AREA 1.4 • PESTICIDES, MANAGEMENT FOR FORESTRY ECOSYSTEMS • RESEARCH ARTICLE

Dissipation of four forest-use herbicides at high latitudes

Mike Newton & Elizabeth C. Cole & Ian J. Tinsley

Received: 28 February 2008 /Accepted: 10 August 2008 /Published online: 28 August 2008# Springer-Verlag 2008

AbstractBackground, aim, and scope Large-scale deforestation isoccurring in subarctic North America following clearing bysalvage logging or insect attack. Numerous shrubs, herbs,and deciduous tree species tend to dominate areas on whichstands of white spruce have grown. In the absence of eco-nomically advantageous mechanical methods, several her-bicides have value in efforts to reforest by planting whitespruce. Glyphosate, imazapyr, triclopyr, and hexazinone areall capable of selectively removing many competing species,but there is concern about whether they would degradenaturally or persist owing to the frigid climate.Materials and methods We established test plots with all fourherbicides in upland and river bottom sites at 65°N and 58°Nlatitudes. The northern site has extremely cold winters, withsoils that freeze to a depth of 1–2 m, and precipitation of275 mm/year. The southern site has heavy rain and snowfall,amounting to 2,250 mm/year evenly distributed. Soil seldomfreezes deeply. On each test plot, one of the four herbicideswas applied at twice the normal operational use rate tofacilitate detection. They were applied at the normal timing,with hexazinone, imazapyr, and triclopyr applied in June andglyphosate applied in fall. Soils were sampled immediately

after treatment and those samples used as references fordissipation data gathered over the next 11–14 months fromsoil 0- to 15- and 15- to 45-cm depths.Results Dissipation rates did not follow first-order ratesbecause freezing conditions slowed most microbial activity.All products dissipated to close to or below detection limitswithin the time of the study. Dissipation from vegetation wassubstantially more rapid and depended on the nature of theplants treated as well as the product used. While soil residuesdissipated more slowly than in temperate regions, they diddisplay consistent dissipation patterns during above-freezingconditions and also the influence of microbial activity. Mobi-lity was very limited with all products but hexazinone.Discussion These products dissipate during summer in highlatitudes much as they would in temperate climates. Winterchanges are small, but are not unlike some changes reportedelsewhere under freezing conditions. Unlike many otherstudies, soil water did not influence dissipation heavily, butthe high latitude and semi-arid climate also did not createseverely droughty soils. Residues in plants were much higherthan those in soils, but denatured the vegetation quickly,leading to unsuitability for forage in any case.Conclusions Low toxicity of these products and theirmetabolites combined with consistent dissipation and lowmobility suggest that toxic hazard of their use at high latitudesneed not be a matter of serious concern to humans, terrestrialwildlife, or aquatic systems. They are safe for use in manage-ment and rehabilitation of boreal forests when used properly.Recommendations and perspectives Dissipation at ratesapproaching those in warmer climates offer a hypothesisthat microflora native to high latitudes may be adapted todestruction of such molecules at lower temperatures thanmay be indicated by experiments with microflora adaptedto warmer climates. Residues pose no observable risk towildlife or humans in the area of use when products areapplied properly.

Environ Sci Pollut Res (2008) 15:573–583DOI 10.1007/s11356-008-0039-7

Electronic supplementary material The online version of this article(doi:10.1007/s11356-008-0039-7) contains supplementary material,which is available for authorized users.

Responsible editor: Alvin L. Young

M. Newton (*) : E. C. ColeDepartment of Forest Science, Oregon State University,Corvallis, OR 97331, USAe-mail: mike.newton@oregonstate.edu

I. J. TinsleyDepartment of Environmental and Molecular Toxicology,Oregon State University,Corvallis, OR 97331, USA

Keywords Dissipation . Forest herbicides . Forest-useherbicides . Frigid climate . Glyphosate . Hexazinone .

High latitudes . Imazapyr . Pesticides . Residues . Soils .

Subarctic . Triclopyr .Wildlife forage

1 Background, aim, and scope

Several events in the past decade have stimulated an interest inthe rehabilitation of the northern boreal forests in Alaska andnorthern Canada. Specifically, a massive outbreak of sprucebeetle (Dendroctonus rufipennis) has killed most merchant-able white and Lutz spruce (Picea glauca, including varietylutzii) timber on 1.2 million hectares of the most productivespruce forests in interior and southcentral Alaska. Moreover,there has been a major increase in harvest of boreal forests innorthern Canada as the result of reductions in harvests onpublic lands in the lower 48 states of the USA. Both of theseincreases in forest harvests, whether for salvage or for pri-mary product utilization, have been accompanied by reali-zation of importance of immediate reforestation on recentlycutover lands. Cole and Newton (1997), Cole et al. (1999),and Lieffers et al. (1993) have recently reported that theheavy cover of grasses, shrubs, and hardwoods typical ofdisturbed sites in the boreal region are severe competitors ofplanted spruce and birch seedlings. They also observed that avariety of herbicide products are of value for maximizingsuccess of reforestation efforts. The efficacy and labor effi-ciency of chemical vegetation control in these labor-deficientforest zones suggests that herbicides be considered as po-tential tools for enhancing regeneration. A general lack ofinformation regarding behavior of herbicide residues in highlatitudes has inhibited general acceptance of these tools.

Numerous studies of herbicide dissipation have includedenvironmental features likely to be important in behavior ofresidues in the boreal forest. While first-order breakdownapproximates patterns for many herbicides under warmtemperate conditions, verification in very cold soils is needed.Low temperatures are reported to be associated with pro-longed residue life (Parker and Doxtader 1983; Choi et al.1988; Newton et al. 1990; Johnson et al. 1995), yet Schmielet al. (2004) report respiration throughout winters in frozenarctic soils and Meyer et al. (2004) have isolated strains ofPseudomonas that grow on a range of cold substrates andalso adapt to various carbon sources. First-order dissipationcurves for herbicide residues may vary because of the sea-sonal changes in temperature and their influence on metabolicrates. Acid soils are reputed to decrease both mobility ofresidues and dissipation rates (Johnson et al. 1995; Stougaardet al. 1990). High organic concentrations in soils are asso-ciated with high levels of sorption, hence immobility (Benoitet al. 2008), but may be positively associated with de-gradation rates (Stougaard et al. 1990). Soil moisture content

is reported as influential on both persistence and mobility ofresidues (Parker and Doxtader 1983; Choi et al. 1988). Thelatter factor may be particularly important in Alaska sincecoastal sites are characterized by very intense rainfall year-round, while inland sites are more likely to be semi-arid.Unfortunately, relatively few herbicide residue studies havebeen conducted in forests at latitudes near the northern limitof tree growth, and the combined influence of several factorsnoted above suggest that extrapolation from temperate zonestudies may be inappropriate.

