Article Post-Emergent Control of Nuisance Cones in Fraser Christmas Plantations

Bert Cregg 1,*, Dana Ellison 2 and Jill O’Donnell 3 1 Department of Horticulture and Department of Forestry, Michigan State University, 1066 Bogue Street, East Lansing Michigan, MI 48824, USA 2 Department of Horticulture, Michigan State University, 1066 Bogue Street, East Lansing Michigan, MI 48824, USA; [email protected] 3 Michigan State University Extension, 401 N. Lake Street, Suite 400, Cadillac, MI 49601, USA; [email protected] * Correspondence: [email protected]; Tel.: +1-517-353-0335

 Received: 29 March 2018; Accepted: 27 April 2018; Published: 28 April 2018  Abstract: Heavy cone production by Fraser fir Christmas requires significant labor inputs to remove nuisance cones. We conducted two field trials in collaboration with operational Christmas tree farms to evaluate the effectiveness of post-emergent herbicides to stop the development of newly emergent cone buds. In the first trial (2016), we applied six products (two conventional herbicides and four herbicides labeled for organic production) to trees using back-pack sprayers in operational plantations at four farms in Michigan. Three products; Scythe, Axxe, and Avenger, provided better cone kill than the other products but resulted in phytotoxicity at two locations. In 2017, we applied the three most effective products from the earlier trial at three farms either as single applications or as two applications approximately one week apart. We also evaluated a hand-held mechanical de-coning device at two farms. For all the products and the mechanical device, cone control in the 2017 trial was high (>80%). Phytotoxicity to foliage was low (mean rating, <0.3; 0 = none, 2 = severe) for single applications of the herbicides. Repeated applications increased cone control slightly but also increased risk for phytotoxicity. The mechanical device caused significant damage to shoots and foliage. We attribute the increased product effectiveness and reduced phytotoxicity between the 2016 and 2017 studies to improved coverage and earlier spray timing. Based on the current retail product cost, chemical cone control can be cost-effective compared to handpicking cones if trees have high numbers of cones that can take several minutes to remove. The effect of using surfactants and reducing product rates should be investigated along with mechanized application.

Keywords: coning; herbicides; cone picking; phytotoxicity; pelargonic acid

1. Introduction Precocious cone production is a major production issue for Fraser fir (Abies fraseri (Pursh) Poir.) Christmas tree plantations. In North America, Fraser fir trees begin to produce female cones (strobili) within four or five years after planting (Figure1). Fraser fir trees in native forests, in contrast, typically do not produce female cones until the age of 15 [1]. Cone production increases rapidly during the Christmas tree production cycle and trees that are 3 m tall can produce more than 1000 cones. Cones are a liability in Christmas trees because they disintegrate after maturing in the fall, leaving unsightly stalks that can render trees unsalable. Moreover, cones are a strong sink for photo-assimilates and reduce the amount of carbohydrates available for shoot and needle growth [2]. Because of the impacts of cones on Christmas tree quality and growth, growers pick female cones from trees shortly after cone buds emerge in the spring. Cone picking is labor intensive and results in one of the largest labor costs for Fraser fir Christmas tree producers in North America.

Forests 2018, 9, 233; doi:10.3390/f9050233 www.mdpi.com/journal/forests Forests 2018, 9, x FOR PEER REVIEW 2 of 12 cone buds emerge in the spring. Cone picking is labor intensive and results in one of the largest labor Forests 2018, 9, 233 2 of 12 costs for Fraser fir Christmas tree producers in North America.

Figure 1. Heavy cone production in a Fraser fir fir plantation in Michigan.

