Seyfert–Starburst Composite Model Overview

• The Seyfert–Starburst Composite Model is a framework describing galaxies where both an active nucleus and intense star formation substantially contribute to the emission.
• Diagnostic techniques—including optical BPT diagrams, mid-IR PAH measures, and radio polarimetry—are used to disentangle AGN and starburst influences.
• Recent Bayesian and multimessenger methods quantify emission contributions, revealing intrinsic differences, evolutionary sequences, and dynamic interactions in composite systems.

The Seyfert–Starburst Composite Model denotes a class of descriptions for galaxies in which a Seyfert active galactic nucleus and nuclear or circumnuclear star formation both contribute substantially to the observed emission. In the literature, that composite character is operationalized in several non-exclusive ways: as mixed AGN narrow-line region and H II-region photoionization, as AGN photoionization plus shocks, as a coexistence of AGN jets with starburst superwinds and galactic magnetic fields, and as a multi-zone high-energy system in which coronae, starburst rings, and jet-driven outflows can all contribute. Non-HBLR Seyfert 2 galaxies, transition objects, and systems such as NGC 3079, NGC 1068, NGC 4945, NGC 6764, and IRAS 01072+4954 are recurrent exemplars of this composite regime (Wu et al., 2011).

1. Classification and conceptual scope

A widely used empirical framework is provided by the large sample assembled by Wu, Zhao, and Meng: 45 Seyfert 1 galaxies, 46 hidden broad-line region Seyfert 2 galaxies, 57 non-HBLR Seyfert 2 galaxies, and 22 starburst galaxies, for a total of 170 galaxies. This sample was explicitly designed to probe the AGN–starburst connection, because it contains pure starbursts, unobscured type-1 AGN, and a spectropolarimetrically split Seyfert 2 population in which HBLR Sy2s and non-HBLR Sy2s can be compared directly (Wu et al., 2011).

In that scheme, Sy1s show broad emission lines in direct light, HBLR Sy2s show only narrow lines in direct light but polarized broad H$\alpha$, and non-HBLR Sy2s show no detectable polarized BLR despite reasonably sensitive spectropolarimetry. Wu et al. argue that HBLR Sy2s may be the counterparts of Sy1s at edge-on orientation, whereas many non-HBLR Sy2s are intrinsically different: lower-luminosity AGNs and/or systems where the line emission and energy budget are dominated by starburst processes (Wu et al., 2011). A central consequence is that the composite model is not simply a matter of obscuration; it also involves intrinsic differences in ionization state, stellar population, and relative AGN/starburst power.

A closely related optical class is the “transition object.” In a sample of 229,618 narrow emission-line galaxies, transition objects were defined as “star-forming AGNs”: the AGN is present but its line ratios are strongly affected or hidden by circumnuclear star formation. In that usage, TOs are effectively composite systems, whereas Sy2s are more AGN-dominated in line diagnostics but still retain substantial star-formation signatures in their star-formation histories (Torres-Papaqui et al., 2013).

Individual galaxies illustrate the same point in more localized form. IRAS 01072+4954 was described as a starburst/Seyfert composite galaxy with a Seyfert 1-like X-ray emission and a starburst optical spectrum that lacks broad line emission. Its optical behavior is H II-like, but its X-ray and near-infrared properties indicate an accreting nucleus, showing that the composite category also includes low-mass, low-luminosity AGN whose classical broad-line signatures are weak or heavily diluted by star formation (Valencia-S. et al., 2012).

2. Optical, mid-infrared, and PAH diagnostics

The standard optical starting point is the BPT plane

$$x=\log\left(\frac{[\mathrm{N\,II}]\,\lambda 6584}{\mathrm{H}\alpha}\right), \qquad y=\log\left(\frac{[\mathrm{O\,III}]\,\lambda 5007}{\mathrm{H}\beta}\right).$$

Two classical boundaries are used: $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.47}+1.19$$ for the Kewley et al. “maximum starburst” line, and

$$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.05}+1.30$$

for the Kauffmann et al. empirical star-forming upper boundary. Galaxies below the Kauffmann line are pure star-forming; between Kauffmann and Kewley are composite; above Kewley are AGN-dominated (Wu et al., 2011).

