Open Access

Tree growth in northern and upper treeline ecotones of the circumpolar boreal forest is generally limited by temperature, i.e., trees grow generally more under warm, and less under cold climatic conditions. Based on the assumption that this relationship between tree growth and climate is linear and stable through time, dendroclimatologists use tree rings as natural archives to reconstruct past temperature conditions. Such tree-ring based reconstructions, together with other natural archives (e.g., ice cores and pollen), constitute our understanding of past climatic conditions that reach beyond modern instrumental records. However, a steadily increasing amount of studies reports a recent reduction or loss of the summer temperature signal for several species and sites of the boreal forest. Such a reduction of temperature sensitivity results in temporally unstable climate-tree growth relationships, which challenges the work of dendroclimatologists by potentially leading to miscalibrations of past climatic conditions. On the upside, this shift in the trees’ climate sensitivity might point to a shift in tree growth-limiting factors and thus serve as an early indicator of climate change impacts. There is evidence that this recent reduction in temperature sensitivity might be caused by the observed strong temperature increase at high latitudes, and thus temperature-induced drought stress. Other potential drivers and amplifiers of this phenomenon are differing microsite conditions (dry vs. wet soils) and factors inherent to trees, like genetic properties or age effects. In this PhD thesis, I systematically assessed the effects of frequently discussed drivers of unstable climate-tree growth relationships (climate change, micro-site effects, genetical predisposition) on two representative species of the boreal forest, white spruce in North America and Scots pine in Eurasia, across various temporal and spatial scales. I used classical (tree-ring width) and more novel (wood density, quantitative wood anatomy) dendrochronological proxies to unravel the effects from annual to sub-monthly resolution. More precisely, in chapter I, white spruce clones were compared to non-clones at two treeline sites in Alaska to test whether their growth patterns differ, and whether white spruce clones are generally suitable for dendroclimatic assessments. Clonal reproduction is frequent at treeline due to harsh conditions, but might lead to competition among individuals due to the close proximity among each other, which in turn might obscure their climatic signal. Second, I tested the effect of warmer and drier climatic conditions on the summer temperature signal of Scots pine in Eurasia (chapter II) and on the growing season moisture signal of white spruce in North America (chapter III), respectively. Temperature-induced drought stress is expected to be the most important driver of unstable climate-growth relationships in the boreal forest. I included several sites across latitudinal (50-150 km) and longitudinal (1,000-2,200 km) gradients to cover large parts of the species’ distribution ranges. Since Scots pine covers a wide range of ecological habitats, I additionally tested the effect of dry and wet micro-site conditions on the summer temperature signal of Scots pine in chapter II. Finally, in chapter IV, a systematic literature review was carried out in order to investigate the distribution of unstable climategrowth relationships in global tree-ring studies, and the usage of such series in climate reconstructions. Furthermore, the scientific impact of these potentially inaccurate climate reconstructions was assessed. In this PhD project, warmer and drier climatic conditions led to temporally unstable climate signals in both Scots pine (chapter II) and white spruce (chapter III), as expected. Unstable climate-growth relationships were found for all tested tree-ring proxies and at all sites in North America, and at most sites in Eurasia. Micro-site effects (chapter II) and clonal growth (chapter I) had no significant effect on the climate sensitivity and high-frequency variability of the tested species, but affected absolute growth. The review (chapter IV) revealed that the phenomenon of unstable climate-growth relationships is globally widespread, and occurs independent of tree species, geographic location, and tree-ring and climate proxies. While reconstructions inferred from these unstable relationships are frequent and respective papers have a high impact, the tree-ring community seems to increasingly recognize the challenge of unstable climate-growth relationships. With these findings, this PhD project helped to shed more light on the frequency, underlying drivers, and the impact of unstable climate-growth relationships in boreal forest trees, as well as underlying reaction processes in trees. Above all, this PhD project suggests that the loss of climate sensitivity is caused by a change of growth limiting factors: temperature limitation seems to be suspended in warmer and drier years for Scots pine in Eurasia, and moisture limitation first arises under warm/dry conditions for white spruce in North America. Due to plastic growth responses in trees, the general assumption in dendroclimatology – that climategrowth relationships are stable through time – seems to be incompatible with the principle of limiting factors (one factors is always most growth limiting). To improve the validity of future climate reconstructions, statistical approaches considering synchronously or changing climatic limiting factors need to be promoted, along with attempts to select the best responding trees from a dataset. Furthermore, a better understanding of nonclimatic factors potentially affecting tree growth (e.g., age, disturbance, soil parameters) is needed. A growth reduction of mature and dominant white spruce trees sampled in this PhD project seems likely under future warming conditions, with series of wood cells being valuable early indicators of climate change effects in white spruce. However, inferences cannot be extended to the entire stand due to the applied sample design. Projected climate warming will probably lead to a further reduction of the summer temperature signal in trees of the northern boreal forest, while wider consequences for forest growth and productivity are unclear.

