INTRODUCTION

Observations, experiments and theory all show that climate and climate change influence the location, composition, structure and function of forests in many parts of the world, including the high-latitude forests dominated by boreal species (Körner, 1998; Barber et al., 2000; Lloyd et al., 2011; Henttonen et al., 2014; Henttonen et al., 2016; Reich et al., 2018; Hisano et al., 2019). However, direct evidence of a fingerprint of climate change on trees and forests is scarce, complex and often prone to conflicting interpretation, in part because of the challenge of disentangling changes among multiple synchronous drivers (Körner, 1998; Barber et al., 2000; Lloyd et al., 2011; Henttonen et al., 2014; Henttonen et al., 2016; Hogg et al., 2017; Reich et al., 2018; Hisano et al., 2019). For example, evidence from North America, Europe and Asia suggests both increased and decreased forest growth as a result of a changing climate, with growth often trending higher in colder regions, whereas increasing drought associated with increasing temperature has suppressed growth elsewhere (Barber et al., 2000; Reich and Oleksyn, 2008; Lloyd et al., 2011; Henttonen et al., 2014; Henttonen et al., 2016; Hogg et al., 2017; Reich et al., 2018; Hisano et al., 2019). Similarly, there is considerable evidence of treeline expansion globally, including in regions with cold winter climates, but changing land use (e.g. vegetation management, grazing pressures), and rising CO₂ and nitrogen deposition complicate the interpretation of such patterns in many if not all areas (Harsch et al., 2009; Cudlín et al., 2017; Weiser et al., 2019).

The size and longevity of leaves influence forest function (Schoettle, 1990; Reich et al., 2003; Pensa and Jalkanen, 2005), are influenced by climate (Jalkanen et al., 1995; Reich et al., 2003; Pensa and Jalkanen, 2005; Reich et al., 2014; Ťupek et al., 2015; Wright et al., 2017) and in turn shape responses to climate (Nicotra et al., 2010; Gornish and Prather, 2014; Reich et al., 2014). For example, in boreal conifers conditions that restrict tree growth also alter needle traits associated with growth: reduce needle length, increase longevity, decrease nutrient concentrations and photosynthetic rates (Schoettle, 1990; Reich et al., 2003; Pensa and Jalkanen, 2005); those modifications in turn co-regulate tree growth (Schoettle, 1990; Reich et al., 2003; Pensa and Jalkanen, 2005). Hence, changing environmental conditions over the past century may have impacted tree needle traits, with consequences for forest growth and productivity (Gornish and Prather, 2014; Reich et al., 2014).

Herein we provide evidence of a climate change fingerprint on the foliage of Swedish Scots pine (Pinus sylvestris L.) forests, based on needle sampling along a ≈1000 km latitudinal transect (Fig. 1) early in the 20th century (1914–1915) and repeated early in the 21st century (2012–2017), when all sites had warmed by ≈1 °C (see Figs S1 and S2 in Supporting Information). Scots pine is the most prevalent tree species in Europe and one of the most widespread conifer species on Earth, and plays a considerable role in temperate and boreal carbon cycling as well as in providing vast ecosystem and economic services. In 2012, 2016 and 2017, we sampled Scots pine trees along the transect at 74 sites (Fig. 1 and Table S1). At each site we measured the average length and longevity of needles on mature trees (mean tree height ≈17 m, mean tree age ≈150 years old). We chose those sites and measures because a Swedish scientist, Nils Sylvén, measured those same climate-sensitive traits at 57 sites along the same transect in 1914–1915 (Sylvén, 1917; Fig. 1 and Table S1). We hypothesised that the temperature rise over the 100-year interval separating the sampling campaigns would increase needle length and reduce needle longevity, reflecting and reinforcing the stimulating effect of warming on tree growth at high latitudes.

METHODS

RESULTS

Contrasts across time demonstrate a major shift in needle traits. Analyses of data from 1914–1915 and 2012–2017 (hereafter ≈1915 and ≈2015 for brevity) show that at any given latitude, modern plants growing in ≈1 °C warmer conditions have longer needles and shorter needle longevity than those a century ago (Fig. 2, Table 1). For example (and based on overall latitudinal regressions), at the mid-point in the transect, 64 °N, modern plants have needle longevity that is shorter by 1.7 years, a decrease of 31%, and needle length that is longer by 4.7 mm, an increase of 14%. The change in needle longevity was similar (decrease ≈31%) along the transect, whereas needle length increase was smaller further south (increase of 9% at 59 °N) and larger to the north (increase of 19% at 68 °N).