The study described here is intended to broaden under-standing of dynamics of dissipation of four herbicides likelyto receive significant use in extreme northern forests androadsides of North America. There have been severalnorthern studies on triclopyr (Jotcham et al. 1989; Stephensonet al. 1990; Thompson et al. 2000), hexazinone (Leveille etal. 1996; Helbert 1990; Feng and Navratil 1990), imazapyr(Torstensson and Borjesson 2004) and glyphosate (Feng andThompson 1990; Thompson et al. 2000), all of which havedocumented moderate to rapid dissipation in north temperateclimates. We conducted experiments on these products incolder and wetter environments than have been reported. Wealso provide fate information on imazapyr, for which littlenorthern latitude soil fate research has been reported.

2 Materials and methods

2.1 Field

Our approach to herbicide residue evaluation entailedreplicated arrays of precise applications in environmentslikely to receive vegetation control treatments in forestmanagement. These applications were followed by infrequentsampling of residues for a year or more to calculate long-termdissipation rates. We chose locations for study to reflect bothextremes of temperature and extremes of rainfall within therange of northern forests in order to havemaximum sensitivityto wide-ranging environmental parameters. We also appliedtest rates at twice the applications used operationally so asto facilitate analytical detection, assuming that kinetics ofdissipation are independent of initial dosage within therates of herbicides that do not influence microbial action(Parker and Doxtader 1982).

2.1.1 Site selection

Sites selected for these experiments included an upland southslope and a bottomland flat in the Alaskan Interior and on thecoast of the Gulf of Alaska. All had been logged withoutleaving residual trees. The Interior sites (Fairbanks Uplandand Fairbanks River bottom) were located about 30 km westof Fairbanks (lat 64.67°N, long. 148.21°W) Alaska. The

574 Environ Sci Pollut Res (2008) 15:573–583

climate in this area is characterized by frigid winters andmoderate summers. NOAA records for Fairbanks indicate aprecipitation average of 269 mm/year with a slight seasonalmaximum in August. Soil typically freezes in October andremains frozen until mid-late May. North-facing slopes areunderlain with permafrost in many situations, and deepfreezing occurs in patches on level river bottom sites. Theriver bottom soil did not thaw completely in several locationsuntil after June in the year of application, but the upland soilwas thawed by late May in both years of the study. Soil on theupland site is deep loess consisting of well-drained silts blownfrom the Tanana River, likely mixed with deposits originatingfrom the Alaska Range. Slope is about 15% in a southerlydirection. The flat river bottom site is on deep alluvial siltloam adjacent to the Tanana River. The river bottom siteremained very wet during the study, presumably because oflack of drainage over frozen subsoil.

The coastal site (Windy Bay Upland and River bottom) islocated on the north shore ofWindy Bay, Alaska (lat. 59.22°N,long. 151.22°W) on the southern tip of the Kenai Peninsula.The climate at this site is characterized as one of moderatelycold winters and cool, very wet summers. Average precipita-tion is not known, there being no weather stations nearby.However, we estimate about 1,750 mm/year, mostly as rain,with a weak maximum in fall. We had local rain gauge data forthe period following plot establishment and before freezingweather set in and for 3 months in the following year. Thisgauge established a similar pattern to that observed at two ofthe three NOAA weather stations at Seward from which weestimated total rainfall.

Soils at Windy Bay River bottom are primarily volcanicash 10–15 cm deep from the Alaska Peninsula originatingfrom Mt. Augustine. Soils are moderately well-drained sandyloams to loams, with spots 15−30 cm in depth and withpockets of fine gravel, on an old beach terrace of coarse, little-weathered angular gravel, from metamorphic shales. Theupland soil, of the same ash underlain by unsorted talus overbedrock, is somewhat deeper than that of the river bottom site.The upland soil is well drained and varies from 30% to 45%slope in southerly orientation; the finer textured soil at bothlocations remains very wet because of the heavy precipitationand lack of evaporative loss. The river bottom soil is welldrained and level, but even when saturated does not hold as

much water as the upland site because of its coarser texture.Soil pH for all sites ranged from 4.1 to 4.55.

2.1.2 Experimental procedure

The general experimental design provided at each region andeach local site the establishment of a completely randomizedset of 12 plots on which three replications compared thedissipation rates of glyphosate, triclopyr, imazapyr, and hexa-zinone on separate plots over a dissipation period of a year ormore. Each herbicide was applied at twice the rate of ap-plication at which one might expect to use the product forselective forest weeding and roughly equal to the rate used forsite preparation. All were applied to coastal and interior sitessimilarly (Table 1).

Plot size was 0.008 ha (3.7×21.9 m). Plot orientation wasparallel to contour of slope where possible. Chemicals wereapplied with a nitrogen-powered R&D Sprayer® withhandheld 3.6-m boom equipped with eight 80015 nozzlesand pressure of 200 kPa. All herbicides were applied in waterat 130 l/ha, with 0.25% Entry II® surfactant plus the technicalproducts. Release height was about 1.2 m, varying slightlyaccording to irregularities of topography; most herbs were lessthan 75 cm in height, but some aspen (Populus tremuloides)suckers and elderberry (Sambucus racemosa) shoots wereabove the range of heights in which they would receivecoverage and were clipped before spraying to ensure co-verage of all foliage and stems.

Application at each plot was timed with a stopwatch toconfirm accuracy and to calculate total chemical applied toeach plot. Times were within 15% of calibrated rate on everyplot.

