Management Implications
Management of tree invasions within conservation areas is particularly challenging, as the strategies to be applied in control programs are restricted by regulations aimed to preserve the environment. In the total protected area of Argentina (4,993,722 ha), about 50 invasive woody species have been identified, 12 of which are conifers. As tree invasions are significant and a growing threat to biodiversity, it is critical to explore cost-effective techniques that can be implemented in protected areas with a high value for conservation, particularly in economically challenged and/or developing countries. This work focuses on herbicide application by the drill and fill method as a potential tool to treat invasive tree species, as this method allows targeted application and calibrated doses at low cost and reduced labor times. The treated trees showed high levels of defoliation, even in the larger trees. As such, this method could be useful on medium to large reproductive individuals (diameter at breast height between 20 and 30 cm), which are an important source of propagules and secondary loci formation, but are difficult to control without causing significant disturbance to the surrounding environment. Applying the drill and fill method significantly reduces removal costs when compared with conventional felling, which is the current removal method used in the area. This economically viable approach expands the possibilities for tackling ecological problems, even in scenarios marked by limited funding for environmental policies. So far, herbicides are not allowed to be used in national parks, because there are very few local studies demonstrating their effectiveness and low environmental risk. In this study, the drill and fill method emerges as a practical, applicable and cost-effective tool for the management of invasive trees in areas of high conservation value.
Introduction
Invasion by nonnative woody species is an important cause of ecological degradation and biodiversity loss (Gioria et al. Reference Gioria, Hulme, Richardson and Pyšek2023; Richardson et al. Reference Richardson, Hui, Nuñez and Pauchard2014), resulting in significant economic losses (Diagne et al. Reference Diagne, Leroy, Vaissière, Gozlan, Roiz, Járic, Salles, Bradshaw and Courchamp2021; Fernandez et al. Reference Fernandez, Haubrock, Cuthbert, Heringer, Kourantidou, Hudgins, Angulo, Diagne, Courchamp and Nuñez2023). Some conifers are considered invasive worldwide, and their establishment has been especially successful in many areas in the Southern Hemisphere, such as South Africa, Australia, New Zealand, and South America (Nuñez et al. Reference Nuñez, Chiuffo, Torres, Paul, Dimarco, Raal, Pollicelli, Moyano, García, Van Wilgen, Pauchard and Richardson2017; Richardson Reference Richardson1998; Simberloff et al. Reference Simberloff, Nuñez, Ledgard, Pauchard, Richardson, Sarasola, Van Wilgen, Zalba, Zenni, Bustamante, Peña and Ziller2010). Studies in these regions have documented serious ecological impacts from the invasions, including changes to hydrological cycles (Jobbágy et al. Reference Jobbágy, Acosta and Nosetto2013; Le Maitre et al. Reference Le Maitre, Versfeld and Chapman2000), fire regimes (Brooks et al. Reference Brooks, D’Antonio, Richardson, Grace, Keeley, DiTomaso, Hobbs, Pellant and Pyke2004; Taylor et al. Reference Taylor, Maxwell, McWethy, Pauchard, Nuñez and Whitlock2017), and habitat loss (de Abreu and Durigan Reference Abreu and Durigan2011), as well as economic impacts related to damage and management costs (Duboscq-Carra et al. Reference Duboscq-Carra, Fernandez, Haubrock, Dimarco, Angulo, Ballesteros-Mejia, Diagne, Courchamp and Nuñez2021; Velarde et al. Reference Velarde, Paul, Monge and Yao2017).
The management of woody species invasion typically involves the application of mechanical and/or chemical control methods (Weidlich et al. Reference Weidlich, Flórido, Sorrini and Brancalion2020). For example, in South Africa, mechanical procedures such as tree felling, pruning, or ring-barking are commonly implemented and are particularly efficient in areas with low invasion density (Boast Reference Boast2021). In New Zealand, mechanical methods are also employed. However, significant progress has been made in recent years in the use of herbicides for the management of Pinus species (Gous et al. Reference Gous, Raal, Kimberley and Watt2015a, 2015b; Ledgard Reference Ledgard2009; Rolando et al. Reference Rolando, Richardson, Paul and Somchit2021). Herbicides have the potential to be the most cost-effective approach for the control of large and well-established invasions (Lange et al. Reference Lange, Boast and Kleynhans2022; Nuñez et al. Reference Nuñez, Chiuffo, Torres, Paul, Dimarco, Raal, Pollicelli, Moyano, García, Van Wilgen, Pauchard and Richardson2017), especially in areas where wood extraction is not profitable. The suitability and efficient application of these methods depend on multiple factors ranging from the age of the invasion, the size of the invaded area, characteristics of the terrain (Nuñez et al. Reference Nuñez, Chiuffo, Torres, Paul, Dimarco, Raal, Pollicelli, Moyano, García, Van Wilgen, Pauchard and Richardson2017), financial limitations (Kettenring and Adams Reference Kettenring and Adams2011), and regulations of each country regarding control of invasive species (Wagner et al. Reference Wagner, Antunes, Irvine and Nelson2017).
