New Zealand Journal of Marine and Freshwater Research, 2006, Vol. 40: 531-5460028-8330/06/4004-0531 © The Royal Society of New Zealand 2006

531

Do toxic baits containing sodium fluroacetate (1080) affect fishand invertebrate communities when they fall into streams?

ALASTAIR M. SURENNational Institute of Water and Atmospheric

Research LimitedP.O. Box 8602, RiccartonChristchurch, New Zealand

PAUL LAMBERTNational Institute of Water and Atmospheric

Research Limited28 Johnstone StGreymouth, New Zealand

Abstract Large-scale control of Australianpossums throughout New Zealand uses toxic cerealbaits containing 1080 (sodium fluoroacetate). Thesebaits are often aerially applied over rough terrainwhere ground application is impractical. Manysmall streams flow in these areas, so 1080 baits canpotentially fall into them. The ecological effect of1080 leaching from baits was assessed on freshwaterfish and invertebrate communities in four streamsusing a BACI experimental design. Four sites wereselected in each stream: 10m and 100m below and10m and 100m above where 1080 baits were placed.All sites were monitored 4 days and 1 day beforeand after baits were added, respectively. Separateexperiments were conducted to assess impacts of1080 on native fish and invertebrate communities.Baits were added to each stream to achieve baitdensities 10 × greater than found after normal controloperations. Three species of native fish, longfineels (Anguilla dieffenbachii), koaro (Galaxiasbrevipinnis), and upland bullies (Gobiomorphusbreviceps) were placed into separate cages at eachsite in each stream, and mortality monitored duringthe experiment. Analysis of water samples collectedduring the fish experiment showed that 1080 wasdetected only for 12h, and at low concentrations (c.0.2 μg litre-1), despite the large number of baits placedin each stream. No fish died after addition of 1080

M06005; Online publication date 28 September 2006Received 12 January 2006; accepted 15 May 2006

baits, suggesting that all three species were tolerantto dissolved 1080 at concentrations observed in thisstudy. Invertebrate communities were quantified bysampling 10 replicate rocks at each site. Caddisflies(Helicopsyche, Pycnocentrodes, and Pycnocentria),orthoclad midges, and the mayfly Deleatidiumdominated the community. 1080 had no detectableeffects on the invertebrate community. These resultssuggest that 1080 leaching from submerged baitsin small streams has no demonstrable biologicalimpacts. Based on this finding, the need to maintainbuffer zones around large waterways as somecouncils require is questioned.

Keywords invertebrates; native freshwater fish;sodium fluoroacetate; 1080; toxicity testing; BACIdesign

INTRODUCTION

The introduced Australian brushtail possum(Trichosurus vulpécula) is common throughout NewZealand, where it jeopardises natural ecosystems andacts as a vector for bovine tuberculosis, threateningthe highly valuable dairy and cattle industry (Green2003). As part of extensive nationwide controloperations, the Animal Health Board (AHB) andDepartment of Conservation (DOC) use cereal baitscontaining 0.15% sodium fluoroacetate (compound1080) as a poison to reduce possum densities in areaswhere they pose risks to either native ecosystems ornearby cattle. Baits are made of cereal, dyed greento reduce their attractiveness to birds, and flavouredwith cinnamon to increase their palatability topossums. Two types of baits are commonly producedwhich differ in their degree of weathering: the moreweather resistant Wanganui No. 7 and less resistantRS5 baits. Three bait sizes of each type (c. 2g,6g, and 12 g) are also produced. Although groundapplication of 1080 baits is preferred, this is notpractical in densely forested mountainous areas.Under such conditions, helicopters aerially apply1080 baits along clearly defined flight paths over

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

532 New Zealand Journal of Marine and Freshwater Research, 2006, Vol. 40

prescribed operational areas. Aerial application ismore commonly used in New Zealand than othercountries, which generally mostly use ground-baitingoperations for pest control to minimise adverseeffects on their indigenous mammal populations.Lack of indigenous mammals in New Zealand meansthat such concerns are unwarranted. New Zealandaerial operations are controlled by resource consents,many of which require that rivers >3 m have buffersaround them to minimise accidental contaminationof streams by 1080 baits. However, smaller streamshave no such buffers, and 1080 baits often fall intothese (Suren 2006).

Fish living in streams may be exposed to 1080leaching from submerged baits, with possible adverseimpacts. Previous studies have shown that a widevariety of fish are tolerant to dissolved 1080, evenat high concentrations. King & Penfound (1946)found that bass (presumably a species of Micropterus:Centrarchidae) survived without signs of toxicity inwater containing 370 pig litre"1 of 1080; Batcheler(1978) showed that rainbow trout (Oncorhynchusmykiss: Salmonidae) survived in 580 pig litre"1 of1080 for 24h, and Fagerstone et al. (1994) found thatbluegill sunfish (Lepomis macrochirus: Centrarchidae)showed no signs of mortality over 96 h to 970 pig litre"1

of 1080. 1080 is thus regarded as being either non-toxic, or only slightly toxic to fish, depending on thespecies (Fagerstone et al. 1994). Despite informationon toxicity of 1080 to the above three fish, potentialeffects on New Zealand native fish are unknown.Although 1080 concentrations that fish have beenexperimentally exposed to have ranged from 90 to240 x the maximum concentration detected in watermonitoring programmes (4 pig litre"1) following aerialapplication of 1080 within New Zealand (Eason2002; Green 2003), potential effects of 1080 onnative fish need to be assessed, especially given thecontentious nature of the use of 1080 within NewZealand (Livingston 1994; Suren 2006).

Baits falling into streams may affect aquaticinvertebrates, given the documented toxicity of 1080to terrestrial invertebrates (Eason et al. 1993; Eisler1995; Spurr & Drew 1999). Detritivores such asstoneflies (e.g., species of Acroperla, Austroperla,Zelandoperla), caddisflies (Zelandopsyche), andsome tipulids may be especially vulnerable, as theycould potentially consume fragments of 1080 bait.However, it is highly unlikely that these animalswould consume whole baits, which are considerablylarger than individual animals. By the time baitsfragment, most, or all, of the 1080 would haveleached from them (Suren 2006), so the chance of

invertebrates suffering mortality from consumingfragments is negligible. Nevertheless, freshwaterinvertebrates may be affected by 1080 leaching fromsubmerged baits, especially in small streams. Toxicitytests performed by the United States EnvironmentalProtection Agency have shown that the non-observable effect concentration (NOEC) of 1080for the small freshwater invertebrate Daphnia magnawas 130 mg litre"1 (Fagerstone et al. 1994). At thishigh concentration, the effect of 1080 was regardedas being "practically non-toxic" to Daphnia. Anotherstudy of 1080 toxicity to fourth instar mosquitolarvae {Anopheles quadrimaculatus) showedthat 1080 was among the most toxic 3% of 6000organic compounds screened, with 65% mortality atconcentrations of 100 pig litre"1 over a 48 h period(Deonier et al. 1946). These two concentrations areabout 30 x higher than the maximum concentrationof 1080 detected in streams in New Zealand (Green2003), so it seems unlikely that 1080 baits falling intostreams would have a toxic effect on New Zealandinvertebrates. Despite this assertion, no studies havequantified the toxicity of 1080 to any New Zealandfreshwater invertebrates, nor assessed the effects ofsubmerged baits on freshwater invertebrates.