Application was done on two different dates, corre-sponding to the timing of operational applications (Table 2).It is noteworthy here that glyphosate had one fewer dates ofresidue collection on this account. The dates of applicationare of major importance in that those applied in spring hada whole summer for dissipation before winter, whileglyphosate was applied slightly more than a month beforefreezing temperatures occurred and when soils were coolingrapidly.

Soil samples were taken with contamination-free samplerswith 23-mm core diameter. Residue samples were collected at

Table 1 Rates of herbicide applied at Windy Bay and Fairbanks sites, Alaska

Herbicide Rate at Windy Bay (kg/ha) Rate at Fairbanks (kg/ha)

Hexazinone (Velpar L®) 2.2 2.75Imazapyr (Arsenal Applicator’s Concentrate®) 0.28 0.28Triclopyr (Garlon 4®) 2.2 2.2Glyphosate (Accord®) 2.2 1.65

Rates are shown as acid equivalent (a.e) or active ingredient (a.i., hexazinone only) in kilograms per hectare.

Environ Sci Pollut Res (2008) 15:573–583 575

depths of 0−15 and 15−45 cm in triplicate on each plot. Thetriplicate samples were taken from three 1-m2 microplotsfrom which debris had been removed and an even surfaceexposed to intercept the spray patterns. After extraction, thehole was closed and a wooden stake placed in the hole toprevent sampling at the identical location again, but allsubsequent samples were taken within the microplots.

Each soil core was gathered in a plastic tube that wascapped and sealed at the time of extraction. The triplicatecores from each square meter and each depth were sealed inplastic-lined canvas bags for freezing and shipment. The coresin each bag were later bulked for analysis. Between plots, theextraction equipment was washed with acetone to remove anycontamination from a previous plot. A fourth square meterwas also laid out in each plot for extraction of gravimetric soilwater and organic matter determination. Those samples werekept separate from residue samples.

Vegetation samples were collected on upland sites to reflectthe degree to which application contaminated forage for localwildlife species. The principal forage species used by localherbivores at Fairbanks was aspen suckers. Species at WindyBay largely consisted of salmonberry (Rubus spectabilis),false azalea (Menziesia ferruginea), blueberry (Vacciniumovalifolium), high-bush cranberry (Viburnum edule), and redelderberry. Also present, and not sampled, was fireweed(Epilobium angustifolium). Jointgrass (Calamagrostis cana-densis), on which there was little evidence of herbivory, wascollected to the extent of 20−25% of collections because ofits predominance in the cover on some plots. Plant materialwas frozen as soon as possible and always within 48 h of thetime of collection. Vegetation was shipped in separate sealedbags with the other samples, but isolated from them.

All types of soil and vegetation were required as con-tamination free standards for calibration of analytical equip-ment. One-kilogram samples of each soil and vegetation typewere collected and sealed in plastic bags inside canvas bags.

These were collected and sealed before herbicide containerswere opened at each location. After spraying, initial residuesamples were collected either later in the day after sprayingand drying or early the next day (days after treatment, DAT1).

All samples were frozen as soon as possible, generallywithin 24 h. Even when collecting samples at Windy Bay,which was accessible only by bush aircraft, we were able tomaintain this schedule thanks to freezer facilities at a nearbylogging camp. Samples were moved in insulated boxes fromthe field site, then kept frozen in Fairbanks or Anchorage untilworkers returned to the laboratory. For transportation to thelaboratory in Corvallis, OR, we packed all hard-frozen sam-ples in special insulated shipping containers. These traveledby air with the field workers and were delivered immediatelyto the lab freezer on arrival. Good Laboratory Practices andGood Field Practices were followed throughout sampling,storage, shipping, and analytical procedures.

We established rain gauges in each subregion for measure-ment of summer rainfall. These were read by local cooperatorsduring summer and dismantled during the winter to avoiddamage from freezing.

Soil temperature was recorded at 15- and 30-cm depth byOMNI digital temperature probes in upland sites only andonly from June through August out of concern for damagefrom freezing. Unfortunately, the Windy Bay soil temperaturerecorder failed, and the Fairbanks sensors recorded data foronly the summer. We therefore used air temperature data fromtwo long-term ecological research (LTER) sites that werenearby the Fairbanks locations. The LTER sites also hadrecords of soil temperature, and these provided a reasonablematch to such soil temperatures as we recorded at theFairbanks Upland site. Air and soil temperature data from aLTER Tanana floodplain site offered a reasonable match forour observations of soil thawing when drawing soil samples,hence is deemed reasonably reliable. The only temperatureinformation available for Windy Bay was from a NOAAweather station at Seward, which did not have a soiltemperature monitor. Soil temperature in the area is presum-ably closely tied to temperature of falling rain, hencetemperature of air during wet weather. Overall regressioncorrelations (r2 values) for the Fairbanks LTER soil and airtemperature data ranged from 0.66 to 0.94 for 1991 and1992. For the time period when temperatures are likely to beabove freezing (May 4 to October 21), r2 values were 0.92 to0.98. For consistency, we therefore used air temperature dataas a surrogate for soil temperature in all models ofdissipation rate.

2.2 Analytical

The four herbicides involved in this study were all analyzedwith gas chromatographic techniques. Appendix 1 providesanalytical methods for each chemical. Appendix 1.1 provides

Table 2 Dates of application (DAT-0) and sampling for soil andvegetation, Windy Bay and Fairbanks sites, Alaska

Chemical Date DAT

Windy Bay Fairbanks Windy Bay, Fairbanks

Triclopyr,hexazinone,imazapyr

June 6, 7 May 30 0June 7, 8 May 30 1July 25, 26 July 16, 17 48, 47Oct. 2, 3 Sept. 19, 20 117, 112June 11, 12 June 9, 10 366, 375Sept 6 Aug 25, 26 456, 454

Glyphosate Aug 27 Aug 22, 23 0Aug 27 Aug 23 1Oct 3 Sept 23 36, 30June 11, 12 June 8 284, 286Aug 29, 30 Aug 28 363, 367

576 Environ Sci Pollut Res (2008) 15:573–583

analytical methods for hexazinone, Appendix 1.2 describesmethods for imazapyr, Appendix 1.3 the methods fortriclopyr, and Appendix 1.4 the methods for glyphosate.