Within the AGN branch, Wu et al. showed that HBLR Sy2s populate the upper-left AGN branch, closer to Sy1s, whereas non-HBLR Sy2s populate regions closer to the star-forming sequence, with lower $[\mathrm{O\,III}]/\mathrm{H}\beta$ and often higher $[\mathrm{N\,II}]/\mathrm{H}\alpha$. They adopted the empirical boundary

$$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) = 3.20\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)+1.15,$$

which separates HBLR and non-HBLR Sy2s statistically well: HBLR Sy2s tend to lie to the left, non-HBLR Sy2s to the right (Wu et al., 2011). This makes the composite model more specific than a generic BPT “composite” label: non-HBLR Sy2s occupy an intermediate regime between classical Seyfert excitation and starburst-like excitation.

Mid-infrared diagnostics sharpen that separation. Wu et al. use the 6.2 $\mu$m PAH equivalent width as a tracer of starburst dominance. Starburst galaxies have large PAH EW, typically $$\mathrm{EW}(6.2\,\mu\mathrm{m})\sim 0.5\,\mu\mathrm{m}$$; non-HBLR Sy2s also have large PAH EW, though somewhat lower than pure starbursts; HBLR Sy2s and Sy1s have small PAH EW, $\sim 0.005\,\mu\mathrm{m}$. Combined with high-ionization to low-ionization ratios such as $$x=\log\left(\frac{[\mathrm{N\,II}]\,\lambda 6584}{\mathrm{H}\alpha}\right), \qquad y=\log\left(\frac{[\mathrm{O\,III}]\,\lambda 5007}{\mathrm{H}\beta}\right).$$0, $$x=\log\left(\frac{[\mathrm{N\,II}]\,\lambda 6584}{\mathrm{H}\alpha}\right), \qquad y=\log\left(\frac{[\mathrm{O\,III}]\,\lambda 5007}{\mathrm{H}\beta}\right).$$1, and $$x=\log\left(\frac{[\mathrm{N\,II}]\,\lambda 6584}{\mathrm{H}\alpha}\right), \qquad y=\log\left(\frac{[\mathrm{O\,III}]\,\lambda 5007}{\mathrm{H}\beta}\right).$$2, these quantities divide the parameter space into AGN-dominated, starburst-dominated, and intermediate regions enriched in non-HBLR Sy2s (Wu et al., 2011).

A complementary PAH-based approach uses the 7.7 $$x=\log\left(\frac{[\mathrm{N\,II}]\,\lambda 6584}{\mathrm{H}\alpha}\right), \qquad y=\log\left(\frac{[\mathrm{O\,III}]\,\lambda 5007}{\mathrm{H}\beta}\right).$$3m feature together with

$$x=\log\left(\frac{[\mathrm{N\,II}]\,\lambda 6584}{\mathrm{H}\alpha}\right), \qquad y=\log\left(\frac{[\mathrm{O\,III}]\,\lambda 5007}{\mathrm{H}\beta}\right).$$4

and the FIR–radio parameter

$$x=\log\left(\frac{[\mathrm{N\,II}]\,\lambda 6584}{\mathrm{H}\alpha}\right), \qquad y=\log\left(\frac{[\mathrm{O\,III}]\,\lambda 5007}{\mathrm{H}\beta}\right).$$5

In this scheme, starbursts and ULIRGs occupy an “Active Star Formation Region” with low $$x=\log\left(\frac{[\mathrm{N\,II}]\,\lambda 6584}{\mathrm{H}\alpha}\right), \qquad y=\log\left(\frac{[\mathrm{O\,III}]\,\lambda 5007}{\mathrm{H}\beta}\right).$$6, strong PAH equivalent widths, and high $$x=\log\left(\frac{[\mathrm{N\,II}]\,\lambda 6584}{\mathrm{H}\alpha}\right), \qquad y=\log\left(\frac{[\mathrm{O\,III}]\,\lambda 5007}{\mathrm{H}\beta}\right).$$7, whereas Seyfert 1 galaxies and quasars have higher $$x=\log\left(\frac{[\mathrm{N\,II}]\,\lambda 6584}{\mathrm{H}\alpha}\right), \qquad y=\log\left(\frac{[\mathrm{O\,III}]\,\lambda 5007}{\mathrm{H}\beta}\right).$$8 and weaker PAHs. Seyfert 2 galaxies span a wider range and often overlap the starburst-dominated region, again indicating that strong circumnuclear star formation is common in Seyfert hosts (Higuera-G et al., 2013).

