Climatic determinants of the Carpathian treeline and its projected upward shifts in response to climate change

Mkrtchian, Alexander; Mueller, Daniel Article — Published Version Climatic determinants of the Carpathian treeline and its projected upward shifts in response to climate change Climatic Change Provided in Cooperation with: Leibniz Institute of Agricultural Development in Transition Economies (IAMO), Halle (Saale) Suggested Citation: Mkrtchian, Alexander; Mueller, Daniel (2025) : Climatic determinants of the Carpathian treeline and its projected upward shifts in response to climate change, Climatic Change, ISSN 1573-1480, Springer Nature, Berlin, Vol. 178, Iss. 6, pp. 1-23, https://doi.org/10.1007/s10584-025-03947-y , https://link.springer.com/article/10.1007/s10584-025-03947-y This Version is available at: https://hdl.handle.net/10419/319153 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0 Received: 29 May 2024 / Accepted: 2 May 2025 © The Author(s) 2025 Alexander Mkrtchian [email protected] Daniel Mueller [email protected] 1 Leibniz Institute of Agricultural Development in Transition Economies (IAMO), TheodorLieser-Str. 2, 06120 Halle (Saale), Germany 2 Humboldt-Universität zu Berlin, Unter den Linden 6, 10099 Berlin, Germany Climatic determinants of the Carpathian treeline and its projected upward shifts in response to climate change AlexanderMkrtchian1· DanielMueller1,2 Climatic Change (2025) 178:109 https://doi.org/10.1007/s10584-025-03947-y Abstract Treelines represent a significant ecological boundary in mountainous regions. Changes in temperature and precipitation regimes due to climate change affect the location of treelines, contingent on fine-scale variations in orographic and climatic conditions. Using high-resolution satellite imagery, we identify the climatic treeline — the potential upper limit of forests determined by climatic conditions — in the Carpathian Mountains, one of Europe’s largest contiguous forest ecosystems. We downscale climate variables to a 30-m resolution by applying a polynomial approximation to the regression residuals, incorporating terrain attributes. We then correlate climatic variables with the location of the climatic treeline. The mean temperature of the warmest quarter demonstrates the strongest correlation with treeline location. We find a total area of 1,370 km2 above the current climatic treeline in the Carpathians, which constitutes the climatic envelope for alpine ecosystems. Depending on future climate projections, this area will decrease to 410–515 km2 by 2040, 100–320 km2 by 2060, and 15–290 km2 by 2080. The anticipated upward shift of the treeline jeopardizes the region's rare and endemic alpine species and has substantial ramifications for ecosystems, water balance, and the carbon cycle in the Carpathian Mountains. Our analysis highlights the importance of understanding how climate affects treeline locations for effective ecosystem management and conservation planning in a changing climate. Keywords Carpathian mountains · Forest ecotone · Mountain ecosystems · Ecosystem shift · Global warming · Climate impacts 1 3 Climatic Change (2025) 178:109 1 Introduction Trees disappear above a specific elevation, giving way to alpine meadows or alpine tundra, which have substantially lower aboveground biomass (Holtmeier 2009; Hansson et al. 2023). The line where trees disappear, the treeline, connects the highest patches of closed forest (Paulsen et al. 2000; Körner 2004; Czajka et al. 2015b). Typically, their location is determined by some threshold of tree height and canopy cover (Körner 1998; Wieser and Tausz 2007; Holtmeier 2009; Treml and Migoń 2015). The position of the treelines differs regionally and locally due to climatic and edaphic factors, local disturbance regimes, and anthropogenic land cover changes. Climatic treeline refers to the transitions determined by climatic conditions (Körner 2003; Holtmeier 2009). Climatic treelines exhibit consistent characteristics across various continents and latitudes, serving as crucial ecological divides and critical reference points for mountain life zones (Tranquillini 1979; Körner and Paulsen 2004; Körner et al. 2011). While temperature is widely acknowledged as the critical environmental determinant for the transition from forests to alpine shrubland and grassland, the precise climatic factors and physiological mechanisms governing treeline position remain unclear (Paulsen et al. 2000; Holtmeier and Broll 2007; Wieser and Tausz 2007; Smith et al. 2009). Factors such as precipitation and wind exposure, which influence the duration of snow cover, along with atmospheric CO₂ levels and soil nutrient status, also affect treeline dynamics; however, their impact is either less significant or locally specific. Traditionally, alpine treelines have been associated with a mean air temperature of approximately 10°C during the warmest month in temperate mountains (Grace 2002; Wieser and Tausz 2007; Richardson and Friedland 2009; Körner 2021). This value is substantially lower in tropical regions, where the growing season extends almost throughout the year (Körner and Paulsen 2004). Treelines are also associated with the duration of the growing season (Ellenberg and Leuschner 2010) and maximum daily temperatures in summer (Daubenmire 1954). Körner (Körner 1998, 2021) contends that the mean temperature of the growing season holds global significance in determining the treeline position. He attributes this to a physiological mechanism that inhibits apical meristem activity in response to low temperatures, impeding tree tissue formation. He defines the growing season for trees as the part of the year with average daily temperatures above 0.9⁰C (Körner et al. 2011). It can be assumed that no single overarching biological mechanism or quantitative parameter shapes global treeline positions. Instead, the