We present the first constraints on stellar mass-to-light ratio gradients inan early-type galaxy (ETG) using multiple dynamical tracer populations to modelthe dark and luminous mass structure simultaneously. We combine the kinematicsof the central starlight, two globular cluster populations and satellitegalaxies in a Jeans analysis to obtain new constraints on M87's mass structure,employing a flexible mass model which allows for radial gradients in thestellar mass-to-light ratio. We find that, in the context of our model, aradially declining stellar-mass-to-light ratio is strongly favoured. Modellingthe stellar mass-to-light ratio as following a power law, $\Upsilon_{\star}\sim R^{-\mu}$, we infer a power-law slope $\mu = -0.54 \pm 0.05$; equally,parameterising the stellar-mass-to-light ratio via a central mismatch parameterrelative to a Salpeter IMF, $\alpha$, and scale radius $R_M$, we find $\alpha >1.48$ at $95\%$ confidence and $R_M = 0.35 \pm 0.04$ kpc. We use stellarpopulation modelling of high-resolution 11-band HST photometry to show thatsuch a steep gradient cannot be achieved by variations in only the metallicity,age, dust extinction and star formation history if the stellar initial massfunction (IMF) remains spatially constant. On the other hand, the stellarmass-to-light ratio gradient that we find is consistent with an IMF whose innerslope changes such that it is Salpeter-like in the central $\sim 0.5$ kpc andbecomes Chabrier-like within the stellar effective radius. This adds to recentevidence that the non-universality of the IMF in ETGs may be confined to theircore regions, and points towards a picture in which the stars in these centralregions may have formed in fundamentally different physical conditions.

We present a detailed study of the nuclear star clusters (NSCs) and massiveblack holes (BHs) of four of the nearest low-mass early-type galaxies: M32,NGC205, NGC5012, and NGC5206. We measure dynamical masses of both the BHs andNSCs in these galaxies using Gemini/NIFS or VLT/SINFONI stellar kinematics,Hubble Space Telescope (HST) imaging, and Jeans Anisotropic Models. We detectmassive BHs in M32, NGC5102, and NGC5206, while in NGC205, we find only anupper limit. These BH mass estimates are consistent with previous measurementsin M32 and NGC205, while those in NGC5102 and NGC5206 are estimated for thefirst time, and both found to be $<$$10^6~M_{\odot}$. This adds to just ahandful of galaxies with dynamically measured sub-million $M_{\odot}$ centralBHs. Combining these BH detections with our recent work on NGC404's BH, we findthat 80\% (4/5) of nearby, low-mass ($10^9-10^{10}~M_{\odot}$;$\sigma_{\star}\sim20-70$ km/s) early-type galaxies host BHs. Such a highoccupation fraction suggests the BH seeds formed in the early epoch of cosmicassembly likely resulted in abundant seeds, favoring a low-mass seed mechanismof the remnants, most likely from the first generation of massive stars. Wefind dynamical masses of the NSCs ranging from $2-80\times10^6~M_{\odot}$ andcompare these masses to scaling relations for NSCs based primarily onphotometric mass estimates. Color gradients suggest younger stellar populationslie at the centers of the NSCs in three of the four galaxies (NGC205, NGC5102,and NGC5206), while the morphology of two are complex and are best-fit withmultiple morphological components (NGC5102 and NGC5206). The NSC kinematicsshow they are rotating, especially in M32 and NGC5102($V/\sigma_{\star}\sim0.7$).

We present observational constraints on the stellar populations of twoultra-diffuse galaxies (UDGs) using optical through near-infrared (NIR)spectral energy distribution (SED) fitting. Our analysis is enabled by new$Spitzer$-IRAC 3.6 $\mu$m and 4.5 $\mu$m imaging, archival optical imaging, andthe \texttt{prospector} fully Bayesian SED fitting framework. Our samplecontains one field UDG (DGSAT I), one Virgo cluster UDG (VCC 1287), and oneVirgo cluster dwarf elliptical for comparison (VCC 1122). Independently of SEDfitting, we find that the optical--NIR colors of the three galaxies aresignificantly different from each other. We infer that VCC 1287 has an old($\gtrsim7.7$ Gyr) and surprisingly metal-poor ($[Z/Z_{\odot}]\lesssim-1.0$)stellar population, even after marginalizing over uncertainties on diffuseinterstellar dust. In contrast, the field UDG DGSAT I shows evidence of beingyounger than the Virgo UDG, with an extended star formation history and an ageposterior extending down to $\sim3$ Gyr. The stellar metallicity of DGSAT I issub-solar but higher than that of the Virgo UDG, with$[Z/Z_{\odot}]=-0.63^{+0.35}_{-0.62}$; in the case of exactly zero diffuseinterstellar dust, DGSAT I may even have solar metallicity. Thespectroscopically confirmed globular clusters of VCC 1287 have similaroptical--NIR colors as the UDG itself, with empirical color relationssuggesting sub-solar metallicities and supporting the metal-poor nature of VCC1287. With VCC 1287 and several Coma UDGs, a general picture is emerging wherecluster UDGs may be "failed" galaxies, but the field UDG DGSAT I seems moreconsistent with a stellar feedback-induced expansion scenario. [abridged]