Research onon reducing reducing or or eliminating eliminating cone cone production production in Fraser in Fraser fir has fir focused has focused on (1) pre-emergent on (1) pre‐ (oremergent pro-active) (or pro strategies‐active) to strategies prevent treesto prevent from forming trees from cone forming buds and cone (2) buds post-emergent and (2) post (or‐emergent re-active) techniques(or re‐active) to eliminatetechniques cones to eliminate once they cones have once formed. they Pre-emergent have formed. approaches Pre‐emergent to reducing approaches coning to includereducing influencing coning include environmental influencing factors environmental that control factors coning that control and the coning application and the of application growth of regulatorsplant growth to reducingregulators coning to reducing [3]. For coning example, [3]. the For application example, the of paclobutrazol, application of apaclobutrazol, gibberellic acid a (GA)gibberellic inhibitor, acid reduced (GA) inhibitor, cone production reduced in cone Fraser production fir trees in Christmasin Fraser treefir trees plantations in Christmas in Michigan tree byplantations up to 54% in [4 ].Michigan However, by there up areto drawbacks54% [4]. However, to the pre-emergent there are drawbacks approach to to cone the reduction. pre‐emergent First, treesapproach must to be cone treated reduction. as buds First, are formed trees must along be new treated shoots. as buds All trees are informed plantations along new must shoots. be treated All becausetrees in plantations it is impossible must for be growerstreated because to identify it is which impossible buds for will growers be reproductive to identify and which which buds will will be vegetative.be reproductive Therefore, and which growers will will be likely vegetative. treat many Therefore, trees that growers would will not conelikely even treat without many treatment.trees that Second,would not paclobutrazol cone even treatments without treatment. can be expensive Second, as paclobutrazol require treatments relatively highcan be doses expensive in order as to induceconifers a require response, relatively and rates high increase doses asin treesorder grow to induce larger. a response, and rates increase as trees grow larger.In contrast to pre-emergent cone treatments, post-emergent treatments are intended as a substituteIn contrast for manually to pre‐ pickingemergent cones. cone An treatments, advantage post to post-emergent‐emergent treatments control isare that intended growers onlyas a needsubstitute to treat for treesmanually that producepicking cones. cones, An rather advantage than treating to post‐ allemergent trees as control in a pre-emergent is that growers strategy. only Chemicalneed to treat post-emergent trees that produce control cones, of Fraser rather fir conesthan treating might beall possibletrees as in because a pre‐emergent cone buds strategy. emerge twoChemical to three post weeks‐emergent prior control to vegetative of Fraser bud-break. fir cones might The asynchronybe possible because between cone cone buds bud emerge emergence two andto three vegetative weeks budprior break to vegetative provides abud window‐break. during The asynchrony which cones between are susceptible cone bud to emergence herbicides, and but existingvegetative foliage bud canbreak be provides relatively a tolerant. window Owenduring [5 which] conducted cones trialsare susceptible and found to that herbicides, up to 90% but of emergingexisting foliage cones can could be berelatively killed with tolerant. properly Owen timed [5] conducted herbicide sprays.trials and Owen found tested that productsup to 90% that of includedemerging conventional cones could herbicides,be killed with as well properly as herbicides timed herbicide listed for sprays. use in organicOwen tested production products systems. that Organic-registeredincluded conventional herbicides herbicides, might as be well well as suitedherbicides for this listed application for use in because organic they production have a relatively systems. lowOrganic overall‐registered toxicity. herbicides Many are naturalmight be products well suited such for as this citrus application oils or short-chain because they fatty have acids a thatrelatively work aslow desiccants overall toxicity. [6–9]. This Many suggests are natural they products could potentially such as citrus cause oils “burn-down” or short‐chain of newly fatty acids emerged that coneswork withas desiccants minimal [6–9]. damage This to suggests existing they needles could with potentially thick cuticles. cause “burn‐down” of newly emerged cones with minimal damage to existing needles with thick cuticles.

ForestsForests 20182018, ,99, ,x 233 FOR PEER REVIEW 3 3of of 12 12

The objective of the current project is to develop protocols for post‐emergent control of cones in FraserThe fir Christmas objective of tree the plantations. current project Specific is to develop objectives protocols of the current for post-emergent studies are controlto: (1) determine of cones in theFraser effectiveness fir Christmas of treetreatments plantations. on stopping Specific objectivescone development of the current and studies (2) evaluate are to: (1)the determine severity theof phytotoxicityeffectiveness ofon treatments needles and on shoots. stopping cone development and (2) evaluate the severity of phytotoxicity on needles and shoots. 2. Materials and Methods 2. Materials and Methods 2.1. 2016 Trial 2.1. 2016 Trial 2.1.1. Study Locations 2.1.1. Study Locations This trial included four locations and plots were installed throughout the state of Michigan. This trial included four locations and plots were installed throughout the state of Michigan. Three experiments were located at grower‐cooperator farms and one was located at the Michigan Three experiments were located at grower-cooperator farms and one was located at the Michigan State State University Forestry Tree Research Center in East Lansing (Figure 2). The treatment plots had University Forestry Tree Research Center in East Lansing (Figure2). The treatment plots had tree tree heights that ranged from 1.9 to 3.5 m. heights that ranged from 1.9 to 3.5 m.

Figure 2. Locations of cone herbicide plots installed throughout Michigan’s Lower Peninsula in 2016. Figure 2. Locations of cone herbicide plots installed throughout Michigan’s Lower Peninsula in 2016.