Despite the widespread use of 1080, and itspotential to adversely impact aquatic organisms, onlytwo published reports have attempted to quantify suchimpacts in New Zealand (Taranaki Regional Council1993, 1994). Both studies used qualitative kicksamples to characterise invertebrate communitiesin streams flowing through forested areas where1080 had been applied aerially. Invertebrate sampleswere collected from streams before and after 1080application, and a number of biotic indices calculatedto describe the invertebrate communities. No de-monstrable effects of the aerial 1080 applicationwere observed (Taranaki Regional Council 1993,1994), suggesting that stream invertebrates werenot sensitive to 1080. Biological samples for thesestudies were collected before, and between 32 and42 days after the aerial application of 1080. Thissampling protocol may thus have missed any short-term reductions in invertebrate density arising as aresult of any toxic effect of 1080. Moreover, thesetwo studies made the implicit assumption that theyadequately sampled areas where 1080 contaminationhad occurred. This was, however, not controlledfor, and recent work (Suren 2006) has shown baitsdo not necessarily land in streams flowing in areaswhere 1080 is aerially applied. It could be arguedthat although these monitoring studies demonstratedno adverse impacts of 1080 operations at a catchment

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

Suren & Lambert—Effect of 1080 baits on fish and invertebrates 533

Ngahere

o- \ \

\ V

171°E

Study site J

Greymouth^T^"

-43°s / *

/ ^ J

X

y¿Christchurch

i I

II )

Fig. 1 Site map showing the five small streams selected for study. Three of the streams were used for both the fishand invertebrate study, whereas Wallaby Creek was used for the fish study only, and Small creek for the invertebratestudy only. Within each stream, there were two sites above locations where 1080 (sodium fiuoroacetate) baits wereadded (open circles), and two sites below (closed circles). Note that Crooks creek only had one control site for thefish study as cages had been stolen from the upper control site. For the invertebrate study, two control sites were usedat this stream.

scale, they did not adequately quantify the effectsof 1080 bait in individual streams. Finally, neitherstudy specifically examined the effects of 1080contamination on fish communities.

Continued public concern exists about theenvironmental fate of 1080 (e.g., Laugesen &Hubbard 2002; Speedy 2003), especially in water.The present study was designed as a field-basedexperiment at the scale of individual catchmentsto examine the effect of submerged 1080 baits ontwo trophic levels within streams: freshwater fishand invertebrates.

MATERIALS AND METHODS

Study sitesFive headwater tributary streams of Dead Horse andWallaby creeks in the Mawhera State Forest in theGrey Valley, on the west coast of the South Island,

were selected for the study (Fig. 1). All streams wererelatively remote from human habitation, and small(<3 m wide), so would not have buffers around themin the course of normal 1080 operations. They thusrepresented streams likely to become contaminatedby 1080. Dead Horse tributary, Wallaby Creek andSmall creek flowed through catchments dominatedby plantation pines (Pinus radiata), whereas Crookscreek and Wheelbarrow creek flowed from uppercatchments of native bush (mainly podocarp forest)into lower catchments of pine. The immediateriparian vegetation of all streams was a mixture ofpines, native scrub, and grasses. Boulders, cobblesand gravels dominated the streambed material at allsites. Velocity was measured at 20 vertical cross-sections across transects in each stream (0.4 x depth)using a small Ott meter, at the beginning and endof the experiment, and immediately before 1080baits were added. Transects were located within20 m of where 1080 baits were added. Discharge

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

534 New Zealand Journal of Marine and Freshwater Research, 2006, Vol. 40

in each stream was calculated (to within + 6.9%)using the TDGAUGE (2005) programme. StreampH (to + 0.01 units), and conductivity (to within +4.4%) were measured on each of the four samplingperiods using a TPS WP81 pH/conductivity meter,and temperature and dissolved oxygen (to within +0.02 ppm) were measured using a TPS WP82 oxygensensor. These measurements were taken at the samelocalities as each gauging.

We planned to conduct the fish and invertebrateexperiments simultaneously in March 2004.However, high rainfall and river flows in the regionduring February 2004 resulted in low invertebratedensities that would have been insufficient to detectany potential effects of 1080. The invertebratecomponent of the study was thus delayed until midMay when invertebrate densities had recovered.Wallaby Creek was not used for the invertebrateexperiment, as large amounts of filamentous greenalgae covered the streambed at that time, reducinghabitat conditions for invertebrates. We subsequentlychose a new stream (Small creek) that was nearCrooks creek for this experiment (Fig. 1). The fishexperiments were conducted in a lower 200 m sectionof each stream, while the invertebrate experimentswere conducted in a 200 m section upstream of where1080 baits were placed for the fish experiment, toensure independence of each experiment. Within eachstream, two control sites were selected c. 100 m and10 m upstream from where 1080 baits were placed(sites 1 and 2), and two impact sites c. 10m and100m downstream from these baits (sites 3 and 4).

Field methods—fish studyThe impact of 1080 leaching from baits was examinedon three fish species that are common on the westcoast: longfin eels (Anguilla dieffenbachii Gray(Anguillidae)), koaro (Galaxias brevipinnis Günther(Galaxiidae)) and upland bullies (Gobiomorphusbreviceps (Stokell) (Gobiidae)). These fish werecaught by electric fishing from a number of smallstreams in the Grey Valley. All fish were transportedto experimental streams within 12 h of collection inplastic buckets equipped with battery-operated airpumps to ensure constant aeration.

Fish cages were constructed from 250-litre plasticbarrels cut in half longitudinally and fitted with 2 mmmesh netting on their upstream and downstreamends. A 2mm mesh lid on the top was sealed withVelero. Up to 10 large cobbles (mean largest diam.140 mm) were added to each cage to provide shelter.Four PVC tubes (24mm diam. and 200mm length)were also added to cages holding eels, to provide

additional shelter. Three cages (one cage per species)were deployed in relatively deep runs or pools (upto 40 cm deep) at each of the four sites, which wereexcavated until cages were submerged. Each cagewas anchored to the streambed, and large cobblesplaced on top to weigh them down and ensure theirmesh lids remained closed. Once deployed, eight fishof each species were placed into individual cages atall sites. Aquatic insects that naturally drifted into thecages provided the fish with food. Extra invertebrates(obtained from the streams) were added to the cagesevery 4 days to minimise any food limitation. Allcages at site 1 were stolen from Crooks creek beforethe experiment started, leaving control cages at site2 only.

Fish were added to all cages on 14 March 2004and mortality recorded 1 and 4 days later (15 and 18March 2004). On each occasion, all cobbles and PVCpipes (in the eel cages) were removed from eachcage, which was then emptied into a large collectingbucket. All fish were counted and dead fish recorded.The cobbles and PVC pipes were replaced into eachcage, which was then partially submerged. The fishwere subsequently poured from the collecting bucketback into each cage, the Velcro-sealed mesh lid wasresealed, and the cage repositioned under water.