Herbicide application methods most frequently used to control Pinaceae invasion involve aerial boom sprays, aerial bark application with wands (both applied from a helicopter), and ground basal bark applications, including ring-barking, stem injection, and cut stump applications (Briden et al. Reference Briden, Raal and Gous2014; Ledgard Reference Ledgard2009; Raal Reference Raal2005). Aerial sprayings are highly effective in controlling large areas of dense invasion within short time periods (Briden et al. Reference Briden, Raal and Gous2014). However, their uncalibrated application can generate negative environmental impacts in the surrounding area (Rolando et al. Reference Rolando, Scott, Baillie, Dean, Todoroki and Paul2023) and/or kill non-target native species (Cornish and Burgin Reference Cornish and Burgin2005) due to herbicide drift and runoff (Baillie Reference Baillie2016; Richardson et al. Reference Richardson, Rolando, Hewitt and Kimberley2020). As such, this procedure would not be a viable option in areas of high conservation value, where prioritizing low-impact methods is essential to ensure the regeneration of native plant cover.
A promising method with low impact on the surrounding native plant community is stem injection (Itou et al. Reference Itou, Hayama, Sakai, Tanouchi, Okuda, Kushima and Kajimoto2015; Raal Reference Raal2005), also known as stem poisoning or the “drill and fill” method. Although labor-intensive in the context of treating invaded stands, it is a precise method that allows highly calibrated doses to be applied directly into the tree, thus avoiding “off-target” effects (Boast Reference Boast2021). Once applied, the herbicide is rapidly absorbed and transported through the vascular system to the target tissues (Gous and Richardson Reference Gous and Richardson2008), causing physiological failure, defoliation, and ultimately death. Compared with other chemical methods, stem injection has several favorable features, such as no spillage of chemicals into the environment or on non-target species, as the herbicide remains inside the tree (Boast Reference Boast2021). For the same reason, it can easily be carried out in variable weather conditions. Because the trees are killed standing, the surrounding system remains mostly undisturbed (Macalister Reference Macalister2010) and light gaps that may favor reinvasion are not suddenly generated in the canopy, which can occur with mechanical removal (Paul and Ledgard Reference Paul and Ledgard2009). Dead trees gradually decompose until they eventually collapse after 10 to 15 yr (Boast Reference Boast2021; Ledgard Reference Ledgard2009). Decaying nonnative trees can serve as perches for native plant seed–dispersing birds and provide habitat for native wildlife (Herrera and García Reference Herrera and García2009; Menvielle et al. Reference Menvielle, Arata, Nuñez, Buono and Eglis2020). Mechanical methods to kill standing trees, such as ring-barking, are not always effective, as the tree may repair the damage if the method is improperly implemented (Ledgard Reference Ledgard2009), or are directly ineffective in resprouting species (Burch and Zedaker Reference Burch and Zedaker2003).
The most commonly used systemic herbicides for chemical control of woody species (including Pinaceae) are glyphosate, triclopyr, picloram, and metsulfuron (Ledgard Reference Ledgard2009). Several studies have tested the effectiveness of mixing these herbicides (Gous et al. Reference Gous, Raal and Watt2014, Reference Gous, Raal and Watt2015b) at different rates, also including other chemical compounds such as aminopyralid and dicamba (Gous et al. Reference Gous, Raal and Watt2015b; Rolando et al. Reference Rolando, Gaskin, Horgan and Richardson2020). The efficacy of each herbicide or mixture varies and depends on the species to be controlled, the size of the individual, and the application method used (DiTomaso and Kyser Reference DiTomaso and Kyser2007; Ledgard Reference Ledgard2009). Therefore, it is critical to the successful application of these herbicides to adjust their implementation strategies according to the tree species, environmental factors, and land use (e.g., conservation area or forest plantation).