Within both time periods, Scots pine needle length and needle longevity were strongly related to latitude (Fig. 2, Table 1), as predicted (Jalkanen et al., 1995; Reich et al., 2003; Pensa and Jalkanen, 2005; Reich et al., 2014; Ťupek et al., 2015). Needles in warmer, more southern locations were longer and had shorter lifetimes during both time periods (Fig. 2); likely in response to temperature (Table S2), and consistent with the parallel shifts in temperature and needle traits across the 100-year period.

Examining needle traits in relation to climate (Fig. 3) rather than latitude shows that for both needle traits recently experienced temperatures (MAT_5y) and time period (i.e. early 20th vs. early 21st century) were both highly significant (P < 0.0001; Table S2). Additionally, the effect of temperature on needle traits was similar across space and time, as the interaction of MAT_5y × time (period) was not significant (Table S2). MAP_5y was not significant as a main effect for either needle trait, although responses to MAT_5y were modestly influenced by MAP_5y (Table S2). The changes over time in needle length and longevity were coordinated; needle length and longevity were inversely correlated in both 1915 and 2015 (Fig. S5). Trees in 2015 tended to occupy positions of higher needle length and shorter longevity along a similar trait–trait axis as trees in 1915.

DISCUSSION

Our results indicate the importance of temperature as a driver of both needle traits. The overall temperature effect likely incorporates both direct (through temperature-mediated physiological responses) and indirect (e.g. variation in soil nutrient availability associated with temperature in boreal ecosystems) mechanisms (Reich and Oleksyn, 2004; van Sundert et al., 2018). However, if temperature alone drove needle traits we would expect that recent 5-year temperatures would fully explain variance, eliminating time (century) as a significant variable (i.e. data for 2015 would migrate to higher temperatures along a shared regression relationship with 1915 data). In fact, responses to century-scale temperature rise were larger than would be predicted based on spatial responses to temperature, especially for needle longevity. This suggests that other factors (e.g. rising CO₂ levels, forest management, etc.) may play a role in the changes in needle trait values over the past century (see below).

Much of the observed spatial and temporal variation in needle traits is likely associated with variation in tree growth, because climatic influences on needle traits likely influence growth, and in turn, climate effects on growth also influence needle traits in a parallel direction (Schoettle, 1990; Reich et al., 2003; Pensa and Jalkanen, 2005; Reich et al., 2014). Needle traits that contribute to faster growth (Reich et al., 2003; Reich et al., 2014) tend to be enhanced in a warmer climate (Schoettle, 1990; Reich et al., 2003; Pensa and Jalkanen, 2005; Reich et al., 2014; Ťupek et al., 2015), and when trees grow faster, as is typical for boreal trees in warmer climates (Pensa and Jalkanen, 2005; Reich et al., 2008; Henttonen et al., 2014; Henttonen et al., 2016), they replace needles more frequently, leading to reduced needle longevity and higher levels of associated traits like length or nitrogen concentration (Schoettle, 1990; Reich et al., 2003; Pensa and Jalkanen, 2005; Reich et al., 2014). Results of our study are consistent with these ideas. For trees sampled in 2012, the height and radial growth of trees decreased with latitude (data not shown) and temperature (Fig. S6), but were unrelated to variation in precipitation (data not shown). The higher growth rate in warmer sites is consistent with the greater needle length and shorter longevity at those sites as well. Moreover, for the data in this study, needle length and longevity were significantly related to both tree height and 30-year radial increment, both using raw data (Fig. S7) and also in analyses of covariance models after accounting for variation in stand age. The strong co-variation between needle traits, climate and growth rate, however, makes it difficult to disentangle their independent effects. Nonetheless, we posit that it is likely that warmer temperatures lead to higher growth rates, and that both contribute to longer needle length and faster turnover rate in southern than northern sites at any point in time and in 2015 than in 1915 at any given site; moreover, those shifts in needle traits in turn likely contribute to higher growth rates in more southerly than northerly sites and in 2015 versus 1915.