After the second observation of fish survivalon 18 March 2004, 1080 baits (Wanganui No. 7,mean weight = 6.5 g) were added to each stream. Wewanted to expose the two impact sites, especiallysite 3 (10 m below the baits) to a high level of1080 leaching from baits. Suren (2006) quantifiedthe number of baits falling into streams flowingthrough areas where 1080 had been aerially appliedduring possum control operations, and found thatthe highest number of baits per 10m of stream waseight. We thus added 80 baits (10 x the maximumnumber) into the stream with the highest discharge(Wheelbarrow creek). The other three creeks hadlower discharges, so the number of baits added tothese was calculated according to the ratio of theirdischarge to that of Wheelbarrow creek (Table 1).Baits were counted into three sets of nylon meshbags (mesh size 10 mm) that were submerged andanchored to the streambed at three evenly spacedlocations across each stream to ensure completemixing of leached 1080.

Observations of fish survival at all four siteswere made 1 and 4 days after addition of 1080(19 and 22 March 2004). On the last occasion, allfish were anaesthetised and their lengths measuredto the nearest mm. Individual water samples (1litre) were collected from all sites concurrently

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

Suren & Lambert—Effect of 1080 baits on fish and invertebrates 535

with observations of fish on all four dates fromthe centre of each stream. Water samples were alsocollected from the two impact sites 2, 4, and 8hafter addition of baits, reflecting the rate of 1080leaching from baits (Suren 2006). All water sampleswere frozen (-18°C within 6 h of collection) andanalysed for dissolved 1080 concentration usinggas chromatography, method TLM 005 (LandcareResearch Ltd, see Lyver et al. 2005). The limit ofdetection was 0.001 pig ml"1 (0.1 ppb).

Field methods—invertebrate studyThe effect of 1080 leaching from baits was assessedon invertebrate communities in each stream in May2004. Invertebrates were collected using the "rockrolling" technique (see Death & Winterbourn 1994;Matthei et al. 2000). Here, a net (mesh size 300^m)was placed immediately below individual rocks,which were lifted quickly into the net. Animals oneach rock were either immediately dislodged andcollected in the net, or removed from the rock byscrubbing. This technique was used as the boulderand cobble-dominated streambed precluded useof traditional quantitative techniques (e.g., Starket al. 2001). Ten replicate rocks were randomlysampled in riffles at each site, and rock dimensions(length x width x height) measured after samplingto calculate their surface areas (Biggs & Kilroy2000) to estimate invertebrate densities. A poweranalysis of invertebrate data collected previouslyfrom individual rocks in Dead Horse creek showedthat 10 replicate rocks were sufficient to detect a20% reduction in total invertebrate density, anddensities of common taxa, with an 80% certainty(A. Suren unpubl. data). As invertebrate densitiesin small streams can vary monthly by up to 30%

of the long-term (18-month) average (Suren 1991,unpubl. data), the ability of the present analysisto detect a smaller density reduction (20%) as aresult of 1080 over an 8-day period was consideredsufficiently robust.

Invertebrates were first collected from all sites on20 May 2004, and 4 days later (24 May 2004), afterwhich 1080 baits were added to the streams in thesame manner as for the fish experiment. Dischargein the streams was less than for the fish experiment,so fewer baits were used (Table 1). It was, however,decided not to reduce the number of baits addedto each stream in direct proportion to the lowerstream discharges, as this may have been perceivedas being too conservative. Instead, enough baitswere added so that predicted 1080 concentrationswere almost three times higher than those used inthe fish trials (Table 1). The invertebrate experimenttherefore represented an extreme scenario followingaerial application of 1080. Invertebrate samples werecollected 1 day (25 May 2004) and 4 days (28 May2004) after addition of 1080. Water samples werealso collected from both impact sites in each stream1 and 4 days after addition of baits, and processedin a similar manner as samples collected in the fishexperiment.

All invertebrate samples were preserved inthe field using 50% iso-propanol, and returned tothe laboratory. Samples were processed using amodification of Protocol P3 (Stark et al. 2001),whereby all samples were placed in a smallBogorov tray (see Winterbourn & Gregson1989) and scanned under a stereo-microscope foridentification and enumeration of all invertebrates.This method allowed small animals such as smallchironomids, nematodes, copepods, and ostracods

Table 1 Estimated 1080 (sodium fluoroacetate) concentrations in each of the experimental streams, based on streamdischarge when baits were added, the number of baits added, and assuming complete dissolution in 8h (based onSuren 2006).

Stream

Fish studyCrooks creekDead Horse tributaryWallaby CreekWheelbarrow creekInvertebrate studyCrooks creekDead Horse tributarySmall creekWheelbarrow creek

Discharge when baitsadded (litres s 1)

568651105

1614617

No. of baits

30503080

20201525

1080 weightadded (mg)

288480288768

192192144240

Estimated concentration(jdg litre-1)

0.180.190.200.25

0.420.480.830.49

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

536 New Zealand Journal of Marine and Freshwater Research, 2006, Vol. 40

to be counted accurately. All mollusca and aquaticinsects were identified to genera using standardkeys (e.g., Winterbourn 1973, 1989), except for theChironomidae, which were identified to subfamily.Other taxa were identified to either subclass (e.g.,Crustacea) or family (Oligochaeta).

Statistical analysisTwo-way ANOVA (SPSS 2000) was used todetermine differences in discharge, measured waterquality parameters and fish length between streams,or between control and impact sites. Three-wayANOVA was used to test for differences in stone sizesbetween streams, control and impact sites, or beforeand after 1080 baits were added. Untransformeddata was used for all these analyses except fordischarge data, which was Iog10-transformed beforeanalysis.

A number of metrics were calculated fromthe invertebrate data, including total invertebratedensity, taxonomic richness, densities of the fourmost common and widespread taxa, the number ofEphemeroptera, Plecoptera and Trichoptera (EPT)taxa, and the % of EPT taxa. The MacroinvertebrateCommunity Index (MCI) score, and its quantitativevariant (QMCI: Stark 1985, 1993) were alsocalculated. Although the latter two metrics wereoriginally derived to assess organic pollution in stonystreams (Stark 1985,1993), they are appropriate fordetecting changes to invertebrate communities as aresult of 1080, as changes to community structurewould manifest themselves as a change to calculatedMCI or QMCI scores.

Invertebrate data were analysed by ANOVA usingthe GENSTAT package (GENSTAT 2001). All datawere examined for normality and Iog10-transformedwhere necessary. The study was based on a Before-After-Control-Impact (BACI) design, used to detectchanges in biological communities as a result ofhuman activities (Underwood 1991), where samplescollected above 1080 baits were considered controlsand samples collected below baits were impact sites.Although traditional BACI tests have two mainexperimental effects (usually a time and site effect),our experiment was not straightforward. First, thetwo sites below the 1080 baits represented a dilutiongradient, where the lowermost site could have beenexposed to lower 1080 concentrations than the uppersite as a result of groundwater inputs and possiblebreakdown of 1080. Secondly, invertebrate densitiesmay have recovered in samples collected 4 days afterbait addition as a result of upstream immigration,whereas such immigration would have been much

less 1 day after the introduction of 1080. Becauseof these complications, extra terms were includedin the ANOVA model: (1) to separate potentialdilution effects, the '"site" term was partitioned intoa "Control" versus "Impact" main effect, a "100m"versus "10m" main effect, and a spatial interactionbetween "Control versus Impact" x "100m versus10m"; (2) to separate any recovery in invertebratecommunities as a result of immigration, the "time"term was partitioned into a "Before" versus "After"main effect as well as a "4 day" versus "1 day" maineffect, and a temporal interaction between "Beforeversus After" x "4 day versus 1 day".

