Global changes in our environment, such as climate change, have the potential to greatly influence the spatial distribution of plant species (Lenoir & Svenning, 2015;Pecl et al., 2017;Scheffers et al., 2016).
Such changes in species distributions have already been demonstrated and are predicted to continue through the 21st century (Iverson et al., 2019;Parmesan, 2006;Prasad et al., 2020). A large body of research has been devoted to detecting shifts in tree species ranges, possible proximal causes of such shifts, and their ecosystem level consequences (Iverson et al., 2019;O'Sullivan et al., 2021;Prasad et al., 2020;Wason & Dovciak, 2017). Elevational vegetation gradients along mountain slopes covary with temperature (in addition to variation in precipitation and soils) and thus mountains can be used as natural study regions to detect early changes in forest communities caused by changing climate (Beckage et al., 2008;Brown & Vellend, 2014;Harsch et al., 2009;Jump et al., 2009;O'Sullivan et al., 2021;Tourville et al., 2022). This is also true of mountains in the northeastern United States, which are expected to experience increased temperatures (particularly during winter months) and fluctuating precipitation regimes over the next several decades (Burakowski et al., 2022;Janowiak et al., 2018;Wason, Bevilacqua, & Dovciak, 2017).
The alpine treeline ecotone, which forms the transition (ecotone) from high-elevation conifer forests to krummholz and alpine zones, has often been used as a barometer for the impact of climate change across the globe (Holtmeier & Broll, 2007;Kimball & Weihrauch, 2000;Smith et al., 2009). As the climate warms, treelines are expected to advance upslope modifying (or outright replacing) the unique and diverse alpine plant communities of mountain summits, generally comprised of low-statured species that are ill-suited to compete with trees (Körner, 2012). Indeed, several studies that have examined coniferous forests at high elevations across the globe document upslope species shifts in tree ranges (Greenwood et al., 2014;Harsch et al., 2009;Mathisen et al., 2014;Morley et al., 2018; but see Beloiu et al., 2022). The rate of advance serves as an indicator of both forest and alpine ecosystem sensitivity to a changing climate (Bader et al., 2021;Smith et al., 2009). However, the spatial heterogeneity and complex dynamics of high-elevation treelines, both between and within mountain ranges, suggest that treeline elevation and tree species demographics are sensitive to many different controlling factors, both climatic and non-climatic (Bader et al., 2021;Harsch et al., 2009;Parobeková et al., 2018;Treml & Veblen, 2017;Wiegand et al., 2006). Moreover, studies of montane forests often find that tree expansion does not keep pace with climate warming, suggesting the possibility of dispersal limitation, abiotic (non-climate) limitations, or another inhibitory mechanism (Alexander et al., 2018;Liang et al., 2018;Sittaro et al., 2017).
While stressful climatic conditions or lack of suitable substrate are generally assumed to ultimately prevent tree encroachment to higher elevations, local variation in the appearance and demography of treeline (e.g. treeline form-defined by differences in tree densities and height) can also arise from a complex interaction of other factors (Harsch et al., 2009;Harsch & Bader, 2011;Körner, 1998;Körner, 2012;Sigdel et al., 2020). The multivariate influence of these different factors (e.g. low temperature, rime icing, wind damage, desiccation) can determine treeline forms and can impact treeline advance by affecting tree growth and mortality (of both adults and seedlings) (Harsch & Bader, 2011;Treml & Veblen, 2017). In order to capture and quantify treeline variation that may relate to treeline capacity to shift upslope, Harsch and Bader (2011) identified four archetypical treeline forms based on changes in tree density, cover and height: diffuse, abrupt, island, and krummholz. The more recent classification scheme of Bader et al. (2021) builds on the four treeline forms of Harsch and Bader (2011) by classifying treelines along two continuous axes including changes in tree height (stature: gradual vs. abrupt forms) and density (clustering, 2D pattern: diffuse vs. discrete forms), resulting in more possible treeline forms.
This approach assumes a pattern-process relationship; that is, that treeline patterns emerge from feedbacks with certain combinations of different mechanistic processes (Bader et al., 2021;Treml & Veblen, 2017). For instance, under this scheme, gradual treelines differ from abrupt treelines by their more subtle decline in tree height across the ecotone, likely resulting from growth limiting processes.