Douglas-fir [Pseudotsuga menziesii (Mirb.) Franco] is a Pinaceae species native to North America that is invading large areas within the Patagonian forest of Argentina (Moyano et al. Reference Moyano, Simberloff, Relva and Nuñez2023; Simberloff et al. Reference Simberloff, Nuñez, Ledgard, Pauchard, Richardson, Sarasola, Van Wilgen, Zalba, Zenni, Bustamante, Peña and Ziller2010). It is a particularly difficult tree species to manage, because it is invading a protected area with a high conservation value. As such, control plans must be carefully designed and implemented to avoid impacting the native ecosystem. The available information about how chemical methods could be applied to tree species invading conservation areas is currently scarce and often limited to anecdotal reports by park staff. To date, conventional felling has been the only method applied for the control of P. menziesii in the region, while the economic benefits of using chemical approaches remaining unknown. As such, the objectives of this study were: (1) to test the efficacy of two systemic herbicides applied by drill and fill method on defoliation of P. menziesii trees and (2) to perform an economic analysis for the removal of invasive trees using the drill and fill method and compare it with conventional felling.
Materials and Methods
Study Site
The study was conducted in Isla Victoria (–40.9530100, –71.5381000), Nahuel Huapi National Park, Patagonia, Argentina. The island has an area of ca. 3,710 ha, its extension from northwest to southeast is ca. 18.2 km, and its maximum elevation point is 1,025 m above sea level (256 m above lake level) (Fig. 1). The climate is temperate with two distinct seasons. Mean summer temperature is 16 C, and mean winter temperature is 4.3 C. The mean annual rainfall in the area is 1,300 mm with 70% falling in the cold season (May to September). Winds are predominantly from the west, northwest, and north (Simberloff et al. Reference Simberloff, Relva and Nuñez2002). Most of the island’s surface area is covered with forest co-dominated by Nothofagus dombeyi (Mirb.) Blume (Coihue) and Austrocedrus chilensis (D. Don) Pic. Serm. & Bizzarri (Chilean cedar). In 1925, the Argentina government established a forestry program (Koutché Reference Koutché1942), which resulted in the planting of 62 broadleaved and 73 conifer species (Richardson and Rejmánek Reference Richardson and Rejmánek2004). At that time, tree plantations occupied a total area of 180 ha, of which 150 ha occurred in the central part of the island. After several introduction events, some of these species have succeeded in establishing themselves within the native plant communities (Simberloff et al. Reference Simberloff, Relva and Nuñez2002) and continue to expand within the park (Moyano et al. Reference Moyano, Simberloff, Relva and Nuñez2023). Nearly 80 yr later, P. menziesii was found to be the most abundant nonnative species outside the original plantations (Simberloff et al. Reference Simberloff, Relva and Nuñez2002), successfully invading native forest (Nuñez and Paritsis Reference Nuñez and Paritsis2018).
Location of the study site within Isla Victoria, in northern Patagonia, Argentina. The treated area is outlined in yellow.
Field Trial
The trees to be treated were selected from a stand densely invaded by P. menziesii (∼5,000 stems ha−1), representing about 30 ha in the central area of Isla Victoria (Fig. 1). We selected healthy-looking trees (i.e., without signs of anomalies or damage) growing along similar topography. For the herbicide application, 50-mm-deep holes were drilled with an 8-mm-diameter drill bit around the tree trunk at ∼1.5 m in height and at an ∼45° downward angle to avoid spillage (Fig. 2A). Two different systemic herbicides were tested (Table 1): (1) glyphosate, a nonselective herbicide (Roundup Full II®; composition: glyphosate [N-(phosphonomethyl)] glycine potassium salt), at 662 g L−1; and (2) aminopyralid + triclopyr, a selective herbicide (Tocón Extra®; composition: aminopyralid [4-amino-3,6-dichloropyridin-2-carboxylic acid] + triclopyr butoexyl ester [2-butoxyethyl 3,5,6-tricholoro-2-pyridyl oxyacetate]), at 40 g L−1 + 166.9 g L−1. For each herbicide, two concentrations were tested and defined as “high” and “low,” resulting in four treatment combinations: glyphosate at high concentration (GH), glyphosate at low concentration (GL), aminopyralid + triclopyr at high concentration (ATH), aminopyralid + triclopyr at low concentration (ATL) (Table 1). Glyphosate is usually recommended by manufacturers to control broadleaf weeds as well as annual and perennial grasses. However, it has been shown to be effective in eliminating invasive species such as Salix trees, shrubs, and Pinaceae species using the stem injection method (Australian Weed Management 2003; Ledgard Reference Ledgard2009; unpublished internal reports of Argentina’s National Park Administration). Aminopyralid and triclopyr are commonly used for dicotyledonous woody species by basal application or cut stump (i.e., not injected), and although the manufacturer does not recommend it specifically for conifers, triclopyr has been successfully used for controlling Pinaceae (Briden et al. Reference Briden, Raal and Gous2014; Rolando et al. Reference Rolando, Richardson, Paul and Somchit2021). In Argentina, neither product is label-recommended for injection; however, Tocón Extra® is occasionally used within the national park for cut stump applications.
