There is an urgent need to design management strategies to reduce invasive species spread and impact, but the large spatial and temporal scales of most biological invasions make them challenging environments in which to conduct field studies. In this context, simulation models can play a key role in informing invasive species management. Woody plants are among the most harmful invaders, yet an evidence base to support different management objectives for these species remains poorly developed. Pines (Pinus) have been intensively studied, in terms of demography, dispersal, spread and impact, which makes them an ideal study system to model invasions. Using a multiyear database of observations of an invasive population, we employed an approximate Bayesian computation to fit an individual-based spatially-explicit model to compare management strategies to reduce the spread, population size and impact of a woody invader, Pinus contorta (pine hereafter), on grasslands in Patagonia. We simulated a pine population spreading from a plantation into a grassland over 50 years. Annual control actions in the grasslands started as soon as pines started spreading (when the pines from the plantation become reproductive) or were delayed at 10-year intervals. For control actions, we targeted different pine life stages, prioritized different locations in the landscape, and explored a wide range of available budgets. Removing saplings was the most cost-effective way to reduce pine spread and population size, whereas reducing management delay had a stronger effect on minimizing pine invasion impact on native grassland productivity. Focusing only on invasive adults was ineffective because it was costly, and it allowed a buildup in the population size of other stages which soon became adults (and started spreading seeds).

Synthesis and applications: Our highest-ranking strategies represent management actions to start implementing in the field as part of an adaptive management plan that iteratively evaluates the validity of our simulation model and updates the management recommendations. Our study can be applied to guide management of invasive pines and replicated with any invasive woody species with sufficient data.

We conducted virtual experiments applying a fully factorial design, whereby we ran five replicates of all the combinations of levels of the variables ‘management delay’, ‘stage targeted’, ‘control location’, and ‘total budget’. We simulated a pine population for 50 years, starting from a 7-year-old plantation (already producing seeds), adjacent to a native grassland. Management actions took place exclusively in the grassland, since the original plantation (southernmost eight cells) was assumed to be grown for commercial purposes. We simulated delayed control actions at 0, 10, 20, 30 or 40 years (from the moment pines started spreading, when the pines from the plantation become reproductive) to assess the consequences of delaying management. These incremental delays were based on real world scenarios where pine invasive populations have been spreading for one to four decades without management. Once control actions started, they were carried out annually. We targeted different life stages for control: seedlings (1-2 year old pines), saplings (3-6 years), subadults (7-12 years), adults (13 years or more) or all four stages (see Table 1 for a detailed description of the life stages). In scenarios where all life stages were removed simultaneously, different stages were selected at random until all trees were removed, or the available budget was spent. We selected cells for control actions at different locations of the landscape, according to different prioritization criteria (hereafter control location): 1. The random criterion selects occupied cells for management arbitrarily across the landscape. 2. The invasion front criterion prioritizes recently colonized cells. 3. The pine density criterion prioritizes cells with the highest population size of the targeted stage(s). 4. The impact criterion prioritizes cells which have higher accumulated impact on native grassland productivity. To quantify this impact of pine invasion on grassland productivity, we built density impact curves relating the effect of increasing pine density (for each pine stage) on native grassland productivity, based on data obtained from P. contorta invasions in native grasslands in Northwest Patagonia, Argentina (Moyano et al., 2023). These curves were fitted using an asymptotic function, with one single parameter (beta), which we included in our sensitivity analyses.

We also modified the total available budget, which limits the area of the landscape that can be managed each year (see below), to identify the minimum budget that achieves containment. To do so, first we explored a range of budgets, from 0 US$/year to 10000 US$/year, in increments of 500 US$/year to find approximate levels of investment sufficient to contain the invasive population for each level of management delay (which we expected to increase management costs). Once we identified these (e.g., 500 US$/year was enough for containment with no management delay), we reduced them in intervals of 100$/year to find a more precise minimum containment budget (e.g., 200 US$/year was the minimum budget that contained the invasive pines if management was not delayed).

We calculated management extent (i.e., the number of cells that could be managed every year) based on the cost of mechanical control actions, 2023 US$, of invasive P. contorta populations in National Reserve Malalcahuello, Chilean Patagonia (Naour et al., 2016). From experimental management plots, we obtained the total labor cost of removing individual pines of each stage. We carried out a simulated experiment in RS without management to obtain average densities for each pine stage across the invaded landscape in the absence of management. We used these densities to calculate an average cost per managed cell. For example, to calculate the average cost of removing all adults from a cell, we multiplied the cost of removing each adult by the average number of adults per cell. By dividing the total budget (US$/year) by the cost of removing pines in each cell, we obtained the total number of cells that could be managed per year. These values ranged from 1% to 100% of the invaded area, depending on the targeted stage(s) and the available budget. Smaller pine stages are more difficult to find in the field and, therefore, we assigned an increasing probability of detection to each pine stage, with their corresponding removal percentage. As a result, we assumed that, for each cell selected for management, control actions effectively removed 80% of seedlings (i.e. 20% of seedlings were missed because of their small size), 90% of saplings, and 99% of both subadults and adults.

Annual control actions were applied before dispersal. We assumed that local population size was imperfectly known by drawing values from a normal distribution centered on the true count and specifying a coefficient of variation of 5%. To assess if increasing this coefficient of variation in detected population size affected our response variables, we increased it to 7.5% and 10% and evaluated the proportion of variability in our response variables explained by these changes.  Each simulation was replicated five times. By the end of each simulation (year 50) we calculated three response variables: Pine invasion extent (the total number of grassland cells occupied by reproductive pines), invasive pine population size (the total number of pines of age > 0 years across the grassland) and pine invasion impact on grassland productivity (the accumulated impact of pine invasion on native grassland productivity across the landscape).

References

Moyano, J., Zamora-Nasca, L. B., Caplat, P., García-Díaz, P., Langdon, B., Lambin, X., Montti, L., Pauchard, A., & Nuñez, M. A. (2023). Predicting the impact of invasive trees from different measures of abundance. Journal of Environmental Management, 325, 116480. https://doi.org/10.1016/j.jenvman.2022.116480

Naour, M., García, R., & Pauchard, A. (2016). Rendimientos y factibilidad técnica de tres métodos diferentes de control de la invasión de Pinus contorta al interior de la Reserva Nacional Malalcahuello. Laboratorio de Invasiones Biológicas.