Another main treeline archetype, diffuse treelines, likely forms due to seedling recruitment failure and adult dieback caused by spatial patterns of adverse climatic or edaphic conditions within the ecotone, as opposed to adult mortality above the ecotone in the case of discrete treelines (Bader et al., 2021;Harsch et al., 2009;Harsch & Bader, 2011) (Figure 1; cf. definitions in Methods). Ultimately, adverse climatic conditions will limit tree growth and reproduction (by seed or vegetative layering) at the highest elevations for all treeline forms, but tree growth and survival seem most responsive to climate at gradual treelines, making these treelines particularly indicative and perhaps sensitive to climate change (Greenwood et al., 2014;Körner, 1998;Körner, 2012).
Climate effects on tree growth and survival can be modified by topography. Trees growing on slopes with northerly aspects (in the northern hemisphere) tend to experience colder temperatures but also lower risk of photoinhibition due to less direct exposure to solar radiation (Smith et al., 2003). Trees on aspects exposed to high winds tend to suffer increased cold-related tissue damage and icing, while less-exposed trees may be more protected by increased snow cover (Cairns, 2001;Hofgaard et al., 2009;Holtmeier & Broll, 2010, 2012). Various treeline forms can arise from the confluence of multiple climatological, edaphic (suitable substrate), and topographical factors acting to increase adult and seedling mortality, and to limit tree growth (Harsch & Bader, 2011;Hofgaard et al., 2009;Holtmeier & Broll, 2012).
In order to improve our understanding of the patterns and mechanisms behind climate-induced treeline dynamics, we combined a broad-scale remote sensing approach with fine-scale field data on tree demography, local topography, and climate to better States have led to distinct upslope movement of regional treelines over the last 30-40 years; (H2) The rate of upslope treeline advance has lagged behind the rate of climatic change, as has been suggested elsewhere (Alexander et al., 2018;Davis et al., 2020;Davis & Gedalof, 2018); (H3) Compared to other treeline forms, gradual diffuse treelines are more indicative of climate change and have experienced the greatest upslope shifts; and (H4) Variation in treeline advance can be explained by accumulated growing degree days (AGDD, a proxy for growing season temperatures) and aspect, which can modulate the correlation of climate variables, like AGDD, with the magnitude of treeline advance (i.e. evidence of significant interaction terms between aspect and AGDD.
The study was conducted in two mountainous areas in the northeastern United States, specifically in the Presidential Range of the White Mountains (New Hampshire) and on Katahdin (Maine). Both areas are within the Adirondack-New England highlands characterized by highly variable terrain (generally ranging from 150 to 1220 meters above sea level [m a.s.l.]), rocky spodosol soils, continental forest climate with warm summers and cold and snowy winters (mean annual temperatures between 3°C and 11°C; mean length of the frost-free period ~100 days; mean annual snowfall >2550 mm), and evenly distributed precipitation (annual mean precipitation of 890 mm) (Janowiak et al., 2018). The mean annual temperature has increased by ~1°C since the 1930s in the White Mountains, and by ~1.8°C since the 1890s across the state of Maine (MCC STS, 2020; Murray et al., 2021). Upland forests in the region grade from broadleaf deciduous forests at the lowest elevations, to spruce-fir dominated forests at higher elevations, and ultimately to open alpine communities on the highest peaks. In the northeastern United States, upper elevation treelines are heavily dominated by balsam fir (Abies balsamea (L.) Mill.), black spruce (Picea mariana (Mill.)) and cordate birch (Betula papyrifera var. cordifolia (Regel) Fern.) (Kimball & Weihrauch, 2000). In upland forests of the Northeast, balsam fir is most abundant between 1050 m and 1550 m asl (Bakuzis, 1965).
Both the Presidential Range and Katahdin are prominent features of the Appalachian Mountains in New England (Figure 1). The Presidential Range reaches a maximum elevation of 1917 m asl on Mount Washington, the highest peak in New Hampshire, and features over a dozen separate summits over 1200 m asl (Kimball & Weihrauch, 2000). The Presidential Range routinely experiences highly inclement weather with Mount Washington recording some of the highest wind speeds ever documented on Earth (Kimball & Weihrauch, 2000). The Presidential Range is protected and managed by the US Forest Service as part of the White Mountain National Forest. Katahdin is the highest point in the state of Maine, reaching an elevation of 1606 m asl. Katahdin is situated within Maine's Baxter State Park.