2.1.2.2.1.2. Study Study Design Design and and One One Treatments Treatments AtAt each each location, location, we we selected selected 140 140 trees trees with with at least at least 10 10cones cones per per tree. tree. Cone Cone sizes sizes ranged ranged from from 2.6 to2.6 3.6 to cm 3.6 at cm the at time the time of treatment. of treatment. The Thestudy study was wasinstalled installed as a ascompletely a completely randomized randomized design. design. At eachAt each farm, farm, treatments treatments were were assigned assigned to individual to individual trees trees at atrandom random by by selecting selecting color color-coded‐coded ribbons ribbons (one(one color color per per treatment) treatment) from from a a bucket. bucket. Herbicide Herbicide treatments treatments were were applied applied in in the the third third week week of of May. May. TemperatureTemperature and and wind wind speed speed data data were were compiled compiled from from the the Michigan Michigan State State University University Enviroweather Enviroweather stationsstations located located closest closest to to each each experiment experiment plot plot (Table (Table 11).).

TableTable 1. 1. Locations,Locations, dates, dates, daily daily maximum maximum temperatures, temperatures, and and wind wind speeds speeds at at time time of of post post-emergent‐emergent conecone herbicide herbicide application application at at four four sites sites in in Michigan Michigan in in 2016. 2016.

LocationLocation Application Application Date MaxMax AirAir TempTemp CC WindWind m/sm/s StationStation 11 GPSGPS Allegan, MI May 19 22.6 2.1 Allegan 42.5292° N, 85.8553° W Allegan, MI May 19 22.6 2.1 Allegan 42.5292◦ N, 85.8553◦ W Horton,Horton, MI MI May 18 19.119.1 2.12.1 Leslie Leslie 42.4514° 42.4514◦ N,N, 84.4325° 84.4325◦ WW Sidney,Sidney, MI MI May 17 18.518.5 2.32.3 Entrican Entrican 43.2925° 43.2925◦ N,N, 85.0814° 85.0814◦ WW EastEast Lansing, Lansing, MI MI May 13 17.117.1 4.74.7 MSU MSU-HTRC‐HTRC 42.6406° 42.6406◦ N,N, 84.5153° 84.5153◦ WW 1 Michigan1 Michigan State State University University Enviroweather Enviroweather NetworkNetwork Station. Station.

Forests 2018, 9, 233 4 of 12

Twenty trees were assigned at random to one of seven treatments: Avenger, Axxe, Goal, Reflex, Scythe, WeedZap, and control (Table2). Herbicides were sprayed to run-off with a backpack sprayer on the upper one-third of the tree where cones typically grow.

Table 2. Description of herbicide products and rates applied.

Product Manufacturer Description Type Rate Avenger Cutting Edge Formulations, Buford, GA d-limonene (citrus oil) Organic 250 mL/L Axxe BioSafe Systems, LLC, East Hartford, CT ammonium nonanoate (ammonium salt of pelargonic acid) Organic 100 mL/L Goal Dow AgroSciences LLC, Indianapolis, IN oxyfluorfen (pre/post emergent herbicide) Conventional 50 mL/L Syngenta Crop Protection, LLC, Reflex sodium salt of fomesafen (pre/post emergent herbicide) Conventional 20 mL/L Greensboro, NC Scythe Gowan Company, Yuma, AZ pelargonic acid Organic 50 mL/L WeedZap JH Biotech, Inc., Ventura, CA Cinnamon oil + clove oil Organic 50 mL/L Control na untreated untreated na

2.1.3. Evaluations Each tree was evaluated for cone kill and phytotoxicity (needle browning). The evaluations took place three to four weeks after the herbicide application. Cones were categorized as live (green and growing), damaged (brown or partially brown with evidence of growth), or dead (brown with no evidence of additional growth). Phytotoxicity to foliage for each was evaluated on a 0 to 3 scale: 0 = no damage, 1 = slight (minor tip browning of new growth), 2 = moderate (browning of new and old growth), and 3 = unacceptable (extreme browning of new and old shoot growth), and evaluated on needles in the top one-third of the tree where the herbicides were sprayed.

2.1.4. Statistical Analysis The effects of Farm, Treatment, and Farm × Treatment interaction were tested by analysis variance using the PROC GLM procedure in SAS (SAS Institute Inc., Cary, NC, USA). Where Treatment main effects were significant (p < 0.05), means were separated using Tukey’s studentized range test.

2.2. 2017 Trial

2.2.1. Study Locations The 2017 study had three locations installed on grower-cooperator farms throughout Michigan Forests 2018, 9, x FOR PEER REVIEW 5 of 12 (Figure3). The tree heights in the plot ranged from 2.0 to 4.0 m and there were ten or more cones per tree.