The ANOVA model had three main effects (site,time, stream), as well as all interaction effects. Wewere particularly interested in interaction effectsthat suggested that sites below the 1080 baits weredifferent to those above. The relevant interactionterms that indicated this specific difference were:(1) "Control versus Impact" x "Before versus After"effect; (2) the spatial interaction term x "Beforeversus After"; and (3) the temporal interaction termx "Control versus Impact". These interaction termswere examined for the combined data set, and forinteractions with the "stream" main effect.

The effect of 1080 on invertebrate communitycomposition in the streams was also assessed usinga multivariate approach. First, a matrix of Bray-Curtis similarity coefficients was constructed usingabundance data from all possible pairs of invertebratesamples (Digby & Kempton 1987). Bray-Curtiscoefficients range from 0 (no common taxa betweenthe two sites) to 1 (all species common to both sites,with the same relative abundances). Differencesin Bray-Curtis scores of samples collected at thedifferent times and sites were analysed usingPERMANOVA (Anderson 2001) to assess whetherplacement of 1080 baits changed the downstreaminvertebrate communities relative to those upstream.The PERMANOVA procedure was set to run 5000different permutations to test whether observeddifferences could have arisen owing to chance, orwhether they were indicative of an actual effect.Significant levels for all tests were set at P < 0.05.

RESULTS

Effects on native fishDischarge differed significantly between streams(F(3 8) = 4.67, P < 0.05), and was highest in DeadHorse tributary and lowest in Wallaby Creek (Table2). Discharge decreased slightly during the study,

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

Suren & Lambert—Effect of 1080 baits on fish and invertebrates 537

but this was not significant (f(18) = 0.70, P >0.05). Stream pH ranged from 4.1 to 6.4, and wassignificantly lower in Wheelbarrow creek and DeadHorse tributary (F(3 8) = 54.94, P < 0.001), and higherin Wallaby Creek (Table 2). Stream conductivity waslow, but significantly higher in Wallaby Creek andCrooks creek (Table 2) than the other streams (Fß 8)

= 38.97, P < 0.001). Dissolved O2 was relatively low(c. 75% saturated) and similar in all streams (f^s) =0.35, P > 0.05), whereas temperature was highest inWallaby Creek (F(3 8) = 35.85, P < 0.001, Table 2).

1080 was detected in water samples collectedduring the fish experiment from impact sites onlyfor a short duration (up to 8h) after addition of baits(Fig. 2). 1080 concentrations were higher at sites10m below the baits (mean + 1SD: 0.176 + 0.054]4g litre"1) than sites 100m below this point (0.063 +0.076 ]4g litre"1). No 1080 was found in any samplescollected from impact sites after 12h, or from anycontrol sites.

There was no significant difference in bully length(54 + 9 mm) in any of the streams (F(3 93) = 0.714,P > 0.05), or in cages placed in control or impactsites (F(1 93) = 0.6545, P > 0.05). Mean lengths oflongfin eels differed significantly between streams

Fig. 2 Concentration of 1080 (sodium fluoroacetate) inwater samples (jig litre"1) collected from the four streamsduring the fish experiment. Samples were collected fromsites 10m below (closed symbols) and 100m below (opensymbols) 1080 baits on three occasions before baits wereadded, and six occasions after. Note that some symbolsfrom the 10m site obscure those from the 100m site.

024

0.20

0.16

0.12

0.08

0.04

0.00

^~- 0.24 -

g 0.20

Ç> 016

S 0 1 2

S 0.08

O 0.04CO

Í= 0.00

§ °-24 1

2 0.20

g 0.16

8 0.12

004

0.00

Crooks creek

Dead Horse creek

024

0.2B

0.16

0.12

0.08

0.04

0.00

Wallaby Creek

Wheelbarrow creek

-96 -24 O 2 4 8 12 24 96

Time (h)

Table 2 Summary of the physical and water-quality conditions of the five study streams. Stream discharge wasmeasured at the site where 1080 (sodium fluoroacetate) baits were added, at the beginning, immediately before 1080baits were added, and at the end of each experiment. Other parameters were measured on all sampling occasions, andare averages (± 1 SD, n = 4). All stream names except Wallaby Creek are unofficial.

Stream

Fish studyCrooks creekDead Horse tributaryWallaby CreekWheelbarrow creekInvertebrate studyCrooks creekDead Horse tributarySmall creekWheelbarrow creek

Range of discharge(litres s"1)

56-12586-30551-12778-238

15-1712-156-7

16-18

pH

5.6 ±0.14.7 ± 0.36.2 ±0.24.5 ±0.3

5.9 ±0.35.8 ±0.26.4 ±0.35.6 ±0.2

Conductivity(jiS cm-1)

31 ±225 ±138 ±326 ±2

34 ±632 ±153 ±234 ±1

Dissolved O2

(mg litre"1)

8.1 ±0.47.9 ±0.38.1 ±0.67.9 ±0.2

11.2 + 0.111.9 + 0.110.5 ±0.310.8 ±0.2

Temperature(°C)

12.5 ±0.312.5 ±0.614.7 ± 0.912.5 ±0.5

7.6 ±0.36.5 ±0.78.7 ±0.37.4 ±0.7

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

538 New Zealand Journal of Marine and Freshwater Research, 2006, Vol. 40

CF(3,102) = 15.76, P < 0.001), and were longest atCrooks creek (339 + 17mm) and the same length atthe other streams (256 + 77 mm). Longfin eel lengthsat impact sites below the 1080 baits (256 + 85 mm)were also less than those at control sites (294 + 79mm;F(1 102) = 16.37, P < 0.001). There was no significantdifference in koaro lengths (86 + 21mm) in any ofthe streams (F(3 93) = 1.258, P > 0.05), or in controlor impact sites (F(1 93) = 0.005, P > 0.05).

No fish mortality was observed in any of the cagesexcept at Wheelbarrow creek, where 15 out of 32upland bullies and one koaro had died between thefirst and second sampling occasions (Table 3). Thismortality appeared to be owing to heavy rainfallbetween these time periods, which overturned someof the cages and partially filled them with fine sand.Although some fish (notably eels) escaped fromcages (Table 3), sufficient fish remained in cagesat all impact sites to monitor mortality arising as aresult of 1080 exposure. No fish died in any of thestreams after 1080 baits were added (Table 3).

Effects on invertebrate communitiesDischarge was much lower for the invertebrateexperiment (Table 2), and was significantly lowerin Small creek (F(3S) = 22.16, P < 0.001) than inthe other three streams. Stream pH was significantlylower in Wheelbarrow creek and Dead Horse tributary(F(3 8) = 6.87, P < 0.01), whereas conductivity wassignificantly higher in Small creek (F(3 8) = 62.59,P< 0.001).