diversity of regional and local climates, floral compositions, and the complexity of functional relationships result in distinct variations of altitudinal patterns of treelines (Smith et al. 2009). For separate regions with uniform ecological and topographical characteristics, the assumption that a single mechanism governs treeline location becomes more reasonable. However, the precise mechanisms and related climatic parameters determining regional climatic treeline locations remain an open question for most mountain ecosystems. Climate change, primarily through increased temperatures, prompts upward shifts in treelines, with significant implications for biodiversity, endemic alpine species, water balance, nutrient cycling, and carbon storage (Greenwood and Jump 2014; Hansson et al. 2021). Both climate change and associated treeline shifts affect forestry management and land use systems, such as alpine pastures (Cannone et al. 2007). While ample evidence exists regarding climatically induced treeline ascent since the late twentieth century (Holt1 3 109 Page 2 of 23 Climatic Change (2025) 178:109 meier 2009), these changes manifest slowly due to trees' delayed response to changing climate. Therefore, most currently reported treeline changes resulted mainly from changes in land use practices, including declining transhumance and logging (Gehrig‐Fasel et al. 2007; Weisberg et al. 2013). The patchy understanding of climate change impacts on treeline alterations is unfortunate amidst accelerating climate change and associated extreme weather events, which further increase pressure on mountain ecosystems, potentially reinforcing upward shifts in treeline locations. We investigate the factors influencing climatic treeline position and potential shifts in the treeline locations for the Carpathian Mountains, which span several countries in the center of Europe. The Carpathians contain many ecosystems and climatic zones, and are a hotspot of European biodiversity, hosting numerous rare and endemic species (Mráz and Ronikier 2016). With peak elevation reaching only up to 2,655 m, even modest upward advancements in the treeline could eliminate alpine habitats across many of their mountain ranges. While recent trends in the Carpathian treeline have been assessed with satellite data (Mihai et al. 2007; Martazinova et al. 2011; Weisberg et al. 2013), there is currently a lack of quantification regarding the influence of historical climate and expected future climate change on treeline location in the Carpathians. Our primary objective in this paper is to identify the climatic factors that determine the location of the current climatic treeline in the Carpathians and to project the potential impacts of climate change on future treeline shifts. To accomplish this, we delineated the extant current climatic treeline fragments on high-resolution remote sensing imagery. We downscaled the best available climatic data to assess the fine-scale variations in the determinants for the treeline location. Quantifying the climatic variables inside and outside these fragments revealed the variables with the strongest correlation with the treeline location. We then predicted the potential climatic treeline locations in the entire Carpathian range and the appropriate areas of climatic envelopes for alpine ecosystems based on the threshold value of the most pertinent climatic variable. Finally, we used data from climate change scenarios to project treeline shifts for 2040, 2060, and 2080, considering different Shared Socioeconomic Pathways (SSPs) (Copernicus Climate Change Service 2021). 2 Materials and methods 2.1 Study area The Carpathian Mountains are the easternmost offshoot of the Central European highlands, stretching 1,500 km from East to West over six countries and covering 190,000 km2 (Fig. 1). Most of this region is below the treeline, except for the highest mountain ranges. These ranges are approximately evenly distributed between three main subdivisions of the Carpathian Mountains: the (North-)Western, Eastern, and Southern Carpathians (e.g., Kondracki 1989). We inspected the highest parts of the Carpathians (above 1,300 m) for the location of treeline. At the same time, we used the entire Carpathian region along with adjacent areas, covering a total of 480,000 km2, for downscaling climatic variables. 1 3 Page 3 of 23 109 Climatic Change (2025) 178:109 2.2 Treeline delineation We define the treeline ecotone as the belt at the upper elevational limit of a closed-canopy forest, with a gradual vegetation transition from closed tree stands to open grasslands and shrubs. Diffuse treelines are more often limited by climatic factors, unlike abrupt and krummholz treelines, which are more likely to be shaped by anthropogenic and natural disturbances (Harsch et al. 2009; Hansson et al. 2021; see Fig. {5}). It is indirectly confirmed by gradual diffuse treelines being more responsive to climate warming (Tourville et al. 2023). We manually delineated treeline ecotones on high-resolution Bing aerial orthographic images (Microsoft n.d.) with 15–30 cm spatial resolution and a spatial accuracy of around 2 m. To do so, we digitized tree stands on the respective elevations with visible and rapid but gradual elevational deterioration of tree stature. We aimed at identifying the typical instances of natural climatic treeline throughout the Carpathian Mountains, digitizing tree stands above 1300 m a.s.l. that had characteristic elevational deterioration of tree stature (see Fig. {6}, Appendix). We identified 161 patches of climatic treeline in the highest mountain ranges of the Carpathians, together encompassing an area of nearly 12 km2. Our sample does not represent an exhaustive inventory of the remnants of the extant climatic treeline and may include some erroneous designations, which did not substantially influence our