The position of the Arctic treeline, which is a key regulator of surface energy exchange and carbon cycling, is widely thought to be controlled by temperature. Here, we present evidence that soil nutrient availability, rather than temperature, may be the proximate control on growth of treeline trees at our study site in northwest Alaska. We examined constraints on growth and allocation of white spruce in three contrasting habitats. The habitats had similar aboveground climates, but soil temperature declined from the riverside terrace to the forest to the treeline. We identified six lines of evidence that conflict with the hypothesis of direct temperature control and/or point to the importance of soil nutrient availability. First, the magnitude of aboveground growth declined from the terrace to the forest to the treeline, along gradients of diminishing soil nitrogen (N) availability and needle N concentration. Second, peak rates of branch extension, main stem radial and fine-root growth were generally not coincident with seasonal air and soil temperature maxima. At the treeline, in particular, rates of aboveground and fine-root growth declined well before air and soil temperatures reached their seasonal peaks. Third, in contrast with the hypothesis of temperature-limited growth, growing season average net photosynthesis was positively related to the sum of normalized branch extension, main stem radial and fine-root growth across trees and sites. Fourth, needle nonstructural carbohydrate concentration was significantly higher on the terrace, where growth was greatest. Fifth, annual branch extension growth was positively related to snow depth, consistent with the hypothesis that deeper snow promotes microbial activity and greater soil nutrient availability. Finally, the tree ring record revealed a large growth increase during late 20th-century climate warming on the terrace, where soil N availability is relatively high. Meanwhile, trees in the forest and at the treeline showed progressively smaller growth increases. Our results suggest temperature effects on tree growth at our study sites may be mediated by soil nutrient availability, making responses to climate change more complex and our ability to interpret the tree ring record more challenging than previously thought.

Increment cores from the boreal forest have long been used to reconstruct past climates. However, in recent years, numerous studies have revealed a deterioration of the correlation between temperature and tree growth that is commonly referred to as divergence. In the Brooks Range of northern Alaska, USA, studies of white spruce (Picea glauca) revealed that trees in the west generally showed positive growth trends, while trees in the central and eastern Brooks Range showed mixed and negative trends during late 20th century warming. The growing season climate of the eastern Brooks Range is thought to be drier than the west. On this basis, divergent tree growth in the eastern Brooks Range has been attributed to drought stress. To investigate the hypothesis that drought-induced stomatal closure can explain divergence in the Brooks Range, we synthesized all of the Brooks Range white spruce data available in the International Tree Ring Data Bank (ITRDB) and collected increment cores from our primary sites in each of four watersheds along a west-to-east gradient near the Arctic treeline. For cores from our sites, we measured ring widths and calculated carbon isotope discrimination (δ13C), intrinsic water-use efficiency (iWUE), and needle intercellular CO2 concentration (C(i)) from δ13C in tree-ring alpha-cellulose. We hypothesized that trees exhibiting divergence would show a corresponding decline in δ13C, a decline in C(i), and a strong increase in iWUE. Consistent with the ITRDB data, trees at our western and central sites generally showed an increase in the strength of the temperature-growth correlation during late 20th century warming, while trees at our eastern site showed strong divergence. Divergent tree growth was not, however, associated with declining δ13C. Meanwhile, estimates of C(i) showed a strong increase at all of our study sites, indicating that more substrate was available for photosynthesis in the early 21st than in the early 20th century. Our results, which are corroborated by measurements of xylem sap flux density, needle gas exchange, and measurements of growth and δ13C along moisture gradients within each watershed, suggest that drought-induced stomatal closure is probably not the cause of 20th century divergence in the Brooks Range.

Trees growing near the Arctic treeline have long been used to reconstruct past climates. However, recent studies have shown deterioration of historically strong positive correlations between air temperature and tree growth (known as "divergence"). Divergence has important implications for confidence in paleoclimate reconstructions and ecosystem-atmosphere carbon exchange. Studies in the Brooks Range of northern Alaska showed that white spruce in the west increased growth in response to late 20th century warming, whereas those in the east failed to show a growth increase. In an earlier study across four watersheds in the Brooks Range, we tested and rejected the hypothesis that divergence in the easternmost watershed reflects moisture limitation of growth. Here, using 16 sites distributed across the same four watersheds, we tested an alternative hypothesis, that greater nutrient limitation in the east may have impeded positive growth responses to warming. Climate comparison across the four Brooks Range study watersheds revealed that, although the easternmost watershed generally had a drier growing-season climate, the most consistent difference was that winter air temperature and both winter and summer soil temperatures were much colder in the central and eastern watersheds. Soil nutrient availability, foliar nutrient concentrations, and tree growth were all generally lower in the central and eastern than in the western watersheds. Foliar phosphorus concentration was the best predictor of spatial variation in branch extension growth-a finding that is somewhat inconsistent with the theory that forest productivity on young, glacially derived soils should be strongly nitrogen limited. Experimental fertilization yielded the greatest growth increase in the eastern, an intermediate response in the central, and the smallest growth increase in the western watershed, generally mirroring trends in soil temperature, soil nutrient availability, foliar nutrient concentrations, and growth of control trees. Our results confirm that growth in the easternmost watershed is more nutrient limited and suggest that phosphorus limitation may be at least as important as nitrogen limitation of growth. We hypothesize that cold soil effects on tree access to nutrients might explain divergence in the eastern Brooks Range and elsewhere near the Arctic treeline, particularly in areas with cold winters and widespread permafrost.

Numerous recent studies have argued that moisture limitation is leading to growth declines and mortality of black spruce (Picea mariana) and white spruce (Picea glauca) in the North American boreal forest. A parallel line of research suggests that increasingly common severe wildfires are altering successional pathways and leading to long‐term replacement of spruce forests with those dominated by paper birch (Betula papyrifera, Betula neoalaskana) and trembling aspen (Populus tremuloides). When both conifers and deciduous trees establish after fire, this biome shift hypothesis implicitly assumes that deciduous species will outcompete the conifers, owing to their more rapid vertical growth and because they might be less sensitive to warm and dry conditions. We established a research site in a white spruce‐paper birch forest on an east‐facing slope in interior Alaska and tested the hypothesis that Alaska paper birch are better adapted to warm and dry conditions than white spruce. Over 6 years (2013–2018), we made hourly measurements of microclimate and xylem sap flux of both species. We also collected increment cores and conducted climate‐growth analyses for both species. During our 6‐year study, growing seasons with low volumetric soil water content (SWC) were those that followed shallow winter snowpacks and had limited summer rainfall, not necessarily those with warm air temperature. Both species were sensitive to moisture limitation. The tree‐ring data revealed significant positive effects of cumulative water year precipitation on radial growth, with a steeper slope for paper birch than for white spruce. Radial growth of both species was also positively related to mean water year air temperature. Sap flux density declined progressively for white spruce over the range of observed SWC and abruptly for paper birch when SWC fell below ~15%. Synthesis. Our results show that, while paper birch might be less sensitive to mild drought than white spruce, it may be more sensitive to severe drought, raising questions about the ability of paper birch to outcompete co‐occurring white spruce in a drier climate.