Herbicide treatment applied by drill and fill method in invading Pseudotsuga menziesii trees. (A) Drilling holes into tree bark; (B) injection of herbicide into the vascular system with a syringe; (C) P. menziesii individual (32-cm diameter at breast height) completely defoliated after 12 mo of being injected with glyphosate at high concentration (GH); (D) aerial photo taken from a drone, the left arrows show completely defoliated trees, and the right arrow shows a tree with brown needles. Scale bar: 10 m. Source: Santiago Quiroga.
Herbicide specifications per treatment, including treatment code, commercial product names, manufacturer, active ingredient (ai), concentrations and carriers used to control invasive Pseudotsuga menziesii trees by drill and fill
a Ricardo Gutiérrez 3652, Munro, Argentina. https://cropscience.bayer.com.ar/distribuidores.
b Avenida Libertador 101, Piso 1, 1638, Buenos Aires, Argentina. https://www.corteva.com.ar/contactanos.html.
The dose per tree was adjusted according to the diameter at breast height (DBH; Table 2). For this purpose, DBH was categorized into three classes: class 1, less than 10 cm, 1 hole; class 2, between 10 and 20 cm, 2 holes; and class 3, between 20 and 30 cm, 3 holes (Table 2). Each P. menziesii tree within a DBH class was randomly assigned to one of the four herbicide treatments. Once assigned, each tree hole was filled with 6 ml of herbicide (maximum hole capacity for the specified size; Fig. 2B; Table 2). Thus, the dosages were as follows: trees with DBH < 10 cm were filled with 6 ml per tree, trees with DBH between 10 and 20 cm were filled with 12 ml per tree, and trees with DBH between 20 and 30 cm were filled with 18 ml per tree. The amount of active ingredient in grams (g ai) applied for each volume varied between herbicide types (glyphosate or aminopyralid + triclopyr) and between concentrations (high or low); see Table 2 for details. We obtained five replicates per herbicide type per concentration per DBH class, which amounted to 60 trees. Herbicide treatments were applied in April 2022 (Austral autumn) and surveyed at 6, 12, and 24 mo after application (October 2022, April 2023, and April 2024, respectively). To calculate the proportion of dead foliage, the crown was divided into four sections and assigned a defoliation value by visual inspection, which was then averaged to obtain the final defoliation ratio. Additionally, the health of 15 neighboring trees (five per DBH class) not treated with herbicides or drilled was monitored to exclude the possibility that defoliation was associated with other external factors such as drought or insect or pathogen attack. The greatest bark thickness recorded for treated trees was 12 mm. In addition to DBH, tree height and crown width were measured and used to estimate tree size.
Details of herbicide concentration of active ingredient (ai) applied into Pseudotsuga menziesii trees according to diameter at breast height (DBH) classes and herbicide treatment
a GL, glyphosate at low concentration (33.1 g ai L−1); GH, glyphosate at high concentration (662 g ai L−1); ATL, aminopyralid + triclopyr at low concentration (0.31 g ai L−1 total for both active ingredients); ATH, aminopyralid + triclopyr at high concentration (1.55 g ai L−1 total for both active ingredients).