Each required calibration with spiked samples, determina-tion ofminimum detection limit, and estimated recoverywhenextracted from the kinds of material brought from the field.Details of calibration are shown in Table 6, Appendix 1. Alldeterminations were adjusted for recovery prior to statisticalanalysis.

2.3 Data analysis

In view of differences between thicknesses of soil layerssampled (15 vs. 30 cm), the concentration observed in thedeeper layer was doubled and added to the residue observed inthe surface layer in order to approximate total residue in theprofile. Total residue levels were regressed with the NLINprocedure of SAS® software (Statistical Analysis System,SAS Institute, Inc., Cary, NC, USA). Models were fitted witha variety of forms, ranging from a standard negative ex-ponential decay model to incremental models of decay rateswithin time periods of differing temperatures. Test modelsincluded measures of DAT, temperature, soil water content,soil carbon, and various combinations of those variables. Allmodels with reasonable fit to the data were exponential decayforms, with multiplicative independent variables. DAT foreach sample date, average air temperature during intervalsbetween sample dates, and soil carbon content were theparameters that best explained the fluctuations of dissipationrates during the intervals of the sampling period. The best-fitcombination is shown as Eq. 1:

Residue; mg=kg; atDAT ¼ ICeaDATebAvtemp ecSoilC ð1Þwhere IC is initial concentration, Avtemp is average airtemperature during the interval between sample dates, andSoilC is % soil carbon content.

This model provides composite curves that allow fortemperatures in each interval to influence the slope of thedissipation curve between sampling dates so that each curverepresents the sums of percent dissipation per day timesnumber of days in intervals. In this case, soil carbon did notvary with date, but temperature did. Because dissipationtrends related to soil carbon were not consistent among thefour sites, we believe that soil carbon is serving as a dummyvariable that separates the sites.

3 Results and discussion

3.1 Soils

Most products approached or reached the lower limit ofdetection during the period of these experiments. Residues

generally followed a negative exponential decay curve withchanging rates as determined largely by temperature (Fig. 1).Hexazinone displayed very low but detectable residues in asignificant fraction of treated plots after a year, and someimazapyr residues were observed at the detection limit. At thislevel, such a determination simply notes a trace, and quantitydetermination is imprecise. Triclopyr and glyphosate wereeither non-detectable at the end of the study or were detectedclose to the detection limit in a minor fraction of total samples.

Equation 1 proved to be more applicable than any otherfor fitting all products and locations. Coefficients (Table 3)varied for the different chemicals, but the generalized re-sponse to the environment did not. The general curves wereslightly improved by including percent carbon that mirroredsomething in sites that also influenced rate of loss. The siteswith characteristically high water contents also had above-average carbon content, so there is confounding between soilcarbon and soil moisture (Tables 4 and 5). Soil carbon variedwidely among study sites, hence would be expected toinfluence microbial degradation or immobilization by sorp-tion. Benoit et al. (1998) observed live fungal tissue having amajor role in immobilizing residues of 2,4-D and atrazine.Benoit et al. (2008) reported sorption of herbicides as beingrelated to character of soil organic particles. Despite the varia-tion on soil carbon among our sites and range of precipitationas well as herbicide characteristics, we did not observe a widerange of behavior in dissipation patterns in these forest soils.We did not analyze the soil with enough detail to detectsorption processes at this level.

Soil moisture content did not contribute enough to pre-cision of estimates for all chemicals and all locations toexplain a significant part of variation. However, soil carbonand soil water were highly correlated (r=0.825); hence, wedid not have sufficient data to separate their partial contri-butions to dissipation rate. Windy Bay residues dissipatedslightly more rapidly than those at Fairbanks, and they alsohad higher moisture content though well drained. Theycertainly had leaching potential far beyond that of interiorsites. For glyphosate alone, adding the term temperature*moisture did improve the predictive power of the equation.

The principal deviation between models and observedresidues appeared to be when the model had to fit rapiddegradation during the warm summer and fall period, leadinginto a winter season when degradation is presumably low ornegligible at sub-freezing temperatures. Curves are undoubt-edly triphasic, with fast-slow-fast phases for those applied inspring, but we did not have enough sample dates to definedistinctly the rates during transition periods because eachspring and fall spanned a month-long (or more) period ofrapidly rising or falling temperatures that were not captured byour data set.

We do not attempt to determine half-life. Determination offirst-order kinetics is contingent on steady-state conditions

Environ Sci Pollut Res (2008) 15:573–583 577

that do not exist in multi-season experiments in the subarctic.Dissipation environment in the field is sufficiently variablewith temperature that the apparent half-life changes withseason during the dissipation process. In view of the extremeranges in temperature and rates of change within those ranges,a daily record of temperature and residue level would havebeen required to evaluate the changing half-lives with changesin environment. This was beyond the scope of this study.

It appears that much of the dissipation of these herbicideswas microbial. Based on the observations of aminomethyl-phosphonic acid (AMPA) in samples from plots treated withglyphosate, microbial decomposition is occurring during thelives of those residues (see Fig. 1a). While we do not quantifydecomposition products from the other herbicides, the ex-istence and declining residues of AMPA is thought to bedefinitive evidence of biological degradation (Torstensson etal. 2005). The untimely retirement and deaths of the two

analytical chemists on our team limited the numbers ofAMPA samples with which we could quantify microbial de-composition of glyphosate, but the emergence of positiveAMPA determinations did identify biological processes asoperating during decomposition.

These experiments tend to extend the range of findingsconducted in temperate regions to verify that the frigid arctichas summer environments and perhaps winter conditions inwhich several types of herbicides degrade microbially.Summer air temperatures at latitudes between 59° and 66°Nare well within the range found in cool temperate zones. Soiltemperatures below 15 cm are slow to rise in summer andseldom approach those found further south. The observationthat residues remain close to the surface, where temperaturesare highest, is probably important. Dissipation that is notcomplete during the season of application, as is most likely infall applications, will likely be very slow during the coldwinter. When soils begin to warm in spring, dissipation thenprogresses until residues are no longer detectable. We thusfind that dissipation patterns are likely to be similar to thosefurther south except for the long interruption of winter.

