19) in thenumber of dead black fly larvae encountered inthe drift samples prior to the mortality tests. Asreported in a previous study (Cummins andWilzbach 1988), mortality of the animals heldfor 12 h was always higher among individualsfrom the drift than from the benthos collections,but in neither case was there a significant dif-

B.?'.r. rN e MIcuIcer RIvrn 407SEPTEMBER 1989

Table 2. Mortality and gut fullness of invertebrate taxa from the drifb-benthos partitioner.

Time/sampleNumberof taxa

Number ofindividuals

%mortality

MeareVogut tuU

69.365.758.853.1

Mean 7oforegut

69.27r.451.585.1

Before/drifbAfter/driftBefore/benthosAfter/benthos

8q

9

2930oo

47

20.7t6.76.11.0

ference between the collections taken before andafter treatment. In addition, there was no effecton the feeding (gut fullness) of the invertebratesthat could be attributed to the application of thepesticide. No significant difference was meas-ured before and after B.t.l. release in either thepercentage of individuals with full guts or thosewith greater than half-full foreguts (the latterbeing a reflection of the most recent feedingactivity; see Table 2).

Mortality estimates from substrates: Counts ofIive and dead black flies and midges for mortal-ity estimates came from both artificial and nat-ural substrates because there was variation incolonization by these insects on the differentsubstrates. Mean mortality estimates for thesetwo groups at the different sites are shown inFig. 7. For black flies (Fig. 7A), there was 0%-ortatity at the site upstream from the point ofapplication, and 0% mortality at the 4,500 mdownstream site; these two sites did not differin percent mortality (one-way ANOVA andDuncan's multiple range test, P > 0.05). At sitesranging from 100 to 2,200 m downstream, therewas nearly 100% mortality of black fly larvae,and these sites did not differ (one-way ANOVAand Duncan's multiple range test' P > 0.05). Atthe 3,200 m site, black fly mortality averaged30Vo, and this site differed from all other sites(one-way ANOVA and Duncan's multiple rangetest, P < 0.05).

Rheotanytarsrzs midge mortality was muchlower than black fly mortality at the down-stream sites (Fig. 7B). There was some apparentunexplained midge mortality (possibly owing tohandling), as the mortality estimate for the siteupstream of the release point was about 10%;one-way ANOVA and Duncan's multiple rangetest showed that the mortality at this site wasnot significantly different (P > 0.05) from the300, 1,800 and 2,100 m downstream sites. How-ever, there was a significantly higher mortalityestimate (37%, P < 0.05) at the 100 m down-stream site than at the other sites.

Mortality estimates from benthns.' Results ofSurber samples are shown in Table 3. Therewere sufficient numbers of larvae from the fol-Iowing insect families to compare numbers col-lected between pre- and posttreatment samples:Baetidae, Ephemerellidae and Heptageniidae

a- ? a, 4 Black f l ies

a1 0 0

75

50

25

0

.=(g

oEco)()oo.

- ' 100 1000 2100 3200 4300

1 0 0

80

60

40

20

U

Rheotanytarsus

O , g

-100 1000 2100 3200 4300

Distance from release Point (m)

Fig. ?. Mean percent mortality of black fly andRheotanytarsus larvae on artificial and natural sub-strates in the Betsie River, after release of B.t'i. Meanswith the same letter are not significantly different(one-way ANOVA and Duncan's rnultiple range test,P > 0.05).

(Ephemeroptera); Gomphidae (Odonata); Hy-dropsychidae, Leptoceridae and Philopotamidae(Trichoptera); Elmidae (Coleoptera); and Chi-ronomidae (Diptera). ?-tests of Surber samplespaired by position in the transects showed nosignificant difference (P > 0.05; see Table 3)between pre- and posttreatment samples forthese groups.

Short-term inuertebrate growth experiment:The short-term in situ growth experiment withStenonema sp. larvae showed no measurableeffects of the B.f.i. treatment when the chambersplaced above and below the release site werecompared. Mortality, which was high in boththe boxes above (mean = 64%) and below (mean: 78%) the point of B.t.i. release, was not sig-nificantly different (Mann-Whitney U : 75;nr,nz : 18; P > 0.05). Growth rates (Table 4)

408 JounNnl oF THE Alrrntc.lr.t Mosqurro Coutnol AssocrATroN Vor,. 5, No. 3

Table 3. Results of Surber samples taken before and after B.r.i. application below the site ofapplication inthe Betsie River.

Mean (range)

Taxon Before After t-testTricladida

PlanariidaeOligochaetaGastropoda

PleuroceridaePelecypoda

SphaeriumIsopoda

AsellidaeAmphipoda

GammaridaeEphemeroptera

BaetidaeEphemerellidaeHeptageniidaeSiphlonuridae

OdonataGomphidae

MegalopteraNigronin

TrichopteraHydropsychidaeLeptoceridaePhilopotamidaeGlossosomatidaeHelicopsychidaeHydroptilidaeLimnephilidae

ColeopteraElmidae

DipteraChironomidaeSimuliidae

0.6 (0-2)0.1 (0-1)

r.4 (0-7)

2.0 (0-e)

0.0 (0)

0.2 (o-1)

6.2 (0-13)5.0 (2-10)

11.6 (1-20)0.0 (0)

1.6 (0-4)

0.0 (o)

