The Nearby Star Formation and Supernova Histories Reconstructed from Young Star Clusters

Abstract: We reconstruct the recent star formation and core-collapse supernova (ccSN) histories of the Solar Neighborhood from the past trajectories of young star clusters. Using a \textit{Gaia}-based cluster sample with newly derived ages, masses, and bulk 3D velocities, we integrate orbits backward in an assumed axisymmetric Galactic potential and combine the trajectories with IMF sampling and stellar lifetimes to infer ccSN times and locations over the past 50 Myr. The result is an all-sky, 3D, time-resolved map of nearby ccSN activity for comparison with high-resolution 3D views of the local interstellar medium. The 0--15 Myr map shows strong enhancements toward Orion, Vela, Sco--Cen, and Cepheus, many within present-day cavities and shells. At earlier times, the dominant enhancements trace the Collinder 135, Messier 6, and Alpha Persei cluster families, showing how the remnants of massive star-forming complexes have shaped the recent local feedback history. We recover a bursty star formation history followed by a delayed, smoother ccSN history. Over the last 40 Myr, the mean star formation and ccSN rates are (823~M_\odot~\mathrm{Myr}^{-1}) and (7.7~\mathrm{Myr}^{-1}), respectively, corresponding to a Milky Way rate of (0.55\pm0.03~\mathrm{century}^{-1}). Present-day OB-star catalogs yield rates ranging from agreement with the cluster reconstruction to several times higher. Because the catalogs overlap weakly and require different corrections, we do not rescale the ccSN map. Our reconstruction provides an empirical framework for connecting the recent history of massive-star feedback to the 3D structure and life cycle of gas in the nearby Milky Way.

• The paper reconstructs the Solar Neighborhood’s clustered core-collapse supernova history over 50 Myr by tracing 568 young Gaia clusters, modeling stellar evolution, progenitors, binaries, failed explosions, and runaway stars.
• The reconstruction identifies 9⁺³₋₃ supernovae inside the Local Bubble over 15 Myr, 17⁺⁵₋₄ in Orion–Eridanus, and 132⁺¹¹₋₁₁ associated with the GSH 238+00+09 supershell, broadly matching independent energy and momentum estimates.
• The paper finds bursty, strongly clustered star formation dominated by the Collinder 135, Messier 6, and Alpha Persei families, while the local supernova-rate normalization remains uncertain at 0.55 ± 0.03 per century because OB-star catalogs disagree by up to a factor of five.

Overview and motivation

Core-collapse supernovae (ccSNe) dominate the mechanical energy budget of the interstellar medium (ISM), each releasing $\sim10^{51}$ erg, and their clustered occurrence is widely invoked to explain superbubbles, chimneys, and multiphase structure. The Solar Neighborhood is the only environment where both the ISM and its stellar feedback sources can be resolved in three dimensions, thanks to Gaia astrometry and parsec-scale 3D dust maps. In this paper, Swiggum et al. reconstruct the recent star formation and ccSN history of the Solar Neighborhood by tracing back the orbits of young Gaia star clusters, converting their ages, masses, and trajectories into a probabilistic, time-resolved, all-sky map of ccSN activity over the past 50 Myr (2608.20307). The work extends the cluster-family traceback framework of Swiggum et al.'s earlier analysis, which showed that most young clusters within $\sim1$ kpc descend from three past massive star-forming complexes (the Collinder 135, Messier 6, and Alpha Persei families).

Sample construction and methods

The starting point is the Hunt & Reffert (HR23) all-sky Gaia DR3 open cluster catalog of 7167 clusters, filtered by reliability criteria ($class\_50>0.5$, $cst>5$), a $|X|,|Y|,|Z|\leq1.25$ kpc spatial cut, HR23 ages $<200$ Myr, at least five member radial velocities, and bulk velocity component uncertainties below $5~\mathrm{km~s^{-1}}$. Ages and masses are re-derived with Chronos, a Bayesian isochrone-fitting code using PARSEC models, with extinction priors drawn from 3D dust maps (Edenhofer et al. 2024, Bayestar19, DECaPS). Notably, 47% of the 2508 fits yield multimodal age posteriors, and the adopted age component is selected by integrated posterior probability. Present-day masses are converted to initial masses via numerical inversion of the cluster disruption law of Almeida et al. (2024), accounting for secular mass loss. The final 2.5 kpc cube sample contains 568 clusters; a 1 kpc cube subsample of 312 clusters is used for rate and clustering statistics.

