top of page


My research interests are broadly focused on observationally characterizing star formation and supermassive black hole growth in the first billion years of cosmic history, as well as connecting galaxy formation to cosmic hydrogen reionization.

Over the past year, my work has largely focused on characterizing faint early galaxies with deep JWST data. This is complementing my thesis work which focused on observations of the brightest galaxies present in the reionization era using telescopes from the optical through far-infrared (Subaru, MMT, VISTA, Spitzer, and ALMA). Because very early galaxies are so distant and therefore generally faint, this extremely luminous subset offered a valuable preview on the properties of reionization-era galaxies. With the advent of JWST, I am now extending these studies to a much fainter regime to better understand the galaxy population generally thought to drive reionization. 

My first-author and co-authored publication lists can be found with ADS.

Star Formation in the First Billion Years

Screenshot 2023-11-10 at 11.05.33 AM.png
Screenshot 2023-11-10 at 11.07.30 AM.png

The incredible sensitivity and wavelength regime of JWST provide an unprecedented opportunity to finally understand the star-forming and ionizing properties of very faint z>6 galaxies. Using deep Cycle 1 JWST data, we have begun statistically characterizing these very faint systems in detail (Endsley et al. 2023b, 2023c). 

After identifying a sample of nearly 1,000 galaxies at redshifts z ~ 6-9 (when the universe was ~550-900 million years old), we found extremely strong statistical evidence for striking behavior in how early galaxies form their stars. Several galaxies showed extremely strong rest-optical nebular line emission, implying a very dramatic upturn in their star formation rates over the past ~5 Myr. But there were also galaxies showing light dominated by fairly young stars (~30 Myr old), but no signatures of extremely young stars. These objects seem to have undergone a strong downturn in their star formation rates over the past ~5 Myr. These two populations likely reflect the peak and trough phases expected from very bursty star formation histories.

Combining constraints from the full ~1,000 galaxy sample, we indeed find strong evidence for bursty star formation histories among early galaxies. But moreover, we find that the most UV-luminous early galaxies are very frequently experiencing a burst of star formation, while the much (~10x) fainter galaxies have a more even mix of recently declining and recently rising star formation histories. This gives us two key big-picture results. First, early UV-bright galaxies are UV bright because they recently formed a lot of hot, massive O stars. This helps us understand the high abundance of bright galaxies at even earlier cosmic times (z>10). Second, our results imply that the most UV-luminous galaxies are typically more efficient producers of ionizing photons than the fainter galaxies. But more work is required to really determine whether bright or faint galaxies drove cosmic reionization.

Supermassive Black Hole Growth in the Very Early Universe


Deep wide-area datasets present a unique opportunity to not only identify, but also characterize the most massive galaxies present in the very early Universe. Using the rich multi-wavelength datasets available over the 1.5 deg^2 COSMOS field, we previously identified (Endsley et al. 2022b) an extremely red Lyman-break z~6.8 galaxy (COS-87259) showing bright detections in mid-infrared, far-infrared, and radio data, suggesting this system may host both a very luminous obscured starburst and AGN.


Observations taken in ALMA Cycle 8 revealed an extremely bright and broad [CII] line at z=6.853, indicating that COS-87259 is indeed located in the epoch of reionization and is undergoing extremely rapid obscured star formation with SFR~1300 Msol/yr (Endsley et al. 2023a). The full multi-wavelength data further reveal that this object is an extremely massive (M*~1.7x10^11 Msol) galaxy harboring a hyperluminous heavily-obscured AGN powered by a ~billion solar mass black hole. This suggests this object is effectively a highly obscured version of a very UV luminous quasar which are exceedingly rare at z~6.8 (1 per ~3,000 deg^2). The identification of COS-87259 within the 1.5 deg^2 COSMOS field thus raises the question of whether the most massive (~10^9 Msol) black holes in the early Universe frequently grew in heavily-obscured phases. Future wide-area datasets from Roman, Euclid, X-ray, and far-infrared facilities will  provide a greatly improved understanding of the incidence of heavily-obscured supermassive black hole growth and dust-enshrouded stellar mass buildup within the most massive reionization-era galaxies.

