**9. Summary and Outlook**

Recent comprehensive spectroscopic and imaging surveys have revealed that the Tarantula Nebula hosts the most exceptional massive star population within the Local Group of galaxies, including the most massive stars identified to date, the fastest rotating early-type stars, and the X-ray brightest colliding wind system. In particular, the VFTS survey has revealed an excess of massive stars with respect to a Salpeter IMF [40] and added support from previous results for the importance of close binary evolution in the evolution of massive stars [64]. As such, the Tarantula Nebula is the closest analogue to the Hubble Deep Field for the community interested in the evolution of massive stars since its richness provides us with a huge breadth of extreme stars.

The integrated appearance of R136 resembles young extragalactic star clusters, while the integrated nebular properties of NGC 2070 is analogous to extreme Green Pea galaxies at low redshift and star-forming knots in high-redshift galaxies. Typical metallicities of Green Pea galaxies are lower than the LMC, as measured from oxygen nebular lines, while the oxygen content of high-*z* star forming galaxies tends to be similar to those of the Magellanic Clouds. However, *α*/Fe abundances of high redshift galaxies are likely to be higher than young populations in the Milky Way, such that winds from OB stars at high redshift are anticipated to be weaker than in the LMC or SMC [112].

The LMC metallicity is only a factor of two below that of the Solar neighbourhood [3], so ideally we would like to supplement the extensive survey of the Tarantula Nebula with counterparts at significantly lower metallicity. The SMC (1/5 solar) represents our best opportunity to study the formation and evolution of massive stars at a metallicity significantly below that of the LMC. Alas, it does not host as rich a massive star-forming region as the Tarantula, but cumulatively does host a substantial number of O stars so is key towards our improved understanding of massive stars at low metallicity, especially as it can be studied in exquisite detail with current instrumentation.

Since we need to look beyond the SMC to study metal-poor counterparts to the Tarantula Nebula, Figure 11 compares the star-formation rate, metallicity, and distance modulus of Local Group dwarf galaxies. Rates of star formation in metal-poor (≤20% of solar oxygen content) galaxies are significantly lower than the Magellanic Clouds, so there are no rich metal-poor massive star populations elsewhere in the Local Group, and those few that are present are much more distant. The absence of nearby metal-poor counterparts to the Tarantula implies that we currently have to rely on the interpretation of integrated populations in order to understand massive stellar evolution at low metallicity, notably extremely metal-poor dwarf star-forming galaxies I Zw 18 and SBS-0335.

**Figure 11.** Comparison between present-day star formation rates, as measured by H*α* luminosity [113], distance modulus (mag), and oxygen metal content (squares: ≥20% of solar value, triangles: <20% of solar value, for Local Group dwarf galaxies. Metal-poor galaxies possess low star-formation rates, so host small numbers of OB stars, and these are ≥6 magnitudes fainter than Magellanic Cloud counterparts.

In order to test predictions of the metallicity dependence of massive star winds, it is necessary to measure mass-loss properties across a wide range of metallicities. Although theory has been qualitatively supported from the observed wind properties of Milky Way, LMC and SMC early-type stars [82], some issues remain, including weak winds in low luminosity OB stars. Our only opportunity to study individual massive stars below 1/10th of the solar oxygen content is to observe O stars in Sextans A and B with 7% solar [114] or the Sagittarius Dwarf Irregular Galaxy (SgrDIR) with 5% solar [115]. Stellar winds from early-type stars at such low metallicities are anticipated to be much weaker than in metal-rich populations, which has been confirmed by UV spectroscopy. By way of example, Figure 12 compares the far UV spectrum of *ξ* Per (O7.5 III) with a counterpart in Sextans A [116], revealing negligible wind signatures in the latter (e.g., C IV *λ*1550).

**Figure 12.** Comparison between far UV spectroscopy of mid O giants in metal-rich [117] and metal-deficient [116] environments, illustrating the extreme differences in wind features (e.g., N˙,V *λ*1240, Si IV *λ*1400, C IV *λ*1550) and the iron forest (Fe IV-V).

**Funding:** This research received no external funding.

**Acknowledgments:** This review is dedicated to the memory of Nolan Walborn, from whom the author learnt a grea<sup>t</sup> deal about the Tarantula Nebula. Thanks to Leisa Townsley and Patrick Broos for access to T-ReX point source results prior to publication, Miriam Garcia for the UV spectrum of Sextans A OB 326, Selma De Mink for the VFTS pie chart, Heloise Stevance for the BPASS predictions, and Joachim Bestenlehner for converting *Q* wind density results into mass-loss rates. Feedback from Fabian Schneider, Chris Evans, Joachim Bestenlehner, Andy Pollock, Roberta Humphreys and external referees on an earlier draft is greatly appreciated.

**Conflicts of Interest:** The author declares no conflict of interest.