148.0 (91-229)4.8 (0-22)2.8 (0-6)0.2 (0-1)0.0 (0)0.2 (0-1)1.8 (0-5)

23.0 (10-50)

89.8 (38-173)2.8 (0-8)

0.2 (0-1)0.0 (0)

NDND

ND

ND

ND

ND

1.2 (0-5)

0.2 (0-1)

0.2 (0-1)

70.4 (4-74)9.6 (2-24)

12.6 (r-29)0.2 (0-1)

3.6 (2-6)

1.0 (0-5)

r29.6 (65-242)6.0 (0-12)

12.2 (0-59)0.0 (0)0.2 (0-1)0.0 (0)1.0 (0-4)

35.6 (11-92)

72.6 (4r-r40)0.4 (0-2\

1.69NS0.32NS0.90NS

ND

0.06NS

ND

0.69NS0.85NS0.43NS

NDNDNDND

0.54NS

0.67NSND

Weight bysite/time

ND, paired t-test not done; NS, t-test not significant.

Table 4. Weight gain and growth rate of Stenonemasp. Iarvae (Ephemeroptera: Heptageniidae) held forsix days (138 degree-days) in enclosures above and

below the point ofB.t.i. release. Relative growth rate(RGR, Waldbauer 1968) is given as percent body

_ wetght tncrease pe

RGR

Mean weightin mg (SE) Per day

InitialFinal aboveFinal below

were also high in both sets of chambers, beingslightly greater in those below the treatmentsite.

Fish studies: The electrofishing collectionswere dominated by rock bass and creek chub(Sernotilus atromaculatus). These were the onlyspecies that were taken in significant numbersin every sample (Table 5), although 21 speciesof fish were collected. A few brook trout (Sol-

uelinus fontinalis) and brown trout (Salmotrutta) were taken, but their distributions werehighly restricted to areas at groundwater seepsbecause of the warm water temperatures. Ingeneral, fish were more abundant at the above-release stretch both before and after treatment,than at the below-release stretch before andafber treatment. There was no significant differ-ence in the relative numbers offish caughtbelowthe B.t.i. release point when collections madebefore and after application were compared(Mann-Whitney U : L'74; n1,n2: 19,19; P >0.05). Similarly, there was no significant differ-ence in the relative numbers offish caught abovethe B.t.i. release point when collections madebefore and after application were compared(Mann-Whitney U : I72; n1,n2 = 19,19; P >0.05). Also, the variations between the upstreamand downstream collections were indistinguish-able from those observed between the samplestaken in the control area above the release pointbefore and after treatment.

Using rock bass as a test species, length-

Perdegree-day

0.60.60.8

51 0.77818 1.89612 2.679

13.913.918.1

B.r.r. rN e Mlcrucln Rrvon 409SEPTEMBER 1989

Table 5. Fish diversity and abundance at samplesites in the Betsie River. Numbers are totals of twopasses ofelectrofishing collections from 100 meter

stretches above and below the B.t.i. application site.Collections were made at both sites in the day prior

_ t", a"d th" t*o d.y" a

Above re- Below re-Iease lease

TaxonBe- Af- Be- Af-fore ter fore ter

PetromyzonidaeLarnpetra lamottei

SalmonidaeSaluelirurs fontinalisSalmo trutta

UmbridaeUmbralimi

CyprinidaeNotropis cornutusN. atherinoidesS emotilus atr o maculatusClinostomus elongatusRhinichthys atratuhtsR. cataractae

CatostomidaeC dto s to n7u.s c orntne r s oni

IctaluridaeIctalurus natalisI. nebulosusNofurus sp.

CentrarchidaeAmbloplites rupestrisLepomis gibbosusMicropterus salmoides

PercidaePerca flauescensEtheostoma caeruleumE. nigrum

CottidaeCottus cognatus

weight relationships were compared using fishfrom both the free-ranging population and thegroups held in cages. The length-weight rela-tionships for all samples conformed best to anexponential fit (Rz values, 0.97-0.98; see Fig. 8).There was no mortality of caged rock bass.There was no significant difference in length-weight relationships of free-ranging rock basscollected below the application site before andafter application of B.t.i. (Fig. 8A; Mann-Whit-ney IJ : 42; n1,n2: 9,11; P > 0.05). Nor wasthere a significant difference in length-weightrelationships of caged rock bass held above orbelow the application site (Fig. 88; Mann-Whit-ney IJ : 26i nt,nz: 8,10; P > 0.05). A fewindividual fish could be specifically identifiedpermitting comparisons of weight changes overthe 3-day holding period. In all cases, the rockbass lost between 0.8-1.0 g regardless ofwhetherthey were located above or below the point ofB.t.i. release.