It is important to note that although shorter needle longevity can indicate stress, such as from air pollution (Kozlov and Niemelä, 1999), the typically shorter longevity of needles on faster growing trees in more favourable conditions is associated with a less conservative growth strategy and is an indication of the rate of whole-plant and foliar metabolism (Reich et al., 2003; Ťupek et al., 2015); for instance, climatic factors impact needle production, and it is thought that needle longevity varies in order to maintain constancy in the overall number of needles or length of foliated shoots (Schoettle, 1990; Pensa and Jalkanen, 2005). Thus reduced needle longevity over space or time in this study should not be considered a negative attribute or a stress response.

As noted above, if changing temperatures were entirely responsible for the large shifts over time in needle traits documented in Fig. 2 – and assuming that relationships to temperature were identical across time as across space – the data from both time periods would be predicted to occupy a similar trait envelope in their relationship of traits to 5-year recent MAT. That is not the case (Fig. 3): even at a common experienced MAT, modern plants have slightly longer needles with considerably shorter lifetimes than plants in 1915. Thus, Scots pine needles have changed more in a century than would be expected solely due to rising temperatures. Other global and regional environmental change factors, and forest management, could have contributed to these century-scale changes in needle traits.

For example, rising CO₂, increasing precipitation and nitrogen deposition could have simultaneously increased tree growth, needle length and needle turnover (decreasing needle longevity) beyond shifts attributable to temperature. That is because in addition to 21st century Sweden being warmer than in ≈1915, it has also experienced higher nitrogen deposition and higher CO₂ as well as become rainier. Collectively, this environmental transformation would encourage higher pace-of-life traits and faster tree growth, potentially contributing to larger changes in needle length and needle longevity than predicted based solely on rising temperatures.

First, nitrogen deposition has increased in Sweden (Karlsson et al., 2012), which could increase growth and needle N concentrations, both likely to be associated with longer needles and shorter needle lifespans (Schoettle, 1990; Reich et al., 2003; Pensa and Jalkanen, 2005; Reich et al., 2014). However, nitrogen deposition has not increased uniformly across Sweden (Karlsson et al., 2012). Its highest levels and greatest geographic variation are in southern Sweden, largely outside the location of the region surveyed in our report, whereas the levels of nitrogen deposition are relatively low, with modest spatial variation, in northern Sweden. Thus, if nitrogen deposition was the main driving factor in increased growth and associated needle trait changes over the century, differences from ≈1915 to ≈2015 should have been negligible in the far north, and large further south, which was not the case (see Fig. 2). This suggests a modest role for nitrogen deposition in the century-scale changes observed in needle traits.

Second, rising CO₂ might also be increasing tree growth (Norby and Zak, 2011), which could contribute to the observed shifts towards faster pace-of-life traits. However, elevated CO₂ tends to cause declining needle N concentration (Laitinen et al., 2000), and common, strong inverse relationships between needle longevity and N concentration (Reich et al., 2003; Reich et al., 2014) suggest that rising CO₂ might indirectly increase, not decrease, longevity. In fact, the very few studies of leaf longevity with rising CO₂ (albeit with angiosperms, not gymnosperms) do suggest longer lifespan (Craine and Reich, 2001). Hence CO₂ might influence longevity in two off-setting ways, by increasing growth (which would tend to reduce needle longevity) and by decreasing N concentration (which would increase needle longevity). Whether these mechanisms were at work for Swedish pine and if so, which of these two mechanisms had greater impact are uncertain.

Third, increasing precipitation over the past 100 years hypothetically might have contributed to faster growth rate and thus to increasing needle length and decreasing longevity. In a drier region in Switzerland, irrigation resulted in increased needle length in Scots pine but had no effect on needle longevity (Feichtinger et al., 2015). In our study, the nearly complete insensitivity of needle traits and tree growth to spatial or temporal variation in precipitation suggests that if important at all, precipitation was likely a very minor contributor to the observed century-scale changes in needle traits. This likely resulted from the study focus on a cold-limited and mesic biome; we do not suggest such effects would be immaterial in drier regions.