results. The distribution of the patches reveals three distinct clusters, each corresponding to a major subdivision of the Carpathian Mountains (Fig. 1). Delineated treeline patches are distributed almost uniformly in the highest mountain ranges of the three subdivisions. Substantial disparities in treeline elevation were observed among the subdivisions, primarily along a north–south gradient, indicating that the characteristic elevation of treeline patches changes with changing climatic conditions (Table {3}). Fig.1 Study area 1 3 109 Page 4 of 23 Climatic Change (2025) 178:109 2.3 Climatic variables and their spatial downscaling We utilized monthly climate data from 1970 to 2000 at 1 km2 resolution from WorldClim 2.1 (Fick and Hijmans 2017). We derived 12 climatic variables to serve as determinants for treeline location. Using climatic values for the end of the last century is justified by the lag of several decades in trees' response to changing climate. We selected variables that capture various aspects of the thermal regime, namely the annual mean temperature, mean temperature of the warmest quarter of the year, mean temperature of the warmest month (July), maximal temperature of the warmest month, and minimal temperature of the coldest month (the latter two variables reflect thermal and cold stress, respectively). Most of these variables indicate thermal resources available to plants during the most active growth phases, and all correlate with the altitude of treelines (Grace 2002; Körner and Paulsen 2004). We further incorporated the mean October temperature, which has been demonstrated to significantly predict treeline movement in the Northern Hemisphere (Hansson et al. 2023). We included growing season duration, growing season mean temperature, and accumulated growing degree days (AGDD). The latter is the cumulative sum of daily temperatures throughout the growing season that surpass a predefined threshold (Hansson et al. 2021). We used two threshold options: 0.9 °C, which defines the growing season in high-altitude treeline studies (Körner and Paulsen 2004; Paulsen and Körner 2014), and 5 °C, which captures the growing season definition of trees (Körner 2003). To accomplish this, we temporally downscaled the time series of monthly mean temperatures to daily time series with a cubic smoothing spline and calculated the duration of the growing seasons as the days when the temperature exceeded the thresholds, the mean temperature during the growing seasons, and the AGDD. We used elevation in meters above sea level (m a.s.l.) as another explanatory variable to verify if temperature-related variables were better predictors than differences in elevation alone. WorldClim data have already been used in the global analysis of treeline factors and distributions, but the 1 km resolution compromises the accuracy of the analysis (Körner et al. 2011; Haesen et al. 2023). We thus spatially downscaled the climate data to a resolution of 30 m with a spatial variability model. This model accounts for local terrain effects, such as temperature changes with elevation and potential influences of terrain aspect. It also addresses large-scale spatial variability, including temperature changes in the north–south direction and the impact of continentality. Our downscaling approach for the 12 climatic variables involved the following steps: 1) Regression models link the climatic variables to elevation and its two derivative components in the X and Y directions, which account for terrain aspect influences. Elevation data was sourced from the Shuttle Radar Topography Mission (SRTM) digital elevation model (version 3), available in 1-arcsecond (~ 30 m) resolution (NASA 2014). We used Akaike and Bayesian information criteria to decide whether to include terrain aspects in each variable's model. 2) The regression models predict the variables at a 30-m resolution. We then calculated the differences (residuals) between these predictions and the original 1-km resolution values of the variables. These residuals are likely due to the remaining large-scale spatial variability not accounted for by terrain attributes. 1 3 Page 5 of 23 109 Climatic Change (2025) 178:109 3) The residuals were smoothed using orthogonal polynomials in the form of a 6th-order trend surface. We then added the emerging large-scale patterns to the regression models' predictions. The final prediction, therefore, combines the fine details contingent on local terrain conditions with the broader patterns associated with latitude and largescale geographical features. We evaluated the downscaling accuracy with weather station data from the Global Historical Climatology Network (GHCN) dataset (Peterson and Vose 1997). Within the study region, we identified 28 weather stations with a minimum of 67% valid monthly records from 1970 to 1984 and from 1985 to 1999 (Fig. 1). We computed the respective climatic variables for these stations and determined the discrepancies between the observed weather station records, our downscaled data, and the original 1 km WorldClim data. This comparison enabled to examine how much our downscaled data enhanced the original 1 km WorldClim dataset. 