chemical ecology the push–pull tactic for mitigation of ... · science, policy, and management,...

CHEMICAL ECOLOGY

The Push–Pull Tactic for Mitigation of Mountain Pine Beetle(Coleoptera: Curculionidae) Damage in Lodgepole and

Whitebark Pines

NANCY E. GILLETTE,1 CONSTANCE J. MEHMEL,2 SYLVIA R. MORI,3 JEFFREY N. WEBSTER,4

DAVID L. WOOD,5 NADIR ERBILGIN,6 AND DONALD R. OWEN7

Environ. Entomol. 41(6): 1575Ð1586 (2012); DOI: http://dx.doi.org/10.1603/EN11315

ABSTRACT In an attempt to improve semiochemical-based treatments for protecting forest standsfrom bark beetle attack, we compared pushÐpull versus push-only tactics for protecting lodgepole pine(Pinus contortaDouglas ex Loudon) and whitebark pine (Pinus albicaulisEngelm.) stands from attackby mountain pine beetle (Dendroctonus ponderosaeHopkins) in two studies. The Þrst was conductedon replicated 4.04-ha plots in lodgepole pine stands (California, 2008) and the second on 0.81-ha plotsin whitebark pine stands (Washington, 2010). In both studies, D. ponderosae population levels weremoderate to severe. The treatments were 1) push-only (D. ponderosae antiaggregant semiochemicalsalone); 2) pushÐpull (D. ponderosae antiaggregants plus perimeter traps placed at regular intervals,baited with four-component D. ponderosae aggregation pheromone); and 3) untreated controls. Weinstalled monitoring traps baited with two-component D. ponderosae lures inside each plot to assesseffect of treatments on beetle ßight. In California, fewer beetles were collected in pushÐpull treatedplots than in control plots, but push-only did not have a signiÞcant effect on trap catch. Both treatmentssigniÞcantly reduced the rate of mass and strip attacks by D. ponderosae, but the difference in attackrates between pushÐpull and push-only was not signiÞcant. In Washington, both pushÐpull andpush-only treatments signiÞcantly reduced numbers of beetles caught in traps. Differences betweenattack rates in treated and control plots in Washington were not signiÞcant, but the push-onlytreatment reduced attack rates by 30% compared with both the control and pushÐpull treatment. Weconclude that, at these spatial scales and beetle densities, push-only may be preferable for mitigatingD. ponderosae attack because it is much less expensive, simpler, and adding trap-out does not appearto improve efÞcacy.

KEY WORDS Pinus albicaulis, Pinus contorta, semiochemicals, verbenone, green leaf volatiles

The mountain pine beetle, Dendroctonus ponderosaeHopkins (Coleoptera: Curculionidae: Scolytinae), isthe principal insect pest of pines in western NorthAmerica, particularly lodgepole pine, Pinus contortaDouglas ex Loudon (Furniss and Carolin 1977, Wood

et al. 2003). Lodgepole pine is a geographically wide-spread conifer species spanning a northÐsouth rangefrom the Yukon Territory to Baja California andfrom the PaciÞc Ocean in the west to the Black Hillsof South Dakota in the east (Lotan and CritchÞeld1990). Lodgepole pine forests recently have beendevastated by expansive D. ponderosae outbreaksthat threaten to turn boreal forest carbon sinks intocarbon sources (Logan and Powell 2001, Kurz et al.2008).

More recently, D. ponderosae attacks in whitebarkpine (Pinus albicaulis Engelm.) stands also have es-calated drastically (Logan and Powell 2001, Perkinsand Roberts 2003, Gibson 2006). Whitebark pine is anecologically important subalpine conifer that serves asa keystone species in high-elevation areas of the Cas-cade Ranges, the northern Rocky Mountains, and theSierra Nevada regions of the northwestern UnitedStates and western Canada (Tomback et al. 2001,Schwandt 2006). Whitebark pine seeds are an impor-tant source of overwintering nourishment for grizzlybears (Ursus arctos horribilis Ord.), and studies have

The use of trade names and identiÞcation of Þrms or corporationsdoes not constitute an ofÞcial endorsement or approval by the U.S.Government of any product or service to the exclusion of others thatmay be suitable.

1 Corresponding author: Nancy E. Gillette, USDA Forest Service,PaciÞc SW Research Station, 800 Buchanan St., Albany, CA 94710(e-mail: [email protected]).

2 USDA Forest Service, Forest Insect and Disease Center, 1133North Western Ave., Wenatchee, WA 98801.

3 USDA Forest Service, PaciÞc SW Research Station, 800 BuchananSt., Albany, CA 94710.

4 Webster Forestry Consulting, 7205 Granada Drive, Redding, CA96002.

5 University of California, Berkeley, Department of EnvironmentalScience, Policy, and Management, Berkeley, CA 94720.

6 Department of Renewable Resources, Faculty of Life, Agricul-tural, and Environmental Sciences, University of Alberta, Edmonton,Alberta T6G 2E3 Canada.

7 California Department of Forestry and Fire Protection, 6105 Air-port Rd., Redding, CA 96002.

0046-225X/12/1575Ð1586$04.00/0 � 2012 Entomological Society of America

shown them to be crucial for the survival of gestatingcubs and other wildlife (Pease and Mattson 1999, Rob-bins et al. 2006). Whitebark pines are facing a doublethreat, however, from the combination of an intro-duced pathogen, white pine blister rust (Cronartiumribicola A. Dietr), and D. ponderosae (Tomback et al.2001). Trees infected with blister rust appear to bemore susceptible than uninfected trees to attack byD.ponderosae when beetle populations are low(Schwandt and Kegley 2004), and although D. pon-derosae preferentially attacks older whitebark pines,blister rust is particularly damaging to seedlings. Thus,both the seed-bearing older trees and the new regen-eration are suffering heavy losses. Furthermore,warming climates have resulted in a shift in high-elevationD. ponderosae voltinism from semivoltine tounivoltine (Bentz and Schen-Langenheim 2007), sooutbreaks ofD. ponderosae in whitebark pines, whichwere once rare, are now common (Schwandt 2006).Regeneration of this species is so severely compro-mised by the combination of beetles, disease, andwarming climate that a petition has been Þled with theU.S. Fish and Wildlife Service to list it as an endan-gered species (Federal Register, http://www.fws.gov/mountain-prairie/species/plants/whitebarkpine/TempFR07192010.pdf). To help maintain whitebarkpine populations, recent efforts have focused on col-lection of seeds from rust-resistant trees for screeningfor propagation in seed orchards (Schoettle andSniezko 2007). In spite of these efforts to conserveresistant genotypes, many of the naturally rust-resistant trees are at risk of being killed by D. pon-derosae before they can be identiÞed, screened, andincorporated into a genetic resistance program.There is therefore much interest in the develop-ment of a safe, effective method for protecting thisimportant resource.