Economic Analysis
To determine the cost-effectiveness of the drill and fill method and conventional felling, an assessment of the costs of treating 10,000 medium to large invasive trees (DBH between 20 and 30 cm) was carried out. To calculate the costs, we first estimated the time it takes a single person to remove 30 trees between 20 and 30 cm DBH with a chainsaw or to inject herbicides. The estimated time was then extrapolated to a total of 10,000 trees. We ensured that the sets of trees were in similar terrain conditions for both estimates (felling/injecting). A list of required items was made and classified into four categories: labor, safety clothing, equipment, and tree residues. Because chemical control requires a competent professional to elaborate the prescription for herbicide purchase and application (Menvielle et al. Reference Menvielle, Arata, Nuñez, Buono and Eglis2020), this item was included as “professional service”. Those fees were then estimated based on “technical field consultation/advice cost per hour” from the website of the Professional Council of Agricultural Engineering of Rio Negro, Argentina (https://cpiarn.org.ar). As for the operators, the estimated payment was calculated by consulting the amount that Argentina’s National Parks Administration pays its operators per hour of work in this particular region (Internal Document No. 206/2024). The prices of each item within the three remaining categories, safety clothing, equipment, and tree residues, were taken from Mercado Libre (www.mercadolibre.com.ar), which is the main platform dedicated to e-commerce in Latin America. Prices were converted to U.S. dollars considering that US$1 is equivalent to 1,005 pesos ARG according to the Argentinian National Bank (www.bna.com.ar) at time of writing.
Data Analysis
To analyze differences in the proportion of dead foliage between treatments, a nested generalized linear model (Zuur et al. Reference Zuur, Hilbe and Ieno2013) with a beta error distribution and logit link were used. Because this distribution only allows values between zero and one, the proportion data were transformed according to Smithson and Verkuilen (Reference Smithson and Verkuilen2006) to exclude extreme values (Equation 1).
$Y_{\rm a}= (Y_{\rm b}\times (N-1) + 0.5)/N$
where Y a is the transformed proportion value, Y b is the raw proportion value (containing zeros and ones) and N is the sample size. Because the tested concentrations are nested within herbicide types, herbicide treatment was used as a four-level explanatory variable (i.e., GL, GH, ATL, ATH). Statistical models were fit for each postapplication time surveyed (6, 12, and 24 mo). During model selection, the explanatory variables herbicide treatment, herbicide dose (three levels: 6, 12, and 18 ml per tree), tree height (covariate), DBH (covariate), and crown width (covariate) were included. Model selection was based on the Akaike’s information criterion (AIC; Akaike Reference Akaike, Parzen, Tanabe and Kitagawa1998) and the Akaike weights (Burnham and Anderson Reference Burnham and Anderson2002). The best-fit models for each subset according to AIC only included the variables herbicide treatment and herbicide dose (Supplementary Table S1). Differences between explanatory variables were determined by an analysis of deviance using a chi-square test and pairwise post hoc multiple comparison using Tukey’s Honestly Significant Difference (HSD) test. Because all the untreated trees had no defoliation (i.e., 0%), they were excluded from the statistical analyses. All the statistical analyses were performed using R v. 4.3.1 (R Core Team 2023) and the packages glmmTMB (Brooks et al. Reference Brooks, Kristensen, Van Benthem, Magnusson and Berg2017), car (Fox and Weisberg Reference Fox and Weisberg2019), ggplot2 (Wickham Reference Wickham2016), and emmeans (Lenth Reference Lenth2023).
Results and Discussion
Effectiveness of the Drill and Fill Method
The proportion of dead foliage was significantly affected by the interaction between herbicide treatment and dose for 6 and 12 mo after application (Fig. 3; Table 3; Supplementary Table S2). After 24 mo, the herbicide treatment and the dose still showed considerable effects on tree defoliation, yet without a statistically significant interaction (Table 3). Glyphosate at high concentration (GH) caused the greatest proportion of dead foliage on P. menziesii trees at 6, 12, and 24 mo after application (Figs. 2C and D and 3). After 6 mo, all small trees were 100% defoliated within the GH treatment (Fig. 3A), while in medium-sized trees, the average defoliation rate was 0.7, and in large trees, 0.5 (Fig. 3B and C). After 12 mo, the average defoliation for medium and large trees increased from 0.7 to 0.98 and from 0.5 to 0.9, respectively (Fig. 3E and F). The proportions remained the same after 24 mo (Fig. 3H and I). The GL treatment produced a mean defoliation of 0.85 after 12 mo only on the smallest trees (Fig. 3D). However, after 24 mo, medium and large trees within this treatment exhibited a wide range of defoliation, between 0.1 and 1 in medium trees and between 0.15 and 0.6 in large trees (Fig. 3H and I). Compared with glyphosate treatment, ATH produced a reduced effect in the smallest trees after 24 mo (Fig. 3G); however, this was not statistically significant among the different doses (i.e., among the different DBH classes; Fig. 3G–I). Associations between defoliation and other measures of tree size, such as height and crown width, were visually explored, but no trends were detected (data not shown).