Most dissipation studies have been done in agricultural orforested environments in which soils were at 20−30°C duringthe period of dissipation (Parker and Doxtader 1982, 1983) ordone in forests in the southeastern USA where temperatureswere unspecified, but characteristic of the warm summerclimate there (Michael and Neary 1993; Neary et al. 1993;

Table 3 Coefficients for the factors in Eq. 1 that influence estimatesof soil residue for four herbicides

Coefficient a b c

Hexazinone −0.00535 0.0136 −0.0347Triclopyr −0.0166 0.0502 −0.1210Imazapyr −0.00302 0.0498 −0.1571Glyphosate −0.00451 −0.0432 −0.0317

Fig. 1 Residues remaining in top 45 cm of soil profiles, in parts per million (mg/kg soil dry weight) during the year or more following application infour environments of Alaska. Means of residues and modeled estimators of predicted values are shown for the four subregions

578 Environ Sci Pollut Res (2008) 15:573–583

Johnson et al. 1995). Choi et al. (1988) demonstrated thatmicrobial degradation of dimethyl tetrachloroterephthalate,although unrelated structurally to the chemicals discussedhere, was strongly influenced by temperature and that at 10°Chalf-lives were greatly extended over those at 20°C. Based on

the LTER soil temperature data and data from other currentstudies in interior and southcentral Alaska (Cole et al.,unpublished data), soil temperatures at 15 cm may be above10°C for 2 to 3 months, but this is highly variable from year-to-year and strongly dependent on the amount of organicmatter covering the soil. Despite this, a substantial part of thedegradation observed in this study clearly took place whensoil temperatures were at or below 10°C. In view of the airtemperature at Windy Bay seldom exceeding a daily mean of11°C, the soil probably seldom reached that level.

There is abundant evidence of microorganism activity atlow temperatures at high latitudes that could help explainmoderately high rates of loss on cold soils. Fahnestock et al.(1999), Stenrod et al. (2005), and Jarvis et al. (2006) havereported losses of carbon under arctic tundra ecosystems andof herbicides in rail ballast while soils were frozen or nearfreezing. Their findings suggest that inclusion of such lossesin estimates of annual carbon dissipation amount to some17% increase in predicted loss by including dissipation duringthe 240-day period of frozen or near-freezing substrate.Lipson and Monson (1998) observed active roles of plantsand microbes competing for amino acids during freeze-thawevents. This level of microbial activity not only corroboratesthe notion that subfreezing microbial action is an importantadaptation, it also suggests that winter losses of residues arenot unlikely. Mohn et al. (1997) also noted variation inadaptation of microorganisms to cold soils and noted

Table 5 Specific environmental parameters for glyphosate plots oneach site; air temperature and soil moisture at each sampling data, andsoil carbon

Region Site DAT Soil C % Water Air Temp (°C)

WB RB 1 6.9 61.2 9.230 43.0 8.4

290 49.4 −2.7366 72.3 11.4

UP 1 11.4 117.4 9.230 109.4 8.4

290 89.0 −2.7366 112.1 11.4

FB RB 1 4.9 81.0 10.228 57.5 8.8

291 70.8 −10.2366 49.8 16.2

UP 1 1.1 16.0 9.128 15.6 9.7

291 29.5 −7.3366 18.9 15.8

WB Windy Bay, FB Fairbanks, RB river bottom, UP upland

Table 4 Specific environmental measures of each site: mean air temperature in interval before sampling and soil moisture (percent) for eachchemical sampling date for plots involving triclopyr, hexazinone, and imazapyr, with means of soil carbon for each chemical on each site

Site DAT Triclopyr Hexazinone Imazapyr Air Temp (°C)

Soil C % Water Soil C % Water Soil C % Water

WB RB 1 12.0 73.8 10.4 81.6 5.8 68.7 13.349 59.1 72.8 62.2 11.8118 87.5 66.6 61.6 9.5370 71.3 96.1 65.1 −2.7456 72.8 88.9 73.6 11.4

UP 1 8.6 108.3 13.3 129.6 12.9 107.2 13.349 124.6 109.5 77.8 11.8118 119.8 108.0 108.1 9.5370 109.3 119.9 96.1 −2.7456 126.3 129.2 103.3 11.4

FB RB 1 4.2 65.7 6.1 127.2 3.7 58.7 13.649 54.1 60.3 43.5 15.8118 71.9 63.6 44.6 10.8370 66.6 87.0 57.3 −10.2456 63.1 59.6 47.6 16.2

UP 1 0.8 33.4 2.1 38.4 2.6 38.2 13.249 16.8 17.2 12.2 15.8118 15.8 19.7 16.9 11.6370 28.5 30.3 29.5 −7.3456 13.9 15.7 19.1 15.8

WB Windy Bay, FB Fairbanks, RB river bottom, UP upland

Environ Sci Pollut Res (2008) 15:573–583 579

increased rates of microbial degradation of polychlorinatedbiphenyl in soil cultures of organisms that had been exposedto those products previously at low temperatures. We thereforeacknowledge uncertainty during freezing conditions. Whereasour intervals between samplings included some evidence ofdissipation, soil temperature varied from freezing or below toan unknown pattern of change between the time of freezing orthawing and that of measurement, and residues close to thedetection limit did not always show decreases then. Regard-less of this, there appeared to be dissipation when soil wasclose to freezing or at that point much of the time and whenthere was no likelihood of losses due to leaching.

Soil water is required for microbial metabolism. Parker andDoxtader (1983) postulated that microbes responsible for de-composition may require time to reach populations at whichmeasurable degradation might take place and that moisturelevels would positively influence the rate of microbial res-ponse. Relevance of this to our study is unknown becauseonly one site, at Fairbanks upland, could be characterized inany terms as dry and then only for part of one samplinginterval. At Windy Bay, soil was always very moist or wet.