2.4 Verifying relations between climatic variables and treeline location We first calculated the pairwise correlation caoefficients for all the explanatory variables to determine the degree of uniformity and detect possible redundancy. We then superimposed the downscaled climatic variables on the delineated treeline patches to assess their correspondence. We employed two approaches for the purpose. The first explorative approach assumes that variables with a stronger association with the treeline exhibit greater homogeneity within the treeline ecotone compared to the surroundings. To quantitatively assess this correspondence, we calculated the ratios of the standard deviations and coefficients of variation for the climatic variables within all digitized treeline patches to their respective background values. The background standard deviations and coefficients of variation were derived from the elevation belt between 1,300 and 1,900 m a.s.l. outside the patch boundaries. The smaller these ratios for a climatic variable, the stronger the correspondence between this variable and the treeline location. The second approach involved applying two machine learning models to relate the treeline with climatic variables sampled over more than 7 million pixels in the 1,300–1,900 m a.s.l. elevation belt, 18,450 of which were located inside the treeline patches. We used Random Forest (RF) and Gradient Boosting Machine (GBM) models from the R Caret package and automatically fine-tuned model parameters to select the “optimal” model (Kuhn 2008). First, we calculated statistics for RF variable importance of each climatic variable (Genuer et al. 2010) using 20 samples of 2000 data points – 1000 from areas inside and another 1000 outside the treeline patches. We repeated this procedure 20 times to compute means and standard deviations of the variable importance on each of these 20 samples (400 model runs in total). We then calculated statistics for the accuracy measures of RF and GBM models that included different explanatory variables and their combinations, using 50 replicated samples of the 2000 data points. Our aim was to identify the variables and their possible combinations that are most closely associated with treeline location. If one single explanatory variable proves sufficient for the purpose, the present-day climatic treeline can be defined by its threshold value. In addition, it allows identifying areas where disturbances caused the treeline to descend and defining the extent of the climatic envelope for alpine ecosystems above the treeline. 1 3 109 Page 6 of 23 Climatic Change (2025) 178:109 We then applied a separate procedure to estimate the locational accuracy of treeline delineations based on climatic thresholds. Specifically, we calculated the distances between the centroids of our treeline patches and the nearest estimated climatic treeline. Distances were assigned positive values for centroids located below the climatic treeline and negative values for those above it. We summarized these distances using two metrics: the mean distance (to capture directional bias) and the mean squared distance (to quantify overall deviation). Each metric was computed in two ways – first, treating all centroids equally, and second, applying weights proportional to the respective patch area, giving the centroids of larger patches greater influence. We also forecasted future climatic treeline shifts using climate projections. We used CMIP6 downscaled projections for four scenarios of the Shared Socioeconomic Pathways (SSPs) SSP1-2.6, SSP2-4.5, SPP3-7.0 and SSP5-8.5, each averaged over 20-year periods (2021–2040, 2041–2060, 2061–2080, 2081–2100); the scenario data are available through WorldClim (O’Neill et al. 2016; Fick and Hijmans 2017; Riahi et al. 2017). Averages and standard deviations were calculated for 23 global climate models (GCMs) for each SSP and period. We calculated the total area where the value of the indicative climatic variable is below the established threshold for the average of the models to estimate the extent of the future treeline ascent and the corresponding reduction in the climatic envelope for alpine ecosystems. 3 Results 3.1 Downscaling climatic variables The downscaling of climatic variables to a 30 m resolution substantially enhanced their spatial accuracy, as evidenced by comparing weather station data and original WorldClim data (Table 1). The extent of accuracy improvement differed among the variables: Minimal temperature showed only a modest gain, as anticipated, due to commonly occurring inverted vertical temperature gradients at night. The accuracy gain exceeded 20% for 8 out of 12 variables, affirming the effectiveness of the employed downscaling method. Consequently, we used the downscaled climatic layers in the subsequent analyses (see Fig. {7} for an illustration of the downscaling effect). 3.2 Relations between climatic variables and treeline location The values of pairwise correlation coefficients between the climatic variables suggest a high degree of collinearity (Fig. {8}). This is especially true for variables associated with warm-season temperatures (Tquart, TJul, Tmax, GP5sum, GP0.9sum), which exhibit correlation coefficients of 0.99 or higher among themselves. The Tmin variable, characterizing the minimal temperature of the coldest month, is the least correlated with other variables. Standard deviations and coefficients of variation for most climatic variables were significantly smaller within treeline patches than in the background elevation belt (Fig. 2; Table {4}). Only for annual mean temperature and elevation a.s.l. were they roughly equal, with a ratio close to 1. This indicates that most climatic variables are considerably better predictors for the climatic treeline locations than elevation alone. Variables characterizing tem1 3 Page 7 of 23 109 Climatic Change (2025) 178:109 peratures during the warmer part of the year exhibit the smallest standard deviation and coefficient of variation ratios. Conversely, minimal and annual mean temperatures had a significantly poorer match with the treeline location. The mean temperature of the warmest quarter (Tquart) variable exhibited significantly higher importance than other variables in the Random Forest models (Fig. 2; Table {5}). It also demonstrated the largest difference in the coefficient of variation between treeline patches and their background (Fig. 2; Table {4}). Calculated statistics for the accuracy measures of RF and GBM models that consider different explanatory variables