Oceanographic and biogeochemical drivers cause divergent trends in the nitrogen isoscape in a changing Arctic Ocean

Abstract

Nitrogen stable isotopes (δ¹⁵N) are used to study food web and foraging dynamics due to the step-wise enrichment of tissues with increasing trophic level, but they rely on the isoscape baseline that varies markedly in the Arctic due to the interplay between Atlantic- and Pacific-origin waters. Using a hierarchy of simulations with a state-of-the-art ocean-biogeochemical model, we demonstrate that the canonical isotopic gradient of 2-3‰ between the Pacific and Atlantic sectors of the Arctic Ocean has grown to 3-4‰ and will continue to expand under a high emissions climate change scenario by the end of the twenty-first century. δ¹⁵N increases in the Pacific-influenced high Arctic due to increased primary production, while Atlantic sector decreases result from the integrated effects of Atlantic inflow and anthropogenic inputs. While these trends will complicate longitudinal food web studies using δ¹⁵N, they may aid those focussed on movement as the Arctic isoscape becomes more regionally distinct.

Keywords: Biogeochemistry; Food webs; Primary production; Spatial ecology; Stable isotopes; Trophic position.

Fig. 1

Major changes affecting the isoscape of nitrogen in the Arctic Ocean. a Time series of sea ice cover (millions km²) in our reanalysis-driven (solid black line) and emissions-driven (red line) simulations compared with preindustrial control conditions (dashed black line). b Observed spatial change in sea ice concentration 2013–2018 minus 1982–1987 (shading) and reanalysis-driven change (contours). Contours are in 5% intervals. c Time series of integrated change in aeolian reactive nitrogen (N_r) deposited to the global ocean. d Change in N_r deposition between modern (2005) and preindustrial (1850) in the Arctic region. The dashed contour represents 0.1 g N m⁻² year⁻¹

Fig. 2

Annual mean conditions (1970–1990) and historical changes (2009–2019 minus 1970–1990) in surface properties and the nitrogen isoscape. a Concentration of nitrate (NO₃) and b its change (Δ). c Concentration of particulate organic matter (POM) and d its change. e Values of δ¹⁵N (isoscape) of NO₃ and f its change. g Values of δ¹⁵N (isoscape) of POM and h its change. All values come from the reanalysis-driven simulation of historical conditions

Fig. 3

Major environmental drivers of the Arctic Ocean isoscape from the reanalysis-driven simulations. Average values of salinity, N* and particulate organic matter (units nitrogen) over the upper 100 m of the Arctic Ocean over simulation years 1970–2019 ce (a, d, g), their linear multi-decadal trends (b, e, h) and their normalised effect size (unitless) on inter-annual trends in δ¹⁵N_POM (c, f, g). Masked regions in right-hand panels (c, f, i) are those where regression analysis could not be performed with all three variables due to interactive effects between variables (variance inflation factor > 3.0). Stippling in right-hand panels (c, f, i) indicates a significant effect of the variable on δ¹⁵N_POM, where the 95% confidence intervals of the effect size do not intersect zero

Fig. 4

Future changes in surface properties of the Arctic Ocean nitrogen isoscape. Changes (Δ) between mean conditions over upper 100 m between 2081–2100 and 1986–2005. a Sea ice concentration. b Salinity. c Concentration of nitrate (NO₃). d Percent change in the concentration of particulate organic matter (POM) in units of nitrogen. e Values of δ¹⁵N (isoscape) of NO₃. f Values of δ¹⁵N (isoscape) of POM in units of nitrogen. All values come from the emissions-driven simulation from preindustrial to future conditions (1801–2100)

Fig. 5

Major environmental drivers of the Arctic Ocean isoscape from the emissions-driven simulations. Average values of N* and particulate organic matter (units nitrogen) over the upper 100 m of the Arctic Ocean over simulation years 1850–1950 ce (a, d), their linear multi-decadal trends (b, e), and their normalised effect size (unitless) on inter-annual trends in δ¹⁵N_POM (c, f). Masked regions in right-hand panels (c, f) are those where regression analysis could not be performed with all three variables due to interactive effects between variables (variance inflation factor > 3.0). Stippling in right-hand panels (c, f) indicates a significant effect of the variable on δ¹⁵N_POM, where the 95% confidence intervals of the effect size do not intersect zero

Fig. 6

Best predictor of temporal trends in the Arctic isoscape. Results of a multiple linear regression analysis using predictors of salinity (psu), N* (mmol m⁻³), and POM (mmol m⁻³) and the response of δ¹⁵N_POM. All variables were averaged over the upper 100 m and over the year, such that trends are inter-annual. N* = NO₃ − PO₄*16. POM is particulate organic matter in terms of nitrogen. White spaces are regions where the predictor variables showed significant interactive effects (variance inflation factor > 3.0) and where multiple linear regression was not performed. Salinity was removed as a predictor in the emissions-driven scenario due to variance inflation factors between it and N*