FigureFigure 3. Locations 3. Locations of cone of cone herbicide herbicide plots plots installed installed throughout throughout Michigan’s Michigan’s Lower Lower Peninsula Peninsula in 2017. in 2017.

2.2.2. Study Design and Cone Treatments At each location, we selected 160 trees with at least ten cones per tree. The study was installed as a completely randomized design. At each farm, treatments were assigned to individual trees at random by selecting color‐coded ribbons (one color per treatment) from a bucket. The three best performing products from the 2016 study (Avenger, Axxe, Scythe) were chosen for this study. Each product was initially applied to 40 trees at each farm (Table 3).

Table 3. Description of herbicide products and rates applied for the 2017 study.

Treatment Manufacturer Product Description Rate Avenger Cutting Edge Formulations, Avenger d‐limonene (citrus oil) 250 mL/L Avenger ×2 * Buford, GA BioSafe Systems, LLC, East ammonium nonanoate (ammonium Axxe Axxe ×2 Axxe 100 mL/L Hartford, CT salt of pelargonic acid) Scythe Scythe ×2 Gowan Company, Yuma, AZ Scythe pelargonic acid 50 mL/L Mechanical BlossomCinch, Manistee, MI na Mechanical cone removal na Control na na untreated na * Treatments denoted as ×2 were applied a second time, one to two weeks after initial application.

Half of the trees in each herbicide treatment were treated a second time (×2) approximately one week after the initial application. The rate of herbicide applied was the same for the single application and the repeated application. Herbicide treatments were sprayed in May 2017 and were dependent on weather forecasts to avoid rain within 24 h after treatment. Temperature and wind speed data were compiled from the Michigan State University Enviroweather stations located closest to experiment plot (Table 4). All herbicides were sprayed to run off with a backpack sprayer to the top one‐third of the tree where the cones commonly grow.

Table 4. Locations, dates, daily maximum temperatures, and wind speeds at time of application of post‐emergent cone herbicide application at three sites in Michigan in 2017.

Application Max Air Wind Location Station 1 GPS Date Temp C m/s Horton, MI May 3 15.8 2.3 Leslie 42.4514° N, 84.4325° W Horton, MI May 10 20.7 2.3 Leslie Sidney, MI May 3 15.6 1.6 Entrican 43.2925° N, 85.0814° W

Forests 2018, 9, 233 5 of 12

2.2.2. Study Design and Cone Treatments At each location, we selected 160 trees with at least ten cones per tree. The study was installed as a completely randomized design. At each farm, treatments were assigned to individual trees at random by selecting color-coded ribbons (one color per treatment) from a bucket. The three best performing products from the 2016 study (Avenger, Axxe, Scythe) were chosen for this study. Each product was initially applied to 40 trees at each farm (Table3).

Table 3. Description of herbicide products and rates applied for the 2017 study.

Treatment Manufacturer Product Description Rate Cutting Edge Formulations, Avenger Avenger ×2 * Avenger d-limonene (citrus oil) 250 mL/L Buford, GA ammonium nonanoate BioSafe Systems, LLC, East Axxe Axxe ×2 Axxe (ammonium salt of 100 mL/L Hartford, CT pelargonic acid) Scythe Scythe ×2 Gowan Company, Yuma, AZ Scythe pelargonic acid 50 mL/L Mechanical BlossomCinch, Manistee, MI na Mechanical cone removal na Control na na untreated na * Treatments denoted as ×2 were applied a second time, one to two weeks after initial application.

Half of the trees in each herbicide treatment were treated a second time (×2) approximately one week after the initial application. The rate of herbicide applied was the same for the single application and the repeated application. Herbicide treatments were sprayed in May 2017 and were dependent on weather forecasts to avoid rain within 24 h after treatment. Temperature and wind speed data were compiled from the Michigan State University Enviroweather stations located closest to experiment plot (Table4). All herbicides were sprayed to run off with a backpack sprayer to the top one-third of the tree where the cones commonly grow.

Table 4. Locations, dates, daily maximum temperatures, and wind speeds at time of application of post-emergent cone herbicide application at three sites in Michigan in 2017.

Location Application Date Max Air Temp C Wind m/s Station 1 GPS

Horton, MI May 3 15.8 2.3 Leslie 42.4514◦ N, ◦ Horton, MI May 10 20.7 2.3 Leslie 84.4325 W

Sidney, MI May 3 15.6 1.6 Entrican 43.2925◦ N, ◦ Sidney, MI May 16 29.5 3.0 Entrican 85.0814 W

Manton, MI May 15 24.5 1.1 Arlene 44.3900◦ N, ◦ Manton, MI May 22 17.6 4.7 Arlene 85.2867 W 1 Michigan State University Enviroweather Network Station.