The rock bass consumed a wide variety ofprey items, including many crayfish and te^r-res-irial prey, which comprised 37% and 28% oftotal prey biomass, respectively. Caddisfly andmayfly Iarvae comprised 25% and 18% of totalprey items. Prey selectivity was not evident,except with respect to size of prey, as items lessthan 2 mm in Iength were rarely present' Meanprey size, which was estimated from head cap-sule width when items were not intact, was 5 to6 mm. Only 2 larval black flies were identifiedfrom 38 fish guts; both were from fish cagedbelow the release site after B.t.i. application.Dipterans in general comprised less than 1% ofprey biomass. Diet composition of rock bass washighly variable among individuals, both withinand between the above- and below-releasestretches, and before and afber B.t.l. application.Kendall's coefficient of concordance ( W), whichranges from 0 to 1 (perfect concordance), wasused to evaluate similarity in rankings of thefour treatments (above-before, above-after, be-Iow-before, below-after), with respect to dietcomposition classified by taxonomic groupingsand by functional groups of prey items. Ex-pressed either by biomass or density of preyitems, similarity among the treatments in allcases was low and not significant for functionalgroup density (W:0.29), functional group bi-

100

BO

OU

40

20

0

60

40

en

20

1 0

80 100 120 140 160

3 6 0 1

0 0 3 20 1 1 2

1 1 2 20 0 1 0

37 85 19 161 1 0 0 0t 2 7 2 40 0 1 1

0 0 0 11 1 1 32 2 0 0

2 9 1 6 9 1 1t 4 0 01 0 2 0

7 0 4 01 0 0 01 1 0 0

AN40

oF(E

E )

.=

.C

.9o3

o=

OU 80 100 120 140 160

Length in mmFig. 8. Length-weight relationships of (A) free-rang-

ing rock bass collected below the site of B.t.i. releaseboth before and after treatment, and (B) caged rockbass held above and below the site of release.

40

* Pre{reat..--..-o- Post-treat

-- Cagebelow-# Cageabove

410 Jounuel oF THE AunnrclN Mosqurro CourRol AssocrATroN V o L . 5 , N o . 3

omass (Ir7'= 0.15), taxon density (I4l : 0.10),and taxon biomass (W : 0.40), rLspectively (F> 0.05 for all).

DISCUSSION

_ Carry_and effects o/ B.t.i. on bl.ack flies: In theBetsie River, direct counts ofbacterial particles(including viable B.r. j. spores) using DApI and.epifluorescence microscopy revealed that theB.t i. moved downstream as a slug after release(Fig. 2). This slug was detectable as far as 600m below the release site, but by 1,800 m theB.t.i. spores had blended with background bac-terial counts and could not be detected by ourmethod. Black fly mortality was recorded as faras 3,200 m downstream from the release site,suggesting that direct counts of spores in riverwater samples may not accurately rcflect B.t.i.carry or toxicity. This may be because modernliquid formulations contain predominantly crys-talline toxin.

The effectiveness of B.t l. against black flylarvae in the Betsie River was evident bv thesignificant increase in drift density (Fig. BB),and by mortality observed on natural and arti-ficial substrates following treatment (Fig. ?A).Results from the drift/benthos mortality exper-iments and micro-drift studies showed that allblack fly species were susceptible to B.t.i., te-gardless of instar. However, most individual lar-vae were greater than 1 mm in size, as deter-mined by the above studies.

The action of Teknar HP-D was fairly quick;larvae on substrates in the first 1,800 m weremoribund or dead 2 h after treatment. but wedid not observe larval detachment or drift until12 h posttreatment. De Moor and Car (1986)observed that larvae ofben remained attached tosubstrates for several days after treatment, be-fore being washed off by the current. Twenty-four hours after treatment and for the remaining2 days, black fly drift densities were extremel!low, far below pretreatment and above treat-ment levels, indicating a decline in the numberof drifting Iarvae in the treated area. This oc-currence is explained by studies showing thatthe distances traveled by drifting invertebratesis often less than 10 m (Townsend and Hildrew1976). Thus, in this stream section it wouldprobably have taken several days before driftingblack fly larvae recolonized the treated areasfrom upstream sites.

Efficacy of B.t.i. declines as one moves furtherdownstream from the application site (Molloyand Jamnback 1981). Maximum black fly mor-tality rates in our study (L007o) occurred in thefirst 2,200 m downstream from the release site(Fig. 7A). However, a significant decline in themortality rate (to -30%) occurred at the 3,200

m site, and no mortality was recorded at 4.300meters. Whether this decline was due to naturaldilution of the toxin as it moved downstream,or to other factors, is unknown. However. thechanges in river bed and channel morpholorymay have influenced carry. Immediately below2,200 m the profile of the Betsie River ihangedfrom a cobble-gravel bottom, long riffle, shallowpool stream to a sand bottom, deep pool andmeandering stream. A variety of iactors areknown to affect downstream carry of B.t.i. (seereviews by Lacey and Undeen 1986. Mollov1989), two of them being stream profile animorphology (Undeen et al. 1984. Colbo andO'Brien 1984, Lacey and Undeen ig86). BothMolloy and Jamnback (1981) and Colbo andO'Brien (1984) found that "pooling" was a majorimpediment to good carry of B.t j. in th-eirstream systems, and there is some evidence frommortality studies on selected nontarget benthicinsects that B.t.i. may adhere to substrates orbe retained in depositional sediments (Undeenet al. 1984, Back et al. 1985, Molloy 1989). Inthese lower reaches ofour study site, the possible"settling out" and/or "dilution" of B.t.i. causedby its movement through deep pools and contactwith finer sediments could have decreased itscarry.

