We use KMOS Deep Survey (KDS) galaxies, combined with results from a range of spectroscopic studies in the literature, to investigate the evolution of the stellar-mass Tully-Fisher relation since z ~ 4. We determine the slope and normalisation of the local rotation-velocity -- stellar-mass (Vc - $M_{star}$) relationship using a reference sample of local spiral galaxies; thereafter we fix the slope, and focus on the evolution of velocity normalisation with redshift. The rotation-dominated KDS galaxies at z ~ 3.5 have rotation velocities ~ -0.1 dex lower than local reference galaxies at fixed stellar mass. By fitting 16 distant comparison samples spanning 0 < z < 3 (containing ~ 1200 galaxies), we show that the size and sign of the inferred Vc offset depends sensitively on the fraction of the parent samples used in the Tully-Fisher analysis, and how strictly the criterion of rotation dominated is enforced. Confining attention to subsamples of galaxies that are especially disky results in a consistent positive offset in Vc of ~ +0.1 dex, however these galaxies are not representative of the evolving-disk population at z > 1. We investigate the addition of pressure support, traced by intrinsic velocity dispersion ($sigma_{int}$) to the KDS dynamical mass budget by adopting a total effective velocity of form $V_{tot} = (Vc^{2} + 4.0sigma_{int}^{2})^{0.5}$. The rotation-dominated and dispersion-dominated KDS galaxies fall on the same locus in the total-velocity versus stellar-mass plane, removing the need for debate over the precise selection threshold for rotation-dominated galaxies. The comparison sample offsets are in the range +0.08 to +0.15 dex in total-velocity zero-point (-0.30 to -0.55 dex in stellar-mass zero-point) from the local Tully-Fisher relation at z > 1, consistent with steady evolution of the ratio of dynamical to stellar mass with cosmic time.
Download