A recurrent misconception is that BPT position alone measures the starburst fraction. The later modeling literature shows that this is not generally true once aperture mixing and shocks are allowed.

3. Explicit composite modeling frameworks

A formal mixed-photoionization version of the model was developed for 2766 Seyfert galaxies from SDSS. In that framework, the observed spectrum within the fiber is treated as a linear combination of AGN narrow-line region emission and H II-region emission: $$x=\log\left(\frac{[\mathrm{N\,II}]\,\lambda 6584}{\mathrm{H}\alpha}\right), \qquad y=\log\left(\frac{[\mathrm{O\,III}]\,\lambda 5007}{\mathrm{H}\beta}\right).$$9 where $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.47}+1.19$$0 is defined as the fraction of H$$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.47}+1.19$$1 flux arising from the NLR. The mixed grids share a common metallicity and pressure, fix $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.47}+1.19$$2, and vary $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.47}+1.19$$3, $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.47}+1.19$$4, $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.47}+1.19$$5, and $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.47}+1.19$$6, with separate runs for different AGN continuum peak energies $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.47}+1.19$$7 (Thomas et al., 2018).

This Bayesian framework, implemented with NebulaBayes, quantifies a result that is central to the composite model: the majority of the Balmer line emission in the majority of Seyfert-classified SDSS spectra arises from contaminating H II regions within the fixed-size aperture. Even for spectra with $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.47}+1.19$$8, a median of $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.47}+1.19$$9 of the Balmer flux arises in H II regions, while ionizing spectra with $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.05}+1.30$$0 produce the most plausible distributions of mixing fractions (Thomas et al., 2018). The model therefore turns a qualitative “composite” label into a fitted parameter.

A second framework emphasizes AGN photoionization plus shocks. Detailed composite models calculated with SUMA for 244 local Seyfert 2 nuclei indicate shock velocities in the range of 50–300 km s$$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.05}+1.30$$1, a metallicity interval $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.05}+1.30$$2, and temperature and ionization structures that differ substantially from pure photoionization models. A key result is that shock velocity in AGNs cannot be estimated using standard optical line-ratio diagrams based on integrated spectra (Dors et al., 2020). This suggests that a Seyfert–starburst composite model that ignores shocks can misattribute line-ratio shifts either to star formation or to AGN hardness.

At $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.05}+1.30$$3, detailed SUMA shock + photoionization modeling likewise found that starburst models reproduce most of the data within the observational errors, about half of the object spectra are well fitted by an accreting AGN, and some galaxies show multiple radiation sources, such as starburst+AGN, or a double AGN. In that context, the ratio $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.05}+1.30$$4 was used as the principal constraint on the temperature structure of the emitting gas (Contini, 2016).

More recently, multi-line molecular PCA has extended the composite model into the cold ISM. In NGC 1068, PCA of 13 ALMA molecular-line maps found that one principal component reconstructs an approximate H$$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.05}+1.30$$5 column density map, another captures differences of volume density and/or chemical composition between the circumnuclear disk and the starburst ring, and a third indicates a possible interaction between the AGN outflow and gas in the starburst ring, contributed by CN, C$$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.05}+1.30$$6H, and HCN, which are known to be enhanced by the AGN outflow (Okubo et al., 19 Jun 2025). This suggests that the composite model is now extending from optical apertures and mid-infrared diagnostics to resolved, cloud-scale decomposition of AGN-, starburst-, and shock-dominated gas.

4. Radio outflows, magnetic fields, and composite feedback

In radio work, the composite model becomes dynamical. Full-Stokes polarimetry measures total intensity $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.05}+1.30$$7, linear polarization $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.05}+1.30$$8 and $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) =\frac{0.61}{\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)-0.05}+1.30$$9, degree of polarization

$[\mathrm{O\,III}]/\mathrm{H}\beta$0

polarization angle

$[\mathrm{O\,III}]/\mathrm{H}\beta$1

and rotation measure through

$[\mathrm{O\,III}]/\mathrm{H}\beta$2

These quantities separate jet-related synchrotron structures from disk or halo synchrotron and thermal superwind contributions (Sebastian et al., 2019).