At each farm, we tested a mechanical de-coning device powered by a cordless drill on 20 trees at each farm to remove cones (Figure4). The device uses spinning “fingers” to dislodge cones. We attempted to dislodge as many cones as possible while minimizing obvious damage to needles and shoots. Forests 2018, 9, x FOR PEER REVIEW 6 of 12

Sidney, MI May 16 29.5 3.0 Entrican Manton, MI May 15 24.5 1.1 Arlene 44.3900° N, 85.2867° W Manton, MI May 22 17.6 4.7 Arlene 1 Michigan State University Enviroweather Network Station.

At each farm, we tested a mechanical de‐coning device powered by a cordless drill on 20 trees at each farm to remove cones (Figure 4). The device uses spinning “fingers” to dislodge cones. We attempted to dislodge as many cones as possible while minimizing obvious damage to needles and Forests 2018, 9, 233 6 of 12 shoots.

Figure 4. Dislodging cones with a mechanical de-coningde‐coning device.device.

2.2.3.2.2.3. Evaluations EachEach tree was evaluated for cone kill and phytotoxicity (needle browning). The evaluations were conductedconducted threethree toto fourfour weeksweeks afterafter thethe finalfinal herbicideherbicide application.application. Cones were scoredscored asas livelive or dead (Figure(Figure5 ).5). We We observed observed in in the the 2016 2016 trial trial that that cones cones rated rated as “damaged” as “damaged” often often continued continued to grow, to grow, so they so werethey were counted counted as “live” as “live” for the for 2017 the evaluation. 2017 evaluation. Some Some evidence evidence of frost of damage frost damage was observed was observed on cones on incones the controlin the treatmentcontrol treatment and they and were they scored were as “dead”scored foras the“dead” cone for kill evaluation.the cone kill Phytotoxicity evaluation. evaluationsPhytotoxicity were evaluations narrowed were in 2017narrowed and separated in 2017 and into separated three categories: into three 0categories: = no damage, 0 = no 1 damage, = slight (moderate1 = slight (moderate browning browning to new or to old new shoots), or old and shoots), 2 = unacceptable and 2 = unacceptable (extreme (extreme browning browning to new and to new old shoots),and old andshoots), evaluated and evaluated on needles on in needles the top in one-third the top one of the‐third tree of where the tree the where herbicides the herbicides were sprayed. were sprayed.

Forests 2018, 9, 233 7 of 12 Forests 2018, 9, x FOR PEER REVIEW 7 of 12

Figure 5.5. Example of Fraser firfir Christmas treetree withwith 100%100% conecone kill.kill.

2.2.4.2.2.4. Statistical Analysis TheThe effectseffects of of Farm, Farm, Treatment, Treatment, and and Farm Farm× Treatment × Treatment interaction interaction were tested were by tested analysis by varianceanalysis usingvariance the using PROC the GLM PROC procedure GLM procedure in SAS (SAS in SAS Institute (SAS Inc., Institute Cary, Inc., NC, Cary, USA). NC, Where USA). Treatment Where mainTreatment effects main were effects significant were (significantp < 0.05), means (p < 0.05), were means separated were usingseparated Tukey’s using studentized Tukey’s studentized range test. Inrange addition, test. In the addition, effect the of applying effect of applying the chemical the chemical treatments treatments a second a second time was time tested was tested using using an a priorian a priori contrast contrast comparing comparing the the means means of allof all three three products products applied applied once once versus versus all all three three productsproducts appliedapplied twice.twice.

3.3. Results

3.1.3.1. 2016 Trial Farm,Farm, Treatment,Treatment, andand FarmFarm ×× Treatment interaction affected cone kill and extentextent ofof damagedamage toto needlesneedles andand shoots.shoots. Applications of Avenger andand AxxeAxxe resultedresulted inin thethe fewestfewest live,live, undamagedundamaged conescones (Figure(Figure6 6).). All All products products resulted resulted in in a a high high proportion proportion (50 (50 to to 80%) 80%) of of cones cones that that were were slightly slightly browned browned butbut notnot killed.killed. Avenger, Axxe,Axxe, andand ScytheScythe producedproduced thethe highesthighest cone kill (35 to 50%), whereas Goal, Reflex,Reflex, andand WeedzapWeedzap killed killed less less than than 20% 20% of of cones cones on on treated treated trees. trees.