We defined a treeline during our field surveys (see below) as the boundary at which average tree height decreased below two meters and forest cover became discontinuous (non-overlapping individual canopies) (Greenwood et al., 2014;Körner, 2012;Mathisen et al., 2014). However, because historical imagery did not provide information on tree height, we only considered forest cover when delineating treelines from our acquired imagery. We did use average tree height measured in the field to help ground-truth contemporary treeline classifications based on remote sensing analysis (see below).
Based on Bader et al. (2021) we defined four treeline forms in the field using changes in canopy cover, tree height, and tree density as (i) gradual diffuse, (ii) abrupt discrete, (iii) krummholz-island and (iv) abrupt krummholz (Figure 1) as follows. The gradual diffuse form (i) was characterized by a gradual decrease in tree density and height across the ecotone between closed canopy forest and open alpine zone (scattered, isolated trees may be found in the open alpine environment); the abrupt discrete form (ii) was characterized by tall upright closed canopy forest directly bordering open alpine areas; the krummholz-island form (iii) was characterized by isolated short forest patches surrounded by alpine vegetation or linear forest strips penetrating into alpine zone; and the abrupt krummholz form (iv) was characterized by a continuous cover (overlapping crowns) of stunted, multi-stemmed trees extending beyond a continuous upright forest (>2 m in height). Even though some subjectivity is introduced when classifying treelines into discrete groups, this classification scheme is helpful when documenting broad trends and mechanisms controlling treeline position. Treelines were manually digitized based on observed tree cover for both sets of images, and the resulting polygons (which cover the area above treeline) were converted to raster format at 2 m resolution (all raster pixels within each above treeline polygon had a value of 1). We identified forest cover only as areas with overlapping crowns and seen as green reflectance in historic imagery and red reflectance in contemporary false-colour near-infrared imagery (no visible bare earth or easily identified alpine vegetation). Isolated tree islands >20 m in diameter were also digitized and included in treeline classification. Rasters of areas above treeline were aligned to and multiplied by Lidar-derived digital elevation models (DEMs; 2 m resolution) acquired from New Hampshire and Maine state GIS repositories (Maine GIS, 2021; NH GRANIT, 2021) in order to determine treeline elevations. A total of 400 random sample points (200 for each range, using the ArcGIS random sample point tool) were placed along the outer boundary of the alpine rasters derived from our contemporary imagery, and for each of them we established a paired point at the nearest location along the alpine raster boundary derived from our historic imagery (Appendix S1 in Supporting Information). For all belt transects, treeline form was assigned based on visual assessments (based on changes in tree height and density across the ecotone). Additionally, we visited a majority of our other accessible contemporary random sample points (~80%) in order to assign treeline form and ground-truth remote sensed treeline classifications. For all visited sample points, we took a new GPS point at the field-verified treeline location (continuous canopy cover and at least one individual >2 m in height) nearest to our random sample points (assigned from our treeline delineation procedure). The new points were compared to the original sample point locations and assessed for accuracy (measuring linear distance between points). Eye-level photos of treelines were taken at all sample points to keep a permanent record of treeline appearance. We stress that because tree height could not be extracted or field validated from our historic imagery, some krummholz individuals (<2 m) may have been present above (higher elevation) our treeline delineation using our classification scheme. Out of all 400 sample point pairs across both the Presidentials and Katahdin, 88 were classified as abrupt discrete (22%), 70 as gradual diffuse (17.5%), 84 as krummholz-island (21%), and 162 as abrupt krummholz (40.5%).