and their combinations show that adding other variables didn’t improve the accuracy of GBM models and only marginally improved the accuracy of RF models (Table 2). The reason is the high multicollinearity of explanatory variables, particularly those related to warm season temperatures. We therefore found it appropriate to use the Tquart variable to indicate current and future climatic treeline locations. The threshold value for climatic treeline was determined by averaging the Tquart (mean temperature of the warmest quarter) variable across all treeline patches, resulting in a value of 9.96⁰C (see column 2 of Table {4} for the threshold values of other variables). Based on this parameter, we estimate the total area of the climatic envelope for alpine ecosystems to be 1,370 km2 (considering the mean 1970 to 2000 climate and the current ecosystem distribution). The treeline defined by the Tquart variable threshold aligns well with the observed treeline patches. Of the 161 delineated treeline patches, 81 are intersected by the defined threshold line, with 36 located above it and 44 below. The mean distance from patch cenTable1 Downscaling summary Climatic variable Regression model Root mean squared deviation from weather station data Accuracy gain for downscaled data, % Variables in the model, sign Multiple R-squared Original WorldClim data Downscaled data Tann Elev, AspX, AspY 0.75 0.78 0.73 5.92 Tquart Elev, AspY 0.82 0.61 0.48 21.5 TJul Elev, AspY 0.83 0.71 0.52 26.2 TOct Elev, AspX 0.72 0.54 0.42 20.3 Tmax Elev 0.80 1.04 0.72 30.4 Tmin Elev, AspY 0.51 1.42 1.40 1.40 GP0.9leng Elev 0.60 283.4 204.8 27.7 GP0.9mean Elev, AspY 0.85 0.26 0.21 18.1 GP0.9sum Elev 0.76 31.8 25.7 19.3 GP5leng Elev 0.47 320.9 159.4 50.3 GP5mean Elev, AspX 0.74 0.72 0.31 56.2 GP5sum Elev 0.74 37.0 28.45 23.1 Climatic variables: Tann Annual mean temperature; Tquart Mean temperature of the warmest quarter; TJul Mean July temperature; TOct Mean October temperature; Tmax Mean maximal temperature of the warmest month; Tmin Mean minimal temperature of the coldest month; GP0.9leng Duration of the growing season above 0.9 °C; GP0.9mean Mean temperature of the growing season above 0.9 °C; GP0.9sum AGDD of the growing season above 0.9 °C; GP5leng Duration of the growing season above 5 °C; GP5mean Mean temperature of the growing season above 5 °C; GP5sum AGDD of the growing season above 5 °C. Explanatory terrain attributes (Column 2): Elev – Elevation (m a.s.l.); AspX – elevation derivative in X direction (north–south); AspY – elevation derivative in Y direction (east–west) 1 3 109 Page 8 of 23 Climatic Change (2025) 178:109 2005; Reyes-Fox et al. 2014), which could amplify and even locally outweigh the effects of predicted temperature increases (Higgins and Scheiter 2012). If our assumption that summer temperatures are the leading factor behind climatic treelines is valid, the drastic decrease in the climatic envelope for alpine ecosystems in the Carpathians should be expected by the end of the century. This means that the efforts for the conservation of alpine communities, including their endemic and endangered species, should be spatially directed towards the highest ridges and peaks in the Carpathians, in particular the Tatras and Southern Carpathians, where favorable climatic conditions for these ecosystems would sustain for longer. The choice of SSP scenario significantly influences the projections, with the divergence in the estimated areas above the climatic treeline increasing progressively over time across the four scenarios (Fig. 4). This means that global measures to slow down climate warming are essential for reducing the detrimental effects of climate change on mountain ecosystems. 5 Conclusions Our study reveals the climatic factors that govern the current treeline position in the Carpathians, and projects its upward shifts under climate change. Warm season temperatures are the most influential for the upper bounds of tree growth and the mean temperature of the warmest quarter exhibits the strongest correlation with treeline location among the 12 climatic variables tested in our study. We used the estimated threshold value of the mean temperature of the warmest quarter to assess the current total area of the climatic envelope for alpine ecosystems at 1,370 km2. Using CMIP6 climate projections, we forecast the decrease of alpine ecosystems to less than a quarter of their current size by 2050. In the highest emission scenarios, alpine ecosystems in the Carpathians will almost entirely disappear by 2080. Our work corroborates the importance of reducing greenhouse emissions to safeguard alpine ecosystems but also underscores the need for adaptation measures to support mountain regions. We fill a regional gap in quantifying treeline behavior under climate change. Unfortunately, such estimates are still lacking for other critical mountain systems, such as the neighboring Alps and Dinaric mountains, which share comparable climate and ecosystem properties. We hope our results contribute to more detailed studies that delve into physiological mechanisms and environmental interconnections to facilitate a deeper understanding of the ecological consequences of climate warming. Such knowledge is crucial for improving land use management and adaptation strategies for mountain forestry, agriculture, and nature conservation amidst accelerating climate change. 