**1. Introduction**

In Uruguay, commercially, *Eucalyptus* trees have been planted in different areas of the country and today trees of this genus occupy 726,000 ha [1]. There is still a large area of soils prioritized for forestry by the government that could be planted, according to the current legislation (3.288 million ha, approximately).

Even though different ecoregions of Uruguay offer a variety of conditions for forest ecosystems [2], many of the planted species have enough plasticity to adapt to a wide range of environmental conditions. According to some authors [3,4], *Eucalyptus* species are adequate for studies of climate change adaptability because trees of this genus are planted in a broad range of environments. This indicates that, in general, eucalypts are relatively less affected by climate changes [5] although this is closely related to the temperature of the areas of origin of the different genetic materials [6]. *Eucalyptus grandis* and *E. dunnii* are two of the most planted tree species in Uruguay, showing great adaptation to soils with low fertility and moderate dryness [7]. Furthermore, according to [8], eucalypt species have a high ability to adapt to climate changes due to their relatively short harvest cycles, making it faster to identify optimal genotypes for different conditions [7]. For the Campos region (Southern Brazil, Uruguay, and the center of Argentina occupying an area of 500,000 km2), the evolution of climate has been studied for the last 70 years. The trend of the records shows that there has been an increase in spring-summer rainfall, a decrease in the maximum temperature in summer, an increase in the minimum annual temperature and a reduction in the frost period [9]. The projections made by these authors indicate that there will be a slight increase in rainfall in the summer months and an increase in maximum and minimum temperatures throughout the year. These types of changes do not imply a relocation of the species because the eventual climate alterations do not necessarily represent a limit to their distribution [5]. Predictions for 2080 for the South of Brazil (among other areas) indicate a decrease in the area considered adequate for *E. grandis*, explained by an alteration in the precipitation regime [10], along with an increment in evapotranspiration [11]. On the other hand, the expected increment in the CO2 levels in the atmosphere in the coming decades, with respect to the current one [12], will also affect the physiological activities and plant growth. According to several studies, this increase would determine an acceleration of the growth rates of some species due to a higher photosynthetic rate in situations where there are no limiting factors [13–15]. On the other hand, the response of the eucalypts to the increase in the concentration of CO2 in the atmosphere is related to changes in the ambient temperature so that the increase of both factors improved the growth of greenhouse plants without water deficit [16].

In general, physiological responses to climate change are known at an individual tree level but less known for populations [17], given the interaction of multiple factors involved. There is evidence that an increment in temperature would impact negatively on trees with higher "maintenance costs" (high physiological activity), such as larger trees, compared to smaller ones [3,18]. However, other authors argue that acclimation capacity of forests allow to maintain gross primary production/net primary production ratios [19].

The site requirements of several of the most commercially important eucalyptus species are wide enough to achieve good conditions of adaptation to the conditions of Uruguay in the current climate scenario [17]. The environmental requirements of *Eucalyptus grandis* and *E. dunnii* show that the different regions of the country have favorable site conditions for their installation [20–22] (Table S1 Supplementary Materials).

One way to understand the eventual effect of climate change on the behavior (distribution and/or productivity) of a species is through predictive models of species distribution [18]. One of the inconveniences that arise when applying (Species Distribution Models) SDM is that there are a great number of available alternatives, which, in some cases, provide different results; this complicates the choice of the best option for each case [19]. According to these authors, this kind of situation happens when the priority is to predict the distribution of a species as a function of different scenarios of climate change. Another disadvantage may appear when many predictive environmental variables are used, producing an over-adjustment [20]. Over-adjustments frequently reduce the applicability of the models to a new set of data [21]. One way to overcome this problem is by using ensemble methods (or ensemble techniques), to obtain ensemble models with greater precision than the individual counterparts. An example of these techniques is the biomod2 R package [22], which contains the four most used modelling tools for species' prediction. This model uses a reduced number of predictive variables,

compared with simple models, and has been applied to project the distribution of species in different climate change scenarios [23]. Besides, some authors suggest the use of several types of algorithms (instead of just one) to predict the distribution of species since this allows to reduce uncertainty in marginal habitat situations [24]. Finally, SDM have been applied in this context of prediction of the potential distribution non-native species, particularly invasive non-native species [25]. Less frequent has been its use for introduced forest species with yield potential. However, those models may help us to understand the biological impacts and management practices for environmental suitability.

We hypothesized that the climate change that he expected climate change occurs in the region occupied by Uruguay has effect both on the area planted with the *Eucalyptus* species and on their productivity. The general objective of this work was to predict the current and future potential habitat of *Eucalyptus grandis* and *E. dunnii* in soils prioritized for forestry in Uruguay, under different future climate change scenarios. The specific objectives were (i) to identify the environmental variables that show the greatest association with the probability of occurrence of both species, (ii) to fit a prediction model of current and future habitat for the studied species, and (iii) to identify the regions with the greatest potential for the growth of these species.