Forests 2018, 9, x FOR PEER REVIEW 8 of 12 ForestsForests2018, 92018, 233, 9, x FOR PEER REVIEW 8 of 128 of 12

FigureFigure 6. Mean 6. Mean percentage percentage of of cones cones unaffected unaffected (Live), damaged, damaged, or orkilled killed on onFraser Fraser fir trees fir trees treated treated withwithFigure post-emergent post 6. ‐Meanemergent percentage herbicides herbicides of at cones at four four unaffected locations locations (Live),in Michigan, damaged, 2016. 2016. or Meanskilled Means onwithin Fraser within a groupfir a trees group with treated withthe the samesamewith letter letterpost are‐emergent notare not different different herbicides at atp

0.7) 2-unacceptable 2‐unacceptable at the Allegan injury. farm. VeryAxxe and Reflexlittleand caused Reflexphytotoxicity the caused greatest theoccurred greatest phytotoxity with phytotoxity any (mean product (mean phytotoxicity at Sidney phytotoxicity or East rating ratingLansing. >0.7) >0.7) at at the the Allegan Allegan farm. farm. VeryVery little phytotoxicitylittle phytotoxicity occurred occurred with any with product any product at Sidney at Sidney or East or East Lansing. Lansing.

Figure 7. Mean phytotoxicity score of Fraser fir trees treated with post‐emergent herbicides at four Figure 7. Mean phytotoxicity score of Fraser fir trees treated with post-emergent herbicides at four locationsFigure 7. inMean Michigan, phytotoxicity 2016. Means score within of Fraser a location fir trees with treated the samewith letterpost‐ emergentare not different herbicides at p at< 0.05.four locationslocations in Michigan, in Michigan, 2016. 2016. Means Means within within a a locationlocation with with the the same same letter letter are are not not different different at p < at 0.05.p < 0.05. Means were separated using Tukey’s range test. NOTE: Mean phytotoxicity was zero for Control trees

at all locations. Forests 2018, 9, x FOR PEER REVIEW 9 of 12

Forests 2018,Means9, 233 were separated using Tukey’s range test. NOTE: Mean phytotoxicity was zero for Control trees 9 of 12 at all locations.

3.2. 20173.2. 2017 Trial Trial ConeCone mortality mortality was was affected affected by by (p (p< < 0.05) 0.05) Treatment, Farm, Farm, and and the the interaction interaction of Treatment of Treatment × × FarmFarm (Figure (Figure8 ).8). AllAll chemicalchemical treatments treatments provided provided excellent excellent cone cone control, control, ranging ranging from 80% from cone 80% conemortality for for single single applications applications of of Avenger Avenger and and Axxe Axxe at at Manton Manton to to 100% 100% cone cone control control for for ×2× 2 applicationsapplications of Scythe of Scythe and and Axxe Axxe at Horton.at Horton. Overall, Overall, repeatedrepeated applications of of chemical chemical treatment treatment increasedincreased cone cone mortality mortality by by 7.8% 7.8% compared to to single single applications applications (contrast (contrast of ×1 vs. of ×2× for1 vs. all chemical×2 for all chemicaltreatments treatments was significant was significant at p < at0.05).p < 0.05).The mechanical The mechanical de‐coning de-coning device removed device removed 82% of cones 82% of conesaveraged averaged across across the the two two farms farms where where it itwas was used. used. The The device device was was effective effective at atremoving removing exposed exposed cones but failed to dislodge cones on interior branches. The interaction of farm and cone treatment cones but failed to dislodge cones on interior branches. The interaction of farm and cone treatment was largely due to cone mortality (17%) on control trees associated with a late frost at Sidney, whereas was largely due to cone mortality (17%) on control trees associated with a late frost at Sidney, whereas there was no cone mortality on control trees at the other two locations. there was no cone mortality on control trees at the other two locations.

Figure 8. Mean mortality of Fraser fir cones following cone control treatments at three locations in Michigan, 2017. Means were separated using by Tukey’s range test. * Mechanical cone removal device was not used at Sidney farm. Forests 2018, 9, x FOR PEER REVIEW 10 of 12

Figure 8. Mean mortality of Fraser fir cones following cone control treatments at three locations in Michigan, 2017. Means were separated using by Tukey’s range test. * Mechanical cone removal device Forests 2018was, 9not, 233 used at Sidney farm. 10 of 12