When comparing other field trials that as-sessed the efficacy and carry of different B.t i.formulations in temperate climates (Molloy1989), we noted that the downstream distancein which larval black fly mortality occurred wasvery good in our study. Our comparisons werebased on studies with rivers and streams havingsimilar discharges and/or comparable applica-tion rates as those in the Betsie River (Car andDe Moor 1984, Lacey and Undeen 1984, Backet al. 1985). We observed 100% mortality at2,200 m and 30% at 3,200 m, while in otherstudies mortality never occurred or was not re-corded more than 1,000 m below the applicationpoint. A number of factors could have beenresponsible for these differences. First. the ele-vated stream temperatures (22-24"C) during asevere drought year in Michigan undoubtedlyinfluenced the level of activity of B.t.i. Compar-ative stream temperatures in other trials rangedfrom 9 to 20'C, with most below 15"C. Severalinvestigators (Lacey et al. 1978, Molloy et al.1981, Olejnicek et al. 1985, Lacoursiere andCharpentier 1988) have reported a positive cor-relation between temperature and B.t i. activityagainst black flies, and the high mortality valuesthat we obtained in this study (100%) werepossibly associated with high water temperatureconditions. Second, the formulations amongstudies were different. We used the liquid for-mulation Teknar HP-D, whereas wettable pow-der formulations or older liquid formulations

SEPTEMBER 1989 B.r.r. ru n MtcHrcet RIvnn 4t]-

were used in some of the other studies. Fieldtrials have indicated that WP formulations gen-erally have less carry than liquid ones (Guilletet al. 1982, Lacey and Undeen 1984), based onthe physics of particle size (Molloy et al. 1984).Other treatment parameters (i.e., concentration,application duration), although similar, were notstandard for each study and may have influ-enced carry and mortality. Finally, factors suchas stream discharge, profile (depth-to-width ra-tio) (Undeen et al. 1984), gradient and vegeta-tion (Frommer et al. 1981) also may have actedsolely or in combination to enhance the down-stream carry and activity of. B.t.i. in this study.

Effects of B.t.i. on nontarget organisrns: Weused several different methodologies in thisstudy, including experiments not employed inprevious studies that have assessed the effectsof B.t.i. on target and nontarget organisms. Ma-cro-drift, artificial or natural substrate coloni-zation and Surber or Hess-type sampling havebeen the most common ways of assessing inver-tebrate mortality due to B.t.i. and other insec-ticides (Kingsbury l97l,Lacey and Mulla 1989).For invertebrates, we employed micro- and ma-cro-drift studies, drift&enthos mortality exper-iments, short-term growth and feeding experi-ments, and examination of functional groupcomposition. Few in situ fish studies have everbeen conducted; therefore, we studied frsh spe-cies composition, relative abundance, diet andlength-weight relationships.

Our comprehensive approach to evaluatingthe effects of B.t.i. on nontarget organisms wasdeveloped after carefully reviewing publishedfield studies and discovering that sampling var-iability, rather than the B.t.i., may have beenresponsible for some of the reported populationchanges in nontarget organisms after treatment.This was particularly true for invertebratesnot considered to be susceptible to B.t.i.(e.g., Ephemeroptera, Trichoptera). Molloy andJamnback (1981) noted that large increases innontarget organisms after treatment in theirstudy may have been the result of small samplesize and choice of sampling sites. Posttreatmentincreases in Trichoptera and Ephemeroptera inthe Vaal River, Africa, were due to colonizationof vacated habitats created when simuliid larvaedrifted downstream (Carr and De Moor 1984).Several authors attributed decreases in benthicdensities of nontarget organisms to emergenceand pupation after treatment rather than toB.f.i. (Colbo and Undeen 1980, Burton 1984,3

' g""to", D. K. 1984. Impact of Boci llus thuringien-sis var. israeLensrs in dosages used for black fly (Si-muliidae) control, against target and non-target orga-nisms in the Torch River, Saskatchewan. M. S. Thesis,Department of Entomology, University of Manitoba,Winnipeg.

Gibbs et al. 1986). Although Pistrang and Burger(1984) recorded moderate increases of somedrifbing mayflies and caddisflies following B.t.i.treatmint, there was no spatial control to thestudy and the emergence of some mayflies oc-cu.t"d during the study period. In the abovestudies, increases in drift of nontargets also mayhave been due, in part, to a temporary increasein particles introduced into the stream with theB.t.i. application (Lacey and Mulla 1989) or toeffects of adjuvants and dispersants included inthe product formulation (Pistrang and Burger1984, Lacey and Mulla 1989). Thus, samplingvariability can result from the choice and oper-ation of sampling devices, physical features ofthe environment, field and laboratory sortingprocedures, product ingredients and formula-tion, and biological features of the study orga-nisms themselves (c/. Resh 1979, Merritt et al.1984).

In addition to sampling variability, the spatialand temporal dynamics of a stream require spe-cial consideration in an environmental impactstudy such as this. A design must simultaneouslyinclude both spatial and temporal controls. AsGreen (1979) stated: "If the spatial control ismissing and only before- and after-impact sam-ples from an impacted area are available, oneruns the risk that a sigrrificant change may beunrelated to the impact. . . . If the temporal con-trol is missing, one may not detect that a differ-ence between an area subject to the impact andan area not subject to it existed before the im-pact occurred." Although Green (1979) recom-mended an areas-by-times factorial analysis ofdata collected in an impact study, we advocatea nested analysis, where the temporal informa-tion is a level nested within the spatial level. Anested analysis provides for variation in naturaltemporal changes at different sampling sites andis especially applicable when each site is sampledseveral times before and aft,er the impact occurs.We successfully applied a nested ANOVA to ourmacro-drifb data, and indeed different patternsof drift at the above and below stream sites,before and after treatment with B.t.i., were ob-served for certain taxa. For comparisons whensamples were taken only immediately before andafter treatment with B.t.i. at upstream anddownstream sites, we opted to use simpler sta-tistical comparisons (t-tests, Mann-Whitney U-tests) to examine differences within the "above-

below and before-after" concept.Within the 20 families of aquatic inverte-

brates investigated in this study, some variationin macro-drift was recorded above and below,and before and after treatment (Figs. 3 and 4).These observed increases and decreases in ma-cro-drift patterns can be ascribed to samplingerror, random factors and population processes

