JADES: Evolution of nitrogen abundances in star-forming galaxies from z ~ 1.5-7

TL;DR

This study presents a JWST/JADES census of nitrogen-to-oxygen abundances in star-forming galaxies from $z\sim1.5$ to $7$, using the low-ionisation [N II]$\lambda$6583 line to derive N/O across a large, diverse sample. A Te-based subsample (40 galaxies with [O III]$\lambda$4363) provides direct oxygen abundances and, together with [S II] density constraints, yields robust $N/O$ measurements; the full sample uses strong-line calibrations to extend the analysis to higher metallicities. The authors find that, at low metallicity, the average $\log({\rm N}/{\rm O})$ is elevated by at least $\sim0.1$ dex relative to $z\sim0$ populations, with about 13% of galaxies at $12+\log({\rm O}/{\rm H})<8.0$ showing nitrogen enhancements $\log({\rm N}/{\rm O})>-1.1$. However, no system reaches the extreme $log({\rm N}/{\rm O})>-0.6$ values typical of high‑redshift N III]- and N IV]-emitters, suggesting such enhancements occur only in the most extreme starbursts or under stratified ionisation conditions. The results imply that nitrogen enrichment in young, high‑redshift galaxies is driven largely by current-generation massive stars in low-metallicity environments, while secondary nitrogen production dominates in chemically evolved, higher-metallicity systems on the star-forming main sequence, offering important constraints on ISM enrichment pathways and galaxy evolution with cosmic time.

Abstract

We present nitrogen abundance measurements based on the low-ionisation [NII]6583 emission line for 588 galaxies between 1.5<z<7.0 from the JWST Advanced Deep Extragalactic Survey (JADES). We detect the temperature-sensitive [OIII]4363 auroral line in 40 galaxies in our sample, affording $T_e$-based abundances for this subset. We find that the average N/O abundance ratio in our low-metallicity sample is at least 0.1 dex higher than z ~ 0 samples. In particular, we find significant scatter toward high N/O, with five galaxies being identified with enhanced nitrogen abundances (log(N/O)>-1.1) at low-metallicity (12+log(O/H)<8.0) from $T_e$-based measurements. Meanwhile, applying strong-line abundance measurements to the remainder of our sample reveals a further 14 candidate galaxies passing these abundance cuts, implying that around 13 % of 12+log(O/H)<8.0 galaxies at these redshifts are nitrogen-enhanced at this level. We find that N/O abundance in low-metallicity systems correlates with SFR, surface density of SFR, and surface density of stellar mass at high redshift, while only in high-metallicity systems does a correlation with stellar mass emerge. Despite healthy representation of these `moderately nitrogen-enhanced' galaxies (-1.1<log(N/O)<-0.6), no galaxies in our low-metallicity sample are identified as having log(N/O)>-0.6, abundances that are typical of high-redshift NIII]- and NIV]-emitters. This demonstrates that the extreme nitrogen enhancements seen in some NIII]- and NIV]-emitters are only attained during the most extreme starbursts. This suggests that these elevated abundances are caused by enrichment from young massive stars in extreme environments and that the impact of this enrichment pathway is milder, though still important, for high-redshift systems on the star-forming main sequence.

Figures (17)

