Sample collection and preparation
We used tree cores archived by the Swedish NFI between 1961 and 2018 (ref. 72). The Swedish NFI samples approximately 10,000 plots per year on productive, evergreen-dominated forestlands (range 55–69° N, 11–24° E), from which tree cores are collected and archived. To systematically sample independent tree cores from the archive, we applied a grid of 250 square cells (50 × 50 km) over Sweden. We then identified samples in the archive by filtering their associated database, for cores originating from dominant or co-dominant Norway spruce (P. abies (L.) H. Karst) and Scots pine (P. sylvestris L.) samples from inventory plots categorized as mesic site types, with slopes of 20% slope or less, and from trees in the 41–60-year-old age class. This age class was selected to minimize the likelihood of N fertilization application; forest fertilization treatment in Sweden is applied to a relatively small area and late in stand rotation73. Samples were then randomly selected for each tree species from each grid cell, and for each of 6 sampling decades (1960s–2010s). Specific sampling years were chosen for each decade based on the available number of archived samples: 1961, 1977, 1986–1988, 1996–1998, 2006–2008 and 2016–2018. Three-year periods were sampled for the 1980s, 1990s, 2000s and 2010s because the NFI’s sampling intensity was reduced from this period onwards, requiring an expanded selection period to ensure a comparable number of samples to those available in 1961 and 1977. Some grid cells did not contain trees, and occasionally samples were not available for a given species, grid cell and decade combination, resulting in a total of n = 1,609 samples collected from the archive for analysis. Data for two counties in Sweden (Jämtland and Västernorrland) were previously reported38 and were combined with new data for all other counties in Sweden. For each sample, collected at breast height (1.3 m), we removed the outer bark and cambial layers, and the most recent annual ring to exclude incomplete ring growth during the collection year. Then, to obtain sufficient material for chemical analysis, the subsequent 10-year annual growth segment was separated from the remainder of the core38. Dissections were performed using a No. 11 stainless steel surgical blade under a stereo microscope with ×20 magnification, with an accuracy of 0.01 mm. As the samples represent a 10-year growth segment, we designated each sample by its corresponding intermediate 10-year growth increment year (for example, samples collected in 1961 were representative of the period 1951 to 1960 and were consequently referred to as 1955).
Measurement and analysis of N isotopes
Nitrogen isotope ratios (δ15N) were analysed at the Central Appalachians Stable Isotope Facility (CASIF), University of Maryland Center for Environmental Science (UMCES) Appalachian Laboratory (Frostburg, MD, USA) with a Carlo Erba NC2500 elemental analyser (CE Instruments) interfaced with a Thermo Finnigan Delta V+ isotope-ratio mass spectrometer (IRMS). The Carlo Erba NC2500 Elemental Analyser with Costech zero-blank autosampler modifications permits for analysis of N isotopes in solid organic samples with content less than 0.5% N, such as wood47. From each tree core, a radial slice was precisely sectioned to represent each 10-year segment and chopped with a steel razor. Approximately 10 mg of wood from the radial core slice was then weighed, placed in a tin and analysed for the δ15N value. Samples were analysed with a Carbosorb trap to remove CO 2 in advance of removing water vapour with magnesium perchlorate (MgClO 4 ). The δ15N data were normalized to the Ambient Inhalable Reservoir (AIR) scale using a two-point normalization curve with internal standards, including ground corn, cocoa and caffeine powder calibrated against international standards, USGS40 and USGS41. Analytical precision (1σ) of an internal wood standard (ground pine powder) analysed alongside samples was 0.3‰ for δ15N; atropine powder was used for determining N-content values.
The ratio of heavy (15N) to light (14N) isotopes of samples is expressed in standard delta (δ) notation with reference to a standard of known isotopic ratio.
$${{\rm{\delta }}}^{15}{\rm{N}}=({R}_{\mathrm{sample}}{/R}_{\mathrm{standard}}-1)\times 1,000$$
where R sample and R standard are the ratios of the heavy to light isotopes in the sample and standard, respectively, and are expressed in units of parts per thousand or per mil (‰). The chosen standard for N is AIR. Corresponding wood %N data are reported in Extended Data Fig. 3.
Climate, Nr deposition and forest data
We leveraged external databases for use in the linear mixed-effects model, including the climate parameters mean annual temperature (°C) and atmospheric CO 2 concentrations (ppm); the N deposition parameters NH x and NO y (g m−2); and the forest stand parameters total basal area and stand age. Monthly temperature data were extracted from CRU TS v4.07 (ref. 74) (resolution 0.5°) from which we calculated a 10-year mean annual temperature specific to each wood core sample, using its unique geographic coordinates. In addition to absolute temperature, we also calculated a relative temperature change variable, using 1961 as the reference temperature value for each grid cell, to minimize the large latitudinal gradient in temperature across our study area. We acquired CO 2 data from previously compiled measurements for the period 1951–2004 (ref. 75) and atmospheric CO 2 dry air mole fraction data for the period 2005–2017 from the National Oceanic and Atmospheric Administration (NOAA) Global Monitoring Laboratory recorded at PAL: Pallas-Sammaltunturi, GAW Station, Finland (67.9733° N, 24.1157° E; 565 m above sea level)76. We obtained monthly wet plus dry deposition as NH x and NO y between 1951 and 2017 from the global gridded Inter-Sectoral Impact Model Intercomparison Project (ISIMIP3a; resolution 0.5 × 0.5°)77 dataset to compile 4 variables describing spatially explicit atmospheric Nr deposition: NH x , NO y , NH x :NO y and the 10-year average total N deposition for each sample based on specific geographic location. The forest stand parameter total basal area was used as a proxy for forest biomass variation, collected at the time of each tree-core collection in the field and archived in the Swedish NFI database. The observed increase in total basal area over time can be attributed to the combined influence of forest management and environmental change factors78. Notably, commercial forestry in the north is relatively recent compared with southern Sweden, which has undergone multiple forest rotations79.
Swedish NFI growth data
... continue reading