Radial velocities are compiled from Gaia DR3, APOGEE DR17, GALAH DR4, DESI, Gaia-ESO DR6, and RAVE DR6, combined by inverse-variance-weighted median. Bulk 3D velocities are fit by uncertainty-weighted least squares to all member-star tangential velocities plus available radial velocities, achieving mean uncertainties of $0.85$, $0.92$, and $0.54~\mathrm{km~s^{-1}}$ in $\sim1$0 for the 2.5 kpc sample. Orbits are integrated 100 Myr backward in the axisymmetric MWPotential2014 potential with galpy, with Monte Carlo propagation of phase-space uncertainties (100 realizations per cluster). The authors acknowledge that non-axisymmetric perturbations from spiral arms and molecular clouds can deflect orbits over $\sim1$1 Myr, but note that simulations indicate co-born stars retain correlated orbital changes for up to 0.5 Gyr, so relative cluster motions—the quantity most relevant here—are more robust than absolute positions.

For each cluster, 100 Kroupa IMF realizations over $\sim1$2--$\sim1$3 generate candidate ccSN progenitors ($\sim1$4), with lifetimes interpolated from non-rotating, solar-metallicity PARSEC v2.0 tracks. Simplified prescriptions account for interacting binaries (50% of stars $\sim1$5 receive shortened lifetimes), failed supernovae (20% of progenitors above $\sim1$6 removed), and runaways (10% of exploded progenitors removed). Explosion locations are assigned on the backward-integrated orbit at the progenitor's death time, with isotropic offsets within the catalogued cluster radius. The authors are explicit that the result captures only the clustered component of the ccSN history recoverable from surviving clusters; full binary population synthesis, runaway trajectories, and cluster expansion are not modeled.

The recent ccSN map and its correspondence to ISM cavities

The 0–15 Myr reconstruction shows strong ccSN enhancements toward Orion, Vela, Sco–Cen, Cepheus Far, Camelopardalis OB1, Lacerta, Aquila Rift, Ara/Norma/Circinus, and CMa, many coinciding with present-day dust cavities and shells in the parsec-scale 3D dust map of Edenhofer et al. Quantitatively, the reconstruction places $\sim1$7 ccSNe inside the 3D Local Bubble volume over 15 Myr—slightly below but consistent with the 10–20 SNe long invoked from Hipparcos-era traceback studies and the $\sim1$8 Myr expansion age. Orion–Eridanus contains $\sim1$9 ccSNe over 15 Myr, with associated wind-plus-supernova mechanical energy input of $class\_50>0.5$0 erg, comparable to the $class\_50>0.5$1 erg budget previously estimated for Orion OB1. The IRAS Vela Shell hosts $class\_50>0.5$2 ccSNe over 3 Myr, matching the one to two SNe required by its momentum budget, while the kiloparsec-scale supershell GSH 238+00+09 accumulates $class\_50>0.5$3 ccSNe over 30 Myr, dominated by the older Cr135 and M6 cluster families. This spatial correspondence between reconstructed feedback sites and carved cavities identifies the structures most plausibly shaped by recent massive-star feedback, though the authors caution that the dust map shows the present-day ISM, not the gas distribution at the explosion epochs.

The 50 Myr traceback and the role of cluster families

Extending the map across five 10 Myr lookback windows, the reconstruction shows that at 20–50 Myr ago the ccSN activity is dominated by three separable enhancements tracing the Cr135, M6, and $class\_50>0.5$4~Per cluster families. The Cr135 and M6 enhancements merge between 20–30 and 10–20 Myr ago, and the smaller $class\_50>0.5$5~Vel family begins contributing at their intersection; Sco–Cen emerges as the youngest nearby component of the broader $class\_50>0.5$6~Per family. The event counts rise monotonically toward the present, from $class\_50>0.5$7 ccSNe at 40–50 Myr to $class\_50>0.5$8 in the last 10 Myr within the traceback footprint. The authors note that weaker substructures fade at earlier lookback times, plausibly because dissolved older clusters are progressively absent from the surviving-cluster catalog—a selection effect that bears directly on the completeness of the reconstruction at older epochs.

Star formation history, rates, and feedback power

The reconstructed star formation history is bursty rather than secular. After a dearth before $class\_50>0.5$9 Myr ago—partly, but not wholly, attributable to cluster dissolution, since the sharp drop is not naturally explained by gradual disruption—the SFR rises around 45 Myr ago with the onset of the $cst>5$0~Per and M6 families. Averaged over the last 40 Myr, the present-day 1 kpc cube sample yields a mean SFR of $cst>5$1 and a mean ccSN rate of $cst>5$2, corresponding to $cst>5$3 and $cst>5$4 within the traced volume. Strongest bursts reach $cst>5$5, roughly three times the long-term average, with rise and fall times of a few Myr comparable to molecular-cloud lifetimes in Milky Way-mass simulations. The ccSN rate is smoother than the SFR, as progenitor lifetimes spread each formation burst into a broader feedback episode. The cluster-based ccSN rate converts to a Milky Way-equivalent rate of $cst>5$6 using the Reed (2005) disk-plus-hole extrapolation—an extrapolation that assumes the local rate is representative at the Solar radius and does not correct for progenitors missing from the sample.