Very little movement was observed out of the surface 15-cmsample depth into the 15- to 45-cm depth. No imazapyr wasdetected in the lower level in any sample, in some contrast toTorstensson and Borjesson’s (2004) observations in coarserailroad ballast. Nine percent of the triclopyr detected in thewhole soil column was below 15 cm, and the proportion oftotal residue in the lower-layer sample decreased with time,becoming non-detectable at about the same time as surfaceresidues. Eleven percent of all detected hexazinone was in thedeeper zone, and 7% of glyphosate was in the deeper zone.Each of these products except hexazinone decreased in con-centration relative to surface residues during the course ofsampling; the ratio of upper and deeper residues for hexa-zinone did not change appreciably with time. This suggeststhat residues were immobile soon after being incorporated inthe soil. Helbert (1990), studying herbicides in forest soils ofNewfoundland, also observed that fully 75% of the residuesof several herbicides in any given sampling period were inthe surface organic layer.

Rainfall at the Fairbanks site (about 60 mm in the first3 months after application) was not sufficient to provide thepotential for major movement of soluble chemicals. WindyBay recorded 45 mm rain in the month of June in whichapplication occurred and between 100 and 250 mm per monthin the next 4 months in a climate with very little evaporativepotential. Approximately 375 mm fell in the 2 monthsimmediately following application of glyphosate. Despitethe flow of water through the soil profile, there was very littletransfer of herbicide into the soil horizon below 15 cm for anyof the chemicals. The extrapolated estimate of rainfallsuggests that a meter or so of water went through the soilprofile in the 7 months of winter before our rainfall records

were restarted and during which changes in residues weresmall. Under the circumstances, we observe that all thesechemicals were strongly sorbed.

The only product that showed consistent movement intothe deeper horizon of soil was hexazinone. This material alsowas evaluated for leaching in a separate study (Newton andCole 1997) at Windy Bay on a nearby steep slope with sandysoil. There was evidence there of subsurface movement atdetectable levels 30 m from hectare-sized plots when drivenby 600-mm rainfall in the 60 days following application.Despite this evidence of mobility, hexazinone residues re-mained in surface soils much more than in deeper samples inthe current experiments. This suggests not only a substantialdegree of binding to soil for a major fraction of the residuepresent but also that a small fraction moves readily. Despitethe possibility of mobility, applications of hexazinone atregistered forestry rates should not pose a measurable en-vironmental risk. Hexazinone is very low in toxicity to fishand aquatic fauna (Ahrens 1994), but damage to sensitivespecies of plants, including planted coniferous seedlings,could occur if movement in water caused an accumulation atpoorly drained sites.

Our assumptions of microbial roles rather than leaching inthe dissipations observed are verified only for glyphosate,with which we also observed the decomposition productAMPA (see Fig. 1a), and imazapyr, which did not reach de-tectable levels below 15 cm. The very low rate of triclopyrmovement to below 15 cm suggests that the dissipation islargely microbial. The very low mobility of all chemicalsregardless of precipitation suggests that sorption rates maybe high in extremely cold soils. This degree of immobilityprovides some assurance of negligible off-site activity.

While soil residues behaved with reasonable consistency,the variation within chemicals and locations decreased ourprecision of estimation somewhat. Analytical data for initialresidues varied widely from calculated application rates. Ini-tial determinations ranged from 22% to 313% of the expected,based on amount nominally applied. We anticipated variabil-ity and designed the sampling spots as homogeneously aspossible and returned to nearly the same sample spots at eachdate. Thus, we did not attempt to correct these in view of theuse of the model expressed in Eq. 1 in which the rate ofdissipation is independent of initial concentration.

Microsite variability in deposition patterns undoubtedlycontributed to variation in observed residues. Newton et al.(1990) demonstrated very high coefficients of variation (29−104%) in aerial application deposits as collected by cups of9-cm diameter placed in crowns of shrubs sprayed with verylarge drops. Intuitively, one would not expect a spray patternwith small drops in large numbers, applied by a hand-heldboom, to be so heterogeneous.

The cleared spots in which samples were taken were largeenough so that amounts of treated vegetation falling into

580 Environ Sci Pollut Res (2008) 15:573–583

sample points were small. External sources of contaminationwere minimal. However, at Windy Bay, while there wereintense rains that might have moved surface deposits ofherbicide a few centimeters into very small depressions, thesedid not occur in the first month following application exceptthe fall application of glyphosate. Feng and Navratil (1990)observed that initial deposits as determined from soil coreswith only a surface deposit tend to underestimate actualdeposition, so that later determinations of cores with rain-incorporated residues are likely to display an apparent in-crease in total residue. Perhaps, heterogeneity of depositcombined with occasional occurrence of the phenomenonreported by Feng and Navratil (1990) would account for bothhigh variance in deposit and also apparent increase close tothe limit of detection that we observed with imazapyr inFairbanks. The soil corer we used had a bore diameter ofonly 23 mm, which is a small area of soil surface on whichto measure a deposit. This is only one sixteenth the area ofthe sample collector noted by Newton et al (1990); hence,one might expect a much larger coefficient of variation.Despite large variances in residue levels, there was remark-able consistency in patterns of dissipation among theseclasses of herbicides.

3.2 Vegetation

Initial residues of vegetation ranged around a mean of115 mg/kg per kg/ha of active ingredient applied. Mostresidues had dissipated to low or non-detectable levelsbetween the date of application and the first revisit to eachsite. Exceptions were that 6−14% of initial concentration oftriclopyr were found at DAT 120 in the aspen vegetationtaken at Fairbanks and <1% was observed at Windy Bay.There was also a trace of hexazinone at DAT 395 at bothFairbanks and Windy Bay (Fig. 2).

Dissipation of residues from vegetation was rapid, but notalways total, within our study period. Studies of residue

dissipation in an analogous period have been reportedelsewhere. Newton et al. (1984), Thompson et al. (1994),and Cessma and Waddington (1995) used sampling intervalsclose enough together to examine details of the short-termdissipation curves from vegetation. They observed very rapiddissipation of glyphosate soon after application in a negativeexponential pattern. Had our sample collections been at closerintervals, it is likely that our dissipation curves would havebeen in a similar form, but short-term kinetics were not withinthe objectives of the study. There may be two or moremechanisms that would explain rapid initial loss followed byslower dissipation. One is that microorganisms capable ofdecomposing residues find weakly adsorbed molecules on thesurface of vegetation which are destroyed quickly, but at aslower rate as sorption of remaining residues progresses(Steinberg et al. 1987; Pignatello et al. 1993). Alternatively,surface temperatures on the vegetation were likely elevatedsubstantially more than those of soils, and this may haveresulted in greater summer microbiological activity onvegetation than on soil unless surfaces were desiccated bydeath of tissue under drying summer conditions.