1 3 Page 15 of 23 109 Climatic Change (2025) 178:109 Appendix Fig.6 Two examples of manually identified instances of natural climatic treeline Fig.5 Land use mosaics of forest and grassland patches shaped by humans (top row) and climatic treeline ecotones (bottom row). Natural treelines typically gradually transition from forest to grassland along altitudinal gradients, whereas abrupt forest edges are generally a sign of human or natural disturbances 1 3 109 Page 16 of 23 Climatic Change (2025) 178:109 Fig.7 Fragments of initial (top) and downscaled (bottom) layers of mean July temperature 1 3 Page 17 of 23 109 Climatic Change (2025) 178:109 Table3 Distribution and characteristics of identified treeline patches Carpathian subdivision Number of patches Total area, km2 Mean parcel area, ha Elevation, mean, m Elevation, median, m Elevation, standard deviation, m Elevation, minimal, m Elevation, maximal, m North-West 48 3.19 6.65 1,578 1,554 97 1,384 1,858 East 69 5.25 7.61 1,684 1,695 90 1,453 1,921 South 44 3.54 8.05 1,848 1,845 42 1,701 1,983 Total 161 11.98 7.44 1,704 1,723 131 1,384 1,983 Fig.8 Pairwise correlation coefficients between the climatic variables 1 3 109 Page 18 of 23 Climatic Change (2025) 178:109 Table4 Means, standard deviations and coefficients of variation of downscaled climatic variables within treeline patches and in the background (1300–1900 m a.s.l. elevation belt in Carpathians) Climatic variable Treeline patches In background Ratio of SD Ratio of CV Mean SD CV Mean SD CV Tann, °C 1.52 0.62 0.002 2.24 0.68 0.002 0.92 0.93 Tquart, °C 9.96 0.42 0.001 11.17 0.90 0.003 0.47 0.47 Tjul, °C 10.54 0.44 0.002 11.75 0.90 0.003 0.49 0.49 Toct, °C 3.03 0.44 0.002 4.04 0.80 0.003 0.55 0.55 Tmax, °C 15.17 0.52 0.002 16.62 1.08 0.004 0.49 0.49 Tmin, °C -12.5 0.64 0.002 -11.49 0.98 0.004 0.64 0.65 GP0.9leng, days 191.2 6.93 0.036 206.5 12.63 0.061 0.55 0.59 GP0.9mean, °C 7.62 0.24 0.0008 8.21 0.463 0.002 0.51 0.51 GP0.9sum, AGDD 1243 116.4 0.002 1571 246.1 0.004 0.47 0.51 GP5leng, days 143.9 5.86 0.041 157.5 11.42 0.073 0.51 0.56 GP5mean, °C 9.31 0.30 0.001 10.02 0.54 0.002 0.56 0.56 GP5sum, AGDD 1112 115.8 0.003 1436 247.4 0.006 0.47 0.51 Elevation, m a.s.l 1704 130.5 0.077 1572 131.8 0.094 0.99 0.91 SD standard deviation (patches/background); CV coefficient of variation (patches /background) Climatic variables: Tann Annual mean temperature; Tquart Mean temperature of the warmest quarter; TJul Mean July temperature; TOct Mean October temperature; Tmax Mean maximal temperature of the warmest month; Tmin Mean minimal temperature of the coldest month; GP0.9leng Duration of the growing season above 0.9°C; GP0.9mean Mean temperature of the growing season above 0.9°C; GP0.9sum AGDD of the growing season above 0.9°C; GP5leng Duration of the growing season above 5°C; GP5mean Mean temperature of the growing season above 5°C; GP5sum – AGDD of the growing season above 5°C Variable Variable importance Mean Standard deviation Tann 15.24 9.59 Tquart 99.68 0.4 Tjul 49.09 14.26 Toct 2.23 2.04 Tmax 17.12 9.11 Tmin 14.66 4.19 GP0.9leng 1.04 0.97 GP0.9mean 37.5 12.65 GP0.9sum 3.53 2.25 GP5leng 18.64 8.11 GP5mean 16.54 8.49 GP5sum 22.07 10.13 Table5 Variables importance in a set of Random forest models with 20 subsamples of 2000 data points (see text) 1 3 Page 19 of 23 109 Climatic Change (2025) 178:109 Acknowledgments We wish to acknowledge the anonymous reviewers for their thoughtful comments and constructive suggestions, which greatly improved the quality and clarity of this manuscript. Author contributions Both authors contributed to the study conception and design, data collection and analysis, writing the first draft of the manuscript and its subsequent editing. All authors read and approved the final manuscript. Funding Open Access funding enabled and organized by Projekt DEAL. The authors declare that no funds, grants, or other support were received during the preparation of this manuscript. Data availability The research is based on open data mentioned and cited it the manuscript. Datasets created during the research process are accessible from Zenodo open repository by the following links: Estimated climate treeline in Carpathians: https://zenodo.org/records/11359344 Predicted climatic treeline location in Carpathians: https://zenodo.org/records/11358952 Carpathian treeline samples: https://zenodo.org/records/11320218 Centroids of treeline patches: https://zenodo.org/records/15282320 Code availability Code for testing the significance of climatic variables in ML models: h t t p s : / / g i t h u b . c o m / a l e m k r t / C l i m a t i c _ d a t a / b l o b / m a i n / V a r i a b l e _ t e s t i n g . R Intermediate data and code are available from the corresponding author by request. Declarations Conflict of interest The authors have no relevant financial or non-financial interests to disclose. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted Years SSP 2021–40 2041–60 2061–80 1–2.6 514.42 318.35 288.81 2–4.5 506.15 227.87 110.61 3–7.0 480.01 169.96 40.37 5–8.5 409.93 99.55 14.76 Table7 Estimated areas, km2, above climatic treeline based on downscale averages for 23 GCMs projections of the mean temperature of the warmest quarter (current value estimated at 1,370 km2) Variable Unweighted centroids Weighted centroids Mean distance Root mean squared distance Mean distance Root mean squared distance Tquart 140.78 206.1 144.8 220.4 Tann 113.8 348.7 254.5 380.2 TJul 145.6 210.5 151.9 230.8 Tmax 192.7 440.4 183.7 366.8 GP5summ 147.1 207.0 159.6 232.3 GP0.9mean 140.7 214.9 152.2 243.2 Elevation 331.6 1635.6 520 1247.2 Table6 Distances between patch centroids and the nearest treeline, calculated by treating all centroids equally, and by applying weights to centroids proportional to the areas of the respective patches 1 3 109 Page 20 of 23 Climatic Change (2025) 178:109 by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. References Cannone N, Sgorbati S, Guglielmin M (2007) Unexpected impacts of climate change on alpine vegetation. Front Ecol Environ 5:360–364. h t t p s : / / d o i . o r g / 1 0 . 1 8 9 0 / 1 5 4 0 - 9 2 9 5 ( 2 0 0 7 ) 5 [ 3 6 0 : U I O C C O ] 2 . 