To examine the factors potentially influencing the spatial dynamics of treeline advance, both climatological and topographical variables were extracted for the Presidential Range (Table 1, Appendix S2 in Supporting Information). We could not conduct a similar analysis for Katahdin given the lack of fine-scale climatological data in that area. Elevation was extracted from 2 m state produced DEMs (NH GR ANIT, 2021). Using the Spatial Analyst toolbox in ArcGIS, topographical variables such as slope, aspect, and curvature (measure of convex or concave shape of the terrain ranging between -4 and 4) were extracted from our DEMs. Raw circular aspect data (measured in degrees, 0-360°) were used to visualize the magnitude of treeline advance partitioned across different aspects via a radar plot. Aspect was also converted to radians and linearized for use as a predictor in subsequent multiple linear regressions using the following equation 1 (Cox, 2006):
The resulting aspect values represent a measure of "Eastness," where north and south are both 0, east is 1 and west is -1. Before linearization, aspect values were used to calculate degree difference from prevailing wind (DDPW -290°) and degree difference from south (DDS -180°) variables. DDPW is a proxy for exposure to strong winds that can cause both direct physical damage and damage from icing, as well as a proxy for the potential for snow accumulation. The prevailing wind direction for the Presidential Range (290°) was based on wind measurements from the Mount Washington Observatory (Iowa Environmental Mesonet (IEM, 2021), see Appendix S3 in Supporting Information). DDS is a proxy for the amount of direct solar radiation (in the northern hemisphere). Average monthly mean, maximum, and minimum temperatures as well as annual accumulated growing degree days Category Variable Description Source Climate AGDD Accumulated growing degree days (°C) HOBO Array (AMC) T mean Mean annual temperature (°C) HOBO Array (AMC) T max Mean annual maximum temperature (°C) HOBO Array (AMC) T min Mean annual minimum temperature (°C) HOBO Array (AMC) PPT Mean annual precipitation (cm) PRISM 30-Year Normals Topography Elevation Altitude (m asl) State DEM Curvature Slope shape (concavity vs. convexity) State DEM Slope Surface slope State DEM Aspect Surface aspect (°) State DEM Aspect (DDPW) Degree difference from prevailing wind direction (°) State DEM Aspect (DDS) Degree difference from south (°) State DEM
Note: AGDD and temperature values were derived from interpolated prediction maps using data from AMC HOBO dataloggers recording hourly. PPT was based on PRISM 30-year normal data (i.e. 30-year means of annual values). The remaining variables were based on calculations from a regional 2 m DEM. Note that soil depth measurements are not listed as these could not be interpolated over our entire study area.
logger data; for AGDD calculations we used a base temperature of 4°C, consistent with other studies examining growth patterns of balsam fir, the dominant species within studied treelines (Gavin et al., 2008;Wason & Dovciak, 2017;Wason et al., 2017). AGDD was calculated as the accumulated maximum value of growing degree days (GDD) in a year following equation 2:
Gridded maps (90 m spatial resolution) of mean annual temperature (T mean , between 2007 and 2020) and AGDD for the Presidential Range region were produced using a cokriging interpolation method (Appendix S2, Aznar et al., 2013). To do this, temperature and AGDD response variables were first checked for normality using qq-plots.
Next, correlation between response variables and potential covariates was assessed; both elevation, aspect and curvature were highly
To test our first hypothesis (treelines have advanced upslope, H1), the differences between the elevation of each randomly placed sample point on our contemporary imagery (see Remote sensing analysis) and historic imagery were calculated to quantify mean treeline advance over the intervening time period. Maximum and minimum treeline elevations, as well as total area of new tree cover for both the Presidentials and Katahdin were also extracted. In order to determine the change in treeline position independently of rectification error, differences in elevation between historical and contemporary digitized treelines and at least 50 fixed, unchanging points (mountain peaks, roads, buildings, etc.) within each range were calculated (see Greenwood et al., 2014).
To test our second hypothesis (treeline advance lagged behind elevational changes in climate envelopes over time, H2), the available data (data are not available for Katahdin) on mean annual temperature (T mean ) and annual AGDD for the Presidential Range (both for the summit of Mount Washington and its base at Pinkham Notch) were plotted against the calendar year for all years between imagery acquisitions . The slopes for the linear increase in both T mean and AGDD for the two locations (2)
, 2023, 12, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jbi.14708 by Yale University, Wiley Online Library on [06/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 1999 TOURVILLE et al. AGDD as well as interactions between aspect and several candidate climate predictor variables. Predictors in our full model included both climatological and topographical variables selected based on their relevance to tree growth (see Spatial data processing and Table 1). Since soil depth could not be interpolated across the landscape, it was used in a separate regression to assess its relationship with tree density along field belt transects. A pairwise Pearson correlation matrix for all variables was examined to assess the correlation structure of predictors. Variables were removed from further analysis if they were strongly correlated (i.e. |r| > 0.6) to a more ecologically relevant variable (see Appendix S4). Examination of variance inflation factors (VIF) conducted on a global model including all remaining predictors revealed that the model was acceptable (all VIF values were acceptable <2.0; Belsley et al., 2005). All predictors were scaled and centered. Stepwise model selection based on AIC values was used to compare model performance with different sets of predictors using the "MASS" package (v.7.3-58.3) in R (Venables & Ripley 2002). Assumptions of residual normality and heteroscedasticity were checked using the "olsrr" package (v.0.5.3;Hebbali, 2020). All data analyses were conducted using the R (v4.1.3) statistical programming language (R Core Team, 2021).