Dissolved O2 was higher in the four streams thanfor the fish experiment (Table 2) and was significantlyhigher in Dead Horse tributary (F(3S) = 28.84, P <0.001), whereas water temperature was cooler, andsignificantly lower in Dead Horse tributary (FßS^ =16.65, P < 0.01). Rock sizes (162 + 37mm x 119 +26 mm x 70 + 21 mm) were similar between streams(Fß 624) = 2.58, P > 0.05), and between control andimpact sites (F(1 624) = 1-H. P > 0.05). There wasalso no significant difference in rock sizes collectedbefore or after placement of 1080 baits (F(1624) =0.67, P > 0.05).

Table 3 Fish loss observed during the experiment, based on eight fish of each species (uplandbullies (Gobiomorphus breviceps) (b), eel (Anguilla dieffenbachü) (e), and koaro (Galaxiasbrevipinnis) (k) originally added to individual cages. Fish survival was monitored 1 and 4 daysbefore 1080 (sodium fluoroacetate) baits were added (-4 and -1 d), and 1 and 4 days after (+1and +4 d), at locations above and below bait placement. Cages placed below 1080 baits areshown in italics. Reductions in numbers owing to mortality indicated in bold, other reductionsin numbers were owing to fish escaping from cages. Cages at site 1 in Crooks creek were stolen,so no data were collected (nd).

Stream

Crooks creek

Dead Horse tributary

Wheelbarrow creek

Wallaby Creek

Time(days)

- 4- 1+ 1+ 4- 4- 1+ 1+ 4- 4- 1+ 1+ 4- 4- 1+ 1+ 4

Above 1080 baitsSite l-100m

(b, e, k)

ndndndnd

0,0,00,0,00,0,00,0,00,0,08,2,08,2,08,2,00,0,00,0,00,0,00,0,0

Site2-10m(b, e, k)

0,0,00,0,00,0,00,0,00,0,00,0,01,0,01,0,00,0,03,0,13,0,13,0,10,0,00,0,10,0, 10,0,1

Below 1080 baitsSite 3-10 m

(b, e, k)

0,0,00,0,00,0,00,0,00,0,00,0, 10,0, 11,0, 10,0,02,0,02,0,02,0,00,0,00,0,00,0,00,0,0

Site 4-100 m(b, e, k)

0,0,01,0, 11,0, 11,0, 10,0,00,4,00,4,00,4,00,0,02,0,02,0,02,0,00,0,00,3,00,3,00,3,0

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

Suren & Lambert—Effect of 1080 baits on fish and invertebrates 539

A total of 72 taxa and 87889 individuals werecollected from the rocks during the study, mostcommon of which was the caddisfly Helicopsyche,contributing up to 35% to total density, followedby orthoclad midges (18%), the leptophlebidmayfly Deleatidium (10%), and the caddisfliesPycnocentrodes and Pycnocentria (7% each). Thestonefly Zelandoperla, and the cranefly Aphrophiliawere also relatively common, contributing 3% tototal density. Only 12 taxa were considered common(densities > 1 %), 22 taxa were considered occasional(densities between 0.1 and 1%), and 38 taxa wereconsidered uncommon (densities <0.1%). The mostwidespread taxa were Helicopsyche, Deleatidium,orthoclad midges, and Zelandoperla, which werecollected at all sites. Small creek had the highesttaxonomic richness (65 taxa) and invertebrate density(mean=2755 individuals m~2), whereas Wheelbarrow

creek had the lowest taxonomic richness (50 taxa)and lowest densities (758 individuals m~2).

No 1080 was detected in water samples collectedfrom impact sites 1 day and 4 days after placingbetween 15 and 25 baits in the streams, despite thevery low stream discharge (ranges from 6-17 litress"1, Table 1) which would have minimised dilutioneffects. Predicted 1080 concentrations in the streamsbased on an 8 h leaching rate were up to three timeshigher than those for the fish study (Table 1).

ANO VA showed no large or consistent effects oneither community metrics, or densities of commontaxa. Few statistically significant interaction termswere observed for the six metrics describing theinvertebrate communities that suggested an effect of1080 (Table 4). The invertebrate fauna at all streamshad high MCI scores (121 + 12) and % EPT (65 +25), although both metrics varied widely in the study

Table 4 Results of ANOVA model for effects of 1080 (sodium fluoroacetate) on selected invertebrate metrics,showing the model interaction effects that were indicative of a potential effect. Values in bold are significant (P< 0.05).(MCI, Macroinvertebrate Community Index; QMCI, Quantitative MCI; EPT, Ephemeroptera, Plecoptera, Trichoptera.)

Metric Model effectSums ofsquares

0.0820.6110.0891.664166.1810.0170.0040.0610.35334.3700.2040.0300.0030.12011.0600.0910.0030.0231.67968.661.6940.7430.0181.712325.96615.30155.4064.301397.8091182.5

d.f.

1113

5761113

5761113

5761113

5761113

5761113

576

Meansquares

0.0820.6110.0890.5550.2890.0170.0040.0610.1180.0600.2040.0300.0030.0400.0190.0910.0030.0230.5600.1191.6940.7430.0180.5710.566

615.300155.40064.300

465.933158.303

F ratio

0.282.120.311.92

0.280.071.021.97

10.621.560.162.08

0.760.020.194.69

2.991.310.031.01

3.890.980.412.94

P

0.5930.1460.5800.125

0.5870.7940.3130.117

0.0010.1960.6860.039

0.3830.8840.6440.003

0.0840.2520.8570.388

0.0490.3220.5240.033

Density Control versus Impact x Before versus AfterSpatial interaction x Before versus AfterTemporal interaction x Control versus ImpactStream x Control versus Impact x Before versus AfterError

Richness Control versus Impact x Before versus AfterSpatial interaction x Before versus AfterTemporal interaction x Control versus ImpactStream x Control versus Impact x Before versus AfterError

MCI Control versus Impact x Before versus AfterSpatial interaction x Before versus AfterTemporal interaction x Control versus ImpactStream x Control versus Impact x Before versus AfterError

QMCI Control versus Impact x Before versus AfterSpatial interaction x Before versus AfterTemporal interaction x Control versus ImpactStream x Control versus Impact x Before versus AfterError

EPT Control versus Impact x Before versus AfterSpatial interaction x Before versus AfterTemporal interaction x Control versus ImpactStream X Control versus Impact x Before versus AfterError

% EPT Control versus Impact x Before versus AfterSpatial interaction x Before versus AfterTemporal interaction x Control versus ImpactStream x Control versus Impact x Before versus AfterError

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

540 New Zealand Journal of Marine and Freshwater Research, 2006, Vol. 40

Time

Fig. 3 MCI and % EPT (x ± 1SE, n = 160) data forthe experimental streams at control (open bars) and im-pact (closed bars) sites, before and after 1080 (sodiumfluoroacetate) bait placement. (MCI, MacroinvertebrateCommunity Index; EPT, Ephemeroptera, Plecoptera,Trichoptera.)