0 . C O ; 2 Chen I-C, Hill JK, Ohlemüller R et al (2011) Rapid Range Shifts of Species Associated with High Levels of Climate Warming. Science 333:1024–1026. https://doi.org/10.1126/science.1206432 Conlisk E, Castanha C, Germino MJ et al (2017) Declines in low-elevation subalpine tree populations outpace growth in high-elevation populations with warming. J Ecol 105:1347–1357. h t t p s : / / d o i . o r g / 1 0 . 1 1 1 1 / 1 3 6 5 - 2 7 4 5 . 1 2 7 5 0 Copernicus Climate Change Service (2021) CMIP6 predictions underpinning the C3S decadal prediction prototypes. Climate Data Store (CDS), Copernicus Climate Change Service. h t t p s : / / c d s . c l i m a t e . c o p e r n i c u s . e u Accessed 24 Apr 2025 Czajka B, Łajczak A, Kaczka RJ (2015a) The dynamics of the timberline ecotone on the asymmetric ridge of the Babia Góra Massif, Western Carpathians. Geogr Pol 88:85–102. https://doi.org/10.7163/GPol.0017 Czajka B, Łajczak A, Kaczka RJ, Nicia P (2015b) Timberline in the Carpathians: An overview. Geogr Pol 88:7–34. https://doi.org/10.7163/GPol.0013 Daubenmire R (1954) Alpine Timberlines in the Americas and Their Interpretation. Butler Univ Bot Stud 11:119–136 Davis EL, Brown R, Daniels L et al (2020) Regional variability in the response of alpine treelines to climate change. Clim Change 162:1365–1384. https://doi.org/10.1007/s10584-020-02743-0 Ellenberg H, Leuschner C (2010) Vegetation Mitteleuropas mit den Alpen: in ökologischer, dynamischer und historischer Sicht. 6., vollst. neu bearb. und stark erw. Aufl. Ulmer, Stuttgart. h t t p s : / / d o i . o r g / 1 0 . 3 6 1 9 8 / 9 7 8 3 8 2 5 2 8 1 0 4 5 Fick SE, Hijmans RJ (2017) WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. Int J Climatol 37:4302–4315. https://doi.org/10.1002/joc.5086 Gehrig-Fasel J, Guisan A, Zimmermann NE (2007) Tree line shifts in the Swiss Alps: climate change or land abandonment? J Veg Sci 18:571–582. h t t p s : / / d o i . o r g / 1 0 . 1 1 1 1 / j . 1 6 5 4 - 1 1 0 3 . 2 0 0 7 . t b 0 2 5 7 1 . x Genuer R, Poggi J-M, Tuleau-Malot C (2010) Variable selection using random forests. Pattern Recogn Lett 31:2225–2236. https://doi.org/10.1016/j.patrec.2010.03.014 Grace J (2002) Impacts of climate change on the tree line. Ann Bot 90:537–544. h t t p s : / / d o i . o r g / 1 0 . 1 0 9 3 / a o b / m c f 2 2 2 Greenwood S, Jump AS (2014) Consequences of treeline shifts for the diversity and function of high altitude ecosystems. Arct Antarct Alp Res 46:829–840. https://doi.org/10.1657/1938-4246-46.4.829 Haesen S, Lembrechts JJ, De Frenne P et al (2023) ForestClim —bioclimatic variables for microclimate temperatures of European forests. Glob Change Biol 29:2886–2892. https://doi.org/10.1111/gcb.16678 Handa IT, Körner C, Hättenschwiler S (2005) A test of the carbon limitation hypothesis by in situ CO 2 enrichment and defoliation. Ecology 86:1288–1300. https://doi.org/10.1890/04-0711 Hansson A, Dargusch P, Shulmeister J (2021) A review of modern treeline migration, the factors controlling it and the implications for carbon storage. J Mt Sci 18:291–306. h t t p s : / / d o i . o r g / 1 0 . 1 0 0 7 / s 1 1 6 2 9 - 0 2 0 - 6 2 2 1 - 1 Hansson A, Yang W-H, Dargusch P, Shulmeister J (2023) Investigation of the relationship between treeline migration and changes in temperature and precipitation for the Northern Hemisphere and Sub-regions. Curr Forestry Rep 9:72–100. https://doi.org/10.1007/s40725-023-00180-7 Harsch MA, Hulme PE, McGlone MS, Duncan RP (2009) Are treelines advancing? A global meta-analysis of treeline response to climate warming. Ecol Lett 12:1040–1049. h t t p s : / / d o i . o r g / 1 0 . 1 1 1 1 / j . 1 4 6 1 - 0 2 4 8 . 2 0 0 9 . 0 1 3 5 5 . x Higgins SI, Scheiter S (2012) Atmospheric CO2 forces abrupt vegetation shifts locally, but not globally. Nature 488:209–212. https://doi.org/10.1038/nature11238 Hijmans RJ, Cameron SE, Parra JL et al (2005) Very high-resolution interpolated climate surfaces for global land areas. Int J Climatol 25:1965–1978. https://doi.org/10.1002/joc.1276 Holtmeier F-K (2009) Mountain timberlines: ecology, patchiness, and dynamics, 2nd edn. Springer, Dordrecht. https://doi.org/10.1007/978-1-4020-9705-8 Holtmeier FK, Broll GE (2007) Treeline advance - driving processes and adverse factors. LO 1:1–33. h t t p s : / / d o i . o r g / 1 0 . 3 0 9 7 / L O . 2 0 0 7 0 1 1 3 Page 21 of 23 109 Climatic Change (2025) 178:109 Intergovernmental Panel on Climate Change (IPCC) (2023) Climate change 2021 – the physical science basis: working group I contribution to the sixth assessment report of the intergovernmental panel on climate change, 1st edn. Cambridge University Press. https://doi.org/10.1017/9781009157896 Kaczka RJ, Lempa M, Czajka B et al (2015) The recent timberline changes in the Tatra Mountains: A case study of the Mengusovská Valley (Slovakia) and the Rybi Potok Valley (Poland). Geogr Pol 88:71–83. https://doi.org/10.7163/GPol.0016 Kondracki J (1989) Karpaty [The Carpathians]. Wydawnictwo Szkolne i Pedagogiczne, Warszawa Körner C (1998) Worldwide positions of alpine treelines and their causes. In: Beniston M, Innes JL (eds) The Impacts of Climate Variability on Forests. Springer-Verlag, Berlin/Heidelberg, pp 221–229 Körner C (2003) Alpine Plant Life. Springer, Berlin Heidelberg, Berlin, Heidelberg Körner C (2004) Mountain biodiversity, its causes and function. AMBIO: J Hum Environ 33:11. h t t p s : / / d o i . o r g / 1 0 . 1 0 0 7 / 0 0 4 4 - 7 4 4 7 - 3 3 . s p 1 3 . 1 1 Körner C (2021) Alpine plant life: functional plant ecology of high mountain ecosystems, 3rd edn. Springer, Cham, Switzerland Körner C, Paulsen J (2004) A world-wide study of high altitude treeline temperatures: Study of high altitude treeline temperatures. J Biogeogr 31:713–732. h t t p s : / / d o i . o r g / 1 0 . 1 1 1 1 / j . 1 3 6 5 - 2 6 9 9 . 2 0 0 3 . 0 1 0 4 3 . x Körner C, Paulsen J, Spehn EM (2011) A definition of mountains and their bioclimatic belts for global comparisons of biodiversity data. Alp Botany 121:73. https://doi.org/10.1007/s00035-011-0094-4 Kuhn M (2008) Building predictive models in R using the Caret package. J Stat Soft 28:1–26. h t t p s : / / d o i . o r g / 1 0 . 