The environmental consequences of these out-breaks have stimulated much productive research fo-cused on semiochemical treatments to mitigate D.ponderosae-caused tree mortality, but at high beetlepopulation levels such treatments are not consistentlyas effective as desired (Bentz et al. 2005, Progar 2005).In an attempt to increase the efÞcacy of antiaggrega-tion semiochemicals such as verbenone for protectinglodgepole pines fromD.ponderosae,various additionalmethods have been investigated, including the use ofconcentration and containment, where beetles arelured into a discrete area and then killed, and pushÐpull strategies, where beetles are lured to attractant-baited traps and also repelled with antiaggregants(Borden et al. 2006). A large body of research hasyielded useful approaches to deploying verbenone(Borden et al. 1983a,b; Gray and Borden 1989; Borden1997; Kegley et al. 2003; Gibson and Kegley 2004;Kegley and Gibson 2004, 2009; Bentz et al. 2005; Gil-lette et al. 2006, 2009) and verbenone with nonhostvolatiles (Borden et al. 2003, Kegley et al. 2010) for theprotection of lodgepole and whitebark pines from D.ponderosae. Most of this research, however, has fo-cused on the deployment of bubblecap and pouchrelease devices that must be applied by hand, which

can be particularly challenging in remote, alpine siteswhere snow-cover does not melt in time to permitaccess by ground before beetle ßight occurs. Morerecently, aerial applications of verbenone-releasingßakes have been shown to be effective for mitigationof D. ponderosae-caused tree mortality in lodgepolepine (Gillette et al. 2009) and whitebark pine (Gilletteet al. 2012) stands.

Although these studies showed considerable prom-ise for the use of antiaggregants against low to mod-erate beetle population levels, concern has been ex-pressed about the efÞcacy of this technology foroutbreak beetle population levels, which typicallyhave been difÞcult to control with any treatmentmethods that have been tested except for insecticideapplications (Fettig et al. 2006). For that reason, wechose to assess a pushÐpull approach where an attrac-tive lure is used in conjunction with an antiaggregant,in the hope of achieving better control than with theantiaggregation semiochemical treatments alone(Borden 1997, Cook et al. 2007). Previous pushÐpull orconcentration approaches for mitigating bark beetledamage mostly have compared pushÐpull with pull-only strategies by using baited trap trees that areintended to be logged to protect adjacent antiag-gregant-treated stands (Lindgren and Borden 1993,Borden et al. 2006). Although the trap-tree approachhas been largely successful, on most public lands in theUnited States the intentional sacriÞce of living treesfor such purposes may be prohibited by regulation,public policy, or both. Shea and Neustein (1995) re-port possible success with a method using only baitedtraps in conjunction with antiattractant semiochemi-cals for control of Ips paraconfususLanier. That study,however, had unreplicated treatments and no controlsand therefore did not account for the record-breakingrainfall at the outset of the study that may have mit-igated beetle-caused tree mortality. Nevertheless, thisapproach may have promise for mitigation of D. pon-derosae infestations without the problems associatedwith trap trees. We therefore chose to test the alter-native of combining beetle trap-out with an antiag-gregation treatment, wherein we deployed a perime-ter ring of attractant-baited traps around theantiaggregant-treated stands. Instead of comparingpushÐpull with the use of baited trap trees, we focusedon a comparison of pushÐpull with push-only, becausethis approach does not risk killing healthy trees andpast studies have already documented the consider-able efÞcacy achievable with push-only (Gillette andMunson 2009).

Materials and Methods

Semiochemical treatments consisted of nonbiode-gradable plastic verbenone ßakes in the Þrst study(California 2008) and a combination of biodegradableverbenone and biodegradable GLV (green leaf vola-tile) ßakes (1-hexanol andZ-3-hexenol) in the secondstudy (Washington 2010). We added the GLVs in thesecond study because work by Kegley and Gibson(2009) and Kegley et al. (2010) indicated their prom-

1576 ENVIRONMENTAL ENTOMOLOGY Vol. 41, no. 6

ise for enhancing efÞcacy of verbenone. This studywas not designed to compare efÞcacy of GLVs incombination with verbenone, so that aspect is notdiscussed further in this study. Because the Þrst studyfailed to show sufÞcient added tree protection byusing widely spaced (47.2 m) perimeter traps baitedwithD. ponderosae aggregation pheromones, we usedsmaller plots with closer trap spacing in the secondstudy. The Þrst study was conducted with Þve repli-cates, but funding limitations constrained us to threereplicates in the second study. Beetle populationswere at full outbreak levels in California and early-outbreak levels in Washington, with 44% of trees incontrol plots killed in the California study, and 4%(nearly 10 trees/ha) in control plots in the Washing-ton study.Semiochemical Formulations.The verbenone (4, 6,

6-trimethylbicyclo(3.1)hept-3-en-2-one) formulationused in California in 2008 was DISRUPT Micro-FlakeVerbenone Bark Beetle Anti-Aggregant ßakes (Her-con Environmental, Emigsville, PA). These consistedof verbenone-releasing ßakes, 3.2 by 3.2 mm2, formu-lated to contain �15% verbenone in a central layer ofplastisol bounded by two thin layers of polymer lam-inate. Thus, 1 kg of ßakes contained �150 g of ver-benone. In Washington in 2010, we used a combina-tion of DISRUPT Bio-Flake Verbenone Bark BeetleAnti-Aggregant ßakes and DISRUPT Bio-Flake GLVBark Beetle Anti-Aggregant ßakes (both made of abiodegradable polymer of roughly 3.2 by 3.2 mm2

ßakes, with � 15% active ingredient in the verbenoneformulation and � 10% of each of the active ingredi-ents, 1-hexanol (N-hexyl-alcohol) and Z-3-hexenol(cis-3-hexen-1-ol), in the GLV formulation). We ap-plied verbenone at the rate of 1,101 g/ha in California in2008. InWashingtonin2010weappliedverbenoneat therate of 741 g/ha and GLVs at 370 g/ha (Table 1).Study Site, Experimental Design, and PheromoneApplication (California 2008). The study was con-ducted on the Goosenest Ranger District of the Klam-ath National Forest in Siskiyou County, CA (41�50�26.849� N, 122� 13�29.25� W, elevation 1,330 m) in astand consisting primarily of lodgepole pine with anadmixture of ponderosa pine. We selected Þfteen4.04-ha plots, at least 500 m apart, with similar standand stocking levels and similar rates of D. ponderosaeinfestation. We then randomly assigned each of thethree treatments (push-only, pushÐpull, and control)(Table 1) to one third of the plots, yielding Þve rep-