(MCI scores from 88 to 186; % EPT from 10 to 100).These metrics both showed statistically significant"Control versus Impact" x "Before versus After"effects (Table 4). Average MCI scores increasedslightly (by 2.8) at control sites before and afteraddition of baits, whereas this metric decreasedslightly (by 2) at the impact sites (Fig. 3). Calculated% EPT scores decreased by 11 % at impact sites, anddecreased by 8% at control sites (Fig. 3).

Patterns in MCI and QMCI scores, and for the %EPT differed between streams (Table 4). MCI scoresdecreased in Dead Horse tributary at impact siteswhen compared with control sites (Fig. 4), howeverthe reduction (7) was negligible when compared tothe natural range of MCI scores at control sites in thisstream (88-150). MCI scores in Small creek increasedat control sites, but decreased slightly at impact sites(Fig. 4). MCI scores at Crooks creek remained similarat all sites throughout the study (Fig. 4), whereasscores at Wheelbarrow creek decreased by a similarvalue in both control and impact sites. QMCI scoresdecreased more at impact sites than control sites at

90

10 -,

Crookscreek

Dead Horsetributary

Smallcreek

Wheelbarrowcreek

Study stream

Fig. 4 MCI and QMCI scores, and % EPT (x ± 1SE, n= 40) data for individual streams at control (open bars)and impact (closed bars) sites, before and after 1080(sodium fluoroacetate) bait placement. (MCI, Macroin-vertebrate Community Index; QMCI, Quantitative MCI;EPT, Ephemeroptera, Plecoptera, Trichoptera.)

Crooks creek and Dead Horse tributary, whereas atthe other two sites, QMCI scores decreased moreat control sites than at impact sites (Fig. 4). The %EPT at impact sites in Crooks creek and Dead Horsetributary declined more at impact sites (by 10% and29%, respectively) than at control sites, where itdeclined by 6% and 15%, respectively. The % EPTdeclined by a similar value at Wheelbarrow creek(c. 2%) and Small creek (c. 6%) at both control andimpact sites (Fig. 4).

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

Suren & Lambert—Effect of 1080 baits on fish and invertebrates 541

% 100

8

O 50

T

T

T

Time

Fig. 5 Densities of orthoclad midges (x ± 1 SE, n = 160)at Control versus Impact sites, 10 m (closed symbols)and 100m downstream (open symbols) from the 1080(sodium fluoroacetate) baits, before and after 1080 baitplacement.

Treatment

Fig. 6 Densities of Deleatidium nymphs (x ± 1SE, n =160) 1 day (closed bars) and 4 days (open bars) beforeand after placement of 1080 (sodium fluoroacetate) baitsin streams, at Control versus Impact sites.

Table 5 Results of ANOVA model for effects of 1080 (sodium fluoroacetate) on densities of the four most commonand widespread taxa found in the study streams, showing the model interaction effects that were indicative of a potentialeffect. Values in bold are significant (P < 0.05).

Taxa Model effectSums ofsquares d.f.

Meansquares F ratio

Deleatidium Control versus Impact X Before versus AfterSpatial interaction X Before versus AfterTemporal interaction X Control versus ImpactStream X Control versus Impact X Before versus AfterError

Helicopsyche Control versus Impact X Before versus AfterSpatial interaction X Before versus AfterTemporal interaction X Control versus ImpactStream X Control versus Impact X Before versus AfterError

Orthocladiinae Control versus Impact X Before versus AfterSpatial interaction X Before versus AfterTemporal interaction X Control versus ImpactStream X Control versus Impact X Before versus AfterError

Zelandoperla Control versus Impact X Before versus AfterSpatial interaction X Before versus AfterTemporal interaction X Control versus ImpactStream X Control versus Impact X Before versus AfterError

0.0710.6021.8237.667247.7090.4862.5970.2664.513532.7130.0363.7180.5946.773578.7291.9010.0670.3809.380490.246

1113

5761113

5761113

5761113

576

0.0710.6021.8232.5560.4300.4862.5970.2661.5040.9250.0363.7180.5942.2581.0051.9010.0670.3803.1270.851

0.161.404.245.94

0.5262.8080.2881.627

0.0363.7000.5912.247

2.2340.0780.4473.674

0.6850.2370.040

<0.001

0.4690.0940.5920.182

0.8510.0500.8390.082

0.1360.3770.5040.012

Densities of the four most common and widespreadtaxa showed few statistically significant interactionterms that suggested an adverse effect of 1080(Table 5). Helicopsyche densities were unaffectedby the 1080 baits, whereas Orthocladiinae midgedensities showed a significant effect only for theSpatial interaction x Before versus After, with an

increase at all sites following addition of 1080(Fig. 5). Densities increased by 65% and 48% atthe impact sites 100 m and 10 m downstream ofthe baits, respectively, which was greater than theobserved increase in midge density at the controlsites 10 m upstream of the baits (Fig. 5). Deleatidiumdensities showed a significant Temporal interaction

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

542 New Zealand Journal of Marine and Freshwater Research, 2006, Vol. 40

160 -, Fig. 7 Densities of Deleatidiumand Zelandoperla (x ± 1 SE, n = 40)collected from individual streamsfrom control (open bars) and im-pact (closed bars) sites, before andafter 1080 (sodium fluoroacetate)bait placement.

Crookscreek

Dead Horsetributary

Smallcreek

Wheelbarrowcreek

Study stream

x Control versus Impact effect (Table 5). Densitiesincreased at all sites before and after addition of1080 baits, except for the impact sites 4 days afterthe baits were added, when densities decreased by18% (Fig. 6).

Patterns in Deleatidium and Zelandoperladensities differed between streams (Table 5). Delea-tidium densities decreased slightly in Crooks andSmall creeks at impact sites when compared withcontrol sites, whereas in Dead Horse tributary,Deleatidium densities remained the same at impactsites before and after addition of 1080 baits (Fig. 7).Densities increased at both control and impact sites atWheelbarrow creek. Zelandoperla densities changedlittle at both control and impact sites during the studyat Crooks and Wheelbarrow creeks, whereas densities

increased slightly at impact sites at both Small creekand Dead Horse tributary after the addition of 1080baits (Fig. 7). Densities of this taxa increased morein the control sites during the study.

DISCUSSION

Effects on native fishWe placed 80 Wanganui No. 7 baits in Wheelbarrowcreek, and monitored fish mortality at a site c. 10mbelow this. Use of 80 baits represented an "extreme"instance of contamination by 1080 baits over thisshort distance. Similarly high numbers of baits wereplaced in the other three streams, proportional totheir discharge. If all the 1080 had leached from

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

Suren & Lambert—Effect of 1080 baits on fish and invertebrates 543

the baits within 8h, estimated 1080 concentrationswould have ranged from 0.18 to 0.25 pig litre"1 (Table1), similar to what was observed. The highest 1080concentration at Wheelbarrow and Wallaby creeks(0.26 pig litre"1) was almost 10 x lower than theMinistry of Health's upper value of 2 pig litre"1,despite the very large quantities of baits placed ina small area in each stream. These results highlightthe rapid rate at which 1080 leaches from submergedbaits (see Suren 2006), and also support assertionsby Eason et al. (1999) that 1080 leaching from baitsis diluted to toxicologically insignificant amounts,even when sampled a short distance away from largenumbers of baits submerged in small streams. Wealso found that 1080 concentrations were almostthree times higher at sites 10m below the baits thansites 100 m downstream. Although this downstreamdilution effect may have reflected incomplete mixingof 1080 by 10 m, we consider it unlikely, as 1080baits were placed at three places across each streamin relatively narrow sections, and stream turbulencewas high as it flowed over the course substrate. Morelikely reasons for the lower 1080 concentrations atthe 100 m site include groundwater inputs to thestream, and losses of 1080 from the stream from eitherbacterial breakdown or losses to groundwater.