1 8 6 3 7 / j s s . v 0 2 8 . i 0 5 Lyu L, Büntgen U, Li M et al (2025) Intensifying neighbouring tree competition suppresses tree growth at the eastern Tibetan tree line. Funct Ecol 1365–2435:70040. https://doi.org/10.1111/1365-2435.70040 Malanson GP, Resler LM, Bader MY et al (2011) Mountain treelines: a roadmap for research orientation. Arct Antarct Alp Res 43:167–177. https://doi.org/10.1657/1938-4246-43.2.167 Malinovsky K (1980) Vegetation in Ukrainian Carpathian highlands [in Ukrainian]. Naukova dumka, Kiev Martazinova V, Ivanova O, Shandra O (2011) Climate and treeline dynamics in the Ukrainian Carpathians Mts. Folia Oecologica 38:65–71 Microsoft (n.d.) Bing Maps Aerial Imagery. https://www.bing.com/maps. Accessed 2 May 2024 Mihai B, Savulescu I, Sandric I (2007) Change detection analysis (1986–2002) of vegetation cover in Romania: a study of alpine, subalpine, and forest landscapes in the Iezer Mountains, Southern Carpathians. Mt Res Dev 27:250–258. https://doi.org/10.1659/mred.0645 Moncrieff GR, Bond WJ, Higgins SI (2016) Revising the biome concept for understanding and predicting global change impacts. J Biogeogr 43:863–873. https://doi.org/10.1111/jbi.12701 Mráz P, Ronikier M (2016) Biogeography of the Carpathians: evolutionary and spatial facets of biodiversity. Biol J Linn Soc 119:528–559. https://doi.org/10.1111/bij.12918 NASA (2014) SRTMGL1 V003: SRTM 1 Arc-SecondGlobal [Data set]. NASA EOSDIS Land Processes Distributed Active Archive Center (LP DAAC). h t t p s : / / d o i . o r g / 1 0 . 5 0 6 7 / M E a S U R E s / S R T M / S R T M G L 1 . 0 0 3 O’Neill BC, Tebaldi C, Van Vuuren DP et al (2016) The scenario model intercomparison project (ScenarioMIP) for CMIP6. Geosci Model Dev 9:3461–3482. https://doi.org/10.5194/gmd-9-3461-2016 Paulsen J, Körner C (2014) A climate-based model to predict potential treeline position around the globe. Alp Botany 124:1–12. https://doi.org/10.1007/s00035-014-0124-0 Paulsen J, Weber UM, Körner Ch (2000) Tree growth near treeline: abrupt or gradual reduction with altitude? Arct Antarct Alp Res 32:14–20. h t t p s : / / d o i . o r g / 1 0 . 1 0 8 0 / 1 5 2 3 0 4 3 0 . 2 0 0 0 . 1 2 0 0 3 3 3 4 Peterson TC, Vose RS (1997) An overview of the global historical climatology network temperature database. Bull Amer Meteor Soc 78:2837–2849. h t t p s : / / d o i . o r g / 1 0 . 1 1 7 5 / 1 5 2 0 - 0 4 7 7 ( 1 9 9 7 ) 0 7 8 / 2 8 3 7 : A O O T G H / 2 . 0 . C O ; 2 Quervain A (1904) Die Hebung der atmosphärischen lsothermenin der Schweizer Alpen und ihre Beziehung zu deren Höhengrenzen. Gerlands Beitr Geophys 6:481–533 Reyes-Fox M, Steltzer H, Trlica MJ et al (2014) Elevated CO2 further lengthens growing season under warming conditions. Nature 510:259–262. https://doi.org/10.1038/nature13207 Riahi K, Van Vuuren DP, Kriegler E et al (2017) The shared socioeconomic pathways and their energy, land use, and greenhouse gas emissions implications: an overview. Glob Environ Chang 42:153–168. h t t p s : / / d o i . o r g / 1 0 . 1 0 1 6 / j . g l o e n v c h a . 2 0 1 6 . 0 5 . 0 0 9 Richardson AD, Friedland AJ (2009) A review of the theories to explain arctic and alpine treelines around the world. J Sustain for 28:218–242. https://doi.org/10.1080/10549810802626456 Smith WK, Germino MJ, Johnson DM, Reinhardt K (2009) The altitude of alpine treeline: a bellwether of climate change effects. Bot Rev 75:163–190. https://doi.org/10.1007/s12229-009-9030-3 Svenning J-C, Skov F (2007) Could the tree diversity pattern in Europe be generated by postglacial dispersal limitation? Ecol Letters 10:453–460. h t t p s : / / d o i . o r g / 1 0 . 1 1 1 1 / j . 1 4 6 1 - 0 2 4 8 . 2 0 0 7 . 0 1 0 3 8 . x 1 3 109 Page 22 of 23 Climatic Change (2025) 178:109 Szwagrzyk J (2015) Structure of the forest ecotone in the Babia Góra Massif, Western Carpathians. Geogr Pol 88:103–113. https://doi.org/10.7163/GPol.0018 Thuiller W, Lavorel S, Araújo MB et al (2005) Climate change threats to plant diversity in Europe. Proc Natl Acad Sci USA 102:8245–8250. https://doi.org/10.1073/pnas.0409902102 Tinner W, Kaltenrieder P (2005) Rapid responses of high-mountain vegetation to early Holocene environmental changes in the Swiss Alps. J Ecology 93:936–947. h t t p s : / / d o i . o r g / 1 0 . 1 1 1 1 / j . 1 3 6 5 - 2 7 4 5 . 2 0 0 5 . 0 1 0 2 3 . x Tourville J, Publicover D, Dovciak M (2023) Forests on the move: tracking climate-related treeline changes in mountains of the northeastern United States. J Biogeogr 50:1993–2007. h t t p s : / / d o i . o r g / 1 0 . 1 1 1 1 / j b i . 1 4 7 0 8 Tranquillini W (1979) Physiological Ecology of the Alpine Timberline: Tree Existence at High Altitudes with Special Reference to the European Alps. Springer, Berlin Heidelberg, Berlin, Heidelberg Treml V, Migoń P (2015) Controlling factors limiting timberline position and shifts in the Sudetes: a review. Geogr Pol 88:55–70. https://doi.org/10.7163/GPol.0015 Vincze I, Orbán I, Birks HH et al (2017) Holocene treeline and timberline changes in the South Carpathians (Romania): climatic and anthropogenic drivers on the southern slopes of the Retezat Mountains. The Holocene 27:1613–1630. https://doi.org/10.1177/0959683617702227 Weisberg PJ, Shandra O, Becker ME (2013) Landscape influences on recent timberline shifts in the Carpathian Mountains: abiotic influences modulate effects of land-use change. Arct Antarct Alp Res 45:404– 414. https://doi.org/10.1657/1938-4246-45.3.404 Wieser G, Tausz M (2007) Current Concepts for Treelife Limitation at the Upper Timberline. In: Wieser G, Tausz M (eds) Trees at their Upper Limit. Springer Netherlands, Dordrecht, pp 1–18. h t t p s : / / d o i . o r g / 1 0 . 1 0 0 7 / 1 - 4 0 2 0 - 5 0 7 4 - 7 _ 1 Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 1 3 Page 23 of 23 109