licates per treatment (see Fig. 1 for a schematic dia-gram of the pushÐpull treatment). Both push-only andpushÐpull treatments received the “push” (ver-benone) treatment, and the pushÐpull treatment hadan additional “pull” treatment consisting of trapsbaited with D. ponderosae aggregation pheromone.The push-only treatment consisted of an aerial appli-cation of MicroFlake Verbenone on 23 June 2008,using methods described in Gillette et al. (2009). ThepushÐpull treatment consisted of identical pheromoneapplications followed by the installation of a perimeterline of Intercept panel traps (Advanced PheromoneTechnologies, Marylhurst, OR) baited with the four-component D. ponderosae aggregation pheromone, ablend of trans-verbenol, exo-brevicomin, myrcene,and terpinolene (ConTech International, Inc., Delta,BC, Canada). The perimeter traps were spaced 45.7 mapart. We established a 2.02-ha core plot nested withineach treated 4.04-ha treated or control plot for mea-surements of stand structure and beetle attack rates, tominimize inßuence of edge effects on those variables(Fig. 1).Study Site, Experimental Design, and PheromoneApplication (Washington 2010). This study was con-ducted on the Chelan Ranger District of theOkanogan-Wenatchee National Forest (48� 39�36.50�N, 119� 55�33.57� W) in a stand consisting primarily ofwhitebark pine with an admixture of lodgepole pine,subalpine Þr, western larch, and Engelmann spruce.We selected nine 0.81-ha plots, at least 400 m apart,attempting to locate plots with similar stand stockinglevels and similar rates of D. ponderosae infestation.We then randomly assigned each of the three treat-ments (push-only, pushÐpull, and control) to onethird of the plots (see Fig. 2 for a schematic diagramof the pushÐpull treatment), for three replicates pertreatment in total. As above, both push-only and pushÐpull treatments received the “push” (verbenone)treatment, and the pushÐpull treatment had an addi-tional “pull” treatment consisting of traps baited withD. ponderosae aggregation pheromone. Biodegradableßakes were applied on 22 July 2010 by a Þve-personcrew using broadcast spreaders with slot augers cali-brated to dispense evenly and at the desired rate as asimulated aerial application (methods described indetail in Gillette et al. 2012). The perimeter traps werespaced 14.4 m apart. We established a 0.40-ha core plotnested within each treated 0.81-ha plot for measure-ments as described below (Fig. 2).

Table 1. Summary description of treatments conducted at Klamath National Forest, CA in 2008 and Okanogan-Wenatchee NationalForest, WA in 2010

Location, yr, andtreatment (code)

Anti-aggregation semiochemical(s), rate, andtype of application (aerial or ground)

Plot size Trap spacing

CA/2008 (CT) None 4.04 ha 45.7 mCA/2008 (PO) Verbenone, 1,101 g/ha (aerial) 4.04 ha 45.7 mCA/2008 (PP) Verbenone, 1,101 g/ha (aerial) 4.04 ha 45.7 mWA/2010 (CT) None 0.81 ha 14.4 mWA/2010 (PO) Verbenone, 741 g/ha � GLVs, 370 g/ha (ground) 0.81 ha 14.4 mWA/2010 (PP) Verbenone, 741 g/ha � GLVs, 370 g/ha (ground) 0.81 ha 14.4 m

CA is California, WA is Washington, CT is control, PO is push-pnly, PP is pushÐpull, GLV is green leaf volatiles.

December 2012 GILLETTE ET AL.: EVALUATION OF PUSHÐPULL TACTIC FOR D. ponderosae 1577

Beetle Flight, Stand Structure, and Beetle AttackRate Measurements. Four Intercept panel traps wereinstalled in each plot immediately after pheromoneapplications to monitor beetle ßight into treated anduntreated plots. In an effort to avoid the potentialconfounding effect of inducing D. ponderosae attack

on nearby pines by the baited traps (and the conse-quent release of natural beetle pheromone by attack-ing beetles) the traps were suspended on nonhosttrees or shrubs as far away from hosts as possible (nocloser than 5 m). The traps were baited with D. pon-derosae aggregation pheromone, a two-part blend of

Fig. 1. Plot lay-out for pushÐpull treatments, Klamath National Forest, CA, in 2008. Open stars indicate traps baited withfour-componentD. ponderosae aggregation pheromone, and black stars indicate monitoring traps baited with two-componentD. ponderosae aggregation pheromone. Central square indicates core plot for measurements of D. ponderosae attack rates(2007 and 2008) and stand structure and composition.

Fig. 2. Plot lay-out for pushÐpull treatments, Okanogan-Wenatchee National Forest, WA, in 2010. Open stars indicatetraps baited with four-componentD.ponderosae aggregation pheromone, and black stars indicate monitoring traps baited withtwo-componentD. ponderosae aggregation pheromone. Central square indicates core plot for measurements ofD. ponderosaeattack rates (2009 and 2010) and stand structure and composition.


trans-verbenol and exo-brevicomin (ConTech Inter-national, Inc.). In an attempt to reduce the releaserates of the lures and the associated potential forbeetle attack on nearby host trees, bait componentswere placed in a semipermeable zip-lock sandwichbag with half the surface area covered with Mylar tape(3M Stationery Products Division, St. Paul, MN). Col-lection cups attached to the trap bottoms containedVaportape insecticide-releasing strips (Hercon En-vironmental, Emigsville PA) to reduce losses of re-sponding D. ponderosae to trapped predators.Trapped insects were collected weekly for 10 wkafter application and were shipped to the Universityof California, Berkeley, for identiÞcation to specieslevel. Voucher specimens were deposited in theEssig Museum of Entomology, University of Cali-fornia, Berkeley.

Stand characteristics, including posttreatment bee-tle attack and previous-year host mortality, were mea-sured on 15 September 2008 in California, and on12Ð13October2010 inWashington.All live trees �10.2cm in diameter at breast height were tallied by speciesand measured for diameter at breast height. Attackrates were recorded for all host trees attacked by D.ponderosae in 2007Ð2008 in California and 2009Ð2010in Washington. Attacked trees were classiÞed by yearof attack (current: pitch tubes and boring dust onotherwise green trees, immature brood; previous: nee-dles ranging from pale green to brilliant orange, oftenwith mature brood; older: needles ranging from dullorange to fallen, no live brood) and type of attack(mass: circumferential attack, strip: attack that doesnot girdle the tree, or pitch-out: unsuccessful attack).We assumed that mass-attacked trees ultimatelywould be killed by the beetle attack because they arecompletely girdled; this attack density thereforeserves as a surrogate for tree mortality. Beetle popu-lation trends commonly are assessed using the ratio of“green-attacked” (attacked in current year, foliage stillgreen) to “red-attacked” (attacked in the previousyear, with foliage discolored), because of the over-riding importance of infested brood trees to the risk ofcurrent-year D. ponderosae attack in a given locality(Wulder et al. 2009). This ratio was used to assesstrends in beetle attack rates in the analyses.Statistical Analysis. Tree attack rates and beetle

capture rates, like most counts data, are typically over-dispersed Poisson-distributed (i.e., the variance is pro-portional to the mean, and there are many cells withlow counts) (McCulloch and Searle 2001). Newersoftware packages enable analysis of this type of datausing the Poisson and overdispersed Poisson distribu-tions (e.g., SAS version 9.2, Cary, NC), with this ca-pability it is particularly important to account for theoverdispersion that arises in most counts data as arandom effect in the statistical models used (Wartonand Hui 2011). We therefore examined each set ofresponse data to assess whether it Þt the overdispersedPoisson distribution, and because all of them did Þtthat distribution, we used it in the analytical modelsfor assessment of treatment effects shown below. Forthe analysis of tree attack rates, it also has been shown