Despite placing a large number of baits in eachstream and detecting 1080 in water, no fish diedafter application of 1080 baits. The only mortalityobserved was that of 15 upland bullies and of anindividual koaro as a result of a flood in Wheelbarrowcreek that occurred before placement of 1080 baits.Lack of mortality of the three native fish speciesagrees with previous work showing that fish aretolerant to high 1080 concentrations (>370 pig litre"1 ;King & Penfound 1946; Batcheler 1978; Fagerstoneet al. 1994). The highest of these experimental testconcentrations was c. 3700 times higher than thehighest concentration achieved in our study. Toreach similar concentrations in small streams witha low discharge of 105 litres s"1 (as observed inWheelbarrow creek when 1080 baits were added)and assuming complete dissolution of 1080 afteronly 4h would require a total of 152160 baits, or973kg of the 6.5g Wanganui No. 7 baits. Such ascenario could only occur if a full hopper bucketcarrying 1080 baits fell from a helicopter into astream, all baits became submerged, and were leftfor 4h. The chance of this occurring is minuscule.Moreover, standard operating conditions imposed on1080 operations require any accidental spillages tobe cleaned up as soon as possible, further reducingthe likelihood of such a scenario.

Effects on invertebratesStream discharges during the invertebrate studywere lower than for the fish study, so fewer baitswere used. Despite the use of fewer baits, 1080concentrations were predicted to be higher thanfor the fish experiment, reflecting the reducedstream discharge and use of proportionally morebaits. Notwithstanding these predictions, thewater sampling protocol did not detect any 1080.Lack of detection reflected the rapid leaching rateof 1080, as the first samples were collected 24 hafter addition of baits—well after all the 1080 hadleached from the baits (Suren 2006). Predicted 1080concentrations (assuming an 8h leaching rate) wouldhave ranged from 0.42 to 0.83 pig litre"1; within thetop 0.8% of 1080 concentrations recorded in water(Eason 2002). Given the close agreement betweenobserved and predicted 1080 concentrations for thefish experiment, we believe these concentrationswere attained in the invertebrate study. As such, theinvertebrate study represented a worse-case scenarioof 1080 contamination encountered within NewZealand streams.

Because 1080 is a toxin, densities of some inver-tebrate taxa were expected to decline followingaddition of baits as a result of mortality. Thismortality would have reduced at least some of thecalculated metrics at the impact sites, especiallythose closest to baits. Such reductions in metricswere not observed, and 1080 had little consistentdemonstrable effect on metrics such as the MCIor %EPT. Although statistical differences wereobserved for some of the metrics, these differenceswere ecologically insignificant, and well withinranges of the metrics observed at control sites. Forexample, the small reduction in MCI scores at impactsites after addition of 1080 baits was insignificant(2) when considering the natural ranges of MCIscores observed at control sites (ranges from 88 to186). The magnitude of the difference in the % EPTbetween Control versus Impact sites (3%) was alsoecologically insignificant when compared to therange of %EPT at control sites (64-72%).

Examination of densities of individual taxaconfirmed a lack of adverse effects of 1080 at theconcentrations found in this experiment. In Crooksand Small creeks, Deleatidium densities decreasedby c. 27% at impact sites following addition of 1080baits. However, long-term studies of invertebratecommunities in pristine headwater streams showthat mean monthly densities of Deleatidium canvary up to 30% of the long-term (18-month) mean(Suren 1991, unpubl. data). We thus consider the

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

544 New Zealand Journal of Marine and Freshwater Research, 2006, Vol. 40

observed reduction in Deleatidium density at thesetwo streams below 1080 baits more likely reflectednatural variation, and not the presence of 1080 baits.Moreover, Deleatidium densities increased at sitesbelow 1080 baits in Wheelbarrow creek, a resultinconsistent with the effects of a toxin. Finally, themultivariate analysis detected no consistent changesto invertebrate community structure followingapplication of 1080 baits in the study streams.

Management implicationsGiven public concern about 1080 in the environ-ment—especially when it falls into streams—theresults and implications of this study are expectedto be of interest to both advocates and opponentsof the aerial application of 1080 baits. Consentconditions controlling aerial applications of 1080usually implement buffer strips around larger rivers,presumably for protection of aquatic ecosystemsand/or ensuring potential drinking water suppliesremain uncontaminated. There is currently nonational standard that we are aware of for imposingbuffers around waterways during 1080 operations.Some regional councils require buffers around allwaterways, others have buffers around all waterways>3 m wide, whereas other councils have no buffersaround waterways that are not used for drinkingwater (Suren 2006). If one of the intents of bufferzones around streams is to minimise potential adverseeffects of 1080 on the environment, then placingbuffers on only streams >3 m wide will not preventsmaller headwater streams from being exposed toaccidental contamination of 1080 baits.

The dilution capacity of a stream would beproportional to its discharge, so it could be arguedthat buffers should also be placed around smallerstreams as they would have a lower capacity to dilutepotential toxins. However, requiring buffers aroundsmall streams within an operational area would reducethe cost-effectiveness of aerial 1080 operations,as large areas of catchments would become no-drop zones, increasing the operational difficulty.Additionally, buffer zones around waterways maycreate refuge zones for the target species. Creationof such refuge zones may diminish the overalleffectiveness of 1080 operations in specific areas.

Our experiments were performed in small streams<3m wide, which could have been subject topotential contamination by 1080 baits during aerialoperations. Despite placing large quantities of baitinto these streams, we detected no adverse effectson three native fish species or natural invertebratecommunities exposed to 1080 leaching from these

baits. If no adverse effects of 1080 were detected insmall streams with a low dilution capacity, then therewould be far less potential for adverse effects on biotain larger streams with a greater dilution capacity.Placing buffer zones around larger streams for thepurpose of ecosystem protection therefore seemsunjustified. Consent granting authorities shouldconsequently consider removing requirements ofbuffer zones around large rivers. An exception tothis recommendation may be when rivers are usedfor human drinking water or for stock supply, assome members of the public may still need assurancethat all steps are being taken to prevent 1080 baitslanding in these streams.

Finally, although our experiments were conductedover only a short-term period and failed to detectlonger-term or sub-lethal impacts, our detailed water-monitoring programme during the fish experimentshowed unequivocally that 1080 was rapidly lost fromsubmerged baits within 12h. This rapid loss of 1080from submerged baits was reinforced by absenceof detectable 1080 in water samples collected 24hafter addition of baits in the invertebrate experiment.Moreover, 1080 is broken down relatively quickly bybacteria present in water or associated with aquaticplants (Ogilvie et al. 1996). We consequently contendthat no long-term effects of 1080 would be possibleto fish or invertebrate communities as a result ofminute quantities of 1080 leaching from submergedbaits as the chemical would not be present in theenvironment for long enough.