Characterizing Ionized Bubbles in the Reionization Era



Right Ascension

Thanks to a variety of observational efforts over the past decade, we have begun to understand that reionization mainly took place at z>6. However, very little remains known about the sizes of ionized bubbles formed during reionization as well as the galaxy overdensities which drove their growth. Fortunately, we can use spectroscopy to begin identifying and studying these bubbles. In a nutshell, Lyman-alpha is far easier to detect from early galaxies situated within ionized bubbles. Because of this effect, regions of the early Universe where galaxies show relatively strong Lyman-alpha emission are likely volumes where a large ionized bubble has formed. Once we identify these bubbles, we can begin characterizing their size as well as the surrounding galaxy overdensities to better understand how reionization happened.

In Endsley & Stark (2022), we presented ultra-deep MMT/Binospec observations of several galaxies surrounding a tentative ionized bubble we previously identified in Endsley et al. (2021b). We detected Lyman-alpha emission at z=6.70-6.88 in 9 of 10 observed galaxies, revealing the the large-scale volume spanned by these sources is highly overdense with >3x the average number density of UV-bright galaxies. We also verified that Lyman-alpha emission is unusually strong from galaxies situated in this volume, suggesting the presence of a large (R~3 physical Mpc) ionized bubble. These results are consistent with the expectation that the largest ionized bubbles formed around strong galaxy overdensities due to the excess ionized photons produced within these regions. Upcoming facilities (e.g., Roman and the GMT) will enable us to characterize many more and even larger ionized bubbles in the early Universe, thereby delivering detailed insight into how reionization happened.

The Ionizing Efficiency of Reionization-Era Galaxies


One of the key questions for understanding how galaxies contributed to reionization is how efficiently they produced ionizing photons (i.e. light with enough energy to ionize hydrogen) and what fraction of that light was able to escape from early galaxies. Recent studies have shown that the rest-optical [OIII]+Hβ emission lines are valuable probes of the ionizing efficiency (both production and escape) of early galaxies. In Endsley et al. (2021a), we used a set of highly luminous galaxies to infer the [OIII]+Hβ EW distribution at z~7, finding that a significant fraction of z~7 galaxies are likely to be highly efficient ionizing agents, far more than at even z~2.

Lyman-Alpha in the Reionization Era

Chart 2.png

Lyman-alpha (an emission line caused by hydrogen recombination) can act as a valuable probe of reionization, constraining how quickly reionization occurred and where it first took place. Since 2019, I have been leading an MMT/Binospec program aimed at characterizing Lyman-alpha emission in very luminous z~7 galaxies. The first results from this program were presented in Endsley et al. (2021a) where we describe a Lyman-alpha detection from a very luminous z=6.85 galaxy with extremely strong [OIII]+Hβ emission (~4000 Angstrom EW). I am also actively involved in a large ALMA program (REBELS) targeting [CII] and dust emission in a large sample of extremely luminous z>6.5 galaxies, several of which possess Lyman-alpha constraints from our Binospec program. With these datasets, I am investigating the properties of Lyman-alpha emission in massive z~7 galaxies to gain general insight on reionization. Stay tuned for more information!

Clustering Measurement Predictions at z=4-10 with JWST

Chart 3.png

Within a small fraction of a second after the Big Bang, a process referred to as inflation created large-scale ripples across the Universe, resulting in regions that were either overdense or underdense. Because galaxies represent locations of the Universe where a large amount of mass has congregated together, they preferentially reside in overdense regions and are therefore clustered together in space. By measuring how clustered these galaxies are, we can infer the total amount of mass they hold (i.e. the halo mass) and compare that to the amount of mass we see locked into stars (i.e. the stellar mass) to calculate their integrated star formation efficiency. In this work, we used an empirical model to predict that planned Cycle 1 surveys with JWST (e.g. JADES and CEERS) would be able to measure the clustering of galaxies at z=4-10 with >5σ significance, and thereby enable the inference of halo masses with <0.3 dex precision at these early epochs (about <10% of the current age of the Universe). This suggests that early JWST observations will deliver the first precise picture of how the stellar-halo mass relation evolves in the reionization era. 

bottom of page