to be important to include the previous-year attackrate as a covariate in the analysis because it serves asa measure of emerging beetle populations that areattacking nearby trees in the current year, and thus itincreases the precision of the estimates (Wulder et al.2009).Tree Attack in California.We tested several over-

dispersed Poisson regression models from the familyof Generalized Linear Models (GLM, McCulloch andSearle 2001) to estimate the proportion of attackedtrees in 2008 for three treatment levels (Control,PushÐpull, and Push-Only). The best model, based onthe Akaike AIC diagnostic (Burnham and Anderson1998), has treatment effect and percent infested treesin the prior untreated year (2007) as Þxed effects. Itis usually important to include the prior year infesta-tion rate in such models because it serves as an indi-cator of the numbers of beetles emerging and attack-ing trees in the current year, which can vary amongplots (Wulder et al. 2009), so using it as a covariateincreases precision of the estimates. The Poissonmodel is as follows:

Expected[#2008 attacked tressij]

#2008 target treesij

� eTi� a*log(rate2007ij) � �ij [1]

Where target trees are deÞned asD.ponderosaehosttrees of a size to be susceptible to attack; i� 1, 2, 3, fortreatment; 1 � control, 2 � PushÐpull, 3 � Push-Only;j � 1, 2, 3, 4, 5 for Þve plots per treatment; #2008attacked trees � number of target trees attacked in2008; #2008 target trees � number of target treesin 2008; rate2007 � proportion of target trees attackedin the prior year; � is the overdispersion error toaccount for plot effect. We assumed #2008 attackedtrees for a given plot to have the Poisson distributionwith expected value given by (1). We used the SAS(version 9.2) GENMOD procedure to estimate theparameter Ti and the coefÞcient a. We used theLikelihood Ratio test with the Bonferroni adjust-ment to test pairwise comparisons between thetreatment levels.Beetle Trap Catch in California. A Mixed Gener-

alized Linear model for overdispersed Poisson re-sponses with plot as a random effect also was used tomodel the number of beetles per trap placed at thefour corners of each site to compare the three treat-ments levels at each weekly period with the Likeli-hood Ratio test. SAS GENMOD procedure and theBonferroni adjustment to test pairwise comparisonsbetween the treatment levels were used for the esti-mation and comparison tests.Statistical Model.

Expected[2008 beetle countijkl]

� eTreati*Datej� �kl [2]

Where 2008 beetle countijk is the number of beetlestrapped in 2008 on corner l of plot k, for treatmentlevel i on day j (j � 1,2,.,5 sampling periods). Thebeetle count in plot k is assumed to have an overdis-


persed Poisson distribution with expected value givenby (2).Tree Attack in Washington. We used the overdis-

persed Poisson regression from the family of GLMmodels to estimate the ratio of number of attackedtrees in 2010/number of attacked trees in 2009 forthree treatment levels (Control, PushÐpull, and Push-only). After an exploratory analysis using the AkaikeAIC diagnostic, the best Poisson model was as follows:

Expected [#2010 attacked treesij]

rate2009ik*#2010 target treesik� eTi� �ik [3]

where i � 1, 2, 3, for treatment: 1 � control, 2 �Pull-push, 3 � Push-only; j � 1Ð3, for three plots pertreatment; #2010 attacked trees � number of targettrees attacked in 2010; #2010 target trees � number oftarget trees in 2010; rate2009 � proportion of targettrees attacked in the prior year; � is the overdispersionerror to account for plot effect. We assume #2010attacked trees for a given plot to have the Poissondistribution with expected value given by (3). Weused SAS (version 9.2) GENMOD procedure to esti-mate the parameter Ti.We used the Likelihood Ratiotest with Bonferroni adjustment to test pairwise com-parisons between the treatment levels.Beetle Trap Catch in Washington. A Mixed GLM

model for overdispersed Poisson with plot as a randomeffect also was used to model the number of beetlesper trap placed at the four corners of each site tocompare the three treatments levels at each weeklyperiod with the Likelihood Ratio test. We used SAS

GENMOD procedures and the Bonferroni adjustmentto test pairwise comparisons between the treatmentlevels for the estimation and comparison tests.Statistical Model.

Expexted[2010 beetle countijkl]

� eTreati*Datej� �kl [4]

Where 2010 beetle countijk is the number of beetlestrapped in 2010 on corner l of plot k, for treatmentlevel i on day j (j � 1, 2,É, Þve sampling periods). Thebeetle count in plot k is assumed to have an overdis-persed Poisson distribution with expected value givenby (4).

Results

California.None of the stand structure or previous-year D. ponderosae attack rate measurements weresigniÞcantly different among treatments (Table 2,Suppl. Table 1) although in some cases the differenceswere rather large in absolute terms. PushÐpull signif-icantly reduced the number of beetles trapped in coreplots as compared with the controls (Figs. 3 and 4),and although the difference between push-only andthe control was not signiÞcant, it was substantial inabsolute terms. The difference in trap catch betweenpushÐpull and push-only was not signiÞcant at � �0.05. At peak beetle ßight (Fig. 3), traps in controlplots caught �350 D. ponderosae beetles, whereaspush-only caught 104 beetles and pushÐpull caught 25

Table 2. Summary of stand conditions at time of treatment in 2008 and 2007 attack rates, Klamath National Forest, CA in 2008

TRMean stems/ha, all spp.,

2008

Mean stems/ha, host spp.,

2007


2008

MeanBA all spp.,2008 m2/ha

MeanBA host spp.,2008, m2/ha

Mean %mass � stripattack, 2007


Mean DBH,all spp.,2008, cm

Mean DBH,host spp.,2008, cm

CT 259.3 183.8 158.6 16.2 9.8 37.8 55.7 26.0 26.6PP 339.8 167.2 137.8 26.9 8.8 39.6 31.1 28.1 28.2PO 271.0 143.8 127.1 18.6 6.5 26.5 19.2 26.8 24.1

TR is treatment, BA is basal area, DBH is diameter at breast height, CT is control, PO is push-only, PP is pushÐpull.