The polarimetric comparison between Seyfert galaxies and starbursts shows a consistent morphological difference: many Seyfert galaxies exhibit bubble-shaped lobes, higher fractional polarization, and multiple misaligned structures, whereas starburst galaxies without AGN show diffuse radio halos aligned with the minor axis and no edge-brightened, bubble-like lobes. NGC 2639 provides an episodic AGN example, with an east–west pair of radio lobes plus an additional north–south pair of highly polarized outer lobes, interpreted as switch-off and restart of the central engine with a changed spin axis; the spectral age of the older lobes is $[\mathrm{O\,III}]/\mathrm{H}\beta$3 Myr (Sebastian et al., 2019).

NGC 3079 is the archetypal Seyfert–starburst composite in this literature. It contains a Seyfert nucleus with a mildly relativistic jet, a strong circumnuclear starburst driving superwinds, and dynamo-generated large-scale magnetic fields. Its 8-shaped radio lobes are bubble-like and edge-brightened; the northern ring reaches fractional polarization $[\mathrm{O\,III}]/\mathrm{H}\beta$4 at 5 GHz, the magnetic fields align with the linear extents of the optically thin filaments, and the rotation-measure images show inversion across the lobe (Sebastian et al., 2019). The preferred interpretation is not a simple starburst wind and not a classical, freely propagating jet. Rather, the jet provides relativistic plasma, the starburst superwind transports it outward, and galactic magnetic fields organize it into filamentary rings and bubbles.

NGC 6764 provides a related, lower-scale example. VLBA imaging detected a core–jet structure and a possible weak counterjet component at 1.6 GHz; the core brightness temperature is $[\mathrm{O\,III}]/\mathrm{H}\beta$5 K, the emission appears to come from a synchrotron jet, and a precessing jet model can explain the radio morphology from parsec to sub-kpc scales. The jet seems to be disrupted on scales of hundreds of parsecs, possibly due to interaction with and entrainment of the interstellar medium gas, which subsequently leads to the formation of bubbles. The inferred 12–21 Myr timescale for inflating the southern bubble corresponds roughly to the starburst episode that took place 15–50 Myr ago (Kharb et al., 2010).

In practice, the radio version of the Seyfert–Starburst Composite Model is a three-component flow: AGN jet energy, starburst superwind transport, and magnetic-field confinement or guidance.

5. Evolutionary interpretations and low-mass composite nuclei

Wu et al. proposed an explicit sequence

$[\mathrm{O\,III}]/\mathrm{H}\beta$6

The observational basis includes stellar-population synthesis on a subset of nuclei, with light-weighted ages $[\mathrm{O\,III}]/\mathrm{H}\beta$7 for starbursts, $[\mathrm{O\,III}]/\mathrm{H}\beta$8 for non-HBLR Sy2s, and $[\mathrm{O\,III}]/\mathrm{H}\beta$9 for HBLR Sy2/Sy1 nuclei. In parallel, $[\mathrm{N\,II}]/\mathrm{H}\alpha$0 increases from starbursts to non-HBLR Sy2s and then decreases toward HBLR Sy2s and Sy1s, while ionization indicators such as $[\mathrm{N\,II}]/\mathrm{H}\alpha$1, $[\mathrm{N\,II}]/\mathrm{H}\alpha$2, and $[\mathrm{N\,II}]/\mathrm{H}\alpha$3 increase from starburst to non-HBLR Sy2 to HBLR/Sy1 (Wu et al., 2011). Orientation explains the HBLR Sy2 $[\mathrm{N\,II}]/\mathrm{H}\alpha$4 Sy1 relation, but an evolutionary sequence is invoked for starburst $[\mathrm{N\,II}]/\mathrm{H}\alpha$5 non-HBLR Sy2 $[\mathrm{N\,II}]/\mathrm{H}\alpha$6 HBLR/Sy1.

A population-level SFH study gives a different but related view. In narrow emission-line galaxies, the star-formation histories vary systematically with mass and bulge prominence: LINERs reside in massive early-type galaxies, Sy2s and TOs are hosted by intermediate-mass galaxies with intermediate morphological types, and SFGs are found in lower-mass late-type spirals. Except for the Sy2s, the more massive galaxies formed the bulk of their stars more rapidly than the less massive ones. The Sy2s formed their stars more slowly and show presently an excess in star formation. The maximum in star-formation rate in the past increases with the virial mass within the aperture, and the virial mass within the aperture increases from the SFGs to the TOs, to the Sy2s, culminating in the LINERs (Torres-Papaqui et al., 2013). This suggests that composite Seyfert phases are tied to bulge growth rather than to abrupt AGN-driven truncation of star formation.