Consistent with H3, the magnitude of advance for the gradual dif- 2018) and historic aerial imagery (Presidentials-1978;Katahdin-1991).
Range Mean elevation change (m ± SE) Rate of treeline advance (m/decade) Area of forest gained (ha) Area of forest gained relative to historic (%) Presidentials 11.77 ± 1.7 2.9 71.5 2.5 Katahdin 8.39 ± 1.1 3.0 10.5 0.8 2018 Max elevation (m) 2018 Min elevation (m) Historic max elevation (m) Historic min elevation (m) Presidentials 1769.6 1111.2 1759.2 1087.4 Katahdin 1429.1 732.7 1410.8 732.7 Change in max elevation (m) Change in min elevation (m) Change from control point (m) Presidentials 10.4 23.8 12.1 Katahdin 18.3 0.0 5.1
Note: The change in maximum and minimum treeline elevations between the two sets of imagery are also shown below, along with the difference between mean contemporary treeline elevation and elevation of our selected control points.
(Figure 3b).
Model comparison procedures selected the best multiple linear regression model that included both climate and topographical variables (elevation, slope, aspect and AGDD) predicting the magnitude of treeline shifts (R 2 adj = 0.32, see Appendix S10 in Supporting Information). Our multiple linear regression model revealed that greater shifts in mean elevation of treeline in the Presidential Range occurred in areas at higher elevations, steeper slopes, and lower AGDD. (Appendix S10). Model comparison between an elevation only and elevation + AGDD model revealed that AGDD explains a significant amount of variation in treeline shifts in addition to elevation (>2 ∆AIC, see Appendix S10). There was also a significant negative interaction term between AGDD and linearized aspect (p = 0.02), indicating that the effect of AGDD on the magnitude of treeline advance is smaller at higher values of linearized aspect (closer to east), consistent with H4.
While DDPW and DDS were not significant variables in our model, nonlinearized aspect plotted with a radar chart revealed greater mean treeline shifts at westerly aspects and smaller shifts at southerly and southeastern aspects (Figure 4a). Caution should be used when interpreting this result however, since there was substantial variation in treeline change with aspect, as evidenced by a one-way ANOVA comparing treeline elevation change between different primary aspects (Fvalue: 0.796; p = 0.592). Contemporary gradual diffuse treelines experienced greater upslope shifts than other treeline forms at similar values of AGDD, however, both gradual diffuse and, although less clearly, abrupt krummholz forms displayed significant upslope advance with lower AGDD, indicated by significant treeline form*AGDD interaction terms (Figure 4b, Appendix S10).
While we did not have appropriate soil data across the studied landscape to include in the multivariate regression model above, we were able to explore how the variation in soil depth may have contributed to changes in treeline location and structure. We found that soil depth decreased sharply and predictably with increasing distance along our survey transects (Presidential Range: p < 0.001, R 2 = 0.93; Katahdin: p < 0.001, R 2 = 0.98; see Appendix S11 in Supporting Information). Importantly, the mean number of trees along the transect (binned in 20 m intervals) decreased steeply and predictably with decreasing soil depth for all treeline forms (Presidential Range: p < 0.001, R 2 = 0.95; Katahdin: p < 0.001, R 2 = 0.95; Appendix S11) adding soil depth as another important variable affecting treelines (either because trees need sufficient substrate to successfully establish or trees increase soil depth through time). mountain ranges in the northern hemisphere (cf. Harsch et al., 2009;Lu et al., 2021). We observed high spatial variability in upslope shifts of treeline elevation for both the Presidentials and Katahdin, likely as a result of complex interactions between tree colonization, colonized by encroaching trees will likely be unsuitable for these species (Körner, 2012). Coupled with the potential loss of persistent winter snowpack under future warming, which serves to insulate alpine plants from severe cold and ice-induced stress, any upslope shifts in treeline positions may threaten vulnerable alpine species further (Burakowski et al., 2022;Inouye, 2020;Kimball et al., 2021;Publicover et al., 2021).