Fig. 3. Number of D. ponderosae beetles collected in monitoring traps (Fig. 1), Klamath National Forest, CA, in 2008.Diamonds indicate responses in control (CT) plots, triangles indicate responses in Push-Only (PO) plots, and circles indicateresponses in PushÐpull plots (PP). Vertical bars indicate standard errors.


beetles. The rate of D. ponderosae attack in 2007 wasa signiÞcant variable in the model (Table 3), as weexpected. The 2008 D. ponderosaemass � strip attackrates for both pushÐpull and push-only were signiÞ-cantly lower, using the Bonferroni approach to main-tain an experiment-wise error rate of 0.05, than that incontrols (Table 4; Fig. 5), and the two pheromonetreatments were not signiÞcantly different from oneanother. The 2007 versus 2008 attack rates per plot(Fig. 5), which is essentially the ratio of green-at-tacked:red-attacked trees for a range ofD. ponderosaeattack rates, indicates that at previous-year infestationlevels of up to 40% of host trees, both treatmentsreduce D. ponderosae attack rate. Push-only was sig-niÞcantly lower than pushÐpull, however, effectivelyreducing attack rate by �50% even when nearly halfthe host trees were mass-attacked the previous year.Washington. The stand conditions and previous-

year attack rates were not signiÞcantly differentamong treatments (Table 5, Suppl. Table 2), but thedifferences were large even if not signiÞcant. These

differences may explain the signiÞcant plot effects forall treatments (Table 6). Both pushÐpull and push-only signiÞcantly reduced the number of beetlestrapped in core plots compared with the control plots(Figs. 6 and 7), and there was no signiÞcant differencebetween the two pheromone treatments. We did notdemonstrate any signiÞcant differences inD. pondero-sae attack rates among treatments (Fig. 8), possiblybecause of excessive heterogeneity in stand condi-tions and previous-year D. ponderosae attack ratesamong treatments (Table 4, Suppl. Table 2), whichwas unavoidable with limited replication. Neverthe-less, the D. ponderosae attack rate in the push-onlytreatment was, overall, about one third lower than thatin either the pushÐpull or the control treatment.

Discussion

We conÞrmed the efÞcacy of DISRUPT Micro-Flake Verbenone Bark Beetle Anti-Aggregant ßakesalone (in contrast to the pushÐpull treatment) to re-duce D. ponderosae damage, even at the very highbeetle population levels that occurred in California.Although the pushÐpull approach clearly reducednumbers of beetles trapped in treated plots in bothCalifornia and Washington, and push-only did so as

Table 4. Comparison of attack rates among various treat-ments, Klamath National Forest, CA in 2008. Values of P lowerthan 0.0167 are significant at a Bonferroni-adjusted exp-wise errorrate of 0.05

Comparison Ratio 95% lower CL 95% upper CL P value

CT vs PP 1.573 1.117 2.215 0.0096CT vs PO 2.256 1.505 3.381 �0.0001PP vs PO 1.435 0.919 2.240 0.112

CT is control, PO is push-only, PP is pushÐpull.

Fig. 4. Comparison of season-long D. ponderosae catches by treatment, Klamath National Forest, CA, in 2008, with 95%conÞdence intervals. Bars with the same letter are not signiÞcantly different at an experiment-wise error rate of 0.05 (CT,Control; PO, Push-Only; PP, PushÐpull).

Table 3. Regression coefficients, Klamath National Forest, CAin 2008

Parameter

CoefÞcient (log scale)

Estimate95% lower

bound95% upper

boundP value

CT 0.672 1.089 0.255 0.0016PP 1.125 1.562 0.687 �0.0001PO 1.486 2.008 0.963 �0.0001Log (proportion of

trees mass � stripattacked, 2007)

0.210 0.001 0.419 0.049

Overdispersion 2.479

P value lower than 0.0167 indicates that the actual mean attack rateis signiÞcantly greater than zero; CT is control, PO is push-only, PPis pushÐpull.


well in Washington, in neither case did the addition ofa perimeter row of attractant-baited traps signiÞcantlyreduce the rate of attack below that in plots treatedonly with antiaggregants. In fact, in California, theaddition of perimeter traps appears to have increasedthe rate of attack in pushÐpull treatments, even thoughthe number of beetles trapped was slightly lowerin the pushÐpull treatment than in the push-only treat-ment. We speculate, however, that the slightly higherattack rate in the pushÐpull treatment was a result ofthe slightly higher number of large trees, higher basalareas, and higher brood tree levels (i.e., 2007-attackedtrees) in the pushÐpull plots (Table 2) rather than anyresult of the pheromone treatment itself.

In Washington, beetle numbers were signiÞcantlyreduced by both treatments, but rate ofD. ponderosaeattack on host trees was not. The rate ofD. ponderosaeattack in push-only treated plots in Washington, al-though not signiÞcantly different from either the con-trol or the pushÐpull treatment, was only about twothirds the rate in pushÐpull and control plots, whereasin pushÐpull plots it was actually slightly (but notsigniÞcantly) higher than in the controls. These Þnd-ings raise questions about the effect of such spatialattributes as plot size, trap spacing, and potential edgeeffects, considering the large difference (a factor of 5)in plot sizes in the two experiments. The difference intrap spacing was more than three-fold, and we hadexpected to see greater efÞcacy with closer trap spac-

ing, but in fact efÞcacy was worse in the study with thecloser spacing, both in terms of beetles trapped andtrees attacked. This outcome suggests that closer spac-ing of perimeter traps may, in fact, attract greaternumbers of beetles into plots rather than simply trap-ping out larger numbers of beetles. Similarly, thesmaller plots likely had larger edge effects (proportionof the plot interiors affected by conditions at plotperimeters) than the larger plots. All of these spatialvariables may have played a role in the beetle trapcatch and attack rate results.

To our knowledge, all of the previous publishedwork with the pushÐpull approach for control of D.ponderosae has been conducted using baited treesrather than baited traps as the attractant treatment(Lindgren and Borden 1993, Borden et al. 2006). Thatapproach, commonly referred to as “concentrationand containment,” is a subset of the pushÐpull ap-proach in which the pest insects are attracted to theirhost plants, attack them, and then are killed when thehost plant is harvested with the insects still withinplant tissues. It is, in effect, an attract-and-kill ap-proach, and was shown to be quite effective whencombined with an antiaggregation pheromone andremoval of infested brood trees (Lindgren and Borden1993, Borden et al. 2006). Although this approach hasbeen shown to work quite reliably (although not in-fallibly), it is appropriate only in lands that are activelymanaged using silvicultural treatments that are incor-

Fig. 5. Attack rate in 2008 as a function of attack rate in 2007 by treatment, with 95% conÞdence intervals (CT, Control;PO, Push-Only; PP, PushÐpull).

Table 5. Summary statistics for stand conditions and 2009 attack rates, Okanogan-Wenatchee National Forest, WA, in 2010

TRMean stems/ha, all spp.,

2010


2009


2010

MeanBA all spp.,m2/ha 2010

MeanBA host spp.,m2/ha 2010



MeanDBH all spp.,

2010, cm

MeanDBH host spp.,

2010, cm

CT 357.5 313.8 292.4 171.4 143.7 23.1 9.9 23.9 24.3PP 374.8 333.6 322.1 172.4 153.8 14.0 7.4 25.0 25.5PO 327.0 281.7 275.1 181.8 158.6 6.6 2.5 24.4 25.1

TR is treatment, BA is basal area, DBH is diameter at breast height, CT is control, PO is push-only, PP is pushÐpull.


porated into pest management implementation. Formany public lands in the United States, regulatoryconstraints make this approach difÞcult to deploy be-cause of the need to sacriÞce some of the living treesin a stand to contain beetle populations and preventthem from attacking other nearby trees.