Proportion of dead foliage as a function of herbicide treatment and herbicide dose at 6, 12, and 24 mo after application into Pseudotsuga menziesii trees. Codes: GL, glyphosate at low concentration (33.1 g ai L −1); GH, glyphosate at high concentration (662 g ai L −1); ATL, aminopyralid + triclopyr at low concentration (0.31 g ai L−1 total for both active ingredients); ATH, aminopyralid + triclopyr at high concentration (1.55 g ai L−1 total for both active ingredients). The solid circles represent the mean with 95% confidence intervals. Sizes of circles represent the three diameter at breast height (DBH) classes. Five trees were treated per DBH class (N = 5). Statistically significant effects determined by Tukey’s Honestly Significant Difference (HSD) test are shown with different letters (P < 0.05). The specific concentrations per volume applied per herbicide are presented in Table 2.
Analysis of deviance summary table (type II chi-square test) for the proportion of dead foliage of Pseudotsuga menziesii trees as a function of herbicide treatment, dose, and their interaction, applied in each month postapplication surveyed (6, 12, and 24 mo)
Estimated Tree Mortality after Herbicide Applications
After two growing seasons postapplication, the trees treated with GH that were 100% defoliated had no new growth. This is a strong indicator that the trees were likely dead. Although P. menziesii individuals experience epicormic regrowth from the trunk (Collier and Turnblom Reference Collier and Turnblom2001), this phenomenon was only observed (at least at our sites) when small trees were damaged or cut with machetes (data not shown). Only 2 of the 15 trees treated with GH maintained a small proportion of green-yellow needles after 24 mo (Fig. 3G–I). One of the trees had a DBH of 19 cm and had 90% of its crown defoliated, yet no new shoots were observed. The other tree, with a DBH of 28 cm, had 50% of the crown defoliated at 12 mo and 70% at 24 mo, and also had no new shoots.
Because there is strong local opposition to using herbicides in conservation areas, both from citizens and park managers, the local authorities only allowed us to conduct our study on a restricted number of trees, at a limited temporal and spatial scale. Regardless, our results provide evidence that herbicide application by drill and fill is a useful tool to control the invasion by P. menziesii in a forest. The highest concentration of glyphosate (662 g ai L−1, GH) killed 90% of the trees, and the lowest concentration (33.1 g ai L−1, GL) killed 30%. That is, a 20-fold lower concentration (GL) killed one-third of the trees compared with GH. This could suggest that concentrations lower than 662 g ai L−1 may also be effective in controlling invasive conifers. Further work should explore the dose–response curves to accurately determine the grams of active ingredient per volume needed to kill other invasive trees for both herbicide types using the drill and fill method. Identifying the optimum dose is essential to avoid the overuse of the product as well as the costs associated with the herbicide and its application.
Environmental Effects of Herbicides Used for Controlling Tree Invasions
Glyphosate is a low environmental risk herbicide (Duke Reference Duke2020; Hagner et al. Reference Hagner, Mikola, Saloniemi, Saikkonen and Helander2019). Nevertheless, its global application has sparked discussions among different community members (Weidlich et al. Reference Weidlich, Flórido, Sorrini and Brancalion2020), mainly triggered by the large-scale overuse of herbicides in agriculture (Pergl et al. Reference Pergl, Härtel, Pyšek and Stejskal2020). Glyphosate has a mean half-life in water and soil of 30 d (Blake and Pallett Reference Blake and Pallett2018), and its degradation is mainly through microbial action (Borggaard and Gimsing Reference Borggaard and Gimsing2008). Once it enters the soil, it binds tightly to organic compounds that transform it into an inactive product (Borggaard and Gimsing Reference Borggaard and Gimsing2008), unable to generate phytotoxicity in non-target plants (Duke Reference Duke2020). Due to its high affinity to soil particles, its movement through surface and ground water is minimal (Saunders and Pezeshki Reference Saunders and Pezeshki2015). The main source of glyphosate entry into the soil is via sprayed droplets (Duke Reference Duke2020); thus, with targeted applications such as the drill and fill method, glyphosate would not cause undesirable effects on ecosystems (Pergl et al. Reference Pergl, Härtel, Pyšek and Stejskal2020). Aminopyralid has a mean half-life of 76 d in soil and 240 d in water, while triclopyr has a mean half-life of 30 d in soil and 25 d in water (Rolando et al. Reference Rolando, Scott, Baillie, Dean, Todoroki and Paul2023). Aminopyralid in particular is highly soluble in water and therefore prone to move into surface water by runoff or groundwater by leaching (Kashuba et al. Reference Kashuba, Kiernan and Costello2005). Both herbicides are mobile in soils (Thompson et al. Reference Thompson, Pitt, Buscarini, Staznik and Thomas2000; Tomco et al. Reference Tomco, Duddleston, Schultz, Hagedorn, Stevenson