4t2 JouRNar, oF THE AMERTcAN Moseurro Cor.rrnor, AssocrATroN V o L . 5 , N o . 3

unrelated to B.t.i. application. We found nosignificant effect of B.t.i. on invertebrate micro-drift composition, for either size class, total driftnumbers, or total biomass when above and belowsites were compared, before and after treatment.Further, no discernible shifts in taxonomic orfunctional group composition could be associ-ated with the B.t.i. application. Short-termgrowth experiments with Stenonemo showed nomeasurable effects of the B.t.i. treatment. andthere was no effect on invertebrate feeding ac-tivity. The high mortality rates of Stenoiemamay have been largely a natural consequenceassociated with the growth of the larvae, asreported previously for Leptophlebia \awae(Cummins and Wilzbach 1988).

The numerous field trials evaluatins the ef-fects of B.t.i. on target and non-targeiinverte-brates (see reviews by Molloy 198g, Lacey andMulla 1989), indicate that only members of thedipteran suborder Nematocera have been shownto be significantly affected. We specifically ex-amined chironomids belonging to the genusRheotanytarsus because of published reports oftheir susceptibility to B.t.i. (Ali, 19g1, Car andDe Moor 1984, De Moor and Car 1986), and thefact that their microhabitat, food and mecha-nism of food acquisition were similar to that ofblack flies (Wallace and Merritt 1980, Coffmanand Ferrington 1984). Our results showed thatafter considering field mortality, approximately27% of.thepopulation of this midge was affectedby B.t.i., and that level of mortality was re-stricted to the first sampling station below thetreatment site (Fig. 7B). Rheotanyforsus mor-tality was much Iower than black fly mortalityat sites farther downstream, and mortality atthese sites was similar to or lower than mortalityobserved at the upstream control site.

No significant effect of B.t.i. on chironomiddrift or midge densities from substrate samplingwas detected in our study; however (except forRheotanytarsus), we did not examine changesbelow the family level. If B.t.i. had caused acutetoxicity and mortality in the resident chiron-omid populations in the Betsie River, we wouldhave expected a similar pattern of macro-driftas was observed for black flies (compare Fig. 3Band Fig. 3D). Back et al. (1985) found no signif-icant effect of B.t.i. on Rheotanyforsus but re-corded reduced posttreatment densities of twoother chironomid genera, Euhiefferella and Po-lypedilum. They also reported mortality in theBlephariceridae (Nematocera) at high dosages,whereas Gibbs et al. (1986) found no adverseeffects of B.t.i. on this family at operationaldosages in the field. Clearly, further research isneeded, using comparable formulations and dos-ages, to properly assess the effects of B.t.i. onthe Nematocera in the field.

There have been few field studies on the ef-fects of B.t.i. on fish in streams. Laboratorvstudies report no effects of B.t.i. against mos-quitofish, rain water killifish, sticklebacks (Gar-cia et al. 1980) or Tilapiasp. (Lebrun and Vlayen1981) at labeled rates; however, with very highexposure rates (4,000 mg/liter), 50% of the Ti-lapia sp. died (LeBrun and Vlayen 1981). In alaboratory study, Fortin et al. (1986) reportedno acute effects on brook trout fry either ex-posed to or fed B.t i. at normal application rates.At a dosage of 3,000 mg/liter, the fry exhibitedsome behavioral changes but no mortality, whileat dosages >4,500 mg/liter (i.e., levels 1,000times greater than operational control dosages)there was observable mortality. Fortin et al.(1986) related this mortality experimentally tothe presence of xylene and other monocyclicaromatic hydrocarbons in the particular formu-lation of B.t j. used, rather than to B.t.i. toxins.Gibbs et al. (1986) found no changes in diets ofslimy sculpin or brook trout in streams in Mainebefore and after treatment with B.t.i. and addi-tionally found that black fly larvae comprised asmall proportion of the diet of these fish. In theNew River of West Virginia, Amrine (1982)similarly found that black fly larvae compriseda small proportion of shiner (Notropls sp.) diets.In a third order New York stream, Molloy (un-published data) found no negative effects ofB.t.i. application on brook trout, brown trout orslimy sculpin densities when sampled after 2.5years of black fly elimination.

In our study, no effect of B.t.i. on numbers orspecies composition of frsh was evident, becausethere was no change in numbers or compositionof the fish community at the downstream stretchthat occurred. The observed overall variation incollections is attributable to other, unknownfactors related to the particular stretches sam-pled and to the prevailing hot, dry weather con-ditions. Detailed studies showed no shifts inIength-weight relationships in wild-caught rockbass, nor any mortality or significant weightchanges in caged rock bass, owing to B.t.i. fueat-ment.

Diet composition of rock bass probably re-flected prey availability. Because black fliescomprised an insignificant portion of rock bassdiets, it is not likely that black fly removal froma section of stream would affect diet compositionor adversely affect fish growth. We concludefrom these data that, under the conditions ofour study, fish were not affected by B.t l. release.