Finally, changes in forest management, including the use of genetically improved seed material (possibly included in our samples from Southern Sweden), higher stocking rates, improved drainage, less reliance on selective harvesting and improved regeneration techniques (Elfving and Tegnhammar, 1996; Henttonen et al., 2016), have likely also increased productivity on average in Sweden over the past century, with those shifts in productivity potentially contributing to century-scale shifts in needle traits being larger than predicted based solely on spatial relations to temperature. For instance, analyses of Swedish and Finnish historical inventory data show increasing tree growth (Elfving and Tegnhammar, 1996; Henttonen et al., 2016), with changing climate as well as changing management considered to have played important roles. However, as the trees we sampled were generally relatively old, in stands with modest, if any, active management, and were likely self-sown, impacts on changing forest management may have been limited in our data.

Uncertainty about the impacts of additional drivers suggests that interpretation of these results must be done carefully. The evidence clearly indicates that across Sweden, pine needles in the current era (≈2015) are longer and live shorter lifetimes than in ≈1915, similar to needle traits of trees in warmer vs. colder regions during both periods. Evidence from within the two time periods, and across them, showing strong relations of needle traits to temperature, strongly suggests that spatial and temporal temperature variation likely is a major driver of the observed variation in both needle traits. Environmental changes in other drivers, such as nitrogen deposition, CO₂ and precipitation may have further increased (in the same direction) the magnitude of the changes in length and longevity of pine needles, but the roles of each potential driver, and the magnitude of their impact, are uncertain.

Climate change-induced modification of the traits, growth rates and vitality of forests globally is almost certain, and as shown for Swedish pine forests, has begun in documentable fashion in high-latitude conifer forests. Needle traits, and how they adjust to changing climate, will help determine future forest productivity and carbon cycling (Gornish and Prather, 2014; Reich et al., 2014). The century-scale shifts we observed in needle length and longevity likely correspond to parallel shifts in needle nitrogen concentration and photosynthetic and growth capacity that might enable trees and forests in high latitudes with generally cold climates and short growing seasons to grow faster and be more productive under today’s elevated temperatures and CO₂ levels (Lloyd et al., 2011; Reich et al., 2003; Pensa and Jalkanen, 2005; Reich et al. 2008; Reich et al., 2014). Note, however, that such shifts are both less likely to have occurred and less likely to enhance Scots pine productivity in warmer areas of its range (Martínez-Vilalta et al., 2008; Reich et al., 2008; Lloyd et al., 2011; Büntgen et al., 2013; Hogg et al., 2017). That is because increased drought and negative effects of heat associated with a changing climate can offset positive benefits of a longer growth season or warmer temperatures on Scots pine performance in locations in Spain and central Europe (Martínez-Vilalta et al, 2008; Reich et al., 2008; Büntgen et al., 2013), and can be confounded by direct effects of rising CO₂. Hence, we should not presume that the needle trait shifts of conifers observed in cold, high-latitude forests (towards longer needles with short lifetimes) would parallel those in warmer, mid-latitude forests; if any prediction is supported, it is the opposite that slower growth and greater stress would lead to shorter needles with long lifetimes. We hope such evidence can be obtained by those working in such regions.

Historical data, as in the case described herein, can be invaluable for quantifying forest change and associated consequences for critical ecosystem services such as carbon storage, but is a rare resource, especially going back more than a few decades. Looking forward for the rest of this century, deploying all tools possible to detect forest change and to understand the underlying mechanisms, will be required to enable knowledge generation that can transfer to management at a rate that can keep up with the pace of environmental change and help maintain sustainable forests.

FUNDING INFORMATION

This research was supported by the National Science Center, Poland (2011/02/A/NZ9/00108).

AUTHORSHIP

JO and PR contributed equally but in distinct ways to this work. They jointly came up with the idea for this study. JO organised, supervised and participated in the field sampling and sample processing; PR analysed the data, and wrote the first and all subsequent drafts of the paper. TW participated in the field sampling and needle processing, and contributed to data analyses and interpretation, and to manuscript development and refinement. All the other authors participated in the field sampling and needle processing, and contributed to manuscript development and revision.