Forests are ecologically important ecosystems, for example, they absorb CO2 from the atmosphere, mitigate climate change, and constitute habitats for the majority of terrestrial flora and fauna. Currently, due to increasing human pressure, forest ecosystems are increasingly subjected to changing environmental conditions, which may alter forest growth to varying degrees. However, how exactly different tree species will respond to climate change remains uncertain and requires further comprehensive studies performed at different spatial scales and using various tree-ring parameters. This dissertation aims to advance the knowledge about tree-ring densitometry and tree responses to climate variability and extremes at different spatial scales, using various tree species. More specifically, the following aims are pursued: (i) to obtain and compare wood density data using different techniques, and to assess variability among laboratories (Chapter I). (ii) To investigate microsite effects on local and regional Scots pine (Pinus sylvestris L.) responses to climate variability (Chapter II) and extremes (Chapter III), using ring width (RW) and latewood blue intensity (LBI) parameters. (iii) To give a general site- and regional-scales overview of Scots pine, pedunculate oak (Quercus robur L.), and European beach (Fagus sylvatica L.) RW responses to climate variability (Chapter IV). (iv) To discuss the challenges which may result from compiling tree ring records from different (micro)sites into large-scale networks. The study area comprises nine coastal dune sites, each represented by two contrasting microsites: dune ridge and bottom (Chapters II and III), and 310 different sites within the south Baltic Sea lowlands (Chapter IV). The dissertation confirms that sample processing and wood density measuring are very important steps, which, if not performed carefully, may result in biases in growth trends, climate-growth responses, and climate reconstructions. The performed experiment proved that the mean levels of different wood density-related parameters are never comparable due to different measurement resolutions between various techniques and laboratories. Further, the study revealed substantial biases using data measured from rings of varying width due to resolution issues, where resolution itself and wood density are lowered for narrow rings compared to wide rings (Chapter I). The (micro)site-specific investigation showed that, depending on the species, different climate variables (temperature, precipitation, or drought) constitute important factors driving tree growth across investigated locations (Chapters II and IV). However, there is evidence that the strength and/or direction of climate-growth responses differ(s) between microsite types (Chapter II) and across sites (Chapter IV). Moreover, climategrowth responses are non-stationary over time regardless of the tree species and tree-ring parameter used in the analysis (Chapters II and IV). There are also differences in RW and LBI responses to extreme events at dune ridge and bottom microsites (Chapter III). The regional-scale investigations revealed that climate-growth responses (strength and non-stationarity) are quite similar to those observed at the local scale. However, compiling RW or LBI measurements into regional networks to study tree responses to extreme events led to weakened signals (Chapter III). The findings presented in Chapters II and IV suggest that the strength, direction, and non-stationary responses are very likely caused by several climatic and non-climatic factors. The mild climate in the south Baltic Sea region presumably does not constitute a leading limiting growth factor, especially for Scots pine, whose distribution extends from southern to northern Europe. Thus, the observed climate-growth responses are usually of weak to moderate strength. In contrast, for other species reaching their distribution limit at the Baltic coast, the climatic signal can be very strong. However, the observed findings also result from the effects of microsite conditions, and potentially other factors (e.g., management, stand dynamic), which all together alter the physiological response of the tree at a local scale. Although climate at the south Baltic Sea coast is mild, extreme climate events may occur and affect tree growth. As demonstrated (Chapter III), extreme climate events affected tree growth across dune sites, however, to varying degrees. The prominent differences in tree responses to extreme climate events were significant at the local scale but averaged out at the regional scale. This is very likely associated with observed microsite differences, where each microsite experiences different drivers and dynamics of extreme growth reductions. This dissertation helped to demonstrate that integrating local tree-ring records into regional networks involves a series of challenges, which arise at different stages of research. In fact, not all possible challenges have been discussed in this dissertation. However, it can be summarized that several steps performed first at the local scale are very important for the quality and certainty of climate-growth responses, tracking tree recovery after extreme events, and potential climate reconstructions at the larger scale. Among them, identification of microsite conditions, sample preparation, and measurement, examination of growth patterns and trends, and identification of a common limiting growth factor are very important. Otherwise, the compilation of various tree-ring data into a single dataset could lead to over- or underestimation of the results and biased interpretations.