The clustering statistics reinforce the bursty picture. Against randomized cluster locations, reconstructed ccSNe are more clustered throughout the last 50 Myr: the different-host nearest-neighbor ratio is 1.5–1.6 over the last 20 Myr and 1.8–2.2 from 20–50 Myr ago, while the fully randomized-location comparison yields ratios of 2.1 (recent) to 2.6–2.9 (20–50 Myr ago). This indicates that the clustering reflects both repeated explosions within individual clusters and the initially compact configuration of sibling clusters within families—consistent with the prediction that bursty star formation should produce strong spatio-temporal clustering of supernovae.

The OB-star normalization problem

A central tension in the paper concerns the absolute rate normalization. Two independent Gaia DR3-based OB-star catalogs are compared against the cluster reconstruction within the 1 kpc cube: the Quintana et al. (Q25) census of 24,706 OB stars within 1 kpc, and the ALS III catalog of 15,542 Galactic OB candidates. Restricted to $cst>5$7, Q25 implies $cst>5$8 and $cst>5$9—within 1.1–1.3 of the cluster-based values—whereas ALS III implies $|X|,|Y|,|Z|\leq1.25$0 and $|X|,|Y|,|Z|\leq1.25$1, factors of 4.4 and 5.0 higher. The two catalogs share only 60 stars out of 731 unique sources in the same volume.

Crossmatching cluster members to both catalogs yields diagnostic but non-convergent corrections. Isochrone-based masses suggest Q25 underestimates masses by a median factor of 1.3 (raising its rate to $|X|,|Y|,|Z|\leq1.25$2), while 79% of matched ALS III $|X|,|Y|,|Z|\leq1.25$3 stars fall below $|X|,|Y|,|Z|\leq1.25$4 (lowering its rate to $|X|,|Y|,|Z|\leq1.25$5, close to the cluster value). Spectroscopic classifications, however, give contamination fractions of 20% (Q25) and 47% (ALS III) and incompleteness fractions of 55–56%, implying corrected rates of 1.26 and $|X|,|Y|,|Z|\leq1.25$6 respectively—nearly a factor of six apart for ALS III depending on the diagnostic. Because the corrections depend so strongly on the adopted reference, the authors decline to rescale the ccSN map, treating the cluster-based rates as conservative lower limits with a plausible normalization bracket of roughly $|X|,|Y|,|Z|\leq1.25$7 to $|X|,|Y|,|Z|\leq1.25$8. They note that regional comparisons (SigMA recovering 98.3% of HR23 Sco–Cen members but 3.5 times as many young stars) support the higher ALS III-like rates, while the present-day ISM—where the cluster-family supernova counts are already broadly consistent with cavity sizes and kinetic-energy budgets—cautions against adopting them, since higher-SFR simulations over-disrupt and over-ionize the local gas.

Limitations and open questions

Several limitations are conceded explicitly. The reconstruction captures only the clustered component of ccSN activity; dispersed star formation and rapidly dispersing clusters would be missed, and the true fraction of star formation occurring outside recognizable clusters is precisely the unresolved question raised by the OB-star disagreement. The apparent SFR drop over the last $|X|,|Y|,|Z|\leq1.25$9 Myr likely reflects embedded clusters underrepresented in Gaia catalogs. Age and mass estimates become ambiguous near the "isochrone blindspot" at 80–100 Myr, marking those intervals as less constrained. The axisymmetric potential, coeval-cluster assumption, time-independent cluster radius, and simplified binary, direct-collapse, and runaway prescriptions all introduce systematic uncertainties that are not fully propagated. The spectroscopic reference samples are not randomly selected, so their contamination and incompleteness fractions may not apply to the full field population. The paper leaves open whether the absolute local star formation rate is closer to the cluster/Q25 normalization or the ALS III normalization, identifying spectroscopic follow-up of candidate OB stars as the decisive missing constraint.

Conclusion

This work delivers an empirical, probabilistic, time-resolved reconstruction of ccSN activity in the Solar Neighborhood over the past 50 Myr, built from 568 young Gaia clusters with re-derived ages, masses, and 3D kinematics. Its principal results are the spatial coincidence of recent ccSN enhancements with known shells and cavities (with event counts consistent with independent energy and momentum budgets), the dominance of the Cr135, M6, and $<200$0~Per cluster families in shaping the local feedback history, a bursty star formation history with peaks roughly three times the 40 Myr mean, and significantly super-random ccSN clustering throughout the interval. The cluster-based Milky Way ccSN rate of $<200$1 sits near the low end of literature estimates, and the unresolved discrepancy among massive-star tracers—spanning a factor of $<200$2 in rate—remains the key open question for pinning down the absolute feedback budget of the local ISM.