Our study was designed to test the hypothesis thatbaited traps deployed in a perimeter line (surroundingthe stand to be protected) might increase the efÞcacyof antiaggregation pheromone alone, thus achievinglevels of control equivalent to those using the con-tainment-and-concentration approach in a pushÐpulltactic. Our study design differed in other ways fromthose of previous studies insofar as those were de-signed to pull beetles into a central, sacriÞcial group oftrees (concentration) rather than inhibiting themfrom entering the stand of interest. In either approach,the “pull” treatment risks attracting too many beetlesinto the area, with resulting spill-over into the standswe wish to protect. It appears that in our two projectsthe scale of the treatments (both plot size and trapspacing) was not adequate for effective trap-out andconsequent tree protection. In addition, we speculatethat baited traps may simply be insufÞcient to attracthost-seeking beetles and to interrupt their response toliving host trees, because other studies have shownsufÞcient interruption by using baited trees (Lindgrenand Borden 1993, Borden et al. 2006). However, we

speculate that baited traps may possibly trap beetlesover a longer period than baited trees, because trapsdo not trigger release of antiaggregation pheromone asdo attacked trees (Wood et al. 1985).

Another question that deserves further investiga-tion is whether, by treating some stands with antiag-gregation pheromones, we are merely herding beetlesonto adjacent untreated stands. Evidence from elec-troantennogram studies coupled with observations ofßight behavior shows that exposure to verbenone re-duces muscle potential in Dendroctonus frontalisZimm. (Dickens and Payne 1978), perhaps explainingthe arrestment of ßying beetles near traps baited withboth attractants and verbenone. McPheron et al.(1997) also demonstrated that verbenone exposuresigniÞcantly reduced the walking speed of anotherbark beetle species, Ips paraconfusus Lanier, in labo-ratory olfactometer tests. Taken together, these re-sults support the hypothesis that verbenone may havea generalized effect on beetle locomotion that couldresult in arrestment rather than repellency. Many an-ecdotal observations of stands adjacent to treated ar-eas support the supposition that treatments do notsimply exacerbate beetle-caused tree mortality in ad-jacent stands (J.N.W., unpublished data). Whateverthe underlying behavioral mechanism, it is clear frommyriad studies that exposure to verbenone reducesthe ability of D. ponderosae to aggregate in sufÞcientnumbers to successfully attack its host trees (Bentz etal. 2005; Gillette et al. 2006, 2009, 2012). It would,nevertheless, be of interest to spatially characterizethe behavioral responses of verbenone-exposed bee-tles in the Þeld, and to determine whether they are stillcapable of aggregating in adjacent areas after exposureto verbenone-treated plots.

In California, the verbenone-releasing ßakes con-sistently reduced D. ponderosae attack rates byroughly 50%, a level that is consistent with long-termD. ponderosae damage control (Coggins et al. 2011).Wulder et al. (2009), using brood tree removal as theonly treatment, attempted to deÞne a D. ponderosae

Table 6. Estimates of regression coefficients for attack rates,Okanogan-Wenatchee National Forest, WA in 2010

Treatments

CoefÞcient

Estimate95% lower

bound95% upper

boundP value

CT 0.039 0.018 0.082 �0.0001PP 0.022 0.011 0.041 �0.0001PO 0.009 0.002 0.0387 �0.0001

P value lower than 0.0167 indicates that the actual mean is signif-icantly greater than zero; CT is control, PO is push-only, PP is pushÐpull.

Fig. 6. Number of D. ponderosae beetles collected in monitoring traps (Fig. 2), Okanogan-Wenatchee National Forest,WA, in 2010. Squares indicate responses in control (CT) plots, triangles indicate responses in PushÐpull plots (PP), and circlesindicate responses in Push-Only (PO) plots. Vertical bars indicate standard errors.


population level that represents a rough threshold orplateau beyond which treatments could be consideredfutile. Following on that work, Coggins et al. (2011)showed that if theD. ponderosae attack rates in treatedstands can be kept below �50%, long-term control ofbeetle damage is feasible. Thus, our Þndings supportverbenone-based intervention to protect lodgepoleand whitebark pine stands fromD. ponderosae-causedtree mortality. We recommend that future tests in-corporate dose-range tests (Miller et al. 1995) andadditional semiochemical components (Kegley andGibson 2009, Kegley et al. 2010) to achieve greaterefÞcacy.

In the face of the epicD. ponderosae outbreaks seenin western North America for the last 10 yr, there hasbeen much discussion regarding the presumed futilityof conducting any treatments at all, whether based onsilvicultural prescriptions, sanitation (brood tree re-

moval), or semiochemical approaches. Our work,along with that of others, shows that these approachescan indeed mitigate losses to D. ponderosae and pre-vent the immediate and complete loss of stands. Thereremains the question of whether cumulative losses,over periods of years or decades, will eliminate naturalstands, result in species conversions in western NorthAmerica, or both, especially under the pressures ofclimate change. Some research suggests that suscep-tibility increases with host density (summarized inAmman and Logan 1998), presumably because thedenser stands suffer more stress, whereas other re-search suggests that susceptibility decreases with hostdensity (Borden et al. 2006, 2007), perhaps becausethe ratio of attacking D. ponderosae to target hosts isdiluted in denser stands. Fettig et al. (2007) reviewedthe existing literature on the effect of silviculturalmanipulations on stand susceptibility to bark beetles,

Fig. 7. Comparison of season-long D. ponderosae catches by treatment, Okanogan-Wenatchee National Forest, WA, in2010, with 95% conÞdence intervals. Bars with the same letter are not signiÞcantly different at an experiment-wise error rateof 0.05 (CT, Control; PO, Push-Only; PP, PushÐpull).

Fig. 8. Ratio of attacks in 2010 to attacks in 2009 (“green-attacked : red-attacked ratio”), with 95% conÞdence intervals;Okanogan-Wenatchee National Forest, WA. Means followed by the same letter are not signiÞcantly different at an exper-iment-wise error rate of 0.05 (CT, Control; PO, Push-Only; PP, PushÐpull).


and concluded that the preponderance of evidencesupports the assumption that wider tree spacing inlodgepole pine stands will help to “beetle-proof” them.To the extent that susceptibility to D. ponderosae de-creases with reduced stand density, it is possible thatby slowing the rate of beetle-caused mortality belowthe critical threshold described by Coggins et al.(2011), we could maintain stands at more beetle-re-sistant densities until the system equilibrates at a levelthat can be sustained for a longer period. In so doing,we would be harnessing the capacity of bark beetlesto thin stands, at a more moderate rate than typicallyoccurs in over-stocked stands subjected to long-termÞre-suppression.