Vegetation sampling posed the problem of locating pro-per living sample material a month or more after treatment.The herbicides tended to kill the susceptible species quicklyenough so that collection of samples of adequate size insome plots was difficult within 45 days following treat-ment. Vegetation that was not injured either was not sus-ceptible to the herbicide or had been missed as the result ofbeing tall enough to be positioned between nozzle patternconvergence; aspen suckers taller than 75 cm were clippedat that height before application. In any case, the samplesdid represent the most available material that would havebeen consumed by herbivores. When the samples weretaken 30−45 days after treatment, residues in dead and livetissue ranged from non-detectable to 14% of initial dosagelevel.

The traces of triclopyr found in aspen suckers on Fairbanksupland sites after an extended period are likely attributable toabsorption into stems or uptake from rained-on soil relativelysoon after application, followed by immobilization in dyingplants. Uptake of hexazinone from soil may have occurred inthe following spring in aspen that did not die from thattreatment. These observations are not inconsistent withfindings from the temperate zone cited above.

Lautenschlager et al. (1992) reviewed implications ofherbicide applications and residues on wildlife in northernconiferous forests. They observed that wildlife of manyspecies typically respond to habitat changes and that habitatsometimes changes for the better, sometimes for worse. Theyalso note that forest herbicide residues have not beenobserved at levels capable of decreasing wildlife populationsthrough direct toxic effects by the herbicides registered forforest use today.

Fig. 2 Herbicide residues in vegetation on upland sites at Fairbanksand at Windy Bay, Alaska following application of four herbicides

Environ Sci Pollut Res (2008) 15:573–583 581

4 Conclusions

Following application to high-latitude reforestation sites,residues of four unrelated herbicide products showed similardissipation patterns in cool wet and frigid semi-arid environ-ments, 59 to 65°N latitude. They provide evidence of pre-dictable degradation from cold soils, including some winterlosses, in dissipation patterns that will generally followtemperature fluctuations. Relatively small deviations fromtemperature-driven dissipation rates are attributable to mois-ture or carbon content of soils within the range encountered ininterior or coastal environments. Most dissipation occurredduring the 3 months of summer, residues were at or neardetection limits within a year, and were immobile.

Movement attributable to leaching was negligible for ima-zapyr in cold soil and restricted to minor vertical movement tothe 15- to 45-cm soil depths for glyphosate and triclopyr.Evidence with hexazinone in high-rainfall areas suggests thatsome lateral movement is possible but that concentrationsharmful to aquatic fauna are extremely unlikely unless wateris trapped and concentrated.

5 Recommendations and perspectives

Dissipation at rates approaching those in warmer climatesoffer a hypothesis that microflora native to high latitudes maybe adapted to destruction of such molecules at lower temper-atures than may be indicated by experiments with microfloraadapted to warmer climates. Residues pose no observable riskto wildlife or humans in the area of use when products areapplied properly.

Acknowledgments Guidance in organization of the research wasprovided by Logan A. Norris, Oregon State University Department ofForest Science, and Ian J. Tinsley, Department of Environmental andMolecular Toxicology, Oregon State University. Gene Johnson,Department of Environmental and Molecular Toxicology, providedsubstantial assistance in analysis of residues. Brian Roth, presently withthe Department of Forestry, University of Florida, Gainesville, providedsubstantial assistance in the field conduct of the experiments in Alaska.Ed Holsten and Ken Zogas, USDA Forest Service Div. of State andPrivate Forestry, Anchorage, AK, provided both personal and admin-istrative support for this project. Funding for the project was providedby the US Department of Agriculture, National Agricultural PesticideImpact Assessment Program. R. A. Werner, USDA Forest Service,Institute of Northern Forestry, Fairbanks, AK, provided administrativesupport. Dr. Ted Alby, of BASF Inc. (then American Cyanamid Co.),Vancouver WA, provided QA support.

References

Ahrens WH (1994) Herbicide handbook, 7th edn. Weed ScienceSociety of America, Champaign, IL

Benoit P, Barriuso E, Calvet R (1998) Biosorption characterization ofherbicides, 2,4-D and atrazine, and two chlorophenols on fungalmycelium. Chemosphere 37:1271–1282

Benoit P, Madrigal I, Preston CM, Chenu C, Barriuso E (2008)Sorption and desorption of non-ionic herbicides onto particulateorganic matter from surface soils under different land uses. Eur JSoil Sci 59:178–189

Cessma AJ, Waddington J (1995) Dissipation of glyphosate and itsmetabolite AMPA in established crested wheatgrass followingspring application. Can J Plant Sci 75:759–762

Choi JS, Fermanian TW, Wehner DJ, Spomer LA (1988) Effect oftemperature, moisture, and soil texture on DCPA degradation.Agron J 80:108–113

Cole EC, Newton M (1997) Competition and the need for vegetationmanagement. In: Reforestation and Vegetation in Central Alaska.State and Private Forestry Special Report R-10-TP-65. USDAForest Service, Anchorage, AK, pp I-1–I-19

Cole EC, Newton M, Youngblood A (1999) Regenerating whitespruce, paper birch, and willow in south-central Alaska. Can JFor Res 29:993–1001

Fahnestock J, Jones MH, Welker JM (1999) Wintertime CO2 effluxfrom arctic soils: implications for annual carbon budgets. GlobBiogeochem Cycles 13:775–779

Feng JC, Navratil S (1990) Sampling for zero-time hexazinone residuesin forest soil dissipation study. Can J For Res 20:1549–1552

Feng JC, Thompson DJ (1990) Fate of glyphosate in a Canadian forestwatershed. 2. Persistence in foliage and soils. J Agric Food Chem38:1118–1125

Helbert S (1990) Behaviour of four soil-active herbicides in a borealpodzol. For Ecol Manage 31:125–152