In summary, this short-term study found nosignificant effects of B.t.i. on major nontargetinvertebrates and fish. The high selectivity,mortality and excellent carry obtained in thisstudy showed that B.t.i. has good potential foruse as a biological control agent against larval

Seprn\,rsnn 1989 B.r.r. rn l MtcrucrN RtveR 41.3

black flies in Michigan streams. Future researchshould be directed at long-term studies in tem-perate climates, as are currently being conductedin the Adirondack Mountains of New York (D.Molloy, personal communication). In addition,studies evaluating the effects of black fly re-moval on stream predators, scavengers and theoverall food web are advised and will be anavenue of research pursued by this Iaboratory inthe future.

ACKNOWLEDGMENTS

We would like to express our sincere thanksto the following individuals for assistance withvarious aspects ofthe study: George Petritz andMike Call, Crystal Mountain Resort, Thomp-sonville; Tom Rohrer and Jack Wuycheck, Sur-face Water Quality Division, Michigan Dept. ofNatural Resources, Lansing; Emily Olds, ChrisEdens, Sylvia Heaton, Cliff Leavitt and PaulMcCann, Dept. of Entomology, Michigan StateUniversity; and Donna Gates, Appalachian En-vironmental Laboratory, University of Mary-land. We also would like to thank Doug Ross,Zoecon Corp., Dallas, Texas, for providing Tek-nar and technical assistance; Peter Adler, Dept.of Entomology, Clemson University, for cytolog-ical identification of black flies; andDan Molloy,New York State Museum, for his expertise inthe area of biological control of black flies, forsharing unpublished data and for the use of hisdrift nets. This study was supported, in part, bythe Northeast Regional Black Fly Project NE-II8, National Institutes of Health Grant AI-21884 awarded to RWM, the Michigan Dept. ofNatural Resources and the Michigan State Uni-versity Cooperative Extension Service and Ag-ricultural Experiment Station.

REFERENCES CITED

AIi, A. 1981. Bacillus thuringiensis serovar. isroel.ensis(ABG-6108) against chironomids and some nontar-get aquatic invertebrates. J. Invertebr. Pathol.38:264-272.

Amrine, J. W. 1982. The New River connection to theblack fly problem in southern West Virginia. WestVirginia Univ. Agr. For. Exp. Stat. Bull. 678:1-30.

APHA-AWWA-WPCF. 1980. Standard methods forthe examination ofwater and wastewater. (15th ed).APHA. Washinpon, DC.

Aronson, A. I., W. Beckman and P. Dunn. 1986.Bacillus thuringiensis and related insect pathogens.Microbiol. Rev. 50:1-24.

Back, C., J. Boisvert, J. O. Lacoursiere and G. Char-pentier. 1985. High-dosage treatment of a Quebecstream with Bacillus thuringiensis serovar. israel.en-srs: efficacy against black fly larvae (Diptera: Si-muliidae) and impact on non-target insects. Can.Entomol. 117: 1523-1534.

Car, M. and F. C. De Moor. 1984. The response ofVaal River drift and benthos to Simuliunt (Diptera:Nematocera) control using Bacillus thuringiensisvat. israel,ensis (H-14). Onderstepoort J' Vet. Res.51:155-160.

Coffman, W. P. and L. C. Ferrington, Jr. 1984. Chi-ronomidae, pp. 551-652. .In; R. W. Merritt and K.W. Cummins (eds), An introduction to the aquaticinsects of North America (2nd ed). Kendall/HuntPubl. Co., Dubuque, Iowa.

Colbo, M. H. and H. O'Brien. 1984. A pilot black fly(Diptera: Simuliidae) control program using Bacillusthuringiensis var. israelensis in Newfoundland. Can.Entomol. 116:1 085-1096.

Colbo, M. H. and A. H. Undeen. 1980. Effect ofBacillus thuringiensis var. israelensis on non-targetinsects in stream trials for control of Simuliidae.Mosq. News 40:368-371.

Cummins, K. W. and M. A. Wilzbach. 1985. Fieldprocedures for analysis of functional feeding groupsof stream invertebrates. Contrib. 1611 AppalachianEnviron. Lab., Univ. Maryland, Frostburg. 18 p.

Cummins, K. W. and M. A. Wilzbach. 1988. Do path-ogens regulate stream invertebrate populations?Verh. Internat. Verein. Limnol. 23 :1232-1243.

De Moor, F. C. and M. Car. 1986. A field evaluationof Bacillus thuringiensis var. israplensis as a biolog-ical control agent for Sirnulium chutteri (Diptera:Nematocera) in the middle Orange River. Onderste-poort J. Vet. Res. 53:43-50.

Dubois, N. R. and F. B. Lewis. 1981. What is Bacillusthuringiensis? J. Arboriculture 7 :233-240.

Fortin, C., D. Lapointe and G. Charpentier. 1986.Susceptibility of brook trout (Saluelinus fontinalis)fry to a Iiquid formulation of Bacillus thuringiensisserovar. israelensis (Teknar@) used for black flycontrol. Can. J. Fish. Aq. Sci. 43:1667-1670.

Fredeen, F. J. H. 1975. Effects of a single injection ofmethoxychlor black fly larvicide on insect larvae ina 161-km section ofthe North Saskatchewan River.Can. Entomol. 107:807-817.