and Seefeldt2016) and can be degraded via photolysis, hydrolysis, and microbial actions (Tu et al. Reference Tu, Hurd and Randall2001; USDOI 2015). In New Zealand, triclopyr was found to be present on the forest floor, retained in litter of dead needles, 2 yr after an area densely invaded by conifers was sprayed (Rolando et al. Reference Rolando, Scott, Baillie, Dean, Todoroki and Paul2023). Moreover, it has been observed that aminopyralid and triclopyr can cause damage to non-target plants via root exudation after basal bark application (Graziano et al. Reference Graziano, Tomco, Seefeldt, Mulder and Redman2022). Although the concentrations found do not pose a risk to the environment in some cases (Rolando et al. Reference Rolando, Scott, Baillie, Dean, Todoroki and Paul2023), these aspects will definitely influence subsequent management decisions (Dickie et al. Reference Dickie, Sprague, Green, Peltzer, Orwin and Sapsford2022). The presence of herbicides in the soil may not only affect non-target plants, but will also impact the success of regeneration or restoration of the system. Addressing these aspects is critical, as the management actions chosen to remove invasive individuals will have a substantial effect on the expected long-term outcomes.
Lessons Learned from Applying Herbicides with the Drill and Fill Method
The insignificant effect of the aminopyralid + triclopyr treatments on tree defoliation was probably due to the less than optimal concentrations of the active ingredient. In Argentina, Tocón Extra® is label-recommended for stump application at 1.5% (0.31 g ai L−1) in woody shrub species. Neither this dose, nor a dose five times higher, was sufficient to cause significant damage to the crown of P. menziesii trees. In the United States, triclopyr is indicated for injection as triethyl amine salt (Kochenderfer et al. Reference Kochenderfer, Kochenderfer and Miller2012); however, this product is not authorized in the region of the present study (SENASA 2024) and therefore is not locally sold. Future studies should address the optimal concentration of the aminopyralid + triclopyr formulation applied by the injection method. Season of application can affect herbicide efficacy, but this variation is usually species specific (Graziano et al. Reference Graziano, Tomco, Seefeldt, Mulder and Redman2022). Application during the growing season is usually recommended for most conifers (Ledgard Reference Ledgard2009); however, early fall application has resulted in significant defoliation rates with glyphosate at high concentrations. This may be due to active transport to the roots as the plant is preparing to enter dormancy before winter. Comparing the effectiveness of injection-applied herbicides as a function of seasonality in temperate forests remains to be assessed.
Economic Analysis
In terms of labor time, the action of injecting or felling 10,000 trees consumes 333 h and 362 h, respectively. Based on our economic analysis (Table 4), it would cost ∼US$2,900 to eliminate 10,000 medium to large trees (DBH between 20 and 30 cm) using the drill and fill method. In comparison, conventional felling would cost ∼ US$3,500 (Table 4). When comparing costs, no substantial differences between the two methods per se were found. This is partially due to the fact that equipment costs (e.g., a drill) are very high in the local market. In this sense, the drill could be replaced with cheaper tools that fulfill the same function, such as a hand drill. This could reduce the cost from US$2,900 to US$2,400, but might increase the operating time and therefore the operators’ pay. Mechanical removal requires the processing of post-felling residues to avoid the accumulation of dry woody material, as dead wood may hinder future regrowth of native plants (Boast Reference Boast2021) or increase the risk of forest fires (Holmes et al. Reference Holmes, Richardson, Van Wilgen and Gelderblom2000). Processing post-felling residues is quite time-consuming and, according to estimations, has a significantly high cost and increases the risk of injury to workers (Dampier et al. Reference Dampier, Bell, St-Amour, Pitt and Luckai2006). Mechanically removed trees are usually chipped in Nahuel Huapi National Park, so we estimated the cost of this particular task. According to the manufacturers, a large chipper can process about 7 trees per hour, so processing 10,000 trees takes 1,428 hours. The machinery rental as well as the fuel needed for this amount of time adds up to a cost of US$120,000 (Table 4). It is worth mentioning that the chippers process small- and medium-sized branches, so the residues from the main trunks must go through a preprocessing that includes delimbing the tree, cutting the trunk into operable pieces, and transporting and stockpiling the segments. The time and cost associated with all those stages are variable and difficult to calculate, as they depend on the distances and challenges presented by the terrain. In summary, the cost of applying the drill and fill method is approximately US$0.29 per tree, while felling and processing the residue would cost approximately US$12 per tree.