Jarvis NJ, Almqvist S, Stenstrom J, Borjesson E, Jonsson E,Gorstensson L (2006) Modelling the leaching of imazapyr in arailway embankment. Pest Manag Sci 62:940–946

Johnson WG, Lavy TL, Gbur ED (1995) Sorption, mobility, anddegradation of triclopyr and 2,4-D on four soils. Weed Sci. 43(4):678–684

Jotcham JR, Smith DW, Stephenson GR (1989) Comparativepersistence and mobility of pyridine and phenoxy herbicides insoil. Weed Technol 3:155–161

Lautenschlager RA, Gjerstad DH, Glover GR, Mitchell RJ, HorsleySB, Zedaker SM, Dougherty PM (eds) (1992) Response ofwildlife to forest herbicide applications in northern coniferousecosystems. Proceedings of the International Conference onForest Vegetation Management. Can J For Res 23:2286–2299

Leveille P, Legris L, Couture J (1996) Residus d’hexazinone dans lesol après une pulverization terreste en milieu forestier. Gouverne-ment du Quebec, Ministere des Resources Naturelles, 19 pp

Lieffers VJ, Macdonald SE, Hogg EH (1993) Ecology and controlstrategies for Calamagrostis canadensis in boreal forest sites.CJFR 23:2070–2077

Lipson DA, Monson RK (1998) Plant-microbe competition for soilamino acids in the alpine tundra: effects of freeze-thaw and dry-rewet events. Oecologia 113:406–414

ManningMJ (1994) Analytical method for glyphosate and AMPA in rawagricultural commodities and their processed fractions. DocumentNo. RES-008-90, Version No. 4. Monsanto, St. Louis, MO

Meyer AF, Lipson DA, Martin AP, Schadt CW, Schmidt SK (2004)Molecular and metabolic characterization of cold tolerantPseudomonas senso stricto. Appl Environ Microbiol 70:483–489

Michael JL, Neary DG (1993) Herbicide dissipation studies insouthern forest ecosystems. Environ Toxicol Chem 12:405–410

Mohn WW, Westerberg K, Cullen WR, Reimer J (1997) Aerobicbiodegradation of biphenyl and polychlorinated biphenyls byArctic soil microorganisms. Appl Environ Microbiol 63:3378–3384

582 Environ Sci Pollut Res (2008) 15:573–583

Monsanto Co. (1991) Analytical method for glyphosate and AMPA insoil. Document No. ES-014-91, Version No. 0. Monsanto, St.Louis, MO

Neary DG, Bush PB, Michael JL (1993) Fate, dissipation andenvironmental effects of pesticides in southern forests: a reviewof a decade of research progress. Environ Toxicol Chem12:411–428

Newton M, Cole EC (1997) Reforestation and vegetation in centralAlaska. Special Report R10-TP-65. Forest Health Protection.USDA Forest Service. Alaska Region, 74 pp

Newton M, Howard KM, Kelpsas BR, Danhaus R, Lottman CM,Dubelman S (1984) Fate of glyphosate in an Oregon forestecosystem. J Agric Food Chem 32:1144–1151

Newton M, Roberts F, Allen A, Kelpsas B, White D, Boyd P (1990)Deposition and dissipation of three herbicides in foliage, litter,and soil of brushfields in southwest Oregon. J Agric Food Chem38:574–583

Parker LW, Doxtader KG (1982) Kinetics of microbial decompositionof 2,4-D in soil: effects of herbicide concentration. J EnvironQual 11(4):679–684

Parker LW, Doxtader KG (1983) Kinetics of the microbial degradationof 2,4-D in soil: effects of temperature and moisture. J EnvironQual 12(4):553–558

Pignatello JJ, Ferrandino FJ, Huang LQ (1993) Elution of aged andfreshly added herbicides from a soil. Environ Sci Technol27:1563–1571

Schmiel JP, Bilbroough C, Welker JM (2004) Increased snow depthaffects microbial activity and nitrogen mineralization in twoarctic tundra communities. Soil Biol Biochem 36:217–227

Steinberg SM, Pignatello JJ, Sawhney BL (1987) Persistence of 1,2-dibromomethane in soils: entrapment in intraparticle micropores.Environ Sci Technol 21:1201–1208

Steinmetz JR (1994) Analytical method for glyphosate and AMPA inraw agricultural commodities and their processed fractions.Document RES-008-90, Version 4. Monsanto, St. Louis, MO

Stenrod M, Eklo OM, Charnay MP, Benoit P (2005) Effect of freezingand thawing on microbial activity and glyphosate degradation intwo Norwegian soils. Pest Manag Sci 61:887–898

Stephenson GR, Solomon KR, Bowlby CS, Liber K (1990)Persistence, leachability, and lateral movement of triclopyr(Garlon) in selected Canadian forest soils. J Agric Food Chem38:584–588

Stougaard RN, Shea PJ, Martin AR (1990) Effect of soil type and pH onadsorption, mobility, and efficacy of imazaquin and imazethapyr.Weed Sci 38:67–73

Thompson DG, Cowell JE, Daniels RJ, Staznik B, MacDonald LM(1989) Liquid chromatographic method for quantitation of gly-phosate and metabolite residues in organic and mineral soils,stream sediments, and hardwood foliage. J Assoc Off Anal Chem72:355–360

Thompson DG, Pitt DG, Buscarini TM, Staznik B, Thomas DR,Kettela E (1994) Initial deposits and persistence of forestherbicide residues in sugar maple (Acer saccharum) foliage.Can J For Res 24:2251–2262

Thompson DG, Pitt DG, Buscarini TM, Staznik B, Thomas DR(2000) Comparative fate of glyphosate and triclopyr herbicides inthe forest floor and mineral soil of an Acadian forest regenerationsite. Can J For 30:1808–1816

Torstensson L, Borjesson E (2004) Use of imazapyr against Equisetumarvense on Swedish railway tracks. Pest Management Science60:565–569

Torstensson L, Borjesson E, Stenstrom J (2005) Efficacy and fate ofglyphosate on Swedish railway embankments. Pest Manag Sci61:881–886

Environ Sci Pollut Res (2008) 15:573–583 583