Acknowledgments

This study was funded by two grants from the USDAForest Service Pesticide Impact Assessment Program to NEGand CJM. We thank: Roger Siemers (USDA FS, KlamathNational Forest) and Mallory Lenz and Randy Niman(USDA FS, Okanogan-Wenatchee National Forest) for as-sistance with plot location; Elizabeth Gonzales and JoyceWong (University of California, Berkeley) for identifying andcounting beetles; Roy Magelssen and Darci Carlson (USDAForest Service, Forest Health Protection, Wenatchee, WA) forassistance with Þeld work; and Gaylord Briggs and Eric Web-ster (Webster Forestry Consulting, Redding, CA) for assis-tance with Þeld work and data entry. Christopher Fettig(USDA FS, PaciÞc Southwest Research Station, Albany, CA),Dan Miller (USDA FS, Southern Research Station, Athens,GA) and Brice McPherson (University of California, Berke-ley, CA) provided very helpful reviews. We thank HankAppleton, USDA FS Forest Health Protection, Washington,DC, for sustained support of bark beetle semiochemical re-search.

References Cited

Amman,G.D., and J.A.Logan. 1998. Silvicultural control ofmountain pine beetle: prescriptions and the inßuence ofmicroclimate. Am. Entomol. 44: 166Ð178.

Bentz,B. J., andG. Schen-Langenheim. 2007. The mountainpine beetle and the whitebark pine waltz: has the musicchanged? pp. 43Ð50. In Whitebark pine: a PaciÞc coastperspective. R6-NR-FHP-2007Ð01. U.S. Department ofAgriculture Forest Service, Portland, OR.

Bentz, B. J., S. Kegley, K. Gibson, and R. Thier. 2005. A testof high-dose verbenone for stand-level protection oflodgepole and whitebark pine from mountain pine beetle(Coleoptera: Curculionidae: Scolytinae) attacks. J. Econ.Entomol. 98: 1614Ð1624.

Borden, J. H. 1997. Disruption of semiochemical-mediatedaggregation in bark beetles, pp. 421Ð438. In R. T. Cardeand A. K. Minks (eds.), Insect pheromone research: newdirections. Chapman & Hall, New York.

Borden, J. H., L. J. Chong, and C. M. Fuchs. 1983a. Theapplication of semiochemicals in post-logging manipula-tion of the mountain pine beetle, Dendroctonus pondero-sae (Coleoptera: Scolytidae). J. Econ. Entomol. 76: 1428Ð1432.

Borden, J. H., L. J. Chong, K.E.G. Pratt, and D. R. Gray.1983b. The application of behavior-modifying chemicalsto contain infestations of the mountain pine beetle,Den-droctonus ponderosae. For. Chron. 59: 235Ð239.

Borden, J. H., L. J. Chong, T. J. Earle, and D.P.W. Huber.2003. Protection of lodgepole pine from attack by themountain pine beetle, Dendroctonus ponderosae (Co-leoptera: Scolytidae) using high doses of verbenone incombination with nonhost bark volatiles. For. Chron. 79:685Ð691.

Borden, J. H., A. L. Birmingham, and J. S. Burleigh. 2006.Evaluation of the pushÐpull tactic against the mountainpine beetle using verbenone and non-host volatiles incombination with pheromone-baited trees. For. Chron.82: 579Ð590.

Borden, J. H., G. S. Sparrow, and N. L. Gervan. 2007. Op-erational success of verbenone against the mountain pinebeetle in a rural community. Arboric. Urban For. 33:318Ð324.

Burnham, K. P., andD. R. Anderson. 1998. Model selectionand inferences: a practical information-theoretical ap-proach, pp. 43Ð54. Springer, New York.

Coggins, S. B., N. C. Coops, M. A. Wulder, C. W. Bater, andS. M. Ortlepp. 2011. Comparing the impacts of mitiga-tion and non-mitigation on mountain pine beetle popu-lations. J. Environ. Manag. 92: 112Ð120.

Cook, S. M., Z. R. Khan, and J. A. Pickett. 2007. The use ofpushÐpull strategies in integrated pest management.Annu. Rev. Entomol. 52: 375Ð400.

Dickens, J. C., and T. L. Payne. 1978. Olfactory-inducedmuscle potentials in Dendroctonus frontalis: effects oftrans-verbenol and verbenone. Cell. Mol. Life Sci. 34:463Ð464.

Fettig, C. J., K. K. Allen, R. R. Borys, J. Christopherson, C. P.Dabney, T. J. Eager, K. E. Gibson, E. G. Hebertson, D. F.Long, A. S. Munson, et al. 2006. Effectiveness of bifen-thrin (onyx) and carbaryl (sevin SL) for protecting in-dividual, high-value conifers from bark beetle attack (Co-leoptera: Curculionidae: Scolytinae) in the westernUnited States. J. Econ. Entomol. 99: 1691Ð1698.

Fettig, C. J., K. D. Klepzig, R. F. Billings, A. S. Munson, T. E.Nebeker, J. F. Negron, and J. T. Nowak. 2007. The ef-fectiveness of vegetation management practices for pre-vention and control of bark beetle infestations in conif-erous forests of the western and southern United States.For. Ecol. Manag. 238: 24Ð53.

Furniss, R. L., and V. M. Carolin. 1977. Western forest in-sects. U.S. Dep. Agric. Forest Service Miscellaneous Pub-lication 273. U.S. Dep. Agric. Forest Service, Washington,DC.

Gibson, K. 2006. Mountain pine beetle conditions in white-bark pine stands in the greater Yellowstone ecosystem,2006. Forest Health Protection Report, Northern Region,U.S. Dep. Agric. Forest Service, numbered report 06Ð03,Missoula, MT.

Gibson, K., and S. Kegley, S. 2004. Testing the efÞcacy ofverbenone in reducing mountain pine beetle attacks insecond-growth ponderosa pine. Forest Health ProtectionReport, Northern Region, U.S. Dep. Agric. Forest Service.

Gillette, N. E., and A. S. Munson. 2009. Semiochemical sab-otage: behavioral chemicals for protection of westernconifers from bark beetles, pp. 85Ð109. In J. L. Hayes, andJ. E. Lundquist (compilers), The western bark beetleresearch group: a unique collaboration with ForestHealth ProtectionÑproceedings of a symposium at the2007 Society of American Foresters conference. Gen.Tech. Rep. PNW-GTR-784. Portland, OR. U.S. Depart-ment of Agriculture, Forest Service, PaciÞc NorthwestResearch Station.

Gillette, N. E., J. D. Stein, D. R. Owen, J. N. Webster, G. O.Fiddler, S. R. Mori, and D. L. Wood. 2006. Verbenone-releasing ßakes protect individual Pinus contorta trees


from attack by Dendroctonus ponderosae and Dendrocto-nus valens (Coleoptera: Curculionidae, Scolytinae). Ag-ric. For. Entomol. 8: 243Ð251.

Gillette, N. E., N. Erbilgin, J. N. Webster, L. Pederson, S. R.Mori, J. D. Stein, D. R. Owen, K. M. Bischel, and D. L.Wood. 2009. Aerially applied verbenone-releasing lam-inated ßakes protect Pinus contorta stands from attack byDendroctonus ponderosae in California and Idaho. For.Ecol. Manag. 257: 1405Ð1412.