Importantly, the climate envelope (mean temperature) in the Presidential Range has shifted ~100 m between 1978-2018, almost an order of magnitude greater than the observed shift of treeline position. We point out that our estimation of climate envelopes were based on lapse rates for one aspect on one mountain (Mt.
Washington), and variation in lapse rates (and temperatures) can be high between various aspects and regionally (Kollas et al., 2014;Minder et al., 2010). For instance, Hutchins et al. (1976) report that moving from southwest to northeast aspects at the same elevation decreased maximum monthly temperatures between 1.7°C and 2.8°C during the growing season in Kentucky. However, even conservative estimates of climate envelope shifts would still be much larger than the observed shifts in treeline position documented here. There is an ample record of lags between climate and vegetation shifts which seems to indicate that the response of vegetation to changes in climate is likely mediated by other drivers, such as topography and biotic interactions (i.e. competition with other woody vegetation and interactions with alpine vegetation, dispersal limitation, etc.) (Greenwood et al., 2014;Lu et al., 2021;Sittaro et al., 2017;Zhu et al., 2012). Given that tree encroachment may also be limited by seed dispersal, and the time needed for seedlings to mature and recruit into larger size classes, we emphasize that care should be taken when using treeline positions as a direct bellwether for ecosystem response to climate change (Sittaro et al., 2017;Tourville et al., 2022).
Treeline advance was most apparent within the gradual diffuse form, consistent with other studies of treeline advance (Greenwood et al., 2014;Harsch et al., 2009). Temperatures cold enough to impose growth limitation (restricting tissue formation), in conjunction with low soil nutrient availability, have been proposed as mechanisms of treeline formation (Greenwood et al., 2014;Harsch et al., 2009;Körner, 1998;Smith et al., 2003). The obser vation that gradual diffuse treelines occurring at higher (both historically and contemporarily, see Appendix S12), colder elevations experienced greater advances might lead to two non-mutually exclusive possibilities; either this treeline form is a (i) direct result of historical treeline advance at high elevations, or (ii) that the gradual diffuse form is very sensitive to low temperatures. Based on the current data, we cannot say to what extent different treeline forms transitioned to the gradual diffuse form between historic and contemporary times (but see Birre et al., 2023). However, taken together, it is possible that physiological thermal thresholds on growth are being overcome via increased regional warming. The potential for the gradual diffuse form to respond more directly to factors controlled by a changing climate (e.g. temperature and growingseason length affecting tree growth and establishment) allows us the opportunity to target specific areas of treeline for monitoring climate-related vegetation changes. Additionally, shallow soil depth to bedrock (e.g. a proxy for the amount of suitable substrate) decreased tree densities in our system, suggesting that exposed bedrock at higher elevations can act to stall shifts in treeline. Processes of soil formation in high-elevation environments tend to be slow, and seedling establishment has been shown to be sensitive to the lack of suitable substrate (Johnson & Yeakley, 2019).
Comparisons among multiple linear regression models revealed that both climatological and topographical variables were correlated with the magnitude and direction of treeline shifts. Treelines which were historically at the highest elevations advanced the furthest (see Appendix S12); areas where temperatures were colder and accumulated degree days were lower. While our modelling suggests that low AGDD at high elevations may have been an important correlate of treeline advance (in addition to elevation; see Appendix S10), we cannot fully discount other unmeasured factors that potentially covary with elevation (i.e. snowpack, humidity, etc.). While the magnitude of treeline change was highly variable across various aspects, in general, greater mean treeline advance was recorded at westerly aspects in the Presidentials and on Katahdin; although when sites were binned into categorical aspects there was no significant difference. Contemporary mean treeline elevation also tended to be higher at these aspects. This was surprising given that these aspects were more exposed to prevailing winds and to the atmospheric jet stream, potentially leading to increased icing from rime ice and less snow cover which can directly physically damage trees, reduce carbon reserves, and prevent further treeline expansion (Kimball et al., 2021;Richardson et al., 2004). At the very least, increasing temperatures and growing season length may help reduce stress due to various aspects of climate (e.g. high levels of wind and icing) for trees that occur near the upper limits of their physical tolerances, particularly so for gradual diffuse and potentially for abrupt krummholz and krummholz-island treeline forms located in more sheltered microhabitats (Greenwood et al., 2014;Kimball et al., 2021). Indeed, infilling and height growth of krummholz-island patches have been observed from comparisons of historic and contemporary eye-level photos in the Presidential Range (Cogbill, personal communication).