Mean phytotoxicity score also varied among farms and cone control treatments, and there was Mean phytotoxicity score also varied among farms and cone control treatments, and there was an interaction effect of farm and treatment (p < 0.05) (Figure 9). The mechanical de‐coning device an interaction effect of farm and treatment (p < 0.05) (Figure9). The mechanical de-coning device resulted in the highest “phytotoxicity” scores at both farms where it was used. Damage from the resulted in the highest “phytotoxicity” scores at both farms where it was used. Damage from the device was associated with needle browning and shoot damage in the upper‐third of tree crowns. device was associated with needle browning and shoot damage in the upper-third of tree crowns. Single applications of post‐emergent herbicides resulted in very little phytotoxicity. The highest Single applications of post-emergent herbicides resulted in very little phytotoxicity. The highest phytotoxicity score for a single chemical treatment was 0.27 at Sidney. Phytotoxicity increased with phytotoxicity score for a single chemical treatment was 0.27 at Sidney. Phytotoxicity increased with repeated applications of herbicide treatments. Averaged across all farms, the mean phytotoxicity repeatedscore increased applications from of 0.05 herbicide for single treatments. applications Averaged to 0.36 acrosswhen treatments all farms, the were mean repeated. phytotoxicity Phytotoxicity score increasedon foliage from was 0.05 particularly for single high applications for repeated to 0.36 applications when treatments of Scythe were and repeated.Avenger at Phytotoxicity Sidney. on foliage was particularly high for repeated applications of Scythe and Avenger at Sidney.

FigureFigure 9. 9.Mean Mean phytotoxicity phytotoxicity score score of of Fraser Fraser fir fir trees trees following following cone cone removal removal treatments treatments at at three three locationslocations in in Michigan, Michigan, 2017. 2017. Phytotoxicity Phytotoxicity was was rated rated from from 0 0 (no (no damage) damage) to to 2 2 (unacceptable (unacceptable browning). browning). Means within a location with the same letter are not different at p < 0.05. Means were separated using

Tukey’s range test. * Mechanical cone removal device was not used at Sidney farm. Forests 2018, 9, 233 11 of 12

4. Discussion Handpicking cones in Fraser fir plantations is one of the single largest labor expenditures for Christmas tree producers in the upper Midwest and other regions of the country where Fraser fir is grown. In Michigan, growers report that picking cones can add over $2.00 in labor costs per tree for 3–4 m tall trees. Growers also report extensive issues with coning in . Coning in that region could potentially impact decisions to deploy grafted seedlings, which can be more physiologically mature than seedling and therefore potentially more prone to coning (Eric Hinesley, personal communication). Results of our current trials indicate that the properly timed application of herbicides can stop the development of up to 95% of cones without phytotoxicity under the best-case scenario (single application of Scythe). Repeating applications of chemical control treatments can improve cone control slightly, but at the expense of increased phytotoxicity and labor and product cost. Treatments applied at or immediately after vegetative budbreak caused damage to new needles, as was the case in 2016; trees at Allegan and Horton had higher phytotoxicity than Sidney and East Lansing. This might be because the application dates were earlier at Sidney and East Lansing, and they are located at a higher latitude than Allegan and Horton, resulting in delayed vegetative bud break. Therefore, proper timing is essential to optimize cone kill while minimizing damage to foliage. Applicators must time treatments carefully in order to target cones soon after they emerge but before vegetative budbreak occurs [10,11]. This means that treatments must be applied in an approximately two-week long window in the spring. We observed that cone control was better overall in the 2017 trial than in the 2016 trial. In particular, we found fewer instances of cones that were damaged but not killed in 2017 than in 2016. We attribute the increased efficacy of the second trial to greater experience with the products, particularly the need to thoroughly wet each cone and applying the products as soon after cone emergence as weather conditions will allow. Mechanical cone removal in the 2017 trial caused a high level of needle and shoot browning at the two trial locations in which it was used. These results point out the difficulty of applying sufficient force to dislodge cones without damaging needles, shoots, and vegetative buds. Investigators should continue to explore options for physical cone removal without hand labor with an emphasis on techniques that reduce unintended damage, e.g., high-pressure water jets. Economically, chemical cone control can compare favorably with hand picking when trees have more than 100 cones. Treating cones in a 2-m tree with post-emergent herbicides with a backpack sprayer requires approximately 45 s per tree. Based on our current application rates, the chemical cost per tree for each application ranges from $0.50 USD per tree for Scythe to $2.00 USD per tree for Avenger. Assuming $15 USD per hour for labor, cone-picking costs on large trees with heavy cone loads range between $0.50 USD to $2.00 USD per tree (2 to 8 min per tree). These product costs are based on retail prices. Growers might be able to reduce material costs by buying in bulk. Two key outstanding questions remain for the use of herbicides for post-emergent cone control. First, do remnant, immature dead cones remaining on trees impact the marketability of the trees? For a given growing season, vegetative growth occurs after cones have emerged, so remnant cones killed by herbicide treatments could be largely hidden by new growth. If current-year growth of new shoots is not sufficient to hide remnant cones, growers may still need to pick cones in the year of harvest. Chemical control could nonetheless be of value in lieu of handpicking in years prior to harvest. A second remaining issue for chemical control of nuisance cones is the development of operational systems. Our experiments and earlier trials [5] have relied on backpack sprayers to treat individual trees. Mechanized spray systems could further reduce labor time and application costs. Owen [12] used a mist-blower to apply post-emergent cone control in North Carolina but found that control was only effective on exterior rows of trees nearest the sprayer. Owen used several rates of Off-Shoot-T (Chemtura Corporation, Middlebury, CT, USA), Off-Shoot-O (Cochran Corporation, Memphis, TN, USA), Axxe (BioSafe Systems, LLC, East Hartford, CT, USA), and Scythe (Gowan Company, Yuma, Forests 2018, 9, 233 12 of 12