Detailed list of costs and inputs required to remove 10,000 invasive Pseudotsuga menziesii trees with diameter at breast height (DBH) between 20 and 30 cm using the standard felling and drill and fill methods
a US$6.64 per hour of work × 362 h for felling and × 333 h for drill and fill.
b Calculated for 0.09 L of chain lubricant, which is the consumption of a 2.4-kW chainsaw at its maximum power for 362 h.
c Calculated for 180 L of Roundup Full II® herbicide, which is required to inject 10,000 trees (0.018 L per tree). The price per liter of herbicide is US$0.32.
d Calculated for 5.02 L of 2t oil (two stroke (2T) oil for engines), which is the consumption of a 2.4-kW chainsaw at its maximum power for 362 h.
e Calculated for 241 L of fuel, which is the consumption of a 2.4-kW chainsaw at maximum power for 362 h.
Unlike the drill and fill method, felling and post-felling action often requires trained workers and heavy and difficult to transport equipment that can be expensive, with its use limited to easily accessible areas (Lange et al. Reference Lange, Boast and Kleynhans2022). This is particularly true in remote areas, natural parks, and ecological reserves. Large-scale removal programs using mechanical methods may have impacts on the surroundings, as falling trees can kill nearby native plants or expose mineral soil (Culliney Reference Culliney2005). Moreover, the disturbance associated with felling can promote the growth of other nonnative invasive species (Hierro et al. Reference Hierro, Villarreal, Eren, Graham and Callaway2006).
Despite its economic advantages, the injection method may not be suitable for treating large, densely invaded stands, as it could pose a potential risk to natural areas. Large-scale applications would favor a major accumulation of dry woody material in the coming years, which could increase the risk of forest fires (Clifford et al. Reference Clifford, Paul and Pearce2013). Furthermore, the residual biomass of dead wood and needles would result in a large input of carbon and nutrients modifying the abiotic properties of the ecosystem (Dickie et al. Reference Dickie, Sprague, Green, Peltzer, Orwin and Sapsford2022). Instead, we propose using the drill and fill method to treat medium to large trees in areas where the target individuals are near water sources, such as wetlands, lakes, or streams, or to stop long-distance dispersal prompted by isolated distant trees. An integrated woody invasive species management program should therefore consider including drill and fill for large isolated specimens while mechanically removing small trees. In protected areas, such as Nahuel Huapi National Park, this approach becomes an excellent option for managing not only P. menziesii invasion, but potentially other Pinaceae species present in the area in moderate to high abundance. Moreover, exploring the effectiveness of injecting other nonnative invasive species of the National Park, including Salix, Acer, Acacia, Sorbus, and Crataegus, would be of considerable interest.
This research underlines the efficacy of drill and fill method in controlling invasive P. menziesii trees. This method achieved a significant defoliation rate among the target individuals 2 yr after the application of glyphosate at high concentration, although the optimal concentration needs to be determined to balance costs and benefits. Drill and fill is a promising tool, both economically and environmentally, for controlling medium to large invasive trees in remote areas with high conservation value. This technique minimizes potential impacts on the ecosystem and eliminates the need for costly waste-removal processes associated with mechanical extraction methods.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/inp.2024.11
Acknowledgments
We thank the associate editor and the anonymous reviewers for their constructive comments on earlier versions of the article. We especially thank F. Merker and R. Páez for their valuable help during the fieldwork and A.L. Llanes for her kind contribution in the provision of field inputs.
Funding
This work was mainly funded by the Rufford Foundation, project ID 30162-1 awarded to MFS. MFS was also funded by a PICT 2016-1412 awarded to MAN and FPT. This scientific research was carried out with the authorization of the National Parks Administration (Autorización de Investigación No. 1696).
Competing interests
The authors declare no competing interests.