Fredeen, F. J. H. 1977. A review of the economicimportance of black flies (Simuliidae) in Canada.Quaest. Entomol. l3:2I9-29.

Fredeen, F. J. H. 1983. Trends in the numbers ofaquatic invertebrates in a large Canadian river dur-ing four years of black fly larviciding with methox-ychlor (Diptera: Simuliidae). Quaest. Entomol.19:53-92.

Fredeen, F. J. H. 1985. Some economic effects ofoutbreaks of black flies (Simulium ft4ggerl (Nichol-son and Mickel)) in Saskatchewan. Quaest. Ento-mol. 21:175-208.

Frommer, R. L., S. C. Hembree, J. H. Nelson. M. P.Remington and P. H. Gibbs. 1981. The evaluationof Bacillus thuringiensis var. israelensis in reducingSimulium uittatum (Diptera: Simuliidae) larvae intheir natural habitat with no extensive aquatic veg-etative growth. Mosq. News 41:339-347.

Garcia, R., B. Des Rochers and W. Tozer. 1981. Stud-ies on Bacillus thuringiensis var. israel,ensis againstmosquito larvae and other organisms. Proc. Calif.Mosq. Vector Control Assoc. 49:25-29.

Gibbs, K. E., F. C. Brautigam, C. S. Stubbs and L. M.Zibilske. 1986. Experimental applications of B.t i.for larval black fly control: persistence and down-

414 JounNel oF THE ArupnrclN Mosquno CoNrnor, Assocrlrron V o L . 5 , N o . 3

stream carry, efficacy, impact on non-target inver-tebrates and fish feeding. Maine Life Sci. Agric.Exp. Stn. Tech. Bull. 123:l-25.

Goldberg, L. J. and J. Margalit. 197?. A bacterial sporedemonstrating rapid larvicidal activity againstAnopheles sergentii, Uranotaenia unguiculata, Culexuniuitattus, Aedes aegypti and, Cubx pipi.ens. Mosq.News 37:355-358.

Green, R. H. 1979. Sampling design and statisticalmethods for environmental biotogists. John Wileyand Sons, New York.

Guillet, P., H. Escaffre and J. M. Prud'Hom. 1982.BaciLlus thuringiensis H-14, a biocontrol agent foronchocerciasis control in West Africa, pp. 460-465,In: Invertebrate pathology and microbial control.Proc. ofthe 3rd Internat. Colloq. on Invertebr. Path.Univ. Sussex, Brighton, U.K.

Hobbie, J. E., R. J. Daley and S. Jasper. 1977. Use ofNuclepore filters for counting bacteria by fluores-cence microscopy. Appl. Environ. Microbiol.33:1225-L228.

Jamnback, H. 1969. Bloodsucking flies and other out-door nuisance arthropods of New York State. Mem.19, New York State Mus. and Sci. Serv.. Albanv.

Kim, K. C. and R. W. Merritt (eds). 198?. Black flies:ecology, population management, and annotatedworld list. Penn. State Univ. Press, University Park,PA.

Kingsbury, P. D. 1975. Monitoring aquatic insect pop-ulations in forest streams exposed to chemical andbiological insecticide applications. Proc. Entomol.Soc. Ontario 106:19-24.

Kondratieff, P. F. and G. M. Simmons, Jr. 1985.Microbial colonization of seston and free bacteriain an impounded river. Hydrobiologia 128:127-133.

Lacey, L. A., H. Escaffre, B. Philippon, A. Seketeliand P. Guillet. 1982. Large river treatment withBacillus thuringi.ensis (H-14) for the control of Si-muliwn damnosum s.I. in the Onchoceriasis ControlProgramme. Z. Tropenmed. Parasitol. 33:97-101.

Lacey,L. A. and M. S. Mulla. 1989. Safety of Bacillusthuringiensis vat. israelensis and Bacillus sphaericusto non-target organisms in the aquatic environment,in press. In: M. Laird, L. A. Lacey and E. W.Davidson (eds). Safety of microbial insecticides.C.R.C. Press, Boca Raton, FL.

Lacey, L. A., M. S. Mulla and H. T. Dulmage. 1978.Some factors affecting the pathogenicity of Bacillusthuringiensis Berliner against black flies. Environ.Entomol. 7:583-588.

Lacey, L. A. and A. H. Undeen. 1984. Effect of for-mulation, concentration, and application time onthe efficacy of Bacillus thuringiensis (H-14) againstblack fly (Diptera: Simuliidae) Iarvae under naturalconditions. J. Econ. Entomol. 77:472-418.

Lacey,L. A. andA. H. Undeen. 1986. Microbial controlof black flies and mosquitoes. Annu. Rev. Entomol.3l:265-296.

Lacey, L. A. and A. H. Undeen. 1987. The biologicalcontrol potential ofpathogens and parasites ofblackflies, pp. 327-340. In: K. C. Kim and R. W. Merritt(eds), Black flies: ecology, population management,and annotated world list. Penn. State Univ. Press,University Park, PA.

Lacoursiere, J. O. and G. Charpentier. 1988. Labora-tory study ofthe influence ofwater temperature and

pH on Bocillus thuringizrxis var. isroelensis effrcacyagainstblack fly larvae (Diptera: Simuliidae). J. Am.Mosq. Control Assoc. 4:64-72.

