Notes. RID: ID in Rafelski et al. (2015), MID: MUSE ID in MXDF (Bacon et al., in prep.), line for z_s: line used to measure spec-z for ZCONF ≥ 1 sources in the v0.8 catalog (Lyα and Abs. indicate Lyα emission and UV absorption lines, respectively; see Sect. 2.2 for 2 ZCONF = 0 sources), z_s: spec-z in MXDF, z_p: photo-z with 95% uncertainties in Rafelski et al. (2015), F775W: F775W magnitude in Rafelski et al. (2015), M₁₅₀₀: absolute UV magnitude with 1σ uncertainties, ZCONF: Confidence level for spec-z for MUSE sources in MXDF, Method: the detection method in the MXDF (the two sources without a detection method are not included in the catalog but discussed in Sect. 2.2), L17: included in Leclercq et al. (2017) or not, Isolated: isolated source or not (nonisolated), and UV-bright: included in the UV-bright sample or not. ^(♢): RID = 4764 has an update in a new version of the MXDF catalog (v0.9, Bacon et al. in prep.) as follows: MID = 8357, line for z_s = Abs., z_s = 3.188, and Method = ODHIN, which do not change the result of our halo test. ^( * ):RID = 7876 and 9944 were assigned for MID = 103. ^{( ⋆ ), ( + )} Sources that have at least one emission line with S/N ≥ 5 (2 ≤ S/N < 5) in the v0.9 catalog. We note that the Lyα emission line was used to measure the spec-z for most of the sources in the v0.8 catalog, though 13 sources have at least one emission line other than Lyα (see ^⋆ and ⁺). It is because Lyα emission typically has the highest-S/N value among all spectral features. We note that it does not mean that our sample is biased s the LAE selection, and that the choices of lines used to measure spec-z are not relevant for our halo test as we took a wide wavelength range when we optimized NBs (Sect. 3.2). The spec-z values are planned to be improved in an updated MXDF catalog.