A low-mass limit of the same phenomenon is provided by IRAS 01072+4954. Near-infrared integral-field spectroscopy found little obscuration toward the nucleus, $[\mathrm{N\,II}]/\mathrm{H}\alpha$7 mag, unresolved hot-dust emission with $[\mathrm{N\,II}]/\mathrm{H}\alpha$8 K, a black-hole mass $[\mathrm{N\,II}]/\mathrm{H}\alpha$9, and a bolometric luminosity $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) = 3.20\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)+1.15,$$0, giving an Eddington ratio $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) = 3.20\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)+1.15,$$1. If scaling relations from more massive systems apply, the broad-line region should produce lines with $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) = 3.20\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)+1.15,$$2, and some indications for such narrow broad-line components were reported in the near-infrared (Valencia-S. et al., 2012). The source therefore seems not to be a True-Seyfert 2, but an extreme case of a narrow-line Seyfert 1 whose AGN is faint enough to be optically overwhelmed by host-galaxy star formation.

A plausible implication is that the “composite” category is not a single evolutionary state. It includes transitional non-HBLR Sy2 systems, aperture-mixed TOs, radio composites in which jets and superwinds coexist, and low-mass nuclei whose AGN signatures are intrinsically weak.

6. High-energy and multimessenger extensions

The composite model has also entered the GeV–TeV and neutrino regime. Fermi/LAT observations of NGC 1068 and NGC 4945 found gamma-ray excesses of 8.3 sigmas and 9.2 sigmas, respectively, with power-law photon indices $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) = 3.20\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)+1.15,$$3 for NGC 1068 and $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) = 3.20\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)+1.15,$$4 for NGC 4945. The high-energy flux of NGC 4945 is consistent with starburst activity, whereas that of NGC 1068 is an order of magnitude above expectations, suggesting dominant emission from the active nucleus (Lenain et al., 2011).

In that high-energy picture, the starburst component is modeled as a hadronic calorimeter in which cosmic rays accelerated by supernovae interact with dense gas, while the Seyfert component can contribute through coronae, inner winds, or jet-driven bubbles. H.E.S.S. observations of the closest composite Seyfert-starburst galaxies—NGC 1068, Circinus, and NGC 4945—found no significant VHE signals. The resulting upper limits constrain theoretical models with different spectral components. For a proton index $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) = 3.20\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)+1.15,$$5, the calorimetric efficiency is constrained to $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) = 3.20\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)+1.15,$$6 for NGC 1068 and $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) = 3.20\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)+1.15,$$7 for NGC 4945; in NGC 1068 and Circinus, lobe models require electron break Lorentz factors $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) = 3.20\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)+1.15,$$8 and $$\log\left(\frac{[\mathrm{O\,III}]}{\mathrm{H}\beta}\right) = 3.20\log\left(\frac{[\mathrm{N\,II}]}{\mathrm{H}\alpha}\right)+1.15,$$9, respectively (Collaboration et al., 23 Mar 2026).

A recent steady-state composite model of NGC 1068 and NGC 7469 makes the multi-zone structure explicit. It treats the system as two distinct non-thermal environments: a homogeneous AGN corona around the supermassive black hole and a circumnuclear starburst ring on kiloparsec scales, with stochastic diffuse acceleration and $\mu$0-pair attenuation included in the corona. The Bayesian parameter study shows that the non-thermal features of NGC 1068 can be explained well, but NGC 7469 exposes a geometric tension: the mm-bump needs to emerge from a coronal size $\mu$1, whereas TeV–PeV neutrino emission requires $\mu$2 (Eichmann et al., 17 Feb 2026). The authors conclude that there is not a unique edge of the corona and that spatial inhomogeneities must matter.

In this high-energy extension, the Seyfert–Starburst Composite Model becomes explicitly multi-messenger. The starburst ring, AGN corona, and jet- or wind-driven outflow are treated as distinct but simultaneous accelerators and radiation zones, and non-detections at VHE are as informative as detections at GeV or in neutrinos.