LeBrun, P. and P. Vlayen. 1981. Etude de la bioactiv-ite comparee et des effects secondaires de Bacillusthuringiensis H 14. Z. Angew. Entomol. 91:15-2b.

Merritt, R. W. and K. W. Cummins (eds). 1984. Anintroduction to the aquatic insects of North America(2nd ed). Kendall/Hunt Publishing Co., Dubuque,IA. 710 pp.

Merritt, R. W., K. W. Cummins and V. H. Resh. 1984.Collecting, sampling and rearing methods foraquatic insects, pp. 11-26. fn; R. W. Merritt and K.W. Cummins (eds), An introduction to the aquaticinsects of North America (2nd ed). Kendall/HuntPublishing Co., Dubuque, IA.

Merritt, R. W. and H. D. Newson. 1978. Ecology andmanagement of arthropod populations in recrea-tional lands, pp. 125-762.1n; G. W. Frankie and C.S. Koehler (eds), Perspectives in urban entomology.Academic Press, New York.

Mohsen, Z. H. and M. S. Mulla. 1982. Field evaluationof Simulium larvicides: effects on target and non-target insects. Environ. Entomol. 11:390-398.

Molloy, D. P. 1989. Progress in the biological controlof black flies with B.t.i., with emphasis on temperateclimates, in press. In: H. de Barjac and D. J. Suth-erland (eds), Bacterial control of mosquitoes andblack flies, Rutgers University Press, New Bruns-wick, NJ.

Molloy, D., R. Gaugler and H. Jamnback. 1981. Fac-tors influencing efficacy of Bacillus thuringicnsisvar. israel,ensis as a biological control agent of blackfly larvae. J. Econ. Entomol. 74:67-64.

Molloy, D. and H. Jamnback. 1981. Field evaluationof Bacillus thuringiensis var. israel.ensi.s as a blackfly biocontrol agent and its effect on nontargetstream insects. J. Econ. Entomol. 74:314-318.

Molloy, D., S. P. Wraight, B. Kaplan, J. Gerardi andP. Peterson. 1984. Laboratory evaluation of com-mercial formulations of Bacillus thuringiensis vat.israelensis against mosquito and black fly larvae. J.Agric. Entomol. 1:161-168.

Newson, H. D.1977. Arthropod problems in recreationareas. Annu. Rev. Entomol. 22:333-53.

Olejnicek, J., V. Matha and J. Weiser. 1985. Theefficacy of Bacilhrs thuringiensis var. israel.etuisagainst larvae of the black fly Odogrnia onnta(Meig.) (Simuliidae) at Iow temperatures. Folia Par-asitol. (Prague) 32:27 7-27 7.

Pistrang, L. A. and J. F. Burger. 1984. Effect ofBacillus thuringiensis var. israelensis on a geneti-cally-defined population ofblack flies (Diptera: Si-muliidae) and associated insects in a montane NewHampshire stream. Can. Entomol. 116:975-981.

Porter, K. G. and Y. S. Feig. 1980. The use of DAPIfor identifying and counting aquatic microflora.Limnol. Oceanogr. 25:943-948.

Resh, V. H. 1979. Sampling variability and life historyfeatures: basic considerations in the design ofaquatic insect studies. J. Fish. Res. Bd. Can. 36:290-311.

Steel, R. G. D. and J. H. Torrie. 1980. Principles andprocedures of statistics (2nd ed). McGraw-Hill, NewYork.

Steelman, C. D. 1976. Effects ofexternal and internal

B.r.r. IN e MIcntcAN Rtvsn 415SEPTEMBER 1989

anthropod parasites on domestic livestock produc-tion. Annu. Rev. Entomol. 21:155-178.

Townsend, C. R. and A. G' Hildrew. 1976. Field ex-periments on the drifting, colonization and contin-uous redistribution of stream benthos' J. Anim.Ecol. 45:759-772.

Undeen, A. H., L. A. Lacey and S. W. Avery. 1984. Asystem for recommending dosage of Bacillus thurin-giensis (H-14) for control of simuliid larvae in smallstteams based upon stream width. Mosq. News44:553-559.

Waldbauer, G. P. 1968. The consumption and utili-zation of food by insects. Adv. Insect Physiol. 5:229-282.

Walker. E. D.. E. J. Olds and R. W. Merritt. 1988.Gut content analysis of mosquito larvae (Diptera:Culicidae) using DAPI stain and epifluorescencemicroscopy. J. Med, Entomol. 25:551-554.

Wallace. J. B. and R. W. Merritt. 1980. Filter-feeding

ecology of aquatic insects. Ann. Rev. Entomol'25:103-132.

Waters, T. F. 1969. Subsampler for dividing large*ulnpi". of stream invertebrate drift. Limnol'Oceanogr. 14:813-815.

Waters, T. F. 19?2. The drift of stream insects. Annu'Rev. Entomol. I7 :253-272.

Wilzbach. M. A. and K' W. Cummins. 1989. An as-sessment of short-term depletion of stream ma-croinvertebrate benthos by drift. Hydrobiologia (inpress).

Wilzbach. M. A.. K. W. Cummins andJ. D. HaII' 1986.Influence of habitat manipulations on interactionsbetween cutthroat trout and invertebrate drift. EcoI-ogy 67:898-911.

Woodwell. G. M.. C. F. Wurster, Jr. and P. A. Isaacson.1967. DDT residues in an east coast estuary: a caseof biological concentration of persistent insecticide.Science 156:821-824.