AZ, USA). Other spray equipment, e.g., an air curtain sprayer, could potentially offer better coverage and control.

5. Conclusions Effective control of nuisance cones in Fraser fir Christmas tree plantations is possible with post-emergent herbicides. Proper timing of treatments, however, is essential in order to optimize cone control and reduce possible phytotoxicity. Cones must be treated shortly after they emerge and before vegetative budbreak. Thorough cone coverage is essential for good control. The application of post emergent herbicides can provide considerable savings over handpicking cones but consumer acceptance of trees with remnant, immature cones is not known.

Author Contributions: B.C., D.E., and J.O. conceived, designed, and performed the experiments; B.C. analyzed the data; D.E. developed presentation graphics; B.C. and D.E. wrote the paper. Acknowledgments: Funding for this project was project was provided by the Michigan Christmas Tree Association and the National Christmas Tree Promotion Board. We gratefully acknowledge Dutchman Tree Farm, Korson’s Tree Farm, Gwinn’s Tree Farm, and Badger Evergreens for the use of their fields for study plots. We thank Ellie Domer and Mary Tuski for assistance with plot installation and field data collection. We thank Eric Hinesley and two anonymous reviewers for their constructive comments on this manuscript. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Burns, R.M.; Honkala, B.H. Silvics of North America. Volume 1. Conifers. In Agriculture Handbook 654; US Department of Agriculture: Washington, DC, USA, 1990; 675p. 2. Powell, G.R. Biennial strobilus production in balsam fir: A review of its morphogenesis and a discussion of its apparent physiological basis. Can. J. For. Res. 1977, 7, 547–555. [CrossRef] 3. Crain, B.A.; Cregg, B.M. Regulation and Management of Cone Induction in Temperate Conifers. For. Sci. 2018, 64, 82–101. [CrossRef] 4. Crain, B.A.; Cregg, B.M. Gibberellic acid inhibitors control height growth and cone production in Abies fraseri. Scandinavian J. For. Res. 2017, 32, 391–396. [CrossRef] 5. Owen, J. Using Herbicides to Interrupt Cone Development on Fraser Fir. In 12th International Christmas Tree Research and Extension Conference; Talgø, V., Fløistad, I.S., Eds.; Norwegian Institute of BioEconomy Research NIBIO: Ås, Norway, 2015; Volume 1, p. 41. 6. Dayan, F.E.; Cantrell, C.L.; Duke, S.O. Natural products in crop protection. Bioorg. Med. Chem. 2009, 17, 4022–4034. [CrossRef][PubMed] 7. Dayan, F.E.; Duke, S.O. Natural compounds as next-generation herbicides. Plant Physiol. 2014, 166, 1090–1105. [CrossRef][PubMed] 8. Coleman, R.; Penner, D. Desiccant activity of short chain fatty acids. Weed Technol. 2006, 20, 410–415. [CrossRef] 9. Coleman, R.; Penner, D. Organic acid enhancement of Pelargonic acid. Weed Technol. 2008, 22, 38–41. [CrossRef] 10. Owens, J.N.; Molder, M. Vegetative bud development and cone differentiation in Abies amabilis. Can. J. Bot. 1977, 55, 992–1008. [CrossRef] 11. Owens, J.N.; Singh, H. Vegetative bud development and the time and method of cone initiation in subalpine fir (Abies lasiocarpa). Can. J. Bot. 1982, 60, 2249–2262. [CrossRef] 12. Owen, J. Fraser fir cone control update. In Abstracts of 13th International Christmas Tree Research and Extension Conference; Icelandic Forest Service: Akureyri, Iceland, 2017; p. 49.

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).