ACKNOWLEDGMENTS

Timberlands West Coast and Ngai Tahu are thanked fortheir permission to work in the Mawhera State Forest.The help of Dale Rowlings and Barry Pétrie (West CoastRegional Council) with placing baits in the streams isacknowledged. Doug Griffin (NIWA Greymouth) andJulian Sykes (NIWA Christchurch) helped with the fishexperiment, and Fred Heine is thanked for his help duringthe invertebrate experiment. Peter Johnstone (AgResearch)is thanked for analysing the data using the modified BACIdesign. Thanks to Ton Snelder and Andrew Davis (NIWAChristchurch), the Technical Advisory Group at theAnimal Health Board, and two anonymous reviewers forcomments. Funding for the study came from the AnimalHealth Board (Contract No. R-80575), and from NIWA.

REFERENCES

Anderson MJ 2001. A new method for non-parametricmultivariate analysis of variance. Austral Ecology26: 32-46.

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

Suren & Lambert—Effect of 1080 baits on fish and invertebrates 545

Batcheler CL 1978. Compound 1080, its properties,effectiveness, dangers, and use. Report toMinister of Forests and Minister of Agricultureand Fisheries. Wellington, New Zealand ForestService.

Biggs BJF, Kilroy C 2000. Stream Periphyton MonitoringManual. Wellington, New Zealand, Ministry forthe Environment. 228 p.

Death RG, Winterbourn MJ 1994. Environmental stabilityand community persistence: a multivariateperspective. Journal of the North AmericanBenthological Society 13: 125-139.

Deonier CC, Jones HA, Incho HH 1946. Organiccompounds effective against larvae of Anophelesquadrimaculatus—laboratory tests. Journal ofEconomic Entomology 39: 459-462.

Digby PGN, Kempton RA 1987. Multivariate analysisof ecological communities. London, Chapman& Hall.

Eason CT 2002. Technical review of sodiummonofluoroacetate (1080) toxicology. Wellington,New Zealand, Animal Health Board. 25 p.

Eason CT, Wickstrom M, Turck P, Wright GRG 1999. Areview of recent regulatory and environmentaltoxicology studies on 1080: results andimplications. New Zealand Journal of Ecology23: 129-137.

Eason CT, Gooneratne R, Wright GRG, Pierce RS,Frampton CM 1993. The fate of sodiummonofluoroacetate (1080) in water, mammals,and invertebrates. Proceedings of the 46th NewZealand Plant Protection Society conference. Pp.297-301.

Eisler, R1995. Sodium monofluoroacetate (1080) hazardsto fish, wildlife, and invertebrates: a synopticreview. Patuxent Wildlife Research Center.Contaminant Hazard Reviews Report No. 30.51 p.

Fagerstone KA, Savarie PJ, Elias DJ, Schafer EWJ 1994.Recent regulatory requirements for pesticideregistration and the status of Compound 1080studies conducted to meet EPA requirements. In:Seawright AA, Eason CT ed. Proceedings of theScience Workshop on 1080, Christchurch. TheRoyal Society of New Zealand MiscellaneousSeries 28: 33-38.

GENSTAT 2001. Genstat 6 Committee, Reference Manual.Oxford, Claredon Press.

Green W. 2003. The use of 1080 to control possums andother pests. A resource document. Wellington, NewZealand, Animal Health Board and Department ofConservation. 49 p.

King JE, Penfound WT 1946. Effects of new herbicideson fish. Science 103: 487.

Laugesen R, Hubbard A 2002. The Big Drop. In: SundayStar Times, 19 May 2002. Auckland, NewZealand. Pp. C1-C2.

Livingstone PG 1994. The use of 1080 in New Zealand.In: Seawright AA, Eason CT ed. Proceedings ofthe science workshop on 1080, Christchurch. TheRoyal Society of New Zealand MiscellaneousSeries 28: 1-9.

Lyver PO'B, Ataria J, Trought K, Fisher P 2005. Sodiumfluoroacetate (1080) residues in longfin eels,Anguilla dieffenbachia, following exposure tocontaminated water and food. New ZealandJournal of Marine and Freshwater Research 39:1243-1252

Matthaei CD, Arbuckle CJ, Townsend CR 2000. Stablesurface stones as refugia for invertebrates duringdisturbance in a New Zealand stream. Journal ofthe North American Benthological Association19: 82-93

Ogilvie SC, Hetzel F, Eason CT 1996. Effect oftemperature on the biodegradation of sodiummonofluoroacetate (1080) in water and inElodea canadensis. Bulletin of EnvironmentalContamination and Toxicology 56: 942-947.

Speedy C 2003. To kill or not to kill. Fish and Game NewZealand 40: 60-63.

SPSS inc. 2000. SYSTAT 10 language reference. Chicago,Illinois, United States. 307 p.

Spurr EB, Drew KW 1999. Invertebrates feeding on baitsused for vertebrate pest control in New Zealand.New Zealand Journal of Ecology 23: 167-173.

Stark JD 1985. A Macroinvertebrate Community Indexof water quality for stony streams. Water & SoilMiscellaneous Publications 87: 1-53.

Stark JD 1993. Performance of the MacroinvertebrateCommunity Index: effects of sampling method,sample replication, water depth, current velocity,and substratum on index values. New ZealandJournal of Marine and Freshwater Research 27:463-478.

Stark JD, Boothroyd IKG, Harding JS, Maxted JR,Scarsbrook MR 2001. Protocols for samplingmacroinvertebrates in wadeable streams. NewZealand Macroinvertebrate Working Group ReportNo. 1.57 p.

Suren AM 1991. Bryophytes as invertebrate habitat in twoNew Zealand alpine streams. Freshwater Biology26: 399-418.

Suren AM 2006. Quantifying contamination of streams by1080 baits, and their fate in water. New ZealandJournal of Marine and Freshwater Research 40:159-167.

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4

546 New Zealand Journal of Marine and Freshwater Research, 2006, Vol. 40

Taranaki Regional Council 1993. Water quality monitoringof Stage 1 of the Department of Conservation/Taranaki Regional Council possum controloperation on Mount Taranaki/Egmont 1993.Taranaki Regional Council Technical Report93-27. 18 p.

Taranaki Regional Council 1994. Water quality monitoringof Stages 2 & 3 of the Department of Conservation/Taranaki Regional Council possum controloperation on Mount Taranaki/Egmont 1993.Taranaki Regional Council Technical Report94-7. 18 p.

TDGAUGE (2005) Tideda Gauge Programme. NIWAChristchurch, New Zealand.

Underwood AJ 1991. Beyond BACI: experimental designsfor detecting human environmental impactson temporal variations in natural populations.Australian Journal of Marine and FreshwaterResearch 42: 569-587.

Winterbourn MJ 1973. A guide to the freshwater Molluscaof New Zealand. Tuatara 20: 141-159.

Winterbourn MJ, Gregson KLD 1989. Guide to theaquatic insects of New Zealand. Bulletin of theEntomological Society of New Zealand 9: 1-96.

Dow

nloa

ded

by [

202.

20.2

.4]

at 1

7:46

19

June

201

4