Gillette, N. E., E. M. Hansen, C. J. Mehmel, S. R. Mori, J. N.Webster, N. Erbilgin, and D. L.Wood. 2012. Area-wideapplication of verbenone-releasing ßakes reduces mor-tality of whitebark pine, Pinus albicaulis, caused by themountain pine beetle, Dendroctonus ponderosae (Co-leoptera: Scolytinae), in Wyoming and Washington, USA.Agric. For. Entomol. (in press).

Gray, D. R., and J. H. Borden. 1989. Containment and con-centration of mountain pine beetle (Coleoptera: Scolyt-idae) infestations with semiochemicals: validation bysampling of baited and surrounding zones. J. Econ. En-tomol. 82: 1399Ð1405.

Kegley, S., and K. Gibson. 2004. Protecting whitebark pinetrees from mountain pine beetle attack using verbenone.Forest Health Protection Report, Northern Region, U.S.Dep. Agric. Forest Service, Missoula, MT.

Kegley, S., and K. Gibson. 2009. IndividualÐtree tests of ver-benone and green-leaf volatiles to protect lodgepole,whitebark and ponderosa pines, 2004Ð2007. ForestHealth Protection Numbered Report 09Ð03, U.S. Dep.Agric. Forest Service, Forest Health Protection, Missoula,MT.

Kegley, S.,K.Gibson,N.Gillette, J.Webster,L.Pederson, andS. Mori. 2010. Individual-tree tests of verbenone ßakes,verbenone pouches, and green leaf volatiles to protectlodgepole pines from mountain pine beetle attack. U.S.Dep. Agric. Forest Service, R-1 Forest Health Protectionnumbered report 10Ð02, U.S. Dep. Agric. FS FHP CoeurdÕAlene Field OfÞce, Coeur dÕAlene, ID.

Kegley, S., K. Gibson, J. Schwandt, andM.Marsden. 2003. Atest of verbenone to protect individual whitebark pinefrom mountain pine beetle attack. U.S. Dep. Agric. ForestService, Forest Health Protection Report, Northern Re-gion, U.S. Dep. Agric. Forest Service, Missoula, MT.

Kurz, W. A., C. C. Dymond, G. Stinson, G. J. Rampley, E. T.Neilson, A. L. Carroll, T. Ebata, and L. Safranyik. 2008.Mountain pine beetle and forest carbon feedback toclimate change. Nature 452: 987Ð990. doi:10.1038/nature06777.

Lindgren, B. S., and J. H. Borden. 1993. Displacement andaggregation of mountain pine beetles,Dendroctonus pon-derosaeHopkins (Coleoptera: Scolytidae), in response totheir antiaggregation and aggregation pheromones. Can.J. For. Res. 23: 286Ð290.

Logan, J. A., and J. A. Powell. 2001. Ghost forests, globalwarming, and the mountain pine beetle (Coleoptera:Scolytidae). Am. Entomol. 47: 160Ð173.

Lotan, J. E., and W. B. Critchfield. 1990. Pinus contortaDougl. ex. Loud, pp. 302Ð315. In R. M. Burns and B. H.Honkala (tech. coords.), Silvics of North America, vol. 1:conifers. Agriculture Handbook 654, USD Forest Service,Washington, DC.

McCulloch, C. E., and S. R. Searle. 2001. Generalized, lin-ear, and mixed models, pp. 15Ð26. Wiley, New York.

McPheron, L. J., S. J. Seybold, A. J. Storer, D. L. Wood, T.Ohtsuka, and I. Kubo. 1997. Effects of enantiomericblend of verbenone on response of Ips paraconfusus tonaturally produced aggregation pheromone in the labo-ratory. J. Chem. Ecol. 23: 2825Ð2839.

Miller, D. R., J. H. Borden, and B. S. Lindgren. 1995. Ver-benone: dose-dependent interruption of pheromone-based attraction of three sympatric species of pine barkbeetles (Coleoptera: Scolytidae). Environ. Entomol. 24:692Ð696.

Pease, C. M., and D. J. Mattson. 1999. Demography of theYellowstone grizzly bears. Ecology 80: 957Ð975.

Perkins, D. L., and D. W. Roberts. 2003. Predictive modelsof whitebark pine mortality from mountain pine beetle.For. Ecol. Manag. 174: 495Ð510.

Progar, R. A. 2005. Five-year operational trial of verbenoneto deter mountain pine beetle (Dendroctonus ponderosae;Coleoptera: Scolytidae) attack of lodgepole pine (Pinuscontorta). Environ. Entomol. 34: 1402Ð1407.

Robbins, C. T., C. C. Schwartz, K. A. Gunther, and C.Servheen. 2006. Grizzlybearnutritionandecology stud-ies in Yellowstone National Park. Yellowstone Sci. 14:19Ð26.

Schoettle, A. W., and R. A. Sniezko. 2007. Proactive inter-vention to sustain high-elevation pine ecosystems threat-ened by white pine blister rust. J. For. Res. 12: 3327Ð3336.

Schwandt, J. W. 2006. Whitebark pine in peril: a case forrestoration. U.S. Department of Agriculture Forest Ser-vice Report R1Ð06-28. (www.fs.fed.us/r1Ðr4/spf/fhp/publications).

Schwandt, J., and S. Kegley. 2004. Mountain pine beetle, blis-ter rust, and their interaction on whitebark pine at TroutLake and Fisher Peak in northern Idaho from 2001Ð2003.U.S. Department of Agriculture Forest Service, R-1 FHPReport 04Ð9. (www.fs.fed.us/r1Ðr4/spf/fhp/publications).

Shea, P. J. andM. Neustein. 1995. Protection of a rare standof Torrey pine from Ips paraconfusus, pp. 39Ð43. In Salomand Hobson (eds.), Application of semiochemicals formanagement of bark beetle infestations: proceedings ofan informal conference. Intermountain Research Station,INT GTR-318, U.S. Department of Agriculture ForestService, Ogden, UT.

Tomback, D. F., S. F. Arno, and R. E. Keane. 2001. White-bark pine communities, ecology and restoration. IslandPress, Washington, DC.

Warton, D., and F. Hui. 2011. The arcsine is asinine: theanalysis of proportions in ecology. Ecology 92: 3Ð10.

Wood, D. L., R. W. Stark, W. E. Waters, W. D. Bedard, andF. W. Cobb. 1985. Treatment tactics and strategies, pp.121Ð139. In W. E. Waters, R. W. Stark, and D. L. Wood(eds.), Integrated pest management in pine-bark beetleecosystems. Wiley, NY.

Wood, D. L., T. W. Koerber, R. F. Scharpf, and A. J. Storer.2003. Pests of the native California conifers. University ofCalifornia Press, Berkeley, CA.

Wulder, M. A., S. M. Ortlepp, J. C. White, N. C. Coops, andS. B. Coggins. 2009. Monitoring the impacts of mountainpine beetle mitigation. For. Ecol. Manag. 258: 1181Ð1187.

Received 8 December 2011; accepted 20 August 2012.


chemical ecology the push–pull tactic for mitigation of ... · science, policy, and management,...

Documents