Importantly, there are several known land-use and anthropogenic impacts that may play a minor role in the treeline dynamics of our study ranges (Publicover et al., 2021). The long-term formal protection of Katahdin has kept anthropogenic influence on the landscape there to a minimum (Publicover et al., 2021). However, the Presidential Range has seen the construction of a toll road and hiking trails to the summit of Mt. Washington as well as hiker huts scattered above the alpine-treeline ecotone, each of which could shape treeline positions at local scales, but not across the full range, vis-à-vis tree removal and damage (Kimball et al., 2021;Publicover et al., 2021). Historical logging and mammal grazing (both commercial and natural) in high-elevation krummholz areas within the region was quite minimal (Capers et al., 2013;Publicover et al., 2021). More relevant to our Ranges is a legacy of acid and nutrient (nitrogen and sulfur) deposition from industrial processes that form distinct east (lower) to west (higher) latitudinal, and high (higher) to low (lower) elevation, gradients, both of which could either inhibit tree growth via soil acidification or improve tree growth via increased nitrogen (N) uptake (Driscoll et al., 2001;Lawrence et al., 2015). Overall, we suspect that N deposition, more so than logging or browsing, has served to enhance tree encroachment in areas detected from our remote sensed approach, although this could only be confirmed with future work on these Ranges.
Treelines experienced greater mean shifts at aspects facing away from the southeast, with less direc t exposure to maximum solar illumination. Trees at southerly aspects may encounter drier environments or suffer increased photoinhibition (oxidative damage) from more direct sunlight, thus reducing overall growth and reproductive potential (Doeweler, 2021;Holtmeier & Broll, 2007). Given that the gradual diffuse treelines experienced the greatest upslope shifts in elevation and are thought to be a direct result of sensitivity to temperature via growth limiting mechanisms, any increase in AGDD (an ecologically meaningful measure of growth potential) at high-elevation sites already close to fundamental limits on tree growth may remove a fundamental limit on treeline expansion (dependent also on slope aspect). Collection of physiological (current height-or radial growth rates from dendrometers) and dendrochronological (historical heightor radial growth rates from tree cores) data from high-elevation trees in this system would help shed some light on patterns and limitations on growth (Etzold et al., 2022;Treml & Veblen, 2017).
As climate change continues to progress, there is an increasing need to better understand the patterns and drivers of geographical variation in treeline characteristics. Treeline locations and demographics (representing ecosystem changes) have the potential to be powerful indicators of the effects of climatic change, serving to provide baseline data to better anticipate changes within alpine ecosystems (Bader et al., 2021;Smith et al., 2009;Wason et al., 2021). However, treeline dynamics are complicated by the fact that treelines are often highly spatially heterogeneous, under the simultaneous influence of multiple controlling factors, and changes in position may lag behind climate warming (Bader et al., 2021;Harsch & Bader, 2011;Sigdel et al., 2020). We have attempted to categorize this variability and highlight specific areas that are the most sensitive to the direct effects imposed by climate change. We suggest that gradual diffuse treelines, particularly in cooler areas with fewer AGDD and a shorter growing season (moderated by aspect) are most likely to be directly limited by low temperatures and are therefore particularly useful to monitor climate-related changes. We emphasize that future work combining more sophisticated time-series remote sensing approaches of treeline change monitoring (using Lidar datasets) with demographic, spatial clustering, and physiological data collected in the field is critical to pursue, as illustrated and advocated by others (Bader et al., 2021;Davis et al., 2020;Feuillet et al., 2020;Morley et al., 2018). Such an approach can provide improved validation and accuracy assessments of treeline location and quantitative form classifications. It is not only critical to explore treeline dynamics for the sake of improved knowledge of threats posed to vulnerable ecosystems and their services, but to scale and integrate multiple regional studies to inform global patterns of ecosystem shifts.
13652699, 2023, 12, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jbi.14708 by Yale University, Wiley Online Library on [06/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License