• Figure 1: Upper panel:$m_{\rm F444W}$ vs. redshift for JADES NIRSpec galaxies entering into this sample. Large circles show the galaxies making it into our final sample (see Section \ref{['sec:data']}) with colours denoting redshift. Green points show the distribution of all JADES galaxies with robust spectroscopic redshifts, while grey points show the portion of the JADES sample for which no reliable redshift was obtained (with the point location given by the photometric redshift). Lower left: Redshift distribution of final sample (colours) vs. all JADES galaxies with a robust spectroscopic redshift and $1.5<z\leq 7$. Lower right: As for lower left, except showing the $m_{\rm F444W}$ distribution. Black dashed lines at $m_{\rm F444W}=27.0$ and $27.5$ show the magnitude cuts for the $m_{\rm F444W}$-limited portion of the JADES NIRSpec selection for Medium/JWST and Deep/JWST respectively (see CurtisLake2025_DR4 for details).
• Figure 2: Example spectra of three galaxies in the $T_e$-based sample. Left: RGB thumbnail showing the location of the three-shutter slitlet. Middle: Section of the NIRSpec spectrum in which key rest-frame optical emission lines from [O ii] $\lambda\lambda$3726, 3729 to [O iii] $\lambda$5007 are observed. Right: Section of the NIRSpec spectrum surrounding H$\alpha$, [N ii] $\lambda$6583, and [S ii]$\lambda\lambda$6716, 6731. Our best-fit model to this complex is shown in green. There is generally overlapping wavelength coverage between different gratings. Across both plot regions, the top panel shows the 2D spectrum from either G140M/F070LP or G395M/F290LP (see label), while the second panel shows the 2D spectrum from G235M/F170LP. In the 1D spectrum, G235M/F170LP is in black and other gratings are in blue.
• Figure 3: Diagnostic diagrams showing the location of galaxies in our sample in emission line ratio space compared to $z\sim0$ galaxies from SDSS Aihara2011. Red points show the galaxies that we discard from our sample due to significant AGN contamination, while the other colours indicate the approximate redshift of each source. The Kewley2001 AGN-SF demarcation line is shown in blue.
• Figure 4: Main panel: JADES $T_e$-based N/O abundance measurements as a function of measured [N ii] $\lambda$6583/[O ii] $\lambda\lambda$3726, 3729 ratio under different $T_e$ and $n_e$ assumptions. All points have $T_e$[O iii] derived from [O iii] $\lambda$4363/$\lambda$5007 measurements, comprising a sample of 22 galaxies. Purple points show abundances derived assuming the $T_e$[O ii] -- $T_e$[O iii] relation of Cataldi2025, while red points show how these change if Pilyugin2009 is adopted instead. Circles denote galaxies for which $n_e$ is measured from the [S ii]$\lambda\lambda$ 6716,6731 doublet, and the blue circles show how the purple points would shift if the $+1\sigma$ value was adopted instead of the best-fit $n_e$ value. Diamonds denote galaxies without $n_e$ constraints -- purple and blue diamonds assume $n_e=100$ & $10^4$ cm$^{-3}$, respectively. Dotted and dashed lines show conversions of [N ii]/[O ii] to N/O for different $T_e$ and $n_e$ values (assuming ICF(${\rm N}^+/{\rm O}^+)=1$). Dark blue solid, light blue solid, green dot-dash, and yellow double-dot-dash lines show the PerezMontero2009, Cataldi2025_NO, HaydenPawson2022, and Florido2022 strong-line calibrations respectively. Top panel: The distribution of [N ii] $\lambda$6583/[O ii] $\lambda\lambda$3726, 3729 ratios among galaxies in which both lines are detected with $S/N>3$ in our JADES strong-line sample.
• Figure 5: Nitrogen-to-oxygen abundance ratio as a function of oxygen abundance for our JADES $T_e$-based sample (blue diamonds). The blue dashed line shows the weighted average N/O for this $12+\log({\rm O}/{\rm H})$$<8.2$ sample. Salmon arrows show how these points would shift if the Pilyugin2009$T_e$[O ii]-$T_e$[O iii] calibration was used instead of the Cataldi2025 adopted for the fiducial values. Purple arrows show the systematic uncertainty arising from a weak or absent $n_e$ constraint (see Section \ref{['sub:Te_based_NO']} for details). Green circles and diamonds show $z\sim0$ H ii regions Pilyugin2012Berg2020. Green squares show $z\sim0$ star-forming galaxies from the CLASSY survey ArellanoCordova2025_CLASSY. Orange '$\times$' marks show the high-N/O-selected subsample of DESI galaxies from Bhattacharya2025. High-redshift N iii]- and N iv]-emitters are shown by the brown hexagons, compiled from Isobe2023MarquesChaves2023Castellano2024Schaerer2024Martinez2025Naidu2025. The grey solid, beige solid, and brown dotted lines show the ${\rm N}/{\rm O}-{\rm O}/{\rm H}$ scaling from Nicholls2017, Cataldi2025_NO, and Scholte2026 respectively, while the purple dashed line shows the average N/O value derived for $z\sim0$ metal-poor dwarf galaxies in Berg2019_CNO_Dwarf